Annals ju ofthe — Missouri Dotanical Garden ЕЕ ми Volume Number Volume 85, Number 1 Winter 1998 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating out- side the Garden will also be accepted. All manuscripts are reviewed by qualified, independent reviewers. Authors should write the Managing Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Commitiee Michael H. Grayum Editor, | Missouri Botanical Garden | Amy Scheuler McPherson Managing Editor, Missouri Botanical Garden Diana Gunter Editorial Assistant, . Missouri Botanical Garden Vicki Couture Secretary | Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical Сосин Коу Е. Gereau Missouri Botanical Garden Peter Goldblatt : Missouri Botanical Garden Gordon McPherson : — Missouri Botanical Garden P. Mick Richardson Missouri Botanical Garden Henk van der Werff Missouri Botanical Garden For subscription information contact ANNALS OF THE MISSOURI GARDEN, % Allen Marketing & Management, Р.О. Box 1897, Lawrence, KS 66044-8897. Subscription price is $120 per volume U.S., $130 Canada & Mexico, $155 all other countries. Four issues per volume. The journal Novon is included in the subscription price of the ANNALS. amcpher@admin.mobot.org (editorial queries) | http://www.mobot.org O Missouri Botanical Garden 1998 The ANNALS OF THE MISSOURI BOTANICAL GARDEN (ISSN 0026-6493) is published quar- terly by the Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, MO 63110. Pe- _ riodicals postage paid at St. Louis, MO and ad- ditional mailing offices. POSTMASTER: Send ad- | dress changes to ÁNNALS OF THE MISSOURI BOTANICAL GARDEN, % Allen Marketing « Management, P.O. Box 1897, Lawrence, KS 66044-8897. The mission of the Missouri Botanical Garden is to discover and share knowledge about plants и their environment, in order to preserve and enrich life. © This paper meets the requirements of ANSI/NISO 239.48-1992 (Permanence of Paper). Volume 85 Number 1 1998 Annals of the Missouri Botanical Garden Y NEW TOOLS FOR INVESTIGATING BIODIVERSITY SYMPOSIUM: INTRODUCTION! P. Mick Richardson? The ready availability of cheaper, faster comput- ers and large amounts of ever-cheaper memory has opened up new possibilities in the study of the vast diversity of living organisms. The 43rd Annual Sys- tematics Symposium introduced a variety of these possibilities to the occupants of an auditorium filled to capacity by the 449 registered participants rep- resenting 11 countries and 35 of the U.S. states. Mike Austin started the morning off with a fine talk about Australian eucalypt forests. After the coffee break, Deborah Clark described some of the studies she and her coworkers have undertaken at La Selva in Costa Rica. The morning session was rounded out by Charles Convis’s talk on the past and future of Geographic Information Systems (GIS) in the bi- ological sciences. GIS are what we now use to study how things occur in space. Amazingly, I saw my first ever GIS system in the middle of a forest in Panama. It was, of course, the machine at La Selva mentioned above. After lunch, David Mladenoff described how GIS, radio collars, and regional databases were be- ing used to define favorable wolf habitat in the Northern Great Lakes region. This was followed by a talk on large-area mapping of biodiversity pre- sented by J. Michael Scott and Michael D. Jen- nings. The final talk of the afternoon was a presen- tation by Hanna Tuomisto about satellite imagery and field studies in Amazonian forests. The after- dinner address, co-authored by Andrew D. Weiss, Claire Kremen, and George E. Schatz, was about GIS and a new national park in Madagascar. Four of the talks are represented by the papers following this introduction. The contents of the other talks have been published elsewhere and can readily be found in various databases via the names of the authors. ! This a nd the four articles that follow it are the proceedings of the 43rd Annual Systematics Symposium of the Missouri Botanical Garden, New Tools for Investigating Biodiversity. The symposium was held 4—5 October 1996 at the „А. Missouri Botanical Garden in St. Louis, Missouri, he symposium was supported in part by the National Science Foundation under grant number DEB-9420140. I thank Bette Loiselle and George Schatz for helping to select a fine diversity of T Kathy Hurlbert and her expert staff for wonderful help in organizing and administrating the symposium, John rs for his fine illustration for the cover of the symposium brochure, and the symposium registrants for being such a 2. group of scientists. ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A. ANN. MISSOURI Bor. САВР. 85: 1. 1998. AN ECOLOGICAL PERSPECTIVE ON BIODIVERSITY INVESTIGATIONS: EXAMPLES FROM AUSTRALIAN EUCALYPT FORESTS! Mike P. Austin? ABSTRACT Australia is a large continent with a relatively small population, and government agencies and research institutions are devoting considerable r lanaging n Australia's biodiversity. isa of data quality, choice of analysis method, ecological theory, and GIS (geographic infor- mation system) use survey design for representative s Exam sles of analytical tools for modeling species distribut additive models (GAM), are 40,000 km? in southeastern New South Wales. opposed to community concepts, is € scussed using examples from recent Australian studies with emphasis on the scie components. The 4. of data quality is examined іп terms of a suitable e data set and the ampling g using results zn a survey of 24,00( presented using data from a 2 need for a ? in northeastern New South Wales. 1, e.g., generalize 4 d models (6 LM) and generalized of 9537 plots a 3 tree species for an area of The necessity for ecological [жна in particular continuum theory аз examined in the context of these results. The interface between ec doni is and a theory is discussed drawing on the results of statistical modeling (GLM) of species richness pa ucalyptus subgenera i in the same area. assessment (CH поп biodiversity The predictive use of GIS (GAM) and ые classification techniques, is demonstrated ME an RA) process for establishing a regional conservation pla collated into a package, BioRap, which also includes methods for e a of priority areas ile is being made in developing new tools. However, theory for ecologic ry processes is urgently needed to ensure effective use of these emerging “р for investigating and managing in mapping vegetation, using statistical modeling application to a comprehensive regional These methods and analytical tools h lave been al, st А key issue facing society is how to conserve our global biodiversity. There is need to use the currently available information now in order to fill the gaps in our conservation strategies. Areas with complementary suites of species and/or rep- resentative types of ecosystems are required. There is also a need to constantly examine how to make better use of available data and to find better methods to convert data into useful infor- mation for policy decisions on conservation. Scott & Jennings (1998, this issue) presents a detailed account of one of the most comprehensive ap- proaches so far. Australia is a large continent with a small pop- ulation, and government agencies and research in- stitutions are devoting considerable resources to the development of new approaches for conserving and managing Australia's biodiversity. To do this, Com- monwealth and State governments have developed major databases and Geographic Information Sys- tems (GIS) to provide biodiversity information on the location, abundance, and dynamics of Austra- lia's native flora and fauna, e.g., the Environmental Resource Information Network (ERIN, Chapman & Busby, 1994). Key issues arising from the use of these tools are how best to answer policy questions, data quality, the suitability of analytic tools, the role of ecological theory, the predictive success of GIS, and how best to make methods available to the wider community of users. This paper focuses on Australian research in this area, in particular on improving information pro- vision methods using modern computer technology. The topics considered are: use of available data, such as herbarium records and vegetation survey data; design of surveys to obtain more cost-effective data; use of statistical modeling and GIS to predict species distributions and richness patterns from survey data; and the need to evaluate methodology against existing theory. ! I thank those colleagues i in the Division of Wildlife and Ecology and in other Australian organizations whose work has ma de this review possible. І also than Meyers, N oops, and С. B. Barnett for preparing the e ? Division of Wildlife and Ecology, Commonwealth. осле and очар Research Organisation, PO Box 84, T. 2602. Australia. т Lyneham, Canberra, A.C. ANN. Missouni Bor. GARD. 85: 2-17. 1998. Volume 85, Number 1 Austin Australian Eucalypt Forests DATA Most developed countries are establishing data- bases of biotic data for creating and evaluating con- servation policies (Chapman & Busby, 1994; Scott & Jennings, 1998; Soberón et al., 1996). Similar methods are also being adopted for developing countries (Hall, 1994). The data are usually based on herbarium or museum records. In Australia, the federal government has established the Environ- mental Resources Information Network (ERIN) to collate, organize, and provide access to the avail- able data. Maximizing the use of existing data is now critical as resources to re-collect data by means of surveys are very limited. As part of this effort the principal herbaria and botanic gardens in Australia have cooperated to produce a common standard for computer-based records systems for specimens. There is a working group that meets regularly to address ongoing applications issues. It is estimated to cost $6 Australian to database a single herbarium or museum record, but several times that to collect specimens using professional staff (Chapman & Busby, 1994). ERIN has devel- oped an extensive hardware and software system to support the aim of providing primary data to iden- tify and characterize regional environmental pat- terns for use in environmental assessment and planning. For handling taxonomic data, ERIN has developed modules for managing taxon names and easily updating them (Taxon), managing individual records of specimens (Specimen), and a Data Dic- tionary and catalogue module for managing data sets including custodianship. These modules and others are linked to a GIS to form what ERIN terms a Spatial Information System (SIS). Chapman and Busby (1994) provided further details of the sys- tem, and there is a website (http://www.erin.gov.au) that also provides a public access system for plant records. The system provides for all types of data and remote-sensing coverage of Australia, but the primary taxonomic record data are a key compo- nent. It is important, however, to recognize that her- barium records suffer from several weaknesses (Hall, 1994; Margules & Austin, 1994; Soberón et al., 1996): the records record presence only, and there is no information about absence; the locations are often poorly recorded; the presence of other species and of environmental variables is inconsis- tently recorded; and the spatial distribution of spec- imens is highly biased. Figure 1 exemplifies the location bias of museum records; it shows the dis- tribution of all suitable records of elapid snakes in Australia (Longmore, 1986). The major roads in re- 112? Collection sites for all species in the At/as of 1. Snakes i es фан (Longmore, 1986). Note the alignment of sites along major roads, especially in the Northern Те. Reprinted with permission. mote areas of the continent are clearly outlined by the record locations. With hindsight it is easy to be critical, and such records were not intended to pro- vide definitive data for regional biogeographic or conservation studies. However, when used for anal- ysis of areas of high biodiversity or endemism, problems can arise; see Tuomisto (1998, this issue) for an example from Amazonia. The minimum data set needed for analysis is presence/absence data and an accurate location for which environmental data can be obtained via a map or GIS. Statistical analysis is precluded by the lack of absence data. How to use presence-only data is a serious problem not int recognized by systematists (Soberón et al., 1996; Margules & Redhead, 1995) when con- ieu conservation issues. However, herbarium collections provide taxonomic precision and verifi- le voucher specimens, which vegetation surveys usually lack. This has led to the development in Australia of two heuristic methods to make maximum use of presence data. The first, BIOCLIM (Nix, 1986; Bus- by, 1986, 1991; now termed BIOMAP (Hutchinson matic variables for the location. The estimates are derived from climatic surfaces calculated using rec- ords from climatic stations. These specimen records are used to estimate the range of each bioclimatic variable within which the species is found. For each location where a specimen of a species is re- corded, the climatic estimates are aggregated to provide a “climate profile” of the taxon. The values for each estimate are ranked in increasing order such that the minimum value, the 5th percentile, Annals of the Missouri Botanical Garden and 95th percentiles, etc., can be defined. This has been done for 12 bioclimatic variables to define the climatic profile (Busby, 1986). By describing the climatic profile for a species as the combination of climatic conditions lying between the 5th and 95th percentiles for 16 climatic variables, a climatic en- velope for the potential occurrence of a species is defined. From this profile, together with a grid of predictions of the bioclimatic variables for a region or continent, a map of the potential occurrence of a species can be generated based on climatic in- formation (Longmore, 1986; Hutchinson et al., 97). The prediction map is only of potential oc- currence because no information on absence is used, and there is no information on other environ- mental or historical factors that might control spe- cies occurrence. There are four essential components to the pro- cedure: (1) a method to produce climatic estimates from the records of climate stations and measure- ments of latitude, longitude, and elevation (see ahba & Wendelburger, 1980; Hutchinson & Bis- chof, 1983; Hutchinson, 1984); (2) existence of a digital elevation model which can be used to gen- erate the climatic predictions for all points in the region; (3) a conceptual model for deciding on an appropriate set of bioclimatic variables relevant to the organisms being studied (Nix, 1986); and (4) a classification algorithm to define the bioclimatic en- velope. Hutchinson et al. (1997) provided an up- to-date presentation of all stages of the approach. Examples of the application of this method are Longmore (1986), for a continental study of elapid snakes; Nix and Switzer (1991), on the potential regional distribution of Australia's rainforest verte- brate fauna; Визђу (1986) on distribution of Могћ- ofagus cunninghamiana (Fagaceae); and Busby (1988) on the 2. of climate change оп Austra- n flora and faun А revised ч HABITAT, has been pub- lished (Walker & Cocks, 1991) that uses a polyg- onal rather than the cruder multidimensional rect- angular definition of the climatic envelope used by BIOCLIM. This procedure provides a more conser- vative (smaller) envelope that takes more account of the actual distribution of presence records in the climate space. It has been applied to estimating the continental distribution of kangaroos (Walker & ocks, 1991). However, BIOCLIM (now BIOMAP) remains the most extensively used of heuristic methods for presence data. See Austin et al. (1994a) for a further review of presence methods. To provide better data, herbarium records should contain precise locations and consistent environ- mental information. A preferable minimum data set is presence/absence data for all species in a stan- dard set of taxa from plots collected as part of a vegetation survey. In any survey, absence is con- ditional on the sampling effort made at a site. Large databases that are capable of supporting statistical modeling can be built up by collating such data from existing surveys (Austin et al., 1990; Leath- wick & Mitchell, 1992). The principal weakness in such data sets is the unknown sampling bias in the original selection of the plots. To make most use of databases they should be ecological in nature rath- er than taxonomic. Margules and Austin (1994) have discussed the requirements for establishing such a database, listing four requirements: (a) a conceptual framework based on ecological theory; (b) field data obtained from sites using survey de- sign principles based explicitly on the conceptual framework; (c) a rationale for determining which measurements should be made at the chosen sites in addition to the floristic records; and (d) appro- priate statistical methods for analyzing survey data and predicting (extrapolating) the regional distri- bution of species from the point records. These au- thors failed to emphasize that this is only possible if the database is linked to a GIS. SuRvEY DESIGN How to obtain a representative sample of the veg- etation variation in a region is a central question for conservation evaluation. Vegetation surveys of large areas are expensive and time-consuming, particular- ly if random or systematic sampling is undertaken in rugged or inaccessible regions (Burbidge, 1991; see also Tuomisto, 1998). Cost-effective methods are required. In Australia, modifications by Austin and Heyligers (1989, 1991) of the gradsect sampling ap proach first proposed by Gillison and Brewer (1985) rovide an example of an explicit, consistent, and repeatable method. Unlike many sampling strategies that produce unbiased estimates of some mean val- пе, e.g., basal area of timber per unit area in the region, vegetation surveys should be directed toward obtaining a representative sample of the range of variation in vegetation composition. The detection of unusual combinations of species is as important as accurate estimates of the average composition of the commonest forest types. The method proposed by Austin and Heyligers (1989) is based on sampling vegetation from all possible combinations of selected environmental variables. The logistics of surveys, e.g., travel time between sampling sites, add consid- erably to the costs. Sampling along a transect is very cost-effective in travel time. If the transect is ori- ented along the steepest environmental gradient in Volume 85, Number 1 1998 Austin 5 Australian Eucalypt Forests 152°E | АГ; P ut X 3teSI- à Figure 2. The position of four gradsects selected for a region of the north coast of New South Wales, showin | ample a ад Ааа fall class n annual rainfall 1000- and alti- tude 180-540 m). Individual squares 22 ] km?. Redrawn from Austin and Heyligers (1991). a region (i.e., a gradsect), then different environ- ments can be sampled with less effort. Where such a gradsect is positioned along an access route, then a very cost-effective although biased survey is ob- tained. Austin and Heyligers (1989, 1991) designed a survey of the forest vegetation of coastal northern New South Wales (NSW) based on the principles outlined above. The area surveyed was 24,000 km?, and the floristic data consisted of presence/absence data of tree species recorded from а 50 х 20 m plot oriented along the contour with estimates of the ranking of the dominant species. The protocol used consisted of seven steps: (1) Identify the major en- vironmental variables influencing the distribution patterns of the vegetation in the study region. For their region these were temperature, rainfall, radia- tion, and nutrients. (2) Recognize a set of variables best suited because of their availability and practi- cality to determine the position and direction of the gradsects. For the north coast region these were al- titude (an easily measured and highly correlated sur- rogate for temperature), mean annual rainfall, and lithology (crude surrogate for soil nutrient content). (3) Select gradsects using these variables and the best available technology. Figure 2 shows the extent to which the four selected gradsects sample one par- ticular combination of altitude and rainfall. (4) Strat- ify the gradsects into geographical segments and stratify the environment within segments to provide Table 1. Example of survey design for a segment. Size of the environmental cells and their sampling frequency for the middle segment of the southern segment. Repro- duced from Austin & Heyligers (1989). Rainfall classes Rock- Altitude type clases 2 3 4 5 6 1 2 4 (x) 3 3 3(1) 45 (1) 4 2 5 1 2 6 3 1 (0) 4 1 (0) 5 1 2(1) 15(1) 26 (1) 2 16 (1) 37 (1) 38(1) 1(0) 8 3 1 (0) 34(1) 22(1) 12 (1) 4 ] (0) 5 1 11 (1) 30(1 2(х) 2 55 (2) 17(1) 28(1) 1(0) 9 3 86 (2) 75 (2) 4(0) 2() 4 1 (0) 26 (1) 5 1 (0) The plain numbers refer to the total number of pixels in an environmental ce 5 plots each from different topographic positions) to b sampled in each cell. An “x” indicates that these cells were not easily accessible and no sample was to be taken. replicate sampling of different environmental com- binations at different locations (Fig. 2). (5) Stratify at the local scale, i.e., within the 1-km resolution used in positioning the gradsects to take account of other important environmental determinants of veg- etation. In this case five topographic positions as a surrogate for solar radiation were sampled within each 1-km gridcell selected. (6) Decide the effort to be spent sampling the rarest environmental combi- nations as compared with increased replication of the commonest combinations. Determine the location of samples by selecting random coordinates and tak- ing the closest suitable cell with adequate access. Table 1 shows an example of the survey design for a particular segment. (7) Review assumptions re- garding importance of environmental variables on which the survey was designed and modify if nec- essary before completing the survey. Austin and Heyligers (1989) modified the survey design after finding that depth to water table had an overriding influence in the coastal lowlands. Annals of the Missouri Botanical Garden The approach of Austin and Heyligers (1989) can be summarized as an SR? strategy, that is, Stratifi- cation, Representation, Replication, and Random- ization. It represents one realization of the first two requirements of Margules and Austin (1994) for es- tablishing а database. A total of 1025 plots were sampled by Austin and Heyligers (1989) equal to an area of 1.025 km”, or approximately 1/24,000th of the study area. The restriction of sampling to grad- sects means that not all locations have an equal chance of being selected in the sample, and there- fore the sample obtained is highly biased. However, the sample is representative, and the design is ex- plicit, consistent, and repeatable, which is not al- ways the case with biodiversity surveys at the pres- ent time. While it is possible to design an SR? survey without a GIS, it is much easier if one is available. Modifications of it have since been used in northern NSW, Australia (Ferrier, pers. comm.), in Sri Lanka (Green & Gunawardena, 1993), and in South Africa (Wessels et al., in press). The approach makes two ecological assumptions. First, that vegetation varies continuously with environment, forming a continuum rather than discrete communities (Austin & Smith, 1989), and therefore all combinations of environ- mental conditions should be sampled. Second, it as- sumes that the major environmental gradients in any given region are known or can reasonably be hy- pothesized. The comparative performance of gradsect sam- pling has been evaluated by Austin and Cawsey (1991) with artificial data and by Wessels et al. (in press) with survey data for birds and dung beetles in South Africa. Both studies support the cost-ef- fectiveness of gradsect sampling against systematic, random, and various purposive methods. Detailed attention has been given to this example of survey design because vegetation survey methods appears to be a neglected topic (Greig-Smith, 1983; Jong- man et al., 1987; Kent & Coker, 1992), though see Noy-Meir (1971) for an early Australian example. A variety of alternative designs have been devel- oped in Australia; see Noy-Meir (1971), Austin and Basinski (1979), Margules and Nicholls (1987), Prober and Austin (1990), McKenzie et al. (1991), and Neave et al. (1996). STATISTICAL MODELING Availability of survey data consisting of pres- ence/absence for species plus information on en- vironmental variables from a GIS allows the pre- diction of species distributions using statistical modeling. Statistical modeling is no longer restrict- ed to quantitative data with normal errors (Mc- Cullagh € Nelder, 1989), as many botanists as- sume. There is currently a wide variety of prediction methods extending well beyond the usu- l statistical methods to neural nets (Aleksander & Morton, 1990; Fitzgerald & Lees, 1992), genetic algorithms (Holland, 1992; Lees, 1994), and deci- 1984; Lees & Ritman, 1991). A recent evaluation of many of these meth- ods (Austin et al., 1994a, 1995; Austin & Meyers, 1996) for analyzing plant ecological data concluded sion trees (Breiman et al., that while most techniques can be found to have advantages under certain circumstances, statistical models perform better with typical vegetation sur- vey data. Franklin (1995) provided review of recent work from a geographer's perspective. The two sta- tistical modeling methods that are currently being actively used are Generalized Linear Models (GLM; McCullagh & Nelder, 1989) and Generalized Ad- ditive Models (GAM; Hastie & Tibshirani, 1990). Examples of the use of GLM with vegetation data are Austin et al. (1990, 1994b) and Leathwick and Mitchell (1992). Nicholls (1989, 1991) provided a detailed discussion with examples of how to use GLM with vegetation survey data. The more recent technique of GAM was introduced to plant ecology by Yee and Mitchell (1991). Leathwick (1995) used it to study the climatic relationships of New Zea- land tree species. Austin and Meyers (1996) com- pared GLM and САМ for Eucalyptus forest species and discussed their role in the management of for- est biodiversity. Recently GAMs have been used for predicting flora and fauna distributions for a large area of northwestern NSW ( NPWS, 1994a, b) and to derive predicted vegetation communities for the south coast of NSW in an unpublished CSIRO consultancy report in 1996. Statistical models such as GLM are used for the prediction of a response variable (or dependent variable) from a set of predictor (or independent) variables. One advantage of GLM over the classical regression method is that it allows error functions other than the normal, and hence the use of density or even binary data is possible. GAM, a non-para- metric technique, has the additional advantage that the mathematical function describing the shape of the curve relating the response variable to a pre- dictor variable need not be specified precisely. A smoothing spline is fitted to the data, and only the number of inflections in the curve need be speci- fied, not whether it is a polynomial or exponential function. The key problem in the model-building process for GLM use with vegetation data has been the shape of the response of a plant species to environmental predictors. Ecological theory is needed to define a Volume 85, Number 1 1998 Austin Australian Eucalypt Forests reasonable set of potential responses. The evidence regarding the existence of the bell-shaped response usually presented in textbooks is ambiguous (Austin & Smith, 1989), and more flexible curves need to be considered. The B-function is one complex func- tion that has been proposed (Austin et al., 1994b). It requires definition of the limits of a species dis- tribution along an environmental gradient within which a variety of skewed or symmetric curves can be represented by B-functions with different param- eter values. Austin et al. (1994b) fitted а B-function for temperature to data for nine species of Eucalyp- tus (Myrtaceae). No species had a symmetric re- sponse shape; all were skewed and the patterns of skewness were dependent on position along the en- vironmental gradient of mean annual temperature (Austin et al., 1994b). Th for a larger set of Eucalyptus species (Austin & Gay- wood, 1994). Their conclusions suggest that species distributions along gradients have well-specified skewed shapes and nonrandom patterns. If these patterns are found in other suitable data sets, then it may be possible to propose rules regarding the biodiversity patterns to be found in vegetation. Data sets are needed where the length of the environmen- tal gradient sampled clearly exceeds the width of the environmental niche of the individual species, oth- erwise the species limits cannot be specified. Failure to appreciate this limitation to the use of B-functions has resulted in controversy (Oksanen, 1997; Austin & Nicholls, 1997; see also Austin & Meyers, 1996). It is the difficulty of specifying the exact form of e results were confirmed Mitchell, 1993; NSW NPWS, 1994a, 1994b; Leath- wick, 1995). GAM, while conferring the advantage of a non-parametric smoothing function, is not with- out problems, e.g., the sensitivity of significance tests (Austin et al., 1995; Austin & Meyers, 1996), and is certainly not without assumptions as asserted by Norton and Mitchell (1993). It is a “current best practice method" for biodiversity analysis but is likely to undergo significant modifications in the near future as further evaluation is done. EUCALYPTUS FASTIGATA: A CASE STUDY IN САМ PREDICTION OF SPECIES DISTRIBUTION The steps in modeling the distribution of a spe- cies based on available plot data from a region in southeastern NSW, Australia, are presented here. The details of the study area have been published previously (Austin et al., ; 1994b). Briefly, it is approximately 40,000 area and runs from just north of latitude 35%S to the Victorian border, and from longitude 148°E to the east coast of Australia. The climate varies con- siderably across the region: mean annual temper- ature ranges from 2.5?C on Mt. Kosciusko at 2200 m to 16.9°C on the northern coastal plain, while mean annual rainfall varies from 480 mm to more than 2000 mm, with marked seasonal differences in rainfall patterns. Eucalyptus fastigata H. Deane & Maiden is a species characteristic of the coastal scarp forests between 400 m and 800 m (Fig. 3), and results of statistical models of its distribution using GLM have been published (Austin, 1992; Austin et al., 1994b) MODELING STEPS 1. Collate available plot data for the defined re- gion. Details of the database and the contribu- tors can be found in Austin et al. (1994b). The current database has 9537 plots with records of the presence/absence for 273 tree species, and the geographical distribution of the plots is shown in Figure 3 2. Select and generate a set of appropriate envi- ronmental predictors. Climatic variables have been generated from a digital elevation model (DEM) using a variety of packages now incor- porated into BioRap (Hutchinson et al., 1997), and values for the plots obtained using a GIS. Those variables based on lithology were derived from a lithology GIS layer. Eleven environmen- tal variables were selected as potential predic- tors. These included eight continuous variables: average summer rainfall, average winter rainfall and rainfall seasonality (ratio of summer/winter rainfall), mean annual temperature, temperature of the hottest month and winter cold index, av- rage summer radiation and average winter ra- diation (kj/m?/day adjusted for slope and as- pect); and three factor or categorical variables: topographic position (6), lithology class (6), and nutrient index (5) (figure in parentheses is the number of classes in the factor). Figure 4 shows the plots mapped into a climate space equivalent to the geographical space shown in Figure 3. . Restrict the data, where the data extend well beyond the environmental niche of a species. For example, E. fastigata does not occur below C or above 16°С mean annual temperature (Fig. 4). Inclusion of those zero values beyond these limits сап complicate the analysis and lead to poor prediction of the occurrence of the species near the limits (Austin & Meyers, 1996). The data are therefore restricted to those obser- vations that occur within limits set by having о Annals of the Missouri Botanical Garden -35.57 ' о о о 1 -36.57] Latitude (decimal degrees) -37.07 present absent 5 T T 148.0 148.5 149.0 T 149.5 T Џ 150.0 150.5 Longitude (decimal degrees) 3. The geographical distribution of Eucalyptus fastigata as determined from the database of 9537 plots for No NSW. Triangles indicate E. fastigata present; dots indicate plots without E. fastigata Mean annual temperature (*C) А present - absent 50 100 150 200 Winter mean monthly rainfall (mm) 250 e 4. The distribution a 22. fastigata in a climate space defined by mean annual temperature and | mean monthly winter rainfall. E. fastigata is absent below 7 numerous plots above and below (e limits. ° and above 16°С mean annual temperature, with Volume 85, Number 1 1998 Austin 9 Australian Eucalypt Forests 150 200 250 300 50 100 Summer mean monthly rainfall (mm) 5 10 15 20 Mean annual temperature (C) 1 2 3 4 5 Topography 15 25 Seasonality 0.5 2 4 6 8 10 12 14 Winter mean daily radiation (kj/m2/day) 1 2 3 4 5 Nutrient index The shape of the GAM response functions for 6 of the 11 predictors for Eucalyptus fastigata. Note that the functions have been fitted only within limits for mean annual temperature and summer mean monthly rainfall. 100 zero values above and below the last posi- tive observation, provided there are additional observations beyond the limits; see Austin and Meyers (1996) for further details. 4. Fita GAM. The model was derived for presence/ absence data for E. fastigata, as predicted from the 11 environmental variables using S-Plus package (Statistical Sciences, 1993), with four degrees of freedom for the continuous predic- tors. All eleven predictors were included in the model. The shapes of the responses differ mark- edly for the different predictors (Fig. 5). 5. Use GIS to predict the distribution of the spe- cies for unsampled areas in the region. This was done using the predictive functions derived from GAMs. The predicted distribution of E. fastigata clearly shows the major zone of occurrence along the coastal scarp (Fig. 6; cf. Fig. 3). These models can be used to investigate current ecological problems of relevance to our future man- agement of biodiversity. For example, where would Eucalyptus fastigata occur if global warming re- sulted in a 2°C rise in regional temperature and local increases in rainfall? The predicted geograph- ical distribution after such a change is shown in Figure 7. Eucalyptus fastigata would undergo a substantial reduction in occurrence on the coastal scarp under such a scenario. Note that this is a static analysis ignoring problems of dispersal, time to equilibrium, and changed competitive interac- tions. The role of environmental niche models in relation to climate change models and physiological growth models was reviewed by Austin (1992), with particular reference to Е. fastigata. The above procedure is explicit, repeatable, and consistent. There are both statistical and ecological research issues still to be resolved about the best procedure. Austin and Meyers (1996; Austin et al., 1995) have examined the performance of GLM, GAM, and regression trees on both real data and simulated data where truth is known. They con- clude that a mixed strategy using both GLM and GAM functions is desirable. They suggest that the best results are as dependent on the availability of suitable ecological and statistical skills as on the particular procedure used. The explicit nature of Annals of the Missouri Botanical Garden ШІ >= 15% and <50% Ш >= 50% Figure 6. The predicted ge pate ‘al distribution of Еисагургиз fastigata in terms of о of occurrence using the GAM functions and a GIS for the coastal zone (outlined area) of New South Wales the models for individual species provides a firm basis on which to build an improved understanding of species distribution patterns. The ad hoc map- ping of vegetation and species based on unknown mental models derived from an unknown arbitrary database imperfectly remembered is no longer ad- equate. However, it must also be remembered that these models are only as good as the data, and the ecological assumptions on which the predictors are selected, and are based on correlation, not causa- tion. SPECIES RICHNESS Statistical models like GLM can be used in other contexts more relevant to evolutionary botany and Volume 85, Number 1 1998 Austin Australian Eucalypt Forests E >= 15% and «5096 Ш >= 50% re 7. Тһе predic ‘ted distribution of Eucalyptus fastigata if mean annual temperature increased by two "pes Figu and rainfall by 20% in coastal and tableland regions and 10% in the western region for the same region as Figure its interface with ecology (Currie, 1991). Austin et al. (1996) investigated patterns of tree species rich- ness in southern NSW using a similar but smaller data set (7208 plots) than that used for Eucalyptus fastigata above. Similar predictors to those of Aus- tin et al. (1994b) were used, namely mean annual temperature, mean annual rainfall, mean annual daily radiation, and four categorical variables (to- pographic position, lithology, nutrient index, and rainfall seasonality). Total tree-species richness for 0.1-ћа plots was predicted as the dependent or re- sponse variable using GLM, with cubic polynomial functions for the continuous variables and interac- tion terms for temperature and rainfall. Regional scale patterns of species richness are predictable from the environment, with mean annual tempera- ture the most important predictor. Maximum spe- cies richness for trees was found in protected gul- lies at temperatures >16°C, with rainfall > mm, on volcanic soils with intermediate or high nu- 12 Annals of the Missouri Botanical Garden optimum environment for numbers of species of we шшш above 2.0 Monocalyptus; any theory of biodiversity and evo- me 1.6-2.0 lution should be able to explain the existence of КМЧ 1.2-1.6 such a pattern in environmental space. Managers 272 0.8 - 1.2 о | КУ 0.4 - 0.8 EC below 0.4 Mean annual temperature (°С) 9 = LL? SKY? 5 |- K 2 i | | | 300 900 1500 2100 Mean annual rainfall (mm) Figure 8. The predicted distribution of species rich- ness for Eucalyptus subg. Monocalyptus in relation to cli- matic predictors on exposed ridges with high radiation and low nutrients, with soft sedimentary lithology. trient levels (Austin et al., 1996). This habitat rep- resents the limited conditions under which the spe- cies-rich warm temperate rainforest species can survive in the fire-prone eucalypt-dominated forests of the region. Various components of tree-species richness can be recognized. For example, there are numerous species of Eucalyptus in the region, and analyses of the species richness patterns were made for two of the subgenera, Monocalyptus and Sym- phyomyrtus. All predictors except seasonality of rainfall were significant for the subgenus Monoca- lyptus, and there was a complex skewed response to temperature and rainfall (Fig. 8). Maximum spe- cies richness for Monocalyptus was predicted for exposed ridges on sediments or granites under low nutrient conditions in temperate climatic condi- tions. The other subgenus, Symphyomyrtus, showed a distinct complementary pattern and insensitivity to radiation and topographic position, but species richness also varied with seasonality of rainfall. High species richness was associated with fertile soils (Austin et al., 1996). There has been consid- erable discussion about the differential behavior of species from these two subgenera and their ability to co-occur (Noble, 1989). The descriptive models obtained by GLM analysis are consistent with the conclusions reached in the literature review of No- ble (1989). For the evolutionary botanist these results pose the question of why there are more species of sub- genus Monocalyptus co-occurring in some environ- ments than others. Figure 8 shows that there is an of biodiversity also need to understand the rela- tionship between species richness and environ- ment. If individual species and species richness are both strongly related to environment, then the con- cept of a regional species pool needs to be re-ex- amined. APPLICATIONS An example of the use to which these computer- based tools are being put in Australia is an unpub- lished consultancy report by the CSIRO Division of Wildlife and Ecology for the NSW National Parks and Wildlife Service. The objective of the consul- tancy was to map the pre-European forest vegeta- tion (pre-1750) at the scale of 1: 100,000, such that the percentage of the pre-European communities still surviving could be estimated. This information would then be used to determine which currently forested areas should be conserved, which logged, and which require further detailed examination. The region concerned was the Southern Coastal Zone of NSW. There was only a limited time avail- able to complete the study. However, the existing database and modeling studies described above, plus an appropriate GIS, provided a suitable basis for undertaking the study using modern methods. The ecological theory on which the study was based assumed that the vegetation formed a contin- uum such that the precise composition of the veg- etation varied continuously, and communities were a function of the frequency of particular environ- mental combinations in the landscape (Austin & Smith, 1989). Estimating individual species distri- utions using GAMs from existing data for relevant environments would allow spatial predictions of distributions for cleared areas. Combining the pre- dictions for individual species for each cell of a GIS gives an expected but continuously varying com- munity composition. This composition data can then be classified using numerical methods to give a consistent description of forest communities for the entire region. The steps involved practical decisions at each stage. These are dependent on the particular fea- tures of each project. These are briefly described here to indicate the types of problems that arise. l. Data. The existing database consisted of 8377 plots. No data were recorded in the database for the northeastern area of the zone. Additional data were collated from that and other regions Volume 85, Number 1 1998 Austin Australian Eucalypt Forests bo о Figure 9. Pre-1750 mapping units. —а (left). —b (right). Lowland Granite Communities (E. tereticornis group Ш Е ргеѕепі O absent E present O absent in the zone. However, this was found to give a very biased sample from the northeastern re- gion; most plots were from warm temperate rain- forest in gullies. An additional survey of the northeastern region was necessary. . Survey. The design was based on the SR? strat- еру, described previously. The gradsect ap- proach was not used, as access was not a major limiting factor and distances were small. Each of the three 1: 100,000 maps containing parts of the northeastern area were used as geographical strata within which environmental combinations based on mean annual temperature, mean an- nual rainfall, and lithology were mapped with a GIS. Sites were selected for second stage sam- pling by topographic position, as previously de- scribed (Austin & Heyligers, 1989). . Modeling. The data finally consisted of 9537 plots. After setting an acceptance criterion of at least 50 presence observations in order to in- clude a species in the GAM modeling, 88 tree species were modeled. To reduce the time taken to model the species, the same model was fitted to all species. The eleven predictors used for Ё. fastigata (see above) were used. Use of such a generic model ignoring significance levels will result in overspecification. The degree to which this reduces the accuracy of the models is the subject of current research. Savannah Woodlands (Eucalyptus melliodora/E. bridgesiana group). 2 4. GIS. A GIS with a 1-ha resolution was available containing all the necessary predictor variables, so predictions for each of the 88 species was possible for each of the 2.7 million cells in the GIS in the 27,000 km? zone. 5. Classification. To provide a community classifi- cation of the zone, the 2.7 million pixels, each characterized by the probability of occurrence of 88 tree species, were x in a numerical classification using ALOC and UPGMA proce- dures in the package PATN (Belbin, 1995). The available computer facilities and time imposed major limitations on the analysis of the large species by pixel matrix. The final stage was a manual reorganization of the classification dendrogram to provide mappable units. Vegetation composition is strongly controlled by aspect in the area, and classification units were grouped into catenary sequences to give spatially coherent units for mapping. Two levels of vegetation classification were recognized: classes roughly cor- responding to formations or alliances, and units ap- proximating communities. Figure 9 gives examples of the class maps obtained. These maps at the finer scale of units, when combined with a land-cover map showing remaining forest areas, were used to decide that a further 100,000 ha of forest needed to be reserved in order to conserve an adequate Annals of th Missouri Botanical Garden representation of the pre-1750 forest communities. Pressey (in press) discusses the relationship be- tween the scientific analysis and the political pro- cess for a similar exercise for northern NSW. When information of this kind is available, then a further stage is reached where methods are need- ed to determine biodiversity priority areas. Explicit criteria are required, but the particular techniques are dependent on the available data. In general terms there are two classes of methods: (1) those that identify a biodiversity attributes (e.g., species) are represent- ed a specified number of times, e.g., once, twice, set of areas in which all selected or three times; (2) those that maximize the amount of biodiversity represented by a given number of & Redhead, 1995). In the first the level of representation is specified arbitrarily, while in the second it is the number of areas that is fixed. This has become a major area of research and in- novation where numerous constraints and trade-offs ave been incorporated into the computer algo- rithms. In Australia, Margules and Nicholls (1987) pioneered an effective computer algorithm for the attribute-representation approach. Subsequently these authors with Pressey explored a number of the options with this approach (Nicholls & Mar- gules, 1993; Pressey & Nicholls, 1989a, b; Pressey, 1994). Faith (1994) has developed a number of ap- proaches to the second class of methods, using areas (Margules measures of dissimilarity and ordination techniques (Faith & Walker, 1994, 1997; Faith & Nicholls, 1997). Two features of the work by these authors are worthy of comment. First, the recognition that species richness per se is not a good criterion for conserving representative biodiversity; it is easily shown with simple examples that selecting the richest site of three may result in conserving fewer species than selecting the two sites each with fewer species. Complementarity of site composition is more important than maximal richness of individual sites. Second, the recognition that sophisticated al- gorithms are only valuable if they can be used with the limited and arbitrary data sets currently avail- able and enhance those data rather than hide their inadequacies. The BioRap manuals (Margules & Redhead, 1995; Boston, 1997; Hutchinson et al., 1997; Faith & Nicholls, 1997; Noble, 1997) pro- vide case studies of the use of alternative methods with various types of data. DISCUSSION Herbarium records, while a primary source of data, have their limitations for analysis of species distributions (Hall, 1994; Austin et al., 1994a; Sob- erón et al., 1996). Data quality is a key issue, and computer routines for examining records of a spe- cies” distribution are an important first step (Chap- man & Busby, 1994). New approaches such as BIOCLIM (Nix. 1986; Busby, 1991) and HABITAT (Walker & Cocks, 1991) are examples of heuristic methods designed to overcome the limitations of presence data. One difficulty is that survey data will age taxonomically. Without voucher specimens, it will not be possible to update survey records to Howev managing biological diversity will require bus ging 8 q take account of taxonomic revisions. data than herbarium presence records provide. Hai- la and Margules (1996) argued strongly that a nec- essary component of any practicable strategy for preserving the biological diversity of the earth is systematic field survey. They noted, however, that modern theoretical ecologists regard surveys as te- dious, mundane activities; yet such data are essen- tial to testing theory. Any survey has implicit in its design a set of ecological assumptions and a set of statistical assumptions; if these are not rec ognized and progressively improved upon, then maximum use will not be made of our limited survey re- sources. This paper has attempted to present some of the Australian experience in this area, but rapid changes are occurring as a result of society's de- mands that decisions be made on the inadequate database that currently exists. Poor survey design and predictive modeling techniques are adding to the difficulties. A major reason for this is that much of the work is appearing in the "gray" literature, and is inaccessible to many conservation scientists who might otherwise use the improved methods and techniques if they were aware of them. This review suffers from this problem in that much of the Aus- tralian work, good and bad, has yet to be published in the international literature and only exists in in- ternal reports or reports published with small num- bers of copies. Electronic publication may solve this problem of access to the literature. Computer technology in various forms, remote- sensing, GIS, and statistical software are being used to create new tools for the study of biodiver- sity. What is more important is that we are finding new ways of thinking about the problems of study- ing biodiversity, whether it is how to design surveys or to develop new theories integrating ecology and evolution to better conserve our flora. Each stage in the study of biodiversity is now the subject of intense investigation in terms of basic research, conservation application, and cost-effectiveness Margules & Austin, 1991). In addition, the results of such biodiversity studies are being incorporated into computer packages designed to facilitate com- —. Volume 85, Number 1 1998 Austin Australian Eucalypt Forests munity decision-making in regional land-use plans (Cocks et al., 1995) and are being actively used in conservation planning (Pressey, in press). А period of evaluation is now needed to determine which of these methods or tools are the best t the present time our immediate pragmatic concern is to make the best possible use of the biodiversity data we currently have to make sen- sible conservation decisions. Margules and his col- leagues, in putting together the BioRap manuals and software for rapid assessment of biodiversity priority areas for the World Bank (with funding from AustAid), have shown how to make use of available data. The opportunity to constantly reit- erate the processes is one of the strongest argu- ments for having computer-based tools for all as- pects of biodiversity study: they can be repeated when necessary. Literature Cited Aleksander, 1. & Н. Morton. 1990. An ег to eural Computing Pi en & Hall, Lonc Austin, M. eling the al niche of mae implican for plant community response to elev ed C 20, levels. Austral. J. Bot. 40: 615—630. Basin 1979. ка physical | wd niques. Pp. 2 n M. P. Austin & K. D. Cocks ined eral Midi I ge Use on the South Coast ai New South Wa es 1. CSIRO, Melbourne. Cawsey. 199]. Sampling strategies cost- ed by ibus Pp. 167-175 in C. R. Margules € M P. Austin (editors), Nature Conservation: Cost Effective icm Surveys and Data Analysis. CSIRO, Mel- bourne . J. Gaywood. 1994. Current problems of en- sebo gradients and species response curves in relation to continuum theory. J. Veg. Sci. 5: 473-482. ——— & . Heyligers. 1989. Vegetation survey de- sign for conservation: Gradsect sampling of forests in ge eastern New South Wales. Biol. Conservation 50: . Vegetation survey design, new ши pea Жы еј Pp. 31–36 in С. R. Margules & M. P. Austin (editors), a Conservation: Cost Effective Bieloglea] Surveys and Data Analysis. CSIRO, Melbourne. . Meyers. 1995. Modeling of Landscape Patterns ма рае esses Using Biological Data. Subpro- ject 4: Real Data Dase Study. Consultancy Report for ERIN, CSIRO Division of Wildlife & Ecology, Canber- ra. 1996. Current approaches to model- ing ia octal niche of eucalypts: Implications for management of forest biodiversity. Forest Ecol. Man- agem. 85: 95-106. —— . O. Nicholls. 1997. To fix or not to fix the species bait. that is the ecological question: Response to Jari Oksanen. J. Veg. Sci. 8: 743-748. & T. M. Smith. 1989. A new model for the con- tinuum ra b aru 83: 35-47. ‚ А. О. Nicholls € С. R. Margules. 1990. Меа- еса of he realized i slave niche: Environmen- 2 ; ed tal niches of five Eucalyptus species. Ecol. Monogr. 60: 161-177. ers & M. D. cpi 1994a. Modeling of 5. Тае and Processes using Biological Data. Sub-project 2: Predičtive Models for Landscape Patterns and Processes. Consultancy Report for ERIN, CSIRO Division of Wildlife ye Ecology, Canberra . G. Pausas & A. O. Nicholls. 1996. Patterns of ide species richness in Aes n to environment in south-eastern New South Wales, Justia. Austral. J. Ecol. 21: Aie 64. icholls, M. D. Doherty & J. A. Meyers. environmental ent by means of a B-function. J. Veg. Sci. 5: ers, D. L. Belbin & M. D. Doherty. 1995. Modeling of Landscape Patterns and Processes using Biological Data. Subproject 5: Simulated Data Case Study. Consultancy Report for EN CSIRO Di- vision of Wildlife & Ecology, Canber Belbin, D. L. 1995. PATN Users Cuide and Technical References. CSIRO Division of Wildlife and Ecology. erra. . 1997. BioRap. Rapid Assessment of Bio- аа, Vol. 1. The BioRap Biological Database. Aus- tralian BioRap Consortium, CSIRO, Canberra. Breiman, L., J. H. Friedman, R. A. Olshen & C. J. Stone. 1984. Classification and Regression т Wadsworth International Group, Belmont, Califor A. A. 1991. Cost 5. on surveys for nature conservation. Pp. 3-6 i Margules & M. P. Austin (editors), Nature ходове Cost Effective 2” al Surveys and Data Analysis. CSIRO, Mel- bourne. Busby. J. R. 1986. A biogeoclimatic analysis of Notho- fagus cunninghamii (Hook.) Oerst. in southeastern Aus- tralia. Austral. J. Ecol. 11: 1-7. 988. Potential impacts of 2” change on Ашық flora and fauna. Pp. 387-398 іп С. 1. Pear- man (editor), pon Planning for Climate Change. CSIRO, Melbour 1991. BIOCL IM—A bioclimate analysis and тей system. Pp. 64-68 in C. R. Marg M. P. Austin (editors), Nature Conservation: Cost Effective Исе Surveys and Data Analysis. CSIRO, Mel- bo bie man, A. D. & J. R. Busby. 1994. Linking plant spe- cies information to continental biodiversity inventory, climate modeling and environmental monitoring. Pp. 179-195 in R. I. Miller (editor), Mapping the Diversity of Nature. PI utr & Hall, London. Cocks, K. D., J. R. Ive & J. L. Clark. 1995. Forest Issues. Processes and Tools for Inventory, Evaluation, Media- iid and bg Map Report of a Case-study of the Bate- ‚ New South Wales, Australia. CSIRO Division of Wildlife and Ecology Project Report, Can- Currie. D. J. 1991. Energy and large-scale patterns o animal- and plant-species richness. Amer. Naturalist 137: 27—49. Faith, D. P. 1994. Phylogenetic pattern and the TN fication of organismal biodiversity. Philos. Trans., Ser. B. 345: 45-58 alker. 1994. DIVERSITY: A Software Package "о Banipling Phylogenetic and Environmental Diversity. Reference Manual and Users Guide. v.2.1. CSIRO Division of Wildlife and Ecology, Canberra. Annals of the Missouri Botanical Garden ----« . 1997. Environmental diversity: On the best-possible use of surrogate data for assessing the relative biodiversity of sets of areas. Biodiversity & Conservation 5: 399-415. . O. Nicholls. 1997. BioRap Rapid Assess- ment of Biodiversity. Vol. 3. Tools for Assessing Biodi- Weed Priority Areas. Australian BioRap Consortium. О, Pac . К. W.& B. G. Lees. 1992. The Application of Neural Networks to the Floristic Classification of Re- mote Sensing and GIS Data in Complex Terrain. Pro- ceedings of the XVII Congress of the International So- ciety for Photogrammetry and Remote Sensing, С. Predictive vegetation mapping: Geo- graphic modeling of biospatial patterns in relation to environmental gradients. Progr. Phys. Geogr. 19: 474— 499. Gillison, А. N. & К. R. W. Brewer. 1985. The use of gradient directed transects or pradene in natural re- 2 servation Evaluati ion of Some Na Lanka. For 227. Sri Lank with UNDP. FAO, and ane Greig-Smith, P. 1983. Quantitative ii Ecology, 3rd. ed. pres Scientific Publicat ations, Oxford. Hall, apping for monographs Baselines for resource еы ment. Pp. 21-35 in R. 1. Miller (edi- i. Mapping the Diversity of Nature. Chapman & Hall, Lo a, in assocation Haila z & C. R. Margules. 1996. Survey research in conservation biology. peser 19: 323-331. Hastie, T. t. Tibshirani. 1990. Generalised Additive Models. Chapman & Hall Holland, J. H. 1992. Genetic algorithms. Sci. Amer. 1984. A Summary of Some Surface Fitting and Contouring Programs for N . Con- sulting Report No. ACT 84/6, CSIRO Division of Math- ematics and Statistics and Division of Water and Land Resources. ‚ Canberra . J. Bischof. 1983. A new method for esti- mating the spatial distribution of mean seasonal and annual jen applied to the Hunter Valley, New South Wales. Austral. Meteorol. Mag. 31: 179-184. =. 7 < d. 1997. BioRap Rapid Assessment of Biodiversity, Vol. 2. Spatial ве Tools. Austra lian BioRap ‚ Consortium, CSIR | . J. F. ter Bud & O. Е уап Топ- ata Analys sis in ак. and Land- p Ecology. [s “екеніңе Kent, M. & P. Coker : MM Description and Analysis: А . 1. Belhaven Press, Lon- don Laathwick. J. R. 1995. Climatic .. of some New Zealand forest tree species. eg. 237-248. & Mitchell. 1992, Forest patiem, climate and vulcanism in central North Island, New Zealand. J. Veg. Sci. 2 —616. 994. Decision Trees, Artificial Neural Net- works and pss Algorithms for Classification of Re- motely Sensed and Ancillary Data. 7th Australian Re- mote Sensing Conference Proceedings, Australia, March 1994. — . Ritman. 1991. Decision-tree and rule-in- Melbourne, duction approach to integration of remotely sensed and GIS data in mapping 1. in disturbed or hilly environments. Environm. Mana 3-8: d m (editor). 1986. Heo of Elapid Snakes of . Australian Flora and Fauna Series, Number 7. Нап of Flora and ше Australian Government 2. Service, Canberr. McCullagh, P. & J. A. Nelder. "logo. Models, 2nd ed. 2. & Hall, Lo McKenzie, Robinson & L. Belbi 1991. Biogeographic survey of the Nullarbor б Austra- lia. Pp. 54—63 in C. R. Margules & M. P. Austin (edi- tors), Nature Conservation: Cost Effective Biological 2-2 Linear ndor . 1991. Nature Conse ке ; Cost Effective Biological Surveys and Data Analysis. CSIRO, Melbourne, Australia. 994. Biological models for monitor- ing species decline : The PAPA and use of data bases. be Теве. Ѕег. В З -75. & A. O. Nicholls. о ues sing the conser- vation oim of remnant habitat ‘islands’: Mallee pate :h- . Hopkins (editors), Nature Conserva- tion: ‘The Role p Remnant Vegetation. Surrey Beatty and Sons, Sydney. edhead. 1995. BioRap. Guidelines for Using the BioRap Methodology and Tools. CSIRO Dick- . T. W. Norton, R. B. Cunningham & H. Biological үш for conservation sin uation: 1. Design of a field survey for diurnal, terrestrial birds in 41. Australia. Tocat Ecol. Managem. 85: 107-122. Nicholls, A. O. How to make biological surveys go further nie Generale Linear Models. Biol. Conser- vation. us –75. ы of the use of a pui models in Meus of survey data for conservation eval- uation. Pp.191-201 in C. R. Margules & M. P. ps en (editors), Nature Conservation: Cost Effective ас Surveys and Data Analysis. 2. Melbourn & C. R. Margules. 1993 el reserve selection algorithm. Biol. (шн 64: 165-169. Nix, H. 19 A как у analysis of Australian el- apid 1 Pp. 4-15 i tralian National Parks and 1989. Ecological traits of Eucalyptus СНеги. subgenera Monocalyptus and Symphyomyrtus. Austral. . 1997. BioRap Rapid Assessment of Biodi- versity, Vol. 4. Tools for Storing and Mapping Spatial Data. Australian — Consortium, CSIRO, телі жн Norton, T. W. & N. D. Mitchell. 1993. Application of generalised additive models to wildlife esed and population viability analysis. /n: Proceedings of the 7th Annual Symposi 993. Ministry of Supply and Services, Vancouver, Canada. Volume 85, Number 1 1998 Austin Australian Eucalypt Forests Noy-Meir, 1. 1971. Multivariate analysis of the semi-arid vegetation in south-eastern Australia: Nodal Ordination by Components Analysis. кии Ecology. Proc. Ecol. Soc. Australia 6: 159. NSW NPWS (New South UM o Parks and Wild- life Service). 1994a. Fauna of North-east NSW a North-east Forests Biodiversity Study Re epo p Report NSW National Parks and Wildlife “ уіс -----. 1994b. Flora of North-east NSW Forests. North- east Forests Biodiversity Study R Oksanen, J. 1997. Why the beta-function cannot be used to estimate skewness of species responses. J. Veg. Sci 8: 147-152. Pressey, R. L. 1994. Ad hoc reservations: Forward or backward steps in developing representative reserve systems. Conservation Biol. 8: 662 3 ----- Algorithms, lapi and timber: An example of the role of science in a public, political врне рго- cess over new conservation areas in production forests. In R. T. Wills, R. J. Hobbs & M. D. Fox (editors), Ecol- ogy for Everyone: Communicating Ecology to Scientists, the Public and the Politicians. Surrey Beatty and Sons, ida d ш press). ——— . O. Nicholls. 1989a. Efficiency in conser- vation uie bet Scoring versus iterative approaches. Biol. Conservation | 199–218. -----. 1989Һ. Application of a numerical algorithm to the selection of reserves in semi-arid New South eia Biol. Conservation 50: 263-278. ias ; . P. Austin Ha bit tat peculiarity s à cause га dui in Eucalyptus paliformis. Austral. J. Ecol. 176: -2 Statistical ean 1993. S-PLUS. Version 3.2. Math- 8. Large-area mapping 5: 34—47. . What satellite imagery and large- scale field studies can tell about biodiversity a in Amazonian forests. Ann. Missouri Bot. Gard. 85: 4 62. Soberón, J.. J. Llorente & H. Benítez. 1996. An inter- national view of И surveys. Апп. Missouri Bot. Gard. 83: 562-573 Wahba, P € ]. Wendelberger. 1980. Some new mathe- matical methods for variational objective analysis using splines and cross-validation. Monthly Weath. Rev. 108: 1122-1143 Walker, P. A. & K. D. Cocks. 1991. HABITAT: A pro- ани for modeling a disjoi nt environmental envelope an xps und species. Global Ecol. Biogeogr. Lett : 108-1 18. | . J., J. D. Grimbeek, M. J. Van der Linde & A. S. Van ¡ei An evaluation of the gradsect biolog- ical survey method. Biodiversity & Conservation (in press). Yee, T. W. & N. D. Mitchell. 1991. — Р models in plant ecology. J. Veg. Sci. 2: DECIPHERING LANDSCAPE Deborah A. Clark?-* MOSAICS OF NEOTROPICAL TREES: GIS AND SYSTEMATIC SAMPLING PROVIDE NEW VIEWS OF TROPICAL RAINFOREST DIVERSITY! ABSTRACT How are tree spec ies within tropical rainforests distributed at the landscape scale? One research site, the La Selva Biological Station in Costa Rica, offers exceptional tools for addressing this d a documented flora, soil and topographic maps, a reserve-wide grid, and a Geographical Information System (GIS). My colleagues and 1 have com- bined these tools with highly replicated systematic sampling over 600 ha of old . to inve stigate patterns of forest composition within this lowland tropical wet forest. This approach has revealed features of within-forest heterogeneity that had remained "invisible" during extensive fieldwork by many rese к at La Selva. Examples are: a doubling in density of the guild of subcanopy ~ canopy ln: го аза flat terrain and increasingly steep topography: strong shifts in density of many palm and tree species over La Selva's limited car of soils and topography; evidence of human harvesting of one palm species from old-growth ore st; and evidence suggesting indigenous human activity deep о the reserve (the co-occurrence of a previously unrecognized zone of alluvial soil, buried charcoal, and ап avocado . These studies have also dne 15 tree species to the known flora of this intensively researched forest. Although Selva's support for such landscape-scale studies is exceptional, even in remote tropical forests it is now possible to НЫ ally sample апа geo-reference information on site variation и ке distributions using newly available Global Positioning Systems. Findings can then be cross-referenced with current and future site data. using a GIS. E such efforts, especially the development of a GIS, require iii investments of time and expertise, the payoff can be a more robust understanding of the distribution of tree diversity and species abundances over tropical 4. landscapes. How many tree species exist in tropical rainfor- One reason for this situation is extreme species ests? Where are they found, and how and why do richness. Those carrying out inventories of lowland their abundances vary spatially? Such information forest in the wet Neotropics typically find 80–300 + is critical for understanding the biodiversity, struc- species of trees = 10 cm in diameter co-occurring ture, and function of this biome and for conserving in a single hectare (Valencia et al., 1994; Gentry, representative sites into the future. Unfortunately, 1988; Foster & Hubbell, 1990; Lieberman et al., these superficially simple and basic questions 1985а). In addition to the sheer numbers of taxa, about the world's tropical rainforests are currently identification is made challenging by the sterile unanswerable. condition of most trees at any given time. Many ! I thank the Andrew W. Mellon Foundation, the National Science Foundation (BSR89-18185 and DEB94- 07581). and the Organization for Tropic val шн s (OTS) for financial support of this research. The La Selva Geographic Infor- mation System and reserve-wide grid were made possible by оо to OTS from the ode Science Foundation, the Environmental Systems eue h Institute, Sun Мк rosystet ns, Inc., and the Andrew W. Mellon Musis OTS also provided invaluable research and logistic al support. David B. Clar k was co-inv nux for all the researc S sum- marized here and provided many key ideas and constructive criticism for this paper. Phil Sollins 1. Же Selva's soil variability. Leonel Campos, William Brenes, and Rosa Sandoval carried ош fieldwork and data ШІЛ with care and expertise. Marco Vinicio Castro, Jenny Juárez, and Jane Read helped with GIS mapping. On-site зе in tree identification was provided by Orlando Vargas. The collections of the Herbario Nacional de Costa Rica d the Instituto Nacional de Biodiversidad de Costa Rica (INBio). and taxonomic expertise contributed by the eer la to the Plants of Costa Rica Project (B. E. Hammel, M. H. Grayum) and INBio М Zamora) were key for this study. as was Alwyn Gentry’s ме: 3) guide to the woody plants of northwest South Americ ? Mailing address: OTS-La Selva Biological Station, INTERLINK-341, PO. Box 02-5635, Miami, Florida 33152, U.S.A.; чай address: Department of Biology, University of Missouri-St. Louis, 8001 Natural Bridge Road, St. Louis, Missour 21-4499, U.S.A. 3 This paper is 5 to the memory of Alwyn H. Gentry (1945-1993), who made extraordinary contributions to the current understanding of the world's tropical forests. ANN. Missouni Bor. САңр. 85: 18-33. 1998. Volume 85, Number 1 1998 Clark 19 Deciphering Landscape Mosaics tropical wet forest plants only fruit and flower ep- isodically, some even supra-annually (cf. Newstrom 1994; Appanah, 1990), and in addition any plot includes many immature stems. Even when re- productive specimens can be obtained, identifying tropical rainforest trees is often not straightforward. et al., Reference collections, keys, and treatments are in- complete and constantly evolving, as are the ac- cepted names for given taxa. Much of the tree di- versity occurs in problematic families, such as Sapotaceae, Myrtaceae, and Lauraceae, with diffi- cult-to-separate species. For these and many other groups, any hope of definitive identification often rests with a few contemporary specialists, who are usually halfway around the world from the study site. Finally, as Gentry (1994) and others have pointed out, whenever a great investment of time and personnel results in distinguishing all tree spe- cies in a tropical rainforest plot, some to many of these turn out to be new to science. The bottom line is that, for most tropical forests, the tree flora remains poorly known. Only a handful of tropical forests have been well studied floristically (e.g., sites in Gentry, 1990; Condit, 1995). Even in the exceptional sites with long histories of plant col- lecting and well developed florulas, new species of trees keep turning up as researchers look carefully at the forest. Deciphering the nature and determinants of tree distributions within the world's tropical rainforests is clearly going to be difficult, but not simply be- cause of these issues of taxonomic complexity and incomplete collecting and monographing (Gentry, 1992). A second obstacle, a corollary to the rich- ness of these tree communities, is the local rarity of most species. For example, in a Costa Rican low- land wet forest (La Selva), 8196 of the tree species inventoried had densities of = 1 individual = 10 cm in diameter per ha (data from Lieberman et al., 1985b). The resulting sample-size limitations man- date innovative approaches for studying the distri- butions of most tree species within these forests. A third challenge is presented by another kind of diversity, the abiotic heterogeneity of tropical for- est landscapes. Those monotonous expanses of green viewed from overflying planes are actually complex mosaics of forest types. Underlying the species-rich tree communities are landscapes of in- terdigitated terrain types. The дары ысы patches differ among themselves in many ways: topography, soil nutrients, and hydrology ны 1964; Austin et al., 1972; Baillie et al., 1987; Kahn, 1987; Gen- try & Ortiz S., 1993; Rodin & Tuomisto, 1993; Tuomisto et al., 1995; Clark et al., 1995; Duivenvoorden, 1996), disturbance histories (e.g., very large blowdowns; Nelson et al., 1994), flooding regimes (Salo & Rasanen, 1989; Foster, 1990), and histories of human intervention (Gordon, 1982; Gó- mez-Pompa & Kaus, 1990; Anderson, 1990; Bush & Colinvaux, 1994). In tandem with the incomplete knowledge of floristics goes a poor understanding of this spatial heterogeneity in tropical rainforest landscapes. Part of this has simply been due to the difficult logistics and the size of the problem—for many areas of tropical rainforest, studies of all of the above factors are still lacking. Efforts to discern the spatial mosaics within these forests will greatly benefit from an interdisciplinary approach. Tropical forest ecologists and plant systematists, particularly those trained in North America, often have little training in soils and geomorphology. Soil scientists and geologists are well prepared to evaluate the spatial variation in these site factors, but usually lack any knowledge of plant systematics. Similarly, anthropologists, historical geographers, and archae- ologists have special skills for assessing current and historical potential human impacts within a for- est. Understanding the patterns and causes of tree distributions within tropical rainforests will require pooling information and insights from these dispa- rate disciplines. In this paper, I describe in-progress research at a rainforest site that offers an unparalleled set of tools for deciphering the spatial mosaics within the tree community. This work has built on three key elements: an extensive site database generated through decades of research in many disciplines; the use of highly replicated, systematic sampling to study the forest at the landscape scale; and syn- thetic analysis of complementary types of spatial information, made possible by a Geographical In- formation System (GIS). I show how this combined approach is revealing levels of spatial heterogeneity within this forest that were previously unrecog- nized, in spite of the extensive research history of the site. I review the process of developing these research tools and then assess the potential appli- cability of these and other promising new methods for extending such studies more generally in trop- ical rainforest. AN EXCEPTIONAL SITE FOR ASSESSING RAINFOREST LANDSCAPES This research was carried out at the La Selva uu Station of the Organization for Tropical Studies (OTS), an international consortium of uni- versities and research institutions. La Selva (Fig. 1) is a 1550-ha reserve located in the Caribbean low- lands of northeast Costa Rica, Central America Annals of the Missouri Botanical Garden | “ Puerto Viejo 2 Д La Selva (ДЕ fe = е e 2N A |) z 47 g ' E P EL m. Vara ' с Мерго Blanca 2 2906 m. Volcán Barva nis os ІІ. па! Рагк to Heredia to San José to Limón Figure l. Location of the La Selva Biological Station, )TS). (10°26'N, 84?00' W; elevation 37-150 m). It is con- tiguous with Braulio Carrillo National Park (47,000 ha), which protects a forested transect extending from the lowlands to 2900 m above sea level. Selva is classified in the Holdridge Life Zone Sys- tem as Tropical Wet Forest (Hartshorn & Hammel, 1994). Mean temperature is 26°C, and mean annual rainfall is 4 m, with every month averaging at least 100 mm of rain (Sanford et al., 1994). Detailed site information is given in McDade et al. (1994) Due to a combination of factors, La Selva offers exceptional research support for landscape-scale studies in tropical rainforest (Clark, 1990). Over nearly 30 years, the station has evolved from a rus- tic farmhouse at the edge of a remote lowland forest into one of the most intensively studied tropical rainforests worldwide. Currently more than 250 re- searchers a year at La Selva carry out studies in the fields of systematics, evolutionary ecology, long- term population dynamics, conservation biology, soil science, forestry, ecosystem ecology, and at- mospheric chemistry. This multifaceted research activity has built up a rich site database and cre- ates opportunities for interdisciplinary studies. The station’s research infrastructure includes four ad- ditional elements that have been critical for our Costa Rica, Central America (map produced by J. Juárez, studies of the spatial heterogeneity of tree distri- butions: a documented flora, a reserve-wide grid system, a detailed soils map, and a site GIS. PLANT IDENTIFICATION For the non-specialist, identifying plants is eas- ier at La Selva than at most tropical rainforest sites. The in-progress La Selva Flora Project (Wilbur, 1986) has produced a checklist of 1852 vascular plant species (R. Wilbur, pers. as including 323 species of trees > 10 cm in diameter (Hart- shorn € Hammel, 1994). Several treatments have been published (see Hartshorn & Hammel, 1994), and the full flora is nearing completion. The station has a small reference herbarium, plus a collection of ca. 4000 plasticized high-contrast xeroxes (meth- od developed by R. B. Foster) of La Selva speci- mens deposited at Duke University Herbarium, headquarters of the La Selva Flora Project. The site’s arboretum includes more than 1300 trees of 230 species. The Station Naturalist (O. Vargas) can provide preliminary identifications for sterile ma- terial of much of the flora. Equally important, at only 1.5-2 hrs. from La Selva, the principal inves- tigators of the “Manual to the Plants of Costa Rica” Volume 85, Number 1 Clark Deciphering Landscape Mosaics project (B. E. Hammel and M. H. Grayum, with collaborator N. Zamora) are based at Costa Rica's National Biodiversity Institute (INBio) and have generously helped with identification of problem- atic taxa. Costa Rica's national herbarium (CR) also maintains a large curated collection in San José. THE LA SELVA GRID In 1991, a remarkable tool for spatially refer- enced field research was installed at La Selva: a reserve-wide grid system. Covering all 1550 ha of La Selva, the grid consists of permanent marker posts at 50 m X 100 m spacing, surveyed (x, y, and z) to decimeter accuracy. The 6000-- points surveyed during the installation of the grid also pro- vided the means to generate a topographic map of the entire reserve. We know of no other tropical rainforest with anything approaching this level of baseline spatial information (the closest would be the growing world network of 50-ha tropical forest plots [Condit, 1995], which are gridded at a 5 m X 9 m spacing). Scientists working in temperate for- ests are likely to take for granted the availability of topographic maps such as the U.S. Geological Survey quadrangles for the United States. Virtually all tropical forest sites, however, lack such funda- mental site data. For this reason, there has been very little spatial referencing and analysis of land- scape composition in tropical field studies. The grid has revolutionized how researchers de- sign and carry out their field studies at La Selva. Now, any organism or observation can be mapped within the forest simply by measuring to the nearest grid post. Most La Selva researchers currently spa- tially reference their field data. This makes it pos- sible to relate their findings to data from other pro- jects and to baseline site information. In addition, the grid provides a basis for systematic, highly rep- licated sampling over large expanses of forest, a prerequisite for assessing how species vary across the landscape. A RESERVE-WIDE SOILS MAP In the early years of research at La Selva, the reserve was considered to include four broad soil types (cf. Hartshorn, 1983): the most fertile sites, currently episodically flooded by the major rivers; *Old Alluvium," higher areas of intermediate fertility interpreted to be river ter- races from the Pleistocene; “Swamp,” the perma- nently or seasonally wet zones within La Selva; and “Residual,” the largest portion of the reserve, with broken topography and infertile soils produced by in-place weathering of the underlying lava flows. “Recent Alluvium," This broad classification scheme, however, was un- supported by landscape-scale soil chemical studies or soil mapping. It is perhaps not surprising that most research at La Selva was carried out without regard to the forest's edaphic variation. In 1987, however, this situation was radically changed when a professional soil survey was car- ried out. The resulting 1:10,000 soil map of La Selva (Sancho & Mata, 1987) may be the most in- tensive soil mapping available for any comparable area of tropical forest worldwide (P. Sollins, pers. comm.). Sancho and Mata used extensive field re- connaissance, analysis of large numbers of grab samples, and more detailed analysis of soil profiles to demarcate 23 soil consociations and 1 complex. La Selva's soils were shown to range from infertile ultisols, a dominant soil type of the world's tropics (Richter & Babbar, 1991), to relatively fertile en- tisols and inceptisols (Sollins et al., 1994). The ad- vent of the soils map and considerable “conscious- ness-raising" of the researcher community by resident soil scientists and ecosystem ecologists have stimulated most current field researchers to factor La Selva's edaphic variation into their stud- ies, just as they stimulated us to investigate the landscape-scale mosaics of tree community com- position within the forest. s valuable as it is, the La Selva soils map is not a static, definitive resource. It is, and should P. constantly evolving as new information comes n. Further, as found for soil maps in other parts of js world (cf. Lathrop et al., 1995; Oberthur et al., 1996), it will always incorporate uncertainties at some scales. Sancho and Mata (1987) did their soil survey before there was a grid, a topographic map, or a reliably surveyed map of the reserve; thus, there were bound to be errors in their delimitation of soil units. Second, their map depicted patterns at a scale of 1:10,000. Such a map inherently in- volves uncertainty at the finer scales of the phe- nomena studied by most field researchers. Soil map units by definition have inclusions of other soil types, and this is necessarily the case in sites like La Selva with substantial soil variation at the very local scale, a condition likely to be general within tropical soils (Richter & Babbar, 1991). Sancho and Mata's consociations are defined as mapping units within which > 75% of the area is the described soil type. Finally, as in plant taxonomy, there are soil splitters and lumpers. Whether La Selva's im- portant soil variation comprises 24 units or many fewer ones (or many more) is a matter for researcher evaluation, and the answer is likely to vary with the organism or phenomenon being studied. For our studies of tree distributions within La Selva, David Annals of the Missouri Botanical Garden • Tubes « Signs ‚ Contours Sheet 15 Hoja EY en ic North e Magnético Map units in meters Unidades en metros 500 Figure 2. Juárez, OTS) 600 800 900 1000 1100 Clark and I modified the soils map both by aggre- gating consociations and by refining Sancho and Mata's (1987) consociation boundaries after carry- ing out intensive grid-based soil sampling (see be- low). THE LA SELVA GIS In 1989 OTS initiated a GIS for the station. The on-site GIS lab includes two Sun workstations that run the GIS software ARC/INFO, a large-format plotter, and a digitizer. Early system development involved major donations of equipment and finan- cial support,' as well as an extensive investment of time in design and set-up by experienced GIS per- sonnel. The station now maintains a full-time GIS manager for data updating and maintenance and to help individual system users. General-use data id ers that have been incorporated into the GIS (e.g Fig. 2) include site topography, stream and river courses, boundaries and trails, the soils map, cur- rent and past land use, locations of study plots, and the 3000-- grid posts. Remote-sensed imagery of the reserve and surrounding region is also being incorporated. Researchers are increasingly using 1200 La Selva Orientation Orientación en La Selva 1300 1400 1500 An example of a field map detailing GIS data for one quadrangle within La Selva (map produced by J. the grid to spatially reference their field data, in- corporate them into the La Selva GIS, and then relate them to the other available data layers. For our studies of tree distributions within La Selva, the 15 was a critical resource. GIS AND SYSTEMATIC SAMPLING REVEAL MOSAICS OF TREE DIVERSITY WITHIN A TROPICAL RAINFOREST Over the last several years, David Clark and I and several collaborators have taken advantage of the research tools at La Selva to investigate how tree community composition varies within the old- growth forest landscape. We began with a study fo- cusing on the large palms, then used the grid and IS to assess the distributions of a core set of tree species under long-term study, and most recently have scaled up to study the spatial variation of overall tree floristics. In all three cases we have evaluated patterns at the scale of multiple 100s of hectares. This landscape-scale focus and an inte- gration of site data with tree distributions have re- vealed previously unrecognized levels of internal heterogeneity within this forest. Volume 85, Number 1 Clark Deciphering Landscape Mosaics Table 1. at La Selva. Tribes and species names from The seven taxa of canopy and subcanopy palms (Arecaceae, subfamily Arecoideae) in old-growth forest Henderson et al. (1995) Vouchers are specimens in the Costa Rican National Herbarium (CR). Summary relationships are from analyses in Clark et al. (1995) Association between:! Densi Occurrence (presence/absence) (stems/ha) and: anc Soil/ Tribe Species [Voucher No.] soil type topography Harvesting Iriarteeae Iriartea deltoidea Ruiz & Pav. + = [local absence] + 7 l. 1968 Socratea exorrhiza (Mart.) H. Wendl. + - [everywhere] = [Stevens 24559] Areceae Prestoea decurrens (H. Wendl. ex Burret) + + [alluvial/flat] - H. E. Moore [Grayum & Jermy 6783] Euterpe precatoria Mart. var. longevaginata? + P [residual/slopes] - (Mart.) Andrew Hend. [Grayum 7813] Geonomeae Welfia regia H. Wendl. ex André + = [everywhere] - [Wiemann & Rich 137] Cocoeae Astrocaryum confertum H. Wendl. ex rare in upland це [de Nevers & Hammel 7820] P odiis alatum H. F. Loomis rare in upland [Stevens 24625] ! + a significant association; — no such association (from Clark et al., 1995). ? “Euterpe macrospadix Oersted” in Clark et al. (1995). ! *Welfia georgii Wendl. ex Burret” in Clark et al. (1995). NON-RANDOM DISTRIBUTIONS OF THE LARGE PALMS Palms are an important component of the La Sel- va old-growth forest. The seven species of subcan- opy and canopy palms (Table 1, henceforth referred to by genus) comprise 2546 of all woody stems — 10 ст in diameter (Lieberman et al., 1985a). This abundant species group has significant impacts on forest structure—the large palms' dense canopies can strongly affect the distribution of light environ- ments in the understory, and the senescing and fall- ing of their massive leaves contributes to the high levels of physical damage to smaller plants below them (cf. Vandermeer, 1977; Clark & Clark, 1989). The fruits of several of these palm species are also important in the diets of diverse mammal species (Levey et al., 1994; Timm, 1994). For all these rea- sons, it is of interest to know how this guild of plants is distributed within La Selva. Does the group as a whole vary in density among different sectors of the landscape? Do any of the component species show non-uniform distributions within the old-growth forest? Approaches to such questions about tropical for- est composition have usually involved evaluating plant species abundances within one to a few plots or transects selected by the researcher as represen- tative of the forest as a whole (a notable exception is the pioneering work of Ashton [1964, and Austin et al. |1972)). Such a design has been to a large degree mandated by both the lack of base maps of site variation and the difficult logistics within such forests. Although enumerating all spe- cies within a plot is indeed the only way to study numerous aspects of plant population structure and dynamics (cf. Dallmeier, 1992), this approach is not suited for generalization to the larger landscape, and the findings can be uninterpretable in terms of specific site variables, such as topography or soil type. One illustration of these issues is given by data from three upland forest inventory plots within La Selva (Hartshorn, 1983), which provided strik- ing evidence of non-uniform distributions of the large palms. In all three plots (2, 4, and 4 ha), the second most abundant tree species = 10 cm in Annals of the Missouri Botanical Garden diameter was a palm, but the species was different in each plot (Iriartea in Plot IIb [Alluvial Soil]; Wel- fia in Plot I [Alluvial Soil plus some Swamp; and lriartea was absent from this plot; Hartshorn 4 Hammel, 1994]; Socratea in Plot III [Residual Soil]; species names of the single representative of each genus at La Selva are given in Table 1). Although these plot data demonstrate that the distribution of large palms varies markedly within La Selva, they are insufficient for assessing how these distribu- tions relate to the forest's soil and topographic vari- ation. The advent of GIS and the grid at La Selva, how- ever, made possible a radically different approach to assessing plant distributions within the old growth. The grid, with its 50 m X 100 m spacing, was a set of precisely located points spread over the entire forest. Using them as sample points, we could systematically sample a large landscape. By assessing palm distributions this way in many wa- tersheds and terrain types, and in different forest stages (gap to mature forest) and topographies with- in each soil type, we could achieve a level of en- vironmental replication difficult to achieve with other methods. To investigate the landscape distributions of large palms at La Selva, we selected a 568-ha sec- tor of the upland (non-swamp) old-growth forest. We then assessed the presence/absence and local abundance of the seven species at each of 516 grid intersections within this study area. We и our field notes regarding streams, swamps, and topog- raphy to refine the Sancho and Mata ee ub map, and we used the GIS to aggregate soil con- sociations into four contrasting units: Alluvium (most fertile, with gentle topography); Streams (the valley soils of the principal streams); Residual soils broken terrain, soils weathered from lava); and Ar- boleda (a problematic area of steep topography, thought to be of intermediate fertility [Sollins et al., 1994]). In the field we also classified each sample point in terms of topographic position. The meth- ods, results, and statistical analyses of this study are detailed by Clark et al. (1995). Here I highlight several findings that demonstrated a spatial sub- structuring of this forest previously invisible to us and dozens of others who have worked at the site for a decade o We found ert ili total abundance of the canopy and subcanopy palms varies greatly across the landscape. Larger individuals of this guild (stems = 10 m tall, all species combined) continuously decrease in density from slope crests, to slopes of decreasing steepness, to slope bases and flat ter- rain. Over this topographic gradient the density of % OF POINTS WHERE PRESENT © EUTERPE PRESTOEA SPECIES m RESIDUAL GARBOLEDA OSTREAMS OALLUVIUM Figure 3. Contrasting soil affinities of Euterpe preca- toria ie Prestoea decurrens (from Clark et al., 1985). Data species are the percent of the vat d points in each “soil type at which any 4. al > 1 m tall was observed. Sample points per soil type (s a descriptions n text): Residual, 329; Arboleda 44; Streams, 65; “or both species the association between Хани der ice |. and soil type was highly sig- nificant (x?, df = 3, Р < 0.0001). large canopy and subcanopy palms declines by a factor of two. Such spatial heterogeneity in the abundance of these palms within La Selva must have widespread impacts on forest structure and on processes as diverse as mammal activity and un- derstory light environments. The species-level distributions of these palms re- vealed further evidence of substructuring within the forest. All five of the species that were not rare in upland forest showed highly significant variation in local density or overall presence/absence patterns with respect to the spatial variation in topography and soils within La Selva (Table 1). One closely related species pair, La Selva’s single species of Prestoea and single species of Euterpe (Table 1), showed strong but contrasting edaphic associations, both with soil type (Fig. 3) and with topographic position. Euterpe was strongly biased toward steep topography and the less fertile soils. Prestoea, in contrast, while nearly omnipresent on gentle topog- raphy, was absent at half the sample points on the soils with steep slopes. The most abundant large palm, the single Welfia species (Table 1), strongly varied in density across the four soil units although it was present everywhere (at 100% of the sample points). The use of GIS to investigate the spatial distri- Volume 85, Number 1 1998 Clark 25 Deciphering Landscape Mosaics butions of these large palms revealed yet another striking feature of the La Selva old-growth forest that was previously unsuspected. From one sector of what had often been considered “virgin forest,” the otherwise omnipresent canopy palm Iriartea (deltoidea; Table 1) is nearly completely absent. The most probable explanation for this anomalous distribution is local removal by historic human har- vesting. Our analyses demonstrated that the pat- terns of presence/absence for both Iriartea and the closely related Socratea (exorrhiza; Table 1) were insensitive to both soil type and topographic posi- tion. Socratea was omnipresent within the old- growth forest. Similarly, Iriartea was omnipresent on all soil types, except in one zone of Alluvium close to former human habitation and easy river access. By interviewing local residents we found that Iriartea was the most sought-after native palm taxon for its large and tasty meristem (“heart-of- palm") and its robust, durable stems, which were used in construction. That Socratea remained in this sector of the forest is probably due both to the bitter taste of its meristem (“palmito amargo") and to its more slender stems, less useful for building. This evidence of prior human impact on tree flo- ristics within this intensively studied sector of the La Selva old growth has changed how the forest is viewed. In addition to being key background infor- mation for many of the studies carried out in this particular sector, this finding has stimulated La Sel- va researchers to be on the alert for evidence of other human impacts within the old growth. GIS REVEALS EDAPHIC ASSOCIATIONS OF CANOPY AND EMERGENT TREES UNDER LONG-TERM STUDY When the grid and GIS were installed at La Sel- va, they offered an opportunity to investigate the spatial distributions of the tree species David Clark and I had under long-term demographic study in the old-growth forest. Since 1982 we have accu- mulated samples of individuals of all post-seedling life history stages (from 50-cm-tall saplings to adults) of nine ecologically contrasting species of canopy and emergent trees at La Selva (voucher numbers are for specimens deposited in the Her- bario Nacional de Costa Rica): Dipteryx panamensis (Pittier) Record & Ме! (Papilionaceae) [R. Robles 1199]; Minquartia guianensis Aubl. (Olacaceae) (С. Herrera 2250]; Lecythis ampla Miers (Lecythida- ceae) [R. Robles 2208]; Hymenolobium mesoameri- canum H. C. Lima (Papilionaceae) [R. Aguilar 19]; Pithecellobium elegans Ducke (Mimosaceae) (В. Hammel 17319]; Hyeronima alchorneoides Allemáo (Euphorbiaceae) [Chacón 751]; Simarouba amara Aubl. (Simaroubaceae) [R. Robles 1670]; Cecropia insignis Liebm. (Cecropiaceae) [W. Burger 11135); and Cecropia obtusifolia Bertol. (Cecropiaceae) [R. Robles 1446]). Our annual measurements of surviv- al, growth, and microsite of > 2800 individuals have enabled us to evaluate these species’ onto- genetic growth patterns, relation to light environ- ments and forest dynamics, and sensitivity to yearly climatic variation (cf. Clark & Clark, 1992, 1994; Clark et al., 1993). Before the advent of the GIS and the grid, however, our understanding of these trees’ relation to the edaphic variation within La Selva was limited to an intuitive sense that Dipteryx was associated with the Alluvium and Pithecellob- ium with the Residual soils. We have recently used the new spatial tools at La Selva to refine our un- derstanding of edaphic variation within the forest and then to assess the distributions of our nine fo- cal tree species with respect to this variation (D. B. Clark et al., 1998) To refine the soils map, we used the grid inter- sections as a framework for systematically sampling the soils within a 573-ha section of the reserve (one soil sample from 50-cm depth at each of 1171 grid points). We arrayed the resulting soil samples geo- graphically and then classified each into a soil type, following the concepts originally developed by San- cho and Mata (1987) and Sollins et al. (1994). For classification, we combined soil color with our field- collected data on each point's topographic position and slope angle and with GIS data for point ele- vation and surrounding terrain (from a kriged dig- ital elevation model of La Selva based on the 000+ surveyed points). We delimited polygons following elevational contour lines around the spa- tial groupings of soil units at grid points, and then digitized the resulting soils map into the La Selva GIS. Although our analysis largely confirmed the previous soil mapping of the reserve, it did result in some significant changes in unit boundaries and a more intuitive geographic relation between site geomorphology (Alvarado I., 1990) and the soils. It also significantly altered our understanding of soil variation in some areas that had been study sites for diverse research studies within La Selva (see below). For our study of the edaphic associations of the nine tree species, we aggregated the upland soil types into three broad units: Old Alluvium (ex- cludes the currently floodable Recent Alluvium), Residual, and Stream Valley. We had previously mapped our tree population samples into the GIS by referring each tree in the field to the grid (by measuring distance and com- pass bearing to the nearest grid post or to the near- est tree that had been so mapped), and then incor- Annals of the Missouri Botanical Garden - Study a individual a ЕС Swam Figure 4. term study populations of our nine Rica. Individual trees are indicated by plac k dots. of the original (Sancho & Mata, 1987) La Selva ma porating these tree location data into the GIS. Although this process was very time consuming (more than 1 person-year of work), it created a per- manent spatially explicit data set for the long-term population ecology studies, and it was a prerequi- site for a tree X soil analysis. For this analysis, we used the GIS to delimit the old-growth areas of La Selva containing our tree samples, overlaid this map on the new soils coverage (Fig. 4), and then used the included grid intersections to generate the expected (background) distributions of both soil and topography within our 216-ha study area. A GIS-generated m of the 216 ha area Gaa 'ated by thick black border) encompassing the long- focal species of canopy and e t Iob soil map (described in the text) is a refined version nergent trees at the La Selva Biological Station, Costa This GIS analysis revealed that, in the upland old- growth forest, all nine of our study species had non- random distributions associated with one or more features of La Selva's edaphic variation. Most of the highly significant associations (P < 5) were with soil type (6 species). Within their preferred soil, three of the nine species also showed highly signif- icant biases with respect to topographic position. As we had suspected, the distribution of Pithecellobium elegans was strongly skewed toward the Residual soils, and that of Dipteryx panamensis was biased toward the Alluvium. Both species, however, showed Volume 85, Number 1 1998 Clark 27 Deciphering Landscape Mosaics additional biases that we had not perceived during many years of fieldwork: P. elegans was preferen- tially on flat ridgetops and biased away from slope bases within the Residual soils, while the distribu- tion of D. panamensis within the Alluvium was bi- ased toward gentle slopes and away from flat terrain. Another discovery was that La Selva's two Cecropia species had strong, contrasting edaphic associations (C. insignis: Stream Valley soils, lower slope angles; C. obtusifolia, higher slope positions, but no bias regarding slope angle or soil type). SCALING UP: THE EDAPHIC ASSOCIATIONS OF THE LA SELVA TREE FLORA We have recently scaled up these approaches to analyze distribution patterns within the total La Selva tree community and their relation to soil type and topography. We once again have used the grid as a basis for systematic, highly replicated sam- pling of old-growth forest, this time including the swamps. At each grid intersection (№ = 1171) in 573 ha of La Selva old growth, we established a 0.01-ha circular plot within which we measured and identified all tree stems of > 10 cm diameter (5127 stems total). This landscape-scale approach has provided new insights about how tree diversity is distributed within La Selva. The sample, which includes 253 of the 323 tree species known from La Selva, also resulted in the addition of 15 new species to the site's tree list. One of these, Caryodendron angustifolium Запа. (Eu- phorbiaceae), represents a genus new to Costa Rica. ese finds are a result of the process of “ecological the challenging work of identifying all trees in a plot, including the non-reproductives. This collecting," approach to tropical forests is very different from that taken by taxonomic specialists, who naturally focus on the plants that are flowering or fruiting when they visit a site (Gentry, 1994). As Gentry pointed out, even some of the more common species found in an all-stem inventory in a given tropical forest can tum out to be new to the site or to science. Additionally, our sampling over the whole landscape turned up nu- merous examples from the “tail of the species distri- bution" (B. E. Hammel, pers. comm.)—those many tree taxa that are locally very rare within any tropical rainforest. Such discoveries improve understanding both of local site biodiversity and of species distri- butions at larger spatial scales within tropical forest regions. In addition to these new finds, use of the GIS for spatial examination of the data set (D. B. Clark et al., unpublished data) has revealed extensive substruc- turing of the tree community across this rainforest landscape (Fig. 5). As with the guild of large palms, some other tree species are strongly associated with certain soils (e.g., Fig. 5A, Castilla elastica Sessé in the Old Alluvium and Recent Alluvium). Additional interesting distribution patterns are evident for many tree species. For example, Lonchocarpus oliganthus F. J. Herm. (Fig. 5B) is associated with the Recent АЈ- luvium, the Swamp soils, and with the Stream Valley soils that meander through the Residual Soil region of La Selva. These soil units range from nutrient-rich (Recent Alluvium) to strongly infertile (Stream Valley soils); this species is able to grow over a wide range of soil fertility but only in sites that are wet or flood- able. Another distribution pattern found for numerous species is that shown by one of the canopy-level Ma- tisia species (Bombacaceae) at La Selva, M. ochro- calyx K. Schum.; although confined to the Residual soils in La Selva, this species occupies only a re- stricted area within this soil type (Fig. 5A). With this type of distribution, factors other than the variation ong major soil units must be playing a role. One possibility is individualistic responses of tree species to combinations of particular soil characteristics, such as certain cations, P, N, pH, soil organic matter, or texture, coupled with substantial variation in these characteristics within soil units. Such complex idio- syncratic edaphic associations were found to charac- terize many tree species in tropical rainforest areas of Sarawak, Borneo (Baillie et al., 1987). To investigate = PEY our next step will be to go beyond the " characterization of soil variation so far used at La Selva and to measure and map individual soil properties across the landscape. We will analyze a large suite of soil characters from each of the sample points used in this study and then combine GIS and multivariate techniques to examine them for associa- tions with the non-random distributions of trees within La Selva. This factor-level approach will greatly aid interpretation of the currently enigmatic distributions of many of the species. In addition to revealing the mosaic nature of the tree community in this tropical rainforest, this GIS- based analysis has also produced new evidence suggestive of past human activity deep within the old-growth forest. Within one of our 0.01-ћа tree inventory plots in well developed forest in La Sel- va's center, 80 m from the nearest trail and more than 1 km from any historic human habitation (Fig. 5a), we found a tree of the species of cultivated avocado, Persea americana Mill. (Lauraceae). This species has been found in very old archaeological deposits (7000 B.C.) in the Americas (Simpson & Connor Ogorzaly, 1986). The site where we en- countered this tree is very close to where charcoal was found within the soil profile by researchers (R. 28 Annals of the Missouri Botanical Garden EN Recent Alluvium [75] Swamp Stream-associated О 200 400 Meters m—— ти Figure 5. indicate ui occurrences bd a pen sper ies in (йе к crosses, Castilla 2. (Моғас PAR Examples of spatial substructuring of the tree community in upland old-growth forest at La Selva. Symbols d шү centered on all g rid i intersec tions (№ = 1171) within istributions in the sample = (this page). Three species’ di . Persea americana (Lauraceae); dots, Matisia ochrocalyx ка pas —B (next page). Occurrences of Lonchocarpus jk сан (Papilionaceae) in the sample plots (crosses Sanford, Jr. & S. Horn) е the reserve's history (Horn & Sanford, 1992, and unpublished). By the original soils map (же & Mata, 1987), the site plots out within the infertile Residual soils that cover most of La Selva, a puzzling location for the kinds of indigenous human activity suggested by the tree and charcoal. However, when we over- laid this collection point on our refined soils map (based on grid-based soil collections, see above), we found this site to be within a previously unrec- ognized zone of (more fertile) Old Alluvium within the Residual soil region. This is a classic case of how such GIS data syntheses can enhance the re- sults of individual studies. Taken by themselves, these lines of evidence are much less interesting than when combined through spatial referencing. SCALES OF SPATIAL VARIATION WITHIN TROPICAL. RAINFORESTS, AND THE Toors TO INVESTIGATE IEM DIFFERENT LEVELS OF WITHIN-FOREST EDAPHIC MOSAICS It is becoming increasingly clear that marked in- ternal heterogeneity is an important characteristic of Volume 85, Number 1 1998 Clark 29 Deciphering Landscape Mosaics SOILS С] Residual | Old Alluvium БЕЙ Recent Alluvium [27] Swamp Stream-associated 0 200 400 Meters Pr Figure 5. the tropical rainforest biome (Ashton, 1964; Austin et al., 1972; Baillie et al., 1987; Kahn, 1987; Gentry & Ortiz S., 1993; Ruokolainen & Tuomisto, 1993; Tuo- misto et al., 1995; Clark et al., 1995; Duivenvoorden, 1996; Clark, present study). Recognizing and inter- preting the spatial mosaics within these forests will be fundamental to both understanding and conserving their great biodiversity. The markedly different scales of important spatial variation, however, will require distinct research approaches. At one scale are structurally distinct forest types that can be distinguished by field reconnaissance, visual inspection from small planes, or interpreta- tion of remote-sensed data. Examples of visually distinct patches that harbor particular floras are Continued. swamp forests dominated by a few species of large palms and the stands of low canopy and small-di- ameter trees found on very infertile white sands across Amazonia. Assessing how such strongly con- trasting patches are distributed within tropical rain- forest landscapes is a vital need for understanding the spatial distribution of biodiversity within the biome (cf. Tuomisto et al., 1995). Evaluation of the community-level biotic differences among these patch types is also needed. Given the dramatic veg- etation shifts between these readily distinguishable forest types (cf. Tuomisto et al., 1995; Duivenvoor- den, 1996; Terborgh et al., 1996), inventorying a few plots or transects within each should adequate- ly indicate major distinctions among them. Annals of the Missouri Botanical Garden А second level of tropical forest mosaics is of those occurring within one of these patch types and thus across much more limited gradients of envi- ronmental variation. This is the scale that we have been principally investigating within La Selva: the internal heterogeneity of an upland tropical forest landscape, in the absence of extreme intra-site con- trasts such as those occurring on adjacent terraces within floodplains or where white sand and clay- dominated soils are interdigitated (cf. Duivenvoor- den, 1996). As we have shown, even at this “with- in-patch" scale, the tree community composition of tropical moist forest can show marked spatial vari- ation. We found much of the internal heterogeneity in the La Selva tree flora to be associated with local variation in soil types and topography, even though the total relief and the total range of soil characters were quite constrained subsets of those found more regionally in the lowland wet tropics. Most of these tree-site associations, even some of the most strik- ing ones, were previously unrecognized, despite the long history of research at this site. For our investigation at La Selva of this more local scale of internal variation within tropical rain- forest, the combined use of landscape-scale system- atic sampling and GIS proved to be critical. By spreading out the sampling over 100s of hectares of forest, we obtained highly replicated, spatially separated observations of tree floristics within each major soil unit and in each type of topographic sit- uation (positions along the ridgetop-to-swale catena and different slope angles). Had we assessed these factors within a single plot or transect, even a very large one, we could not have generalized our find- ings to the larger landscape due to the possibility of sample bias from particular local edaphic or his- toric conditions. The 50 to 100 m intervals between neighboring sampling points also reduced the like- lihood of spatial autocorrelation among samples (cf. Clark et al., 1996). Finally, the GIS enabled us to relate several classes of complementary site infor- mation to the patterns we discovered in the tree distributions. A particular additional strength brought by GIS to such work is that it provides an ongoing link to the site database, even as it is being expanded and refined through time. This combined research approach thus seems well suited to both assessing and interpreting the landscape-scale spa- tial variation in floristics within a given type of tropical rainforest. A further analytical step that could significantly enhance the data interpretation would be to use generalized linear modeling (G to evaluate the simultaneous (combined) effect of multiple site factors (e.g., soil type, topography, and disturbance history) on the tree species” distribu- 1996). tion patterns (cf. Austin et al., RECOGNIZING HUMAN “FOOTPRINTS” In addition to edaphic variation, past human in- terventions, from silviculture to swidden agriculture to selective cutting, are an important potential source of within-landscape heterogeneity in tropical rainforest. This is true even in stands considered old growth, as we found at La Selva. Indeed, an accumulating body of research findings indicates that most tropical forests are likely to have been affected this way (e.g., Gordon, 1982; Anderson, 1990; Gómez-Pompa & Kaus, 1990; Brown et al., 1991; Bush & Colinvaux, 1994; Garcia-Montiel & Scatena, 1994). Given the likely pervasiveness of such impacts, researchers studying the distribution of biodiversity within these forests should explicitly seek indications of human ac 2. in their study sites (Hamburg & Sanford, La Selva, evidence of such .. impacts s we found at can emerge when landscape-scale floristic patterns are assessed for non-random distributions beyond those attributable to the site's edaphic variation. For this kind of question, the combined use of GIS, edaphic surveys, and systematic replicated vege- tation sampling seems a useful approach. CAN IT BE DONE ELSEWHERE? Is this La Selva experience translatable to other, less developed sites in the tropics? One newly available research tool makes the answer to this increasingly yes. The Global Positioning System (GPS) is a technology for field determination of lo- cations by interpretation of satellite signals. A field researcher in any tropical forest can now use a por- table GPS receiver with extendable antenna, run in parallel with а GPS station at their base site, to measure field locations with good to excellent ac- curacy (recent trials under canopy in a suite of U.S. forests produced accuracies of ca. 2-8 m; Deckert & Bolstad, 1996). This technology will also rapidly become both less expensive and more effective (the U.S. administration recently resolved to remove the current system of signal degradation, probably within 10 years). Although using GPS requires sig- nificant training and equipment, the benefits are immense for field researchers. Now, any tree or vegetation type encountered in the most remote tropical forests can be spatially referenced so as to be relocatable by anyone. Thus "ecological collecting," the inventorying and iden- tification of even non-reproductive plants, becomes feasible anywhere, because sterile plants can be Volume 85, Number 1 1998 Clark 31 Deciphering Landscape Mosaics precisely mapped for repeated visits until found fruiting or flowering. Well documented location data made generally available to the research commu- nity will maximize the current and future value of the very limited systematic and ecological work within these forests. We should begin to think of the entire tropical forest biome as a “permanent study plot." This new tool also makes feasible in any tropical rainforest the type of landscape-scale/GIS research we used at La Selva. With GPS any field researcher can carry out highly replicated, systematic sam- pling over 100s of hectares, without depending on a very expensive physical grid such as La Selva's. When site variables as well as vegetation are as- sessed at all sample points, the ground is laid for a synthetic GIS analysis of the relation between floristic patterns and site conditions at the land- scape scale. None of these elements, however, are easy to attain. Those starting from ground zero in a tropical forest site will find that considerable effort is involved in achieving each component of such an approach: the use of GPS in the field; the de- termination of site factors, particularly soil char- acteristics; plant identification; and the analysis of the resulting data with GIS. Our strongest recom- mendation based on our experiences at La Selva is to build multidisciplinary research teams incorpo- rating expertise in all these fields, rather than hav- ing individual researchers trying to develop all the necessary skills. Such a team brings in the needed levels of prior training and experience in these new areas, and thus ensures the quality of data pro- uced. CONCLUSIONS The biodiversity of the world's tropical rainfor- ests is still largely unstudied and unprotected. These are compelling reasons for increasing current research efforts to understand the geographic and local variation of the biota within these ecosystems. Recent studies in different parts of the tropics have demonstrated a high level of within-forest mosai- cism. This is an important new dimension of com- plexity that must be understood if current conser- vation efforts in this biome are to be effective. As shown by our research experiences in one Central American forest site, the combined use of GIS and highly replicated systematic sampling of site and biotic variables over meso-scale landscapes (100— 10,000 ha) is one promising strategy for deepening current understanding of the great spatial variation within the world's tropical rainforests. Literature Cited Alvarado I., С. E. 1990. Características geológicas de la Estación En La Selva, Costa Rica. Tecnol. en Marcha 10: Anderson, A. ч 199 . Extraction and forest management by rural 2 in the Amazon estuary. Рр. in А. B. Anderson (editor), Alternatives to Deforesta- tion. Columbia Univ. Press, New York. . Plant- pollinator interactions in Ma- laysian rain forests. Pp. 85-101 in K. S. Bawa & M. Hadley (editors), Reproductive Ecology of Tropical For- terocarp Forest Oxford Forest. Mem. 25, Oxford Univ. Press, Oxford. 1969. Speciation among t Some deductions in the lig J. Linn. Soc. 1: 155-196. Austin, M. P., P. S. Ashton & P. Greig-Smith. 1972. The application of quantitative methods to vegetation survey A re-examination of rain forest data from Brunei. J. Ecol. 60: 305— ———, J. 5. Puy & A. O. Nicholls. 1996. Patterns of tree species richness in relation to environment in 2. New South Wales, Australia. Austral. J. Ecol. 21: Baillie, I. C., E 9; Ashton M. N. Court, J. A. R. Anderson, A. Fitzpatrick & J. Tinsley. 1987. Site characteris- tics and the distribution of tree species in Mixed Dip- terocarp Forest on Tertiary sediments in central Sara- 20. ropical forest trees: ht of recent evidence. Biol. . J. R. Gillespie & A. E. Lugo. 1991. Bio- mass s of tropical vers of south and southeast Asia. 1. J. Forest Res. 21: 111-117. Bush. м. В. & P. A wt invaux. 1994. Tropical forest disturbance: Balececaliuical кемікті from Darien, Рап- ama. var 75: 1761-1768. Clark, D. A. & D. B. Clark. 1992. Life history diversity 4. апа emergent trees in a neotropical rainforest. Ecol. Monogr. 62: 315-344. & 994. Climate-induced annual vari- ation in canopy [ow growth in a Costa Rican tropical rain forest. J. Ecol. 82: 865-872. . R. Sandoval M. & M. V. Castro C. 1995. Edaphie and un effects on landscape-scale distri- butions of tropical rain forest palms. Ecology 76: 2581— 2594. Clark, D. B. 1990. La Selva Biological Station: A blue- print for stimulating tropical research. Pp. 9-27 in A. Gentry (editor), Four Neotropical Rainforests. Yale Univ. Pr a New Haven, Connecticut. . A. Clark. 1989. The role of physical dam- age in qe seedling mortality regime of a neotropical rain forest. Oikos 55: 225-230. J. M. Read. 1998. Edaphic variation and the mesoscale distribution of tree species in tropical rain = e Ecol. 86: 101-112. M. Rich. 1993. Comparative anal- ysis of chat анов by saplings of nine tree species in neotropical rain forest. Biotropica 25: 397— 407. ‚ S. Weiss « S. Е Oberbauer. 1996. Landsc ed -scale rion of understory light and forest structure in a neotropical lowland rain forest. Canad. J. Forest Res. 26: 747—757 32 Annals of the Missouri Botanical Garden Condit, R. 1995. Research in large, long-term tropical orest plots. Trends Ecol. Evol. 10: 18-22. Long-term monitoring of bi- ological di epi in iropical forest areas. UNESCO Paris. Deckert, C. € P. V. Bolstad. 1996. Forest canopy, terrain, and distance effects on global positioning system point accuracy. Photogramm. Engin. Remote Sensing 62: 317-3: — J. F. 1996. Patterns of tree species rich- in rain forests of the middle 2. area, Colom- bia Amazonia. Biotropica 28: 142-158. Foner R. B. 1990. Long-term change in the successional forest community of the Rio Manu floodplain. Pp. 565- 572 in А. Н. Gentry (editor), Four Neotropical Rainfor- ests. Yale Univ. Press, New Haven, Connecticu & S. P. Hubbell. 1990. The floristic ЕТТИМ of the Barro Colorado Island forest. Pp. 85-98 in Four Neotropical Rainforests. Yale "licut. Garcia-Montiel a. 1994. The effect of human activities on the structure and composition of a оң forest in Puerto Rico. Forest Ecol. Managem. 63: 57—78. Gentry, A. H. 1988. Tree species richness of upper Am- azonian forests. Proc. Natl. Acad. Sci. U.S.A. 85: 156– 159. (editor). 1990. deg dani d Rainforests. Yale Univ. Press, о Haven, Conr 1 Tropical phos 1. ода. naltéma ке their conservational significance. Oikos 63: 19—28. . 1993. A Field Guide to the Families and Genera of Woody Plants of Northwest South America (Colombia, Ecuador, Peru). Conservation International, Washing- ton, D.C. 994. Importance of the Explorer's Inn Reserve ia plots. Pp. 55-57 in R. B. Foster, T. A. Parker Ш, A. Н. Gentry et al. (editors), The Tambopata-Can- 4 1-4 Zone of Southeastern Peru: А Biological Assessment. Conservation International, Washington, D. С. — & R. Ortiz S. 1993. Patrones de composición florística en la Amazonia Peruana. Pp. 155-166 in R. Kalliola, M. Puhakka & W. Danjoy (editors), Amazonia Peruana. PAUT and ONERN, Jyvaskyla, Finlan Gómez-Pompa, A. & A. Kaus. 1990. Traditional manage- ment of tropical forests in Mexico. Pp. 45-64 i Anderson (editor), Alternatives to 2. Calum bia Univ. diem New York. 982. A Panama Forest and Shore. The Boxwood B. Pacific Grove, California. Hamburg, 5 ¿ R. L. Sanford, Jr. 1986. _ | Homo sapiens, and ecology. Bull. Ecol. Soc. Amer. 169-171. Hartshorn, С. S. 1983. Plants. Pp. 118-157 in D. H. Janzen (editor), Costa Rican Natural History. Univ. Chi- cago Press, 2. 0. & B. E. Hammel. 1994. Vegetation types and 2. 2. Pp. 73-89 in L. A. McDade, К. S. . H. A. Hespenheide & С. S. Hartshorn (editors), E Selva: Ecology and Natural History of a Neotropical Rain Forest. Univ. ш Press, Chicago. Henderson, A., G. Galeano & R. Bernal. 1995. Field Guide to the Palms of the a Princeton Univ. Press, Princeton, New Jers Horn, S. P. & R. L. Sanford, Jr. 1992. Holocene fires in Costa Rica. Biotropica 24: 354-361. Kahn, F. 1987. The distribution of palms as a function of local topography in Amazonian terra-firme forests. „Ј. Рр. А J. A. Bognar. 1995. 2. variability of digital soil maps and its impact on onal ecosystem modeling. Ecol. Modelling. 82: 1- T D. J., T. C. Moermond & J. S. Denslow. 1994. Frugivory: An overview. Pp. 282-294 in L. A. McDade, K. S. Bawa, H. A. Hespenheide & G. S. Hartshorn (ed- itors), La Selva: Ecology and Natural History We Neo- tropical Rain Forest. Univ. Chicago Press, Chic Lieberman, D., M. Lieberman, G. S. Hartshorn & R Per alta. 1985b. Growth rates and age-size relationships of tropical wet forest trees in Costa Rica. J. Trop. Ecol. 1(2): 2. Lieberman, М., ieberman, С. S. Hartshorn & R. Per- alta. 1985a. 4. altitudinal variation іп low- land wet tropical forest vegetation. J. Ecol. 73: 505- 516. McDade, L. A., K. S. Bawa, H. A. Hespenheide & G. S Hartshorn 1. 1994. La Selva: Ecology апа Nat- ural History of a Neotropical Rain Forest. Univ. Chicago Press, Chicago. Nelson, B. W., V. Kapos, J. B. Adams, W. J. Oliveira, O. P. G. Braun & I. L. Doamaral. 1994. Forest distur- bance by large blowdowns in the Brazilian Amazon. Ecology 75: 853—858 Newstrom, L. OW. Frankie, H. G. Baker & R. K. Colwell. 1994. Diversity of 2 -term flowering pat- terns. Pp. 142-160 in L. A. McDade, K. S. Bawa, H A. Hespenheide & G. S. Hartshorn (editors), La Selva: Ecology and Natural History of a ене Rain For- est. Univ. Chicago Press, Chica, Oberthur, T., A. Dobermann & H. U. Neue. 1996. How good is a reconnaissance soil map for agronomic pur- poses? Soil Use & Managem. 12: 33-43. Richter, D. D. & L. I. Babbar. 1991. Soil T E in the tropics. Advances Ecol. Res. 21: 315- Ruokolainen, K. & H. Tuomisto. 1993. Е vegetación de terrenos no inundables ш firme) en la чү baja de la Amazonia Peruana. Pp. 139-153 in R. Kalliola, M. Puhakka & W. Danio (lia: Amazonia Peruana. PAUT and ONERN, Jyvaskyla, Finland. Salo, J. & M. Rasanen. 1989. Hierarchy of landscape patterns in western Amazon. Pp. 35-45 іп L. B. Holm- Nielsen, I. C. Nielsen & H. Balslev (editors), Tropical Forests: Botanical Dynamics, Speciation and Diversity. Ác rae Press, New Yor Sancho F. & R. Mata Ch. 1987. su 4. Estación Biológica “La Selva bx SEMEN San José, Costa R Sanfor ‚ Jr., P. Раађу, J. С. Luvall & E. Phillips. 1994. ue geomorphology, and 22. systems. Pp. 19-33 in L. McDade, K. S. Bawa, H. A. Hes- ас С. S. Hartshorn (editors), La a Ecology and Natural History of a Neotropical Rain Forest. Univ. Chicago Press, Chicago. Simpson, B. B. & M. Connor Ogorzaly. 1986. Economic Botany: Plants in Our World. McGraw-Hill, New York. Sollins, P., F. Sancho M., R. Mata Ch. & R. L. Sanford, Jr. 1994. n and soil E ess research. Pp. 34—53 in L. A. CS . H. Hespenheide & G. S. hom lis: [a Selva: Ecology and Natural His- Pic Estudio detallado de 2 Не (ог Volume 85, Number 1 1998 Clark 33 Deciphering Landscape Mosaics tory of a Neotropical Rain Forest. Univ. Chicago Press, Chicago. Terborgh, Ј., R. B. Foster & P. Nunez V. 1996. Tropical tree communities: А test of the но hypoth- esis. 2. 561-507. Timm . The mammal ju Pp. 229-237 in L. A. McDa Pi Е S. Вама, Н. A. Hespenheide & С. S. Hartshorn (editors), La Selva: Ecology and Natural History of a e жін Rain Forest. Univ. Chicago Press, Chica Tuomisto, H., K. Ruokolainen, R. Kalliola, A. Linna, W. Danjoy & Z. Rodriguez. biodiversity. Science 269: Valencia, R., H. Balslev & G. Paz y Mino C. 1994. tree alpha-diversity in Amazonian Ecuador. Biodiversity and Conservation 3: 21—28. Vandermeer, J. H 1995. те Amazonian 63—66 1977. Notes on density dependence in Palmae), a lowland Si: 5 % $ ps P 3 E % ж ics) = z 2 logical Station: Coda Rica—Introduction. Selbyana 9: 91. LARGE-AREA MAPPING OF J. Michael Scott? and Michael D. BIODIVERSITY! Jennings ABSTRACT The age of discovery, description, and classification of biodiversity is entering a new phase. In responding to the conservation imperative, we can now supplement the essential work of systematics with spatially iis icit information s s lages of species. This is possible because of recent conceptual, technical, and organizational ›торгезз in generating synoptic views of the earth's surface and a great deal of its biological content, at multiple scales of thematic as well as geographic resolution. The de velopment of extensive spatial data on species distributions and vegetation types provides us with a framework for: (a) assessing what we know and where we know i ; and (b) stratifying the biological universe so that highe >r-resolution surveys can be more efficiently implemented, cov- ering, for example, geographic adequacy of 2. collections, pun abundance, reproductive success, and genetic dynamics. The land areas involved a ry large, and the questions, such as resolution, scale, classification, and accuracy, are complex. In this paper, we pro 5 . from he United States Gap Analysis Program on the advantages and limitations of mapping the occurrence of terrestrial vertebrate species and dominant land-cover types over large areas as joint ventures and in multi-organizational partnerships, and how these cooperative efforts can be designed to implement results from data development and analyses as on-the-ground actions. Clearly, new frameworks for thinking about mee information as well as organizational cooperation are needed if we are to have any hope of nel ‘umenting the full range of species occurrences and ecological processes in ways meaningful to their management. The Gap Ads experience provides one model for achieving these new frameworks. = b ~ D £ E a ы 7 РА Ф 3 = = 3 = > D т ~ о 1 т = A с т т жи Systematics is the science of describing the fun- ical diversity we need to know: what species there damental units that make up the diversity of life, аге (systematics), how they function (behavioral and classifying organisms in a way that indicates their ecosystem science), how they are distributed in natural relationships. The age of discovery, descrip- space (biogeography), time (population ecology), tion, and classification of biological diversity is far and how they are presently managed (wildlife and from over. New species of chordates, the most thor- — conservation biology). One distinct problem is that oughly described phylum, are still being discov- the properties of biological diversity change as the ered. However, we are entering a new phase of objects (individuals, populations, species, assem- characterizing biological diversity. This new phase Ыареѕ of species) are aggregated or disaggregated is distinguished on the one hand by: (a) progress (Allen & Starr, 1982). in applying concepts relating spatial scale to the А complete Оа inventory of a large area hierarchy of biotic organization and more cooper- may involve, for example, describing the genetic ative relationships among institutions that conduct structure of a species, its behavior, population research, planning, and management of biological sizes, and other metrics such as reproductive suc- resources; and (b) new and powerful technologies cess, mortality, and mutation rates. It must describe for inventorying and monitoring biological diversity. the species’ ecological positions in multiple dimen- On the other hand there are setbacks due to finan- sions (e.g., trophic, community affiliations, habitat, cial limitations and a lack of societal support for — etc.) as well as the processes that maintain the eco- the management practices that it will actually take systems in which a species occurs. The undertaking to maintain the natural diversity of life on earth. must include studies of the biogeography of the Clearly, the level of effort being invested in com- species and the biogeography of its habitats. Fi- pleting the description of most species and subspe- nally, it must include an assessment of the current cies is orders of magnitude less than the level of conservation status of the species and its habitats. human enterprise that results in the collateral dam- Тһе challenge is no less daunting than launching age of extinction and extirpation (Hawken, 1993). а 19th-century expedition to describe the flora and In order to make progress in managing for biolog- fauna of the Amazon Basin. ! We thank the many partners of GAP for their efforts to | ак biologically defensible databases that can be used == -% в. т Ф (2 © = а Y ы < z © = Ur eolo E 7 Survey: Ийе Cooperative Fish and Wildlife Research Unit, Uni- versity of Idaho, Moscow, Idaho 83844- TET * Biological D Division U.S. Ge kids | Gap Analysis Program, 530 S. Asbury St., Suite 1, Moscow, Idaho 83843, ANN. MISSOURI Bor. GARD. 85: 34-47. 1998. Volume 85, Number 1 Scott & Jennings 35 Large-Area Mapping So far, we have been able to document habitat- specific distributions and have obtained some sense of reproductive success over space and time for only a very few species—those that are important recreationally or commercially or those that are rare and popular such as the California condor (Gymnogyps californianus) or whooping crane (Grus canadensis). Even for a group as intensively studied as the birds of North America, there are hundreds of spec es reported in fewer than ten studies in the rimary ornithological literature (J. Ratti & J. M. aem unpublished ms.). study the earth as a biosphere, and the tools we are using, such as remote sensing and geographic in- formation systems, are still developing. The chal- lenge is to think hierarchically (Wiens, 1989) and to link the tools of geographers with those of clas- e have just begun to sical taxonomists and naturalists by building two- way bridges among the disciplines. Only by in- creased interdisciplinary cooperation are we to have some hope of describing and understanding the complexity of nature's diversity and how to bet- ter manage our natural heritage for future genera- tions. We describe a method and its implementation that complements the work of systematics by fo- cusing on two other specific parts of the biodiversity issue: biogeography and land management. The method we describe is now being carried out in the United States as the Gap Analysis Program under the Biological Resources Division of the U.S. Geo- logical Survey (Scott et al., 1993, 1996). We present some background, methods, and results to date. Then we discuss opportunities for improving bio- diversity information through better integration of systematics, ecosystem science, and biogeography. BACKGROUND BIODIVERSITY AND SPATIAL SCALE Biodiversity is *. . . the variability among living organisms from all sources, including, inter alia, terrestrial, marine, and other aquatic ecosystems and the ecological complexes of which they are a part; this includes diversity within species, between species and of ecosystems" (1993 Convention on Biological Diversity, Article 2, as cited in Heywood, 995: 8). By this view, biodiversity is complex and deals with composition, structure, and process of its component parts (Noss & Cooperrider, 1994). Its characterization necessitates a synthetic hierarchi- cal construct. Additionally, when dealing with the spatial or geographic aspects of biological diversity, clear labels and definitions for units that relate bi- ological diversity to geographic extent are neces- a un ry. The four basic principles that underpin the con- cept of hierarchy for ecology are: (1) that systems are defined by measures of their structural com- ponents and by the rates of their processes; (2) sys- tems are ordered according to both their occur- rences in space and the frequencies or rates of their processes over time; (3) larger/slower systems con- strain the occurrences and behaviors of smaller/ faster systems, providing the context within which the smaller/faster systems operate; and (4) the mechanisms or properties by which a system op- erates may not be determined only by a simple ag- gregation of its smaller/faster components, nor by a reduction of its larger/slower components (O'Neill et al., 1986) hen mapping elements of biodiversity over large areas, the relationships among and between the pattern of dominant land-cover types, species diversity, and spatial scale are critical. Measures of species diversity must be expressed relative to bio- geographic units of a determined spatial scale if they are to be meaningful (Levin, 1981). However, confusion about the differences between types of diversity (“thematic resolution”) and cartographic scale is ere (e.g., Short & Hestbeck, о Davis, 1995; Edwards, 1995; Scott et al., 1995). We ЗФ UM seven categories as a Do for describing species diversity in relation to eco- logical patterns and spatial scale (Table 1; Whit- taker, 1960, 1977) The linkage between types of diversity and spa- tial scale makes this framework especially useful. Figure 1 (Stoms & Estes, 1993) shows how four of these categories ("inventory diversities") are used to describe species diversity within sampling units of four approximate sequential sizes and corre- sponding with four hierarchical levels of biotic or- ganization: a single ground sampling point (point diversity), a natural community (alpha diversity), a landscape (gamma diversity), and a large geograph- ic region (epsilon diversity). Three other terms (“dif- ferentiation diversities") are used when comparing e in species composition be- tween individual sampling points (pattern diversity), natural communities (beta diversity), and land- scapes (delta diversity) (Whittaker, 1977). The critical point here is that the magnitude of alterations to land and water characteristics, for- merly limited in spatial extent and pattern so as to be manifest at the levels of populations and spe- the amount of chang cies, is now so extensive that changes are manifest at the levels of natural communities, landscape eco- systems, and global ecosystems (Heywood, 1995; 36 Annals of the Missouri Botanical Garden Table 1. Spatial categories of species diversity (Whittaker, 1977; Stoms & Estes, 1993). Inventory diversities Differentiation diversities 1. Point diversity: А small, or microhabitat, sample of spe- cies diversity from within an alpha unit. Generally 10 to 100 square meters. 3. Alpha diversity: А single within-habitat measure of spe- cies diversity regardless of internal pattern. Generally 0.1 to 1000 hectares. л . Gamma diversity: Тће species diversity of a landscape made up of more than one kind of natural community. Generally, 1000 to 1,000,000 hectares. 7. E ipsilon diversity: The species diversity of a broad region Іі | Е 2. Pattern The points within a community. diversity: change in diversity. between 4. Beta diversity: The change in diversity among different communities of a landscape: an index of between-hab- itat. diversity. 6. Delta diversity: The change in diversity between land- scapes along major climatic or physiographic gradients. of differing landscapes. Generally 1,000,000 to ido don uo hectares. Vitousek et al., 1996; Vitousek et al., 1997). Con- servation efforts implemented at the population and species level alone may no longer be effective when system-wide changes are being forced at the land- scape and global levels of ecosystem functioning. Furthermore, the properties by which a system in- teracts with the agents of change may not be readily identified by an aggregation of a system's smaller components or by a reduction of its larger compo- nents. Information derived from synoptic observa- tions of both the level of biotic organization and the geographic extent at which the changes are being induced is needed (Jennings & Scott, 1993). We believe that providing a rangewide elemental basis for assessing biodiversity conservation using maps of vegetation types and vertebrate species distributions creates sampling frameworks from which unbiased samples for more detailed studies of species occurrences, density, and viability may be made. For the first time, we can be spatially explicit about a suite of species that co-occur in a repeating pattern across the landscape, for exam- ple, those characterized by the dominance of pon- derosa pine (Pinus ponderosa Douglas ex Lawson & C. Lawson). We can understand the extent of its occurrence as context, examine its landscape po- sition, and make inferences about composition, structure, and function that are rangewide. The re- sult is a significant advancement over being limited to conclusions about the ponderosa pine vegetation alliance only from stand-level examinations. In a similar fashion, we may use the distribution map, for example, of a wolverine (Gulo gulo) to ask questions about the representativeness of extan collection records, or view it as a testable hypoth- esis and conduct wolverine surveys to document not only presence/absence but also abundance and re- productive success. We may also use these maps to make more detailed descriptions of its habitat from an unbiased sample of its entire range, all in such a manner that inferences may be made about the wolverine or its habitat (or in the earlier case, pon- derosa pine) rather than simply the study site we chose to sample. PURPOSE The purpose of gap analysis is twofold. The first is to provide regional conservation assessments of native vertebrate species and natural land-cover types. The second is to facilitate the application of this information to land-management activities. These goals are accomplished by (a) mapping the vegetation alliances (FGDC, 1996; Grossman et al., 1994) of the United States; (b) mapping predicted distributions of each native vertebrate species; (c) mapping the existing conservation lands and rank- ing them by their management status; (d) determin- ing the degree of representation that vertebrate spe- cles and land-cover types have in conservation lands; (e) providing this information to the public and those entities charged with land-use research, olicy, planning, and management; and (f) building institutional cooperation in the application of this Volume 85, Number 1 1998 Scott & Jennings 37 Large-Area Mapping information to state and regional management ac- tivities. This. then, provides an objective database of biogeographic information that allows research- ers, planners, and managers to stratify the land sur- face for work at higher resolutions (Scott et al., 1993, 1996), and to understand the regional and continental context of higher-resolution information from smaller areas (Jennings, 199 DEVELOPMENT The term “рар analysis" refers to the process by which species and natural communities not ade- quately represented in conservation lands are iden- tified. These are the “gaps” in our present-day ef- forts to maintain biological diversity, and it is these that are most likely to become endangered with ex- tinction in the future. By understanding what these gaps are and where they are, future conservation crises and conflicts may be avoided. The development of the Gap Analysis Program (GAP) began in 1987 in response to the need to complement species-by-species management of en- dangered species in dealing with broad-spectrum habitat loss (Scott et al., 1987, 1993, 1996). There was a need for synoptic and spatially explicit in- formation on the distribution of each native verte- brate species and natural community, and their management status. At the time, there were no readily available, consistent data that could provide for an understanding of either the context of a sin- gle land management decision or the occurrence of a species’ habitat in the ecological contexts of land- scapes or bioregions. There are many other uses for these data. Most states do not have current maps of land cover, and GAP is the first state- and national-level effort to produce this information at resolutions usable by land managers, planners, scientists, and policy makers (Scott et al., 1987, 1993, 1996). Map showing the distributions of land cover, habitat Ф type, vertebrate species, land management, ог com- binations thereof can be generated regionally or na- tionally. Such information may be used to identify areas that are suitable for development and where other land-use conflicts may be avoided, as well as those areas important for meeting conservation needs. In the years since 1987, significant barriers to mapping elements of biological diversity across large areas have been overcome (Scott et al., 199 A wide range of tools for mapping natural land- cover and habitat types and predicting vertebrate species distributions has emerged, and procedures ave been refined, tested, and further refined. There is still room for improvements; additional de- velopment and testing of some methods at varying spatial and thematic scales (for example, accuracy assessment) and land-cover mapping is still need- ed. COOPERATION The U.S. Geological Surveys GAP is conducted as state-level projects, and currently there are 46 active or completed projects. Although coordinated and primarily funded by the U.S. Geological Sur- vey's Biological Resources Division (formerly the National Biological Service), GAP is made up of over 450 cooperating organizations, including uni- versities, businesses, and state and federal agen- cles. Of equal importance to the technical progress is the way natural resources institutions (private and public) are coalescing around the concept of a stan- dard large-area information base (one. way this may be seen is through the “bottom-up” organization (and funding) nýa characterizes the program). GAP, the largest effort ever mounted to map selected (i.e., vertebrate species and vegetation types) biological resources of the United States, is being carried out cooperatively by state-level projects. The importance of having data sets that are com- parable across state boundaries is in revealing ac- tual patterns of species and vegetation community distribution at scales relevant to both the magni- tude of present-day changes and the multiple levels of biological organization. Such information may be used to identify areas that are suitable for devel- opment and where land-use conflicts may be avoid- ed, as well as those areas important for meeting conservation needs. New frameworks are emerging for both in the new type of information being de- veloped and in the convergent way it is being de- veloped. There is now convergence on mutually recog- nized and systematic definitions for natural com- munities as intrinsic entities and as habitat types, for example, as indicated by the land-cover clas- sification system being proposed for adoption by the Federal Geographic Data Committee (FGDC) and by formation of the Ecological Society of America’s Vegetation Classification Panel. There has been substantial recent progress on methods for mapping alliances of natural communities, as represented by dominant natural vegetation or non-vegetated land- cover types, though it appears that no single meth- od will suffice for all environments (Caicco et al., 1995; Stoms, 1994). common ground on methods for predicting the dis- ere is increasingly more 38 Annals of the Missouri Botanical Garden INVENTORY DIFFERENTIATION DIVERSITIES DIVERSITIES EPSILON | REGIONAL SAMPLING UNITS: = 1 - 100 MILLION HA DELTA | г лашын GRADIENTS SAMPLING UNIT ALPHA IN Ес MMUNITY TYPE DOMAIN: DES UR. TO REGION GAMMA | Mm SAMPLING UNIT * 1,000 TO I Mo UON HA BETA | ENVIRONMENTAL GRADIENTS SAMPLING UNITS: ALPHA IN DIFFERENT COMMUNITIES DOMAIN: COMMUNITY TO LANDSCAPE соу ALPHA! кашын S Кс SAMPLING UN is 2. CLLLLLIIT] PATTERN HERD GRADIENTS | SAMPLING POINTS NS SIUE COMMUNITY DOMAIN: POINT TO COMMUNITY наю аа а 8 tt ttt tt) POINT | MICROHABITAT SAMPLING UNITS: = 0.01 TO 0.1 HA ТІТІТІПГ Figure 1. Diagram showing seven spatial levels of species diversity defined by Whittaker (1977). The lefthand sat emen levels of diversity within a AD sequential set of sample units, or "Inventory Diversities." The righthand column represents categories of spe change in composition between or among sample units of the same spatial level. (From Stoms & Estes, 1993, 1. with permission.) tribution of native vertebrate species (Butterfield et nd, much ex- perience has been gained in the mapping of areas that are managed for biodiversity (Beardsley & Stoms, 1993). Although many issues remain, such as accuracy assessment and appropriate scale and 94; Edwards et al., 1995) resolution, much attention is being brought to bear on them, and the trends are quite positive. Frameworks are now in place in GAP, as well as in other large-scale biological assessments, for gen- erating, archiving, distributing, querying, and ex- perimenting with biological data that cover large areas, and there is a great deal of interest in im- proving the science of these efforts. What might be of greater significance is that consensus on these issues is taking place among state-level institutions as well as among the state and national interests who have responsibility for research and manage- ment of natural resources. The concept underlying this dynamic is that it is Volume 85, Number 1 1998 Scott & Jennings 39 Large-Area Mapping far more important now, while land use decisions concerning millions of hectares are being made dai- ly, to begin with an accounting of the conservation status for the mappable elements of biological di- versity than to put off any real action until perfect methods have been conceptualized, and all ele- ments of biodiversity have been identified and mapped, tested, published, replicated, adopted, dif- fused, and applied. There is simply not the time, money, nor political will to take that path. Today we have the capabilities to build powerful sets of information, imperfect though they may be, that correspond to the multiple levels of biotic organi- zation. And we have the ability to foster the appli- cation of that information, by all concerned, to solve the seemingly inexorable problems of maintaining our biological heritage. It requires that profession- als and their institutions put aside their past dis- ciplinary and institutional differences, assume some risk, and commit to work together with what- ever resources they have. This can result in a lev- eraging of funds and minimizing of duplicate ef- METHODS GAP requires computer-based (digital) maps of: (a) existing natural or semi-natural land cover to the level of community alliances (vegetation types character- ized according to their dominant or co-dominant plant species or, in the absence of a dominant vegetation species, dominant land-cover feature (Grossman et al., 1994)); (b) predicted present-day distributions of na- tive vertebrate species; and (c) public land ownership and private conservation lands. These data layers are analyzed to compare distributions of each native ver- tebrate species, group of species, and community al- liance with the existing network of conservation lands. Results show where the conservation “gaps” are in both land management and in the body of knowledge about species and natural communities. Ап overview of the methods for developing each of these three data sets is presented below (see also Scott et al., 1993; Jennings et al., 1996; Gap Analysis Program World Wide Web home page http://www.gap.uidaho.edu/ gap). LAND COVER Generally, the mapping of land cover is done by delineating areas of relative homogeneity (basic “objects”), then labeling these areas using categories defined by a land-cover classifi- cation system. More detailed attributes of the in- dividual areas are added as more information be- comes available, and a process of validating both cartographic polygon patterns and labels is applied for editing and revising the map. This is done in an iterative fashion, with the results from one step causing re- evaluation of results from another step. For exam- ple. the discovery of attributes for a given mapped polygon may result in adjustment of its boundary. Finally, an assessment of the overall accuracy of the data is conducted. Where the database is ap- propriately maintained, the final assessment of ac- curacy will show where cs ies құй should be made in the next update (Davis et al., Some of the problems with уйы 0 mapping of large areas at the desired spatial and thematic res- olutions (i.e., 1:100,000-scale and community al- liance theme) that have been overcome are: (a) classification of land cover, (b) data 21 (с) delineation of land-cover pattern, (d) ob nter- pretation (Orians, 1993), and (e) pos li э fina. map accuracy. In order to provide meaningful comparisons across large areas, a consistent land- cover classification system is needed. Land-cover classifications must rely on specified attributes such as the structural features of plants, their flo- ristic composition, differentiate categories evenly (Küchler & Zonne- veld, 1988). Although there has been much effort devoted to the classification of vegetation, there has been no previous attempt to apply a detailed clas- sification of natural land cover across the contigu- ous 48 United States at a 1:100,000 scale, al- though Crumpacker et al. (1988), assessed the occurrence of 135 potential vegetation types on fed- eral and Indian lands. In mapping land cover, GAP uses the National Vegetation Classification (FGDC, 1996, 1997; Grossman et al., ; Bourgeron & Engelking et al., 1994; Sneddon et al., 1994; Weak- ley et al., 1996; Loucks, 1995, 1996) The minimum thematic object that Gap Analysis is mapping is the community alliance (Grossman et al., 1994; see Appendix 1 for a sample description of a community alliance), although in practice for some areas, mosaics of undifferentiated alliances (e.g., “oak woodlands" rather than “Quercus gar- ryana alliance”) represent the limit of current ca- pabilities to map land cover across ecoregions and biomes. The alliance corresponds most closely with the units of alpha diversity (a sample representing a community regarded as homogeneous despite its internal pattern) in order to conduct analyses at the beta, gamma, delta, and epsilon levels. A spatial depiction of beta diversity (between-habitat diver- sity) represents the pattern of landscape, or gamma, heterogeneity. For Gap Analysis, the central con- cept is that the structural and floristic characteris- tics of dominant vegetation or (in the absence of or environmental conditions to 40 Annals of the Missouri Botanical Garden vegetation) dominant land features, can be used systematically to delineate and map patterns of beta and gamma diversity. Models of these patterns are important for generating and evaluating landscape- level conservation options. For the delineation of land-cover patterns, the Landsat Thematic Mapper (TM) satellite images serve both as a base map and as a source of spec- tral information for discriminating among land-cov- er types. Although methods for preprocessing the basic TM product used in mapping land cover were variable at the earlier stages, currently state pro- jects use a standard TM product that is geograph- ically registered to within 30 m, corrected for ter- rain distortion and systems errors, and spectrally classified into 240 classes using bands 1, 2, 3, 4, 5, and 7 (see Bara, 1994). No single procedure is appropriate for the delin- eation of land-cover patterns in all environments of the United States (Davis et al., 1995), and a variety of methods are used to delineate land-cover pat- terns by the САР state project analysts I et al., 1991; Davis & Stoms, 1996; Davis et al., А Edwards et al., 1995; Lillisand, 1996; Scott et ә 1993; Slaymaker et al., eated, the resulting objects are interpreted and la- . Аз pattern is delin- beled in an iterative fashion. To recognize vegeta- tion alliances, training images of each type are identified on the ground. Air photos or air videos are being used to train analysts. Additional data sets, such as digital elevation models, temperature and precipitation patterns, and soils maps, are also used. A single, precisely standardized method for pattern delineation is not possible because: (a) veg- etation characteristics differ substantially among biogeographic regions, requiring different ap- proaches, especially for interpretation of remotely data; for example, the use of TM imagery from different seasons may be used singularly in a sensed false color composite format and interpreted visu- ally, or their spectral values may be transformed in а specific way and merged together to reveal pat- terns based on phenotypic distinction (the possible variations are almost endless); (b) the expertise for vegetation typing and mapping is itself also region- al in nature, resulting in different approaches by the state project scientists; (c) many different sources of information are used to render the maps (for example, variability in the date of imaging among TM scenes within a state and wide variation in the availability of information about the occur- rence of dominant cover types from state to state), introducing variability into the product; (d) the cur- rent mapping work is a first generation effort, with significant improvements to the technology being made by the state GAP projects; there is a need to try different methods because an effort of this mag- nitude, extent, and degree of resolution has not been undertaken before; ( e) of necessity, GAP is a collaborative *bottom-up" effort focused on prag- matic, near-term conservation, and at present there is neither the institutional support nor the time to research and develop a single method, achieve con- sensus on such a method, then implement a large "top-down" program. Each map class of the state-level spatial data sets Is tested for accuracy, using independent field data, with the confidence interval carried through further transformations with that data set's meta- data. A detailed review of data quality is undertak- en when edge-matching data from adjacent states. Since the present effort is a first generation one, improved methods are expected to dampen the am- plitude of inter-state variation in later generations as well as increase thematic resolution and accu- racy. А number of land-cover data sets from states that used different methods have been edge- matched with good results (M. Murray, Idaho Co- operative Fish and Wildlife Research Unit, C. Homer, Utah Cooperative Fish and Wildlife Re- search Unit, and R. Redmond, Montana Gap Anal- ysis Project, Missoula, pers. comm.). VERTEBRATE SPECIES DISTRIBUTIONS The objectives for mapping the distributions of vertebrate species are to provide maps of known confidence in order to support analysis of conser- vation status to develop a database of locational records, geographic range, wildlife habitat associ- ations, and predicted distribution of each vertebrate species for the long-term utility for GAP and its cooperators. Most existing information on species distribution has typically been collected at the scale of individ- ual field sites and extrapolated to small-scale range maps for state, regional, or national references and field guides. Lacking for most biogeographic infor- mation on species 18 a meso-scale expression (e.g., 1:100,000) of a detailed distribution map, as com- pared with a general range map depicting broad regional or continental limits. The basic assumption of GAP's predicted species distribution maps is that a species has a high prob- ability of occurring in appropriate habitat types that are within its predicted range. GAP links species’ general ranges to large-area land-cover maps and other physical data, which are intermediate in scale between a known specimen collection site and a field guide range map (see Edwards et al., 1996; Volume 85, Number 1 1998 Scott & Jennings 41 Large-Area Mapping Scott et al., 1993). This approach is derived from the assumption that, for large areas such as states or nations, it is impractical to map the distribution of species at a nominal scale of 1:100,000 only from intensive field surveys. GAP therefore makes use of existing information on range limits and re- fines it to develop spatial statements of the pres- ence and absence of a species in map polygons that represent appropriate habitat as understood from current knowledge of the species and the ability to map its habitat (Scott et al., 1993; Butterfield et al., 1994; Edwards et al., 1995). Predicting species distributions by relating them to environmental features that can be mapped from remotely sensed data is an efficient approach to estimating the distribution and management status of elements of biodiversity. However, no matter what their scale, all range and distribution maps are predictions about the presence of a species in a particular geographic area. The accuracy of those predictions generally improves as the size of the area, length of the sampling period, and intensity of sampling are expanded because greater temporal scale as well as heterogeneity of large areas make it more likely that a species will be found to occur there. GAP maps of predicted distributions are cur- rently intended for use and validation at the land- scape, or gamma, level of diversity (an area made up of more than one kind of natural community, generally, 1000 to 1,000,000 ha; Whittaker, 1977), but new efforts are able to attribute species to “patches” as small as 2 hectares. For some species, such resolution may be desirable to allow more pre- cise estimation of habitat area, while for other spe- cies, such small patches may be biologically mean- ingless. For the majority of species, the ability to map at this resolution probably exceeds our knowl- edge of their ecology. We mapped predicted vertebrate species” occur- rences by first obtaining specimen collections and verified sighting records for specific known Јоса- tions for each species and entering this information into a database. These records are considered as either current (within the past 10 years) or histor- ical (> 10 years old). Second, the general range extent for each species is established from the best available information—frequently field guides. Third, an exhaustive literature search is done to establish the known habitat relationships (vegeta- tion, elevation, lakes, etc.) for each species. Fourth, a habitat relationship model for each species is constructed for use in a geographic information sys- tem (GIS). Fifth, the range units and habitat asso- ciations are integrated into a predicted species-dis- tribution map, with areas attributed by known versus predicted occurrences. Sixth, an expert re- view of the draft maps is conducted, the maps are edited, and all changes are documented (Csuti & Crist, in prep.). The resulting maps are testable hy- potheses, predictions we hope will be improved with better information over time (Fig. 2). This type of database bootstrapping is critical if we are to overcome both sparse data and funding constraints. At the landscape level of resolution, GAP predic- tions of accuracy have ranged from 70% to over 90% for birds, mammals, amphibians, and reptiles (Edwards et al., 1996; Scott et al., 1993; C. Peter- son, Idaho State University, pers. comm.). The pro- cedure works best for species with habitat prefer- ences that can be described in terms of land cover and other mapped features or characteristics. It works for habitat specialists only if their specific habitat requirements are available as mapped fea- tures or are well associated with other mapped characteristics such as land-cover types. An addi- tional caution is that species with very restricted distributions cannot reliably be predicted to occur ropriate habitat within their gen- eral distributional limits. Because of their rarity, these species are often the subject of special atten- tion from state and federal resource agencies. The specific locations where they are known to occur are usually tracked by Natural Heritage Programs (NHPs) and Conservation Data Centers (CDCs). GAP makes use of the data from Heritage Programs and CDCs to report the presence of populations of such species within a mapped unit. For security purposes, the exact locations of these populations are distributed only by the NHPs or CDCs. in seemingly a LAND-OWNERSHIP AND LAND-MANAGEMENT MAPS Since one purpose of GAP is to provide an as- sessment of the conservation status of species and their habitats, maps of lands that are managed for conservation must be compared with the distribu- tions of species and habitats. Most states, however, do not have a current inventory of land-manage- ment status. The first step toward developing a map of conservation lands is to map land-ownership by the major categories of (1) public lands by man- aging agency, (2) voluntarily identified privately owned conservation lands, and (3) all other pri- vately owned lands. Then, as a second step, the attributes for land-management categories are add- ed to these tracts. All non-conservation privately owned lands (category 3 above) are simply labeled “private,” and individual parcel boundaries are not delineated. Land-ownership and land-management maps in- 42 Annals of the Missouri Botanical Garden Study Area Species List clude land parcels that can be reasonably resolved at a 1:100,0 is equivalent to 1 mm? on a 1: 100,000 scale map. Descriptions of how the land-management maps are developed are provided by Scott et al. (1993), Expert Review |Final Animal Modeling Flow Chart literature review, database query Habitat РЕ WHHNMV Wildlife Habitat Relation Model А | Revised WHR or continue These arrows indicate that the available GIS coverages guide collection of th habitat association data, the data values guide a reclass of the coverages to corresponding units, and the resulting information creates the WHRM Data Known locational or existing range maps data Expert Review Geographic Range limits Add Habitat Association Coverages [Land Cover | [Temperature | [Hydrological | Soils | Database query Expert Review Final species Hyperdistribution| Distribution | 'ntersect |Coverage Map all species distributions intersected Figure 2. The gap analysis wildlife habitat relationship model. All coverages intersected Hypercoverage | Intersect S Е raphic maps В е. ingle species _ Stewardship Coverage [see Stewardship section of Handbook Analysis > Phase Beardsley and Stoms (1993), and Edwards et al. 00 scale. Commonly this is 1 ha, which (1995). Land-management is ranked by the four levels shown in Table Over the past year, there has been an ongoing discussion among GAP participants about the ad- Volume 85, Number 1 1998 Scott & Jennings 43 Large-Area Mapping Table 2. The four levels of land management and their definitions. Level Definition — dance events are allow to natural communities. Pi" Areas for w Areas having a management plan in operation to maintain a natural state and within which natural distur- еа to proceed without interference Areas generally managed for natural values, but which may receive uses that degrade the quality of existing 1 legal mandates generally prevent permanent land cover conversions from natural or semi- natural 55 to anthropogenic habitats, such as conversions to agriculture, but which are subject to ex- active uses such as silviculture or mining. 4. Areas managed for intensive human uses. equacy of these definitions. Many feel a larger num- ber of categories that use a wider variety of man- agement activities undertaken on behalf of native species and ecosystems would be more useful. When a greater number of management categories was recognized during the Sierra Nevada ecosystem project and species were rated differently within these categories because of their varying responses to management practices, communication among cooperators was greatly improved (F. Davis, Uni- versity of California, Santa Barbara, pers. comm.). As a result, GAP is exploring a land-management scheme having greater thematic resolution. ANALYSES While there are many ways that the three basic data sets of land cover, vertebrate distributions, and land management may be analyzed, the primary purpose is to identify potential gaps in the existing network of conservation lands. The identification of conservation gaps is intended to provide land stew- ards with the information needed to modify their plans and practices in order to maintain our natural biodiversity and the processes that sustain it, and to avoid conflicts with other uses of the land. The analysis presented here focuses on the basic requirements for a state gap analysis project. These call only for identification of those biotic elements that lack adequate representation in conservation lands rather than the 2. of specific geo- " the gaps. The latter is the selection phase Eo reserve design and graphic locations needed to * requires detailed on-the-ground information con- cerning habitat quality and demographics of the species of interest. The first objective is to deter- mine the representation of each mapped alliance and vertebrate species in each category of land ownership and management status. The second ob- jective is to interpret the analysis in a way that is useful for land stewards in land-use planning and management for conserving those biotic elements. The program provides the data sets and the anal- yses in forms suitable for additional modeling, bio- diversity assessment, and planning activities. intersecting the land-cover and animal (“element”) distribution GIS coverages with the land-ownership and land-man- agement coverage so that the element coverages in- corporate the stewardship boundaries. Then, the statistics from that intersection are used to generate a table reporting the representation of individual elements (species and dominant cover types) in each ownership and management category. Finally, these results are used to generate maps of those elements found to be lacking in their representation in conservation lands, and they are incorporated into a standard final report. Each species and plant community alliance is identified and analyzed sep- arately. Selected groups of elements of interest may also be analyzed. For example, a spatial analysis of species having less than 10%, 20%, and 50% of their distributions in status 1 or 2 (Table 2) land- management areas is provided. Other groups of species of special interest (e.g., endangered species These objectives are met by or declining neotropical migrant bird species, and endemic species, etc.) may also be analyzed for their representation in conservation lands. There are clearly some limitations to this ap- proach. One is that the historical distributions of elements are usually poorly known; measures of present-day distributions usually cannot indicate the extent of loss in historical range (but see Noss et al., ). For example, if an element has al- ready been reduced by 90% from its historic dis- tribution, and gap analysis indicates a 50% occur- rence in management status 1 or 2 areas, the result is that only 4.5% of its historic distribution is rep- resented. Another limitation is that GAP currently does not predict element viability. For most species and plant communities, viability measures such as habitat quality, species abundance, population trends, reproductive success, and mortality at a site Annals of the Missouri Botanical Garden are unknown and cannot be assessed given current knowledge. Therefore, GAP only provides infor- mation on representation with the objective of high- lighting at-risk species and vegetation types that should undergo viability analysis as a next step. GAP process that also includes reserve selection and de- sign. Generally, conservation assessment of animal is the first phase of identifying a three-part species must be used with more caution than as- sessment of land-cover types because land-cover maps are actual, typically with a statistically valid accuracy assessment, while animal distributions are predicted and difficult to validate. Land-cover types are more stationary and change slowly, while ani- mal species are mobile and can expand and con- tract ranges over relatively short time spans; effects of management status on land-cover types are gen- erally easier to predict than effects on animal spe- cies. RESULTS AND DISCUSSION Prior to the development of spatial data by GAP, the information needed to assess the conservation status of all but a few of the most popular vertebrate species was not available in the United States. There were no geographically extensive maps or da- tabases of species distributions or actual dominant vegetation types at cartographic scales usable local land managers. For example, Klopatek et al. (1979) estimated that 34% of the land surface in the United States was subject to some form of in- tensive land use. The authors concluded that 23 of the 106 types of potential (or original) vegetation have been reduced by over 50%. Much more significantly though, they concluded that there were major drawbacks and limitations to their find- ings because no inventory of actual vegetation ex- isted at that time. They relied on general predic- tions of the occurrence and extent of potential vegetation for baseline data and compared those hypothetical data with nonstandardized estimates of county-level land-use practices. The critical infor- mation has, until now, been unavailable at the level of resolution necessary for large-area management of ecological systems. We believe that a comprehensive plan for pro- tecting our nation's biodiversity must include a rep- resentation of species and vegetation communities across their full range of geographical occurrence and ecological expression. The latter is being made possible by the development of standard catalogs and classification of the nation's vegetation types (Bourgeron & Engelking, 1994; 1994; Weakley et al., 1997; neddon et al., Drake & Faber-Lan- gendoen, 1997; FGDC, 1997; ESA Vegetation Clas- sification Panel, in prep.), which is overcoming the lack of a standardized system of vegetation classi- fication (Orians, 1993). There is some confusion as to what represents a reasonable target for species or community conservation. The Endangered Spe- cies Act (ESA) currently stipulates species, sub- species, or distinctive population units. Much of the current debate over reauthorization of the ESA con- cerns the unit of protection, with many asking that we be more restrictive and protect only species or populations for which it can be demonstrated there is no gene flow with other populations. It is the belief of many that we have spent an inordinate amount of effort protecting subspecies and popu- lations, although this is not borne out by the facts (Tear et al., 1993). One suggested conservation tar- get is the natural community or the association in the National Vegetation Classification (FGDC, 1997; Jennings, 1993). However, we are currently unable to synoptically map that level of detail across physiographic provinces, ecoregions, or bi- omes. Examination of coarser levels such as mo- saics of dominant vegetation types suggests that 16 of 30 plant communities evaluated in Utah were at risk (Edwards et al., 1995), and 32 of 71 in Idaho 1995). Thus, even at this coarser level of the GAP map- ping effort, we found perhaps 25—40% of mapped vegetation types were at risk, and with them, other were considered vulnerable (Caicco et al., associated elements of biodiversity. This suggests that major progress toward protecting biodiversity could be made by simply insuring that viable ex- amples of each of the vegetation alliances in North America be managed for their long-term viability. However, we must be cautious. In interior mar- itime coniferous forest in the Pacific. Northwest (Scott et al., unpublished ms.) we found the West- ern Red Cedar had 36% of its acreage in special management status. However, when examining the evenness of the Western Red Cedar forest alliance across its full range of ecological and geographical expression, we found its occurrence in special man- agement areas was biased elevationally and geo- graphically. When we examined the representation of the 16 identified and mapped natural community associations of the Western Red Cedar alliance in Idaho, we found eight with no acreage in special management areas and several with more than 80%. Th formed the decision-making process of how to pro- us, the more detail we have, the more in- ceed with management. This just serves to empha- size the need for a hierarchical approach, spatially and thematically, for evaluating the effectiveness of current conservation efforts. Volume 85, Number 1 1998 Scott & Jennings 45 Large-Area Mapping To date, results from Gap Analysis projects have been reported from Utah, Wyoming, Arkansas, Cal- ifornia, Idaho, Oregon, Massachusetts, and Maine. Information from these areas has been used for land-use planning at several locations in southern California (Crowe, 1996), including Camp Pendle- ton and the Mojave Desert wards, comm.). It has been used to assess the contribution of proposed wilderness areas and new national parks to the lve € of biodiversity (Wright et al., er uses include identifi- cation of new 2. sites апа species and veg- etation types at risk. But perhaps more importantly, it has served as the catalyst for new partnerships, often among individuals and organizations who had little or no history of working together. Partnerships (e.g., Crowe, 1996) forged in the data acquisition and analysis phase of the individual Gap Analysis pers. projects have continued on into the implementation phase of GAP and into other endeavors as well. These partnerships are deepening as we peri- odically update the thematic layers of GAP, and their application for more informed land-use deci- sions becomes an ordinary feature of natural re- Additional information on can be found at : sources management. ap Analysis gap.uidaho.edu/gap. Literature Cited Allen, T. F. & T. B. Starr. 1982. — id Ші. for Ecological Theory. Univ. € ess, Chicag Bara, T. J. (editor). 1994. Multi- 2. Гапа ы teristics Consortium Documentation Notebook. EMAP- Landscape Characterization, EPA Office of Research and Development, Research Triangle Park, North Car- olina. See also, the MRLC home page on the World Wide Web at ии lc. Beardsley, K. & D. M. Stoms. 1993. ompiling a digital map hi areas managed for 2. іп California. eas J. 1: —190. e P. S ). Engelking. 1994. A Preliminary Vegetation 2 of the Western United States. A report prepared by the Western Heritage Taskforce and The Nature Conservancy for the University of Idaho — Fish and Wildlife Research Unit. The Na- ancy Western Regional Office, Boulder, Butterfield, B. R., B. Csuti & J. M. Scott. iri Modeling vertebrate distributions for gap analysis. Pp. 5: R. I. Miller (editor), Mapping the e of Nature. Chapman and Hall, London. Caicco, S. L., J. M. Scott, B. Butterfield & B. Csuti. 1995. A gap analysis management status of the vegetation of Idaho. Conservation of Biological Diversity 9: 498—511. Crowe, R. E Use of gap d in regional plan- ning in 2 California. Pp. 221-238 іп J. M. Scott, T. H. r & F. W. Davis (editors), c Analysis, A Eu Approach to Biodiversity Planning. Ameri- can Society for Photogrammetry and Remote Sensing, Bethesda, Maryland. Crumpacker, D. W., S. W. Hodge, D. Friedley & W. P. Greg, Jr. 1988. A а assessment of the status of major terrestrial and wetland ecosystems of federal and Indian эе in the United States. Conservation Biol. 2(1): 103-115. Davis, F. W. ca The nature of gap analysis. Letter. BioScience 46: 74—7 ч & D. M. Stoms. 1996. Geographic Information Systems Mass of Biodiversity in California. Final re- port. University of California, Santa Barbara, Depart- ment of po sae . Estes, B. Csuti, J. M. Scott, D. эж» М. Рајића, P Prag A. Habd R. Walker . Bueno. C. Cogan & V. G ograp hic poa ation "ar ment of Geography, University of California, Santa Bar ara. Stine, D. M. Stoms, M. I. Borchert & A Hollander. 1995. Gap analysis of the actual E of California: 1. The southwest region. Madrofio 42: 40— Drake, J. & D. Faber-Langendoen. 1997. An Alliance Level Classification of the Vegetation of i Midwesten United States. A The Nature onec Midwest Ке- gional Office, Minneapolis, Min Edwards, T. C., 995. Data КАШ and gap anal- ysis. Letter. e чепсе 46: 7 , С. С. Homer, S. D. Ga A. Falconer, R. D. Ramsey & D. W. Wight. 1995. Utah Gap [кле Ап Environmental Information System. Final project report 95-1. Utah Cooperative Fish and Wildlife Research Unit, Utah State Misi Logan. Falconer, R. D. Ramsey & W. Wright. 1996. j nudo of wildlife habitat re- lationships models estimating spatial distributions on terrestrial vertebrates. Conservation. Biol. 10: 63-270. Eyre, F. H. (editor). 1980. Forest Cover Types of the Unit- ed States and Canada. Society of American Foresters, Washington, D. FGDC (Federal Copla Data Committee). у FGDC Vegetation Classification and Information Stan- dards. U.S. Geological Survey, MS 590 National Center, Reston, Virginia. —— ———. 1997. FGDC Vegetation Classification and In- formation Standards. Federal Geographic Data Commit- tee, U.S. Geological Survey, Reston, Virginia Grossman, D., K. L. Goodin, X. Li, C. Wisnewski, D. Fa- ber-Langendoen, M. Anderson, L. Sneddon, D. Allard, M. Gallyoun & A. Weakley. 1994. Standardized Na- tional аута Classification System. Report by The Nature Conservancy and Environmental Systems Re- search Institute for de NBS/NPS Ж ны Mapping Program. M Biological Service, Denver, Colorado. Hawken, P. 1993. The Ecology of Commerce. Harper Col- lins, New к Heywood H. (edito, 1995. Global Biodiversity As- ridge 993. Natural Terrestrial Co Classi- fication: Assumptions and Definitions. Gap Analysis Technical Bulletin 2. Idaho Cooperative Fish and Wild- life Research Unit, Moscow, Idaho. 199 confluence of biology, ecology, and ge- ograph hé management of biological resources. or Wildlife Soc. Bull. 23(4): 658-662. Annals of the Missouri Botanical Garden 1993. Building a macroscope: How well do places pde for biodiversity match reality? Renewable а urces J. 11: "as Um —— kney, P. Crist & R. Sorbel. 1996. Gap bo Program 1995 and 1995 Status Report. USGS Gap Analysis ay gram, Moscow, Idaho. Кору 1. М 3 Olson, C. J. Emerson & J. C. ТӘ. Lande -use conflicts with natural 4. in ы United e. Environm. Conservation 6 aa )9. Küchler, А. W. & 1. S. Zonneveld (edit any i Vege- tation vmi Kluwer Academic Publishers, "enr Levin, S. A. 1981. The ae of pattern and scale in ecology. Pa Ng 1942-1968 Lillisand. T. M. ‚А site ol for satellite-based land- cover des ds ‘ation in the upper . Рр. 103-118 in J. M. Scott, Т.Н. Tear & F. W. Dav та s), Gap Analysis, A Landscape Approach to Biodiversity Plan- ning. American Society d Photogrammetry and Remote Sensing, Bethesda, Maryland. Loucks, O. L. 1995. ч cial committee on vegetation classification annual report. Bull. Ecol. Soc. Amer. 76: 221-223. 1996. 100 years after Colles: sification for vegetation. Bull. Ecol. Soc. 76. A national clas- Amer. 77 Noss, R. . Y. Cooperrider. 1994. Saving Nature's Legacy. e Press, Washington, D.C — ———, T. E. LaRoe III € J. M. Scott. 1995. .. Ecosystems of the United States: А Pre ^ sessment of Loss and Degradation. Biol. Rep. 28. 1 Dept. of the ines National Biological Service, Wash- a D. L. DeAngelis, J. B. Waide & 1 1986. A hierarchical concept of ecosystems. Monographs in Population Biology No. 23. Princeton Univ. Press, Princeton. Orians, G. Н. 1993. Endangered at what level? Ecol. Applications 20: к +208 Scott, J. M., B. Csuti, J. D. Tea, E. Estes. 1987. Species richness: А geographical do 'h to protecting biodiversity. BioSci лепсе Ж 782- ‚ F. Davis Bet p. Butterfield, “aden son, " Cas А Mn Erchia, T. C. F oe ES 1 Ulliman & G. Wright. 1993. Gap analysis: A geographic approac x to protection of biological di- versi Nuls Wildlife Mo onogr. 123. . D. Jennings, R. G. Wright & B. Csuti. 1995. 51. 'ape approaches to mapping biodiversity. Letter. BioScience e 46: 74—75. Т. Н. Tear & Е. W. Davis (editors). 1996. Gap Analysis, A Landscape Approach to Biodiversity Plan- g. Americ 'an Society for EUREN: and Remote 1995. National biotic re- source inventories and GAP analysis. BioScience 45: 35-539. Slaymaker, D. К. Griffin & J.’ Finn. 1996. Mapping p iduous forests in 4. New England using aerial videography апа hyper clus- tered multitemporal landsat TM imagery. Pp. 8 Scott, T. H. Tear & К. W. Davis (editors), Gap . А Landscape Approach to B . Jones, C. FERE ning. 4! ап Society for Photogrammetry and Remote Sensing, Bethesda, Maryland. Sneddon, L., M. Anderson & K. Metzler. 1994. A Clas- sification and Description of Terrestrial Community Al- liances іп The Nature Conservancy’s Eastern Region: First Approximation. А report prepared for the National Gap Analysis Program, ‘ish and Vildhi Jnit, University . Scale dependence of species richness maps. The Presiona 2... 46: 346-358. ^R 93. A remote sensing research agenda for mapping ne РА biodiversity. Int. J. а ме Ser Pu Е 2 —1860. . T. H., J. SR базы j % Griffith. Status and p for success ~. Es 1993. f the Endangered A look at recovery ш. Science 262: Thomas, К. A. & К. W. Davis. 1996. Applications of gap analysis data in the Ws 1. ^e California. Pp. 209-220 in J. M. Scott, T. H. itors), Gap Analysis, A 2. ‘ape el to Biodi- versity Planning. American Society for Photogrammetry and Remote EU Bethesda. Maryland. Vitousek, P. M., . D'Antonio, L. L. Loope & R. West- brooks. 1996. E al invasions as global environ- mental change. Amer. 84: 486-47 ——— Mooney. J. Lubchenco & J M. Melillo. Homan domination of earth's ecosystems. Sci- 99, \ 5. Landaal € M. Gall- youn. 1996. International Classification of Ecological Communities: Terrestrial Vegetation of the Southeastern United States. Working draft of April 1966. The Nature Conservancy, Southeastern. Regional Office, Hill. North Carolina. Patterson, Chapel —— == & ——— An Alliance Lev "| C 6 'ation of У Vegetation й, the Southeast- ern United States. Conservancy Southea versity of Idaho ni Fish and Wildlife Research Unit. The Nature Conservaney Southeastern Regional Office. C oe Hill, North Carolina. и К. Vegetation of the Siskiyou Moun- ains, Oregon and California. Ecol. Monogr. 30: 279-338. 19 ies diversity in land communities. Evol. Biol. 10: 1-6 Wiens, J. A. 1989. “The Ecology of Bird Communities. Vol 2. Резе esses and Variations. Cambridge Univ. Press. New York. Wright. R. G., J. L. MacCracken & J. Hall. 1994. An ec ologic ‘al evaluation of proposed new conservation ar- eas in Idaho: 4. proposed Idaho national parks. Conservation Biol. 8: 207-216 Appendix 1. А Sample Description of a Community Al- iance Vegetation Type (from Sneddon, 1994) 1. Class: Forest LC. Subelass: Mixed evergreen and deciduous forest І.С.З. Group: Mixed needle-leaved evergreen and cold deciduous forest C.3.N. Subgroup: Natural/Seminatural vegeta- tion (not cultivated I.C.3.N.a. Formation: et < Mixed needle-leave 5? and cold deciduous nce : Tsuga can- асс Бата Већ allegheniensis Forest Alliance ^C ommunity Alliance: adensis—Acer M gen 2. hemlock, sugar maple, yellow rch, forest alliance: Forests of this alliance include me- sic communities known as "ће “hemlock ravine” and “hem- Volume 85, Number 1 1998 Scott & Jennings 47 Large-Area Mapping Communities of this alliance ic ravines, north-facing slopes, and contain substantial lock-northern hardwoods." generally occur i in mesi er of the northern hardwood fees, most commonly Betula allegheniensis, as well as Acer saccharum and Fagus gran- difolia. Tsuga may be dominant, particularly in ravines, and Prunis serotina is a major component in the Allegheny Mountains. Other canopy associates о Betula lenta, occasional Pinus strobus, and P rubens in norther New England. Viburnum alnifolium, "Dieriilla hi Sambucus pubens, and Taxus canadensis occur in these communities; Ahododendron maximum is particularly characteristic in southern representatives of this alliance. Herbaceous flora may be sparse, but generally includes Mitchella repens, Oxalis montana, Lycopodium lucidulum, Streptopus — Hu virginiana, Epigaea repens, and Maiar qud nade SAF 1 pe > 24, p omen аз ,— Yellow Birch is more or less a ith this alliance Regional Distribution! This lanes occurs in all East- given for SAF e 24. Hemlock- Yellow Birch: central and southe sou sin and Michigan, and Cape Breton, south on the Alle- ioe and Catskill Mountains, central New band and the alachians, south discontinuously on the southern при каш (Eyre. 1980). WHAT SATELLITE IMAGERY AND LARGE-SCALE FIELD STUDIES CAN TELL ABOUT BIODIVERSITY PATTERNS IN AMAZONIAN FORESTS! Hanna Tuomisto? ABSTRACT The great problem i in Dep de studies in Amazonia is that the existing data are regionally very biased, 1. the question is about specie species composition among sites. data be extrapolated, anc question of how many sp t are in one hectare. how much th i as a who s : wide areas, even inaccessible ones. The di ; s i atellitè imagery from distribution patterns, local species diversit e surroundings of a few cities and biological stations are relatively well menie, while most of Amazonia still doped со in these respects. The essential questions аге, to what exter с do we most urgently need more data? Quantifying biodiversity is not just a It is also a question of how many diffe y levels, or differences in species diversity and at can th erent habi es there are in a g n to bs studied to document and understand it. Because it would take too much time to дену the thousand or so plant аар ies that can be found in a single hectare of forest, species. This n ma and е patterns in Amazonia. we have developed an inventory method based on indicator es il possible to monitor large areas relatively rapidly and has revealed some intriguing ecological Amazonia comprises a huge block of tropical rainforest, and in spite of the relatively dense net- work of navigable rivers and the active roadbuild- ing in some areas, most parts of it are practically inaccessible, at least within the time and budget limits of an average biological field trip. No wonder, therefore, that biological inventories have been heavily concentrated on very few spots, which have then become famous inventory sites that everyone working in Amazonia wishes to visit. In southern Peru we have Cocha Cashu and Tambopata biolog- ical stations, with high levels of bird and butterfly species richness. In northern Peru we have the Mi- shana and Yanamono sites, which had the highest tree species diversity per hectare (Gentry, 1988) until a few years ago, when the 1-ha tree inventory in Cuyabeno (eastern Ecuador) was completed and set the new world record, 307 species (> 10 cm DBH; Valencia et al., 1994). Ecuador also hosts the second 50-ha tree plot in the Neotropics, which is now being established in the Yasuni National Park and is surpassing diversity estimates (Robin B. Fos- ter, pers. comm.). In Colombian Amazonia there is Araracuara with impressive bird and tree diversi- ties, and in central Brazil there is Manaus, with a high species richness in all Eum These are just a few of the most important s In spite of the dedicated efforts of many field biologists, the total area of forest that has been thor- oughly inventoried at these sites is vanishingly small, only a few square kilometers out of the 5 million in Amazonia, and no one knows what there is between these well-visited sites. The situation is like having to map plant diversity of North America on the basis of a few tree inventory plots made, say, in northern California, Yellowstone National Park, and the surroundings of the Niagara Falls, plus some general collections concentrated along the road between New York City and Washington, D.C. So what can we really learn about biodiversity in This paper is dedicated to the memory of Karl U. Kramer, who described many of the Lindsaea species discussed ls 're, and provided valuable help with species identification. I a grateful to numerous cee for collaboration in the field, especially Kalle Ruokolainen, Abel Sarmiento, Richer Ríos, Alberto ' Melchor Aguilar, Simón Cortegano, Guillermo Criollo, N Illich Arista, Isabel Oré, Elina Lusa, and logistic support. Jaana Vormisto, Risto Kalliola, and K Nestor 1, Gustavo Torres, Césa sar Jaana Vormisto, and to Universidad Nacional de la Amazonia Peruana for valle Ruokolainen gave constructive comments on the manuscript. Garcia, Antonio Layche, Bardales, Lizardo Fachín, 'orres, Mildre Financial ү for the work has been provided by the Academy of Finland, the STD-3 program of the European IDA. Union, and F . of Biology, University of Turku, FIN-20014 Turku, Finland. ANN. MISSOURI Bor. GARD. 85: 48—62. 1998. Volume 85, Number 1 1998 Tuomisto 49 Satellite Imagery and Biodiversity Patterns Amazonia on such a basis? Are the established sites truly the hotspots of biodiversity, or is the high number of species found in them only a function of the high amount of collecting effort that has been invested (Nelson et al., 1990)? In a situation like this, there are many questions one would like to get an answer for. The most fun- damental of all is: To what extent can we generalize the results from the sites we know something about? In the case of plant diversity, this translates into finding out whether there are floristically different vegetation types involved, and if so, how these can e characterized and mapped. Quantifying biodi- versity in general is not just a question of how many species there are in a hectare. It is also a question of how many different habitats there are, how much the floras of the different habitats differ from each other, and how many species there are in a given region as a whole. This is where satellite imagery comes into the picture. Since satellite images cover practically the whole earth, they can provide information on even the most remote rainforest areas, and in such a way that an overview of wide areas can be obtained at a glance. The color patterns in the satellite images are created by local differences in how the ground cover reflects sunlight, which depends on many physical properties, such as vegetation structure, color of plant leaves, presence of surface water, presence of bare soil or rock, and many more. Be- cause of this physical basis, it is a relatively safe assumption that whenever there is a color differ- ence between two parts of a single satellite image, there is also some que difference between the corresponding sites in the field. This is the good news. The bad news is that it is канай impossible to know what these physical differences are unless one has visited the sites in the field. In Amazonia, the only easily recognizable ground cover catego- ries are rivers, cities, roads, and vegetation; also cultivated areas can usually be recognized by their characteristic shape. But to find out more detailed properties of the vegetation (natural or cultivated) at a given site, one needs ground truthing or aerial photographs. Even such a trivial question as wheth- er the vegetation is forest or not cannot be solved on the basis of a satellite image alone; the spatial resolution of the images is not detailed enough to show 1 details like trees. This is why large-scale field studies are needed. Коре, d satellite images soon reveal that even the continuous rainforest is not homogeneous, but rather shows itself as a bewildering mosaic of patches that come in different colors, sizes, and shapes. It is the task of the field studies to find out how each of the reflectance patterns can be inter- preted in ecological and floristic terms. Do the dif- ferences in reflectance correspond to floristic dif- ferences? Does the degree of reflectance difference reveal the degree of floristic difference? Which are the species that actually occur in each of the rec- ognizable patches? How many different habitats are there in the area as a whole? How many species are there in the area as a whole? How are the dif- ferent habitats distributed? How restricted are the distributions of the plant species in relation to the abitats? All these questions can be answered, at least to some degree, if the field studies cover enough area and are carefully planned with the help of the sat- ellite imagery. Field inventories need to include sites in landscape patches with different reflec- tances to document the differences, but they also need to include some sites in similar patches to document how homogeneous these are. And the in- ventories need to be rapid; it is not feasible to spend several months at a single site, if there are hundreds of sites to be inventoried. This paper will present results of studies that have been conducted in the northern part of Peruvian Amazonia with these objectives in mind. Basics ОЕ SATELLITE IMAGERY There is an ever-growing body of literature about satellites and satellite images, but much of it is rather technical and therefore alien to the majority of botanists who are not specialized in this partic- ular field. The purpose of this paper is to give the reader an idea of the potential and the problems involved in the use of satellite imagery in vegeta- tion and biodiversity studies by giving a short in- troduction to those aspects of satellite images that technical accounts can Harris 5 Mather (1987), we Lillesand and Kiefer (199 The ЕН basis of satellite images is quite simple: the satellite carries sensors that scan the ground and record the intensity of reflected sun- and sensor used. In Landsat satellites, which are the most widely used satellites in vegetation stud- ies, each pixel roughly corresponds to a s m by 80 m on the ground when MSS (multispectral scanner) sensors are used, and to 30 m by 30 m when the more advanced TM (thematic mapper) . While scanning the terrain, the sensors essentially measure the average reflectance sensors are use 50 Annals of the Missouri Botanical Garden (MSS 456 7) MSS 123 4 ШЕШ | ТМ 123 4 CHO El 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 blue Nu infrared [Band 1 AAA o ЕЕ кунк ее "nm + rid Е = та! | =H — Pixel — (has a single reflectance value) The bands in Landsat MSS and TM data. The numbering of MSS bands has been die i and Fig there un older images use a matic in its own data file of each pixel on a relative scale and store the mea- sured value into the cell corresponding to that pixel in a spreadsheet database (Fig. 1). In the process, any details smaller than the pixel are lost, and af- terward it is impossible to see on the image if a particular pixel contained, say, a house or a group of trees or open field. If such details are needed, they have to be obtained from other sources, like ield studies or aerial photographs. In the satellite, there are several spreadsheets being filled in simultaneously, each by its own sen- sor that records the reflectance in a certain wave- length band. The MSS is equipped with sensors to a different numbering system (shown in parenthesis) than more recent images. — Sche- ' representation of the же structure in Landsat TM imagery, where each of the seven wavelength bands is rec onda ile. observe four bands, and the TM has sensors for seven bands, including both visible and infrared wavelengths (Fig. 1). Any of the bands in a satellite image can be viewed on a computer screen with the help of an image analysis program. The com- puter is then told to use the spreadsheet corre- sponding to the desired band, for example, the one recording green, and to convert the numerical in- formation back to light intensities, in this case to the intensity of illumination of the pixels on a com- puter screen. The result is a map where those pixels that had a high reflectance for green on the ground are displayed bright on the computer screen, and Volume 85, Number 1 1998 Tuomisto 51 Satellite Imagery and Biodiversity Patterns those that had low reflectance are displayed dark. At this stage, it is possible to choose any color for the display. The original green can be displayed as green, but it can equally well be displayed as red or blue or (most commonly) gray. It is irrelevant which color is chosen, because the information is here conveyed by the relative brightness of the pix- els, not their hue. Different bands show somewhat different infor- mation, because not all surfaces reflect the different wavelengths in a similar way. For example, water tends to absorb infrared, but to have a high reflec- tance for visible wavelengths, especially blue. Therefore open water areas can appear almost black in images created using the infrared bands, but very bright in images that use the bands of visible wavelengths. Vegetation, on the other hand, has high reflectance for infrared, but very low re- flectance for visible wavelengths, especially red and blue, because these are absorbed and used in photosynthesis. Because of these differences, it is usually desirable to view several bands at once to get a better idea of the overall spectral variation in the study area. The easiest way of doing this is to make a color composite using three of the available wavelength bands. Three bands can be used simultaneously be- cause each pixel on a color computer screen can be thought of as a group of three lamps: red, green, and blue. The hue and brightness of a pixel depend on which of these lamps are lit and how brightly they shine. For example, if only blue is lit, the pixel will appear blue; if red and green shine at equal intensities but blue is not lit at all, the result is yellow, and if the intensity of red is increased rel- ative to the intensity of green, the hue becomes progressively more orange and finally red. In the process of making a color composite, the computer produces a map with each of the desired bands, assigns each of these to its own color on the screen, and displays the three maps simultaneously on top of each other to produce a full-color image. On displaying a TM image, for example, the reflec- tance values of band 3 can control the intensities of red on the screen; band 4 can control green, and band 5 can control blue. Obviously there are many other possible combinations that can be used. In fact, because different bands convey different in- formation and hence show partly different patterns in the image, the choice of the band combination is very important. Patterns in an image created with bands 4, 5, and 7 are somewhat different than those in an image created with bands 1, 2, and 3. For vegetation studies, the most useful color composites are often obtained by combining two near-infrared ands with one visible-light band. It is important to keep in mind that the actual colors of the final satellite image product have no absolute meaning: they depend on the arbitrary de- cision on which of the chosen bands was assigned to which of the colors on the computer screen, and as long as the same band combination is used, changing the color assignments makes no differ- ence for the information content of the image al- though the overall color of the image may change drastically. Of course some color combinations look more pleasing than others and are therefore more commonly use In densely vegetated areas with little surface re- lief, such as lowland Amazonia, most of the surface is green. From an airplane, such areas look rather monotonous with few eye-catching features, and so they look from a satellite, too. This is because there are no big differences in surface reflectance from one site to another, and therefore most of the pixels have reflectance values that are very similar and only represent a narrow range of the possible in- tensities that the satellite sensor is capable of re- cording. Consequently, a color composite created with the original satellite data looks relatively ho- mogeneous: there may be a hint of a pattern there, but if the differences in intensity are not big enough, they cannot be confidently recognized or mapped. Image enhancement is the solution to this prob- lem. There are several ways of enhancing an image, but the main purpose of all of them is to make more efficient use of the different light intensities a com- puter is able to display. Instead of using just the narrow range of intensities that were recorded in the original spreadsheets of the satellite image, the computer recalculates the reflectance value for each pixel in such a way that the differences are exaggerated and a wider range of possible intensi- ties is used on the computer screen. The process can be compared to adjusting the contrast on a TV screen: if contrast is too low, all patterns on the screen seem fuzzy; if contrast is too high, details are lost; when contrast is optimized, the patterns become clear and easy to recognize. А more advanced phase in the digital analysis of satellite imagery is the automatic classification of the image to different ground-cover types. There are two principally different methods that can be used to obtain a computer-classified image: super- vised and unsupervised classification. procedures. In supervised classification, the user selects groups of pixels that represent the different ground-cover types in the area, and the computer then assigns Annals of the Missouri Botanical Garden each of the remaining pixels to one of these classes In unsupervised classification, the user defines ei- ther the number of classes she wants to obtain, or the amount of variation to be allowed within any one class. The computer then creates as many classes as are needed and assigns each pixel to one of thes [к ушаш: a general drawback of digital clas- sification is the very low reliability of classification within the rainforest realm (see discussion in Tuo- 1994). Relatively good results can be obtained in areas where there are clear structural misto et al., differences among the ground-cover types, such as exist among savannas, swamps, and closed-canopy forests. However, distinguishing among different kinds of closed-canopy forest is much more diffi- cult, and hence the degree of error is higher. An- other source of error is that, even if the study area is unexplored in the field and it is unknown how many vegetation types there are, the user has to define either the number of classes or the variation allowed within each class prior to analysis. In this way, user-induced bias is easily incorporated into the results, although digital a is often ad- ” vocated as an “objective” met P А great advantage of шы images is that the researcher can be fairly certain that all the pat- terns that are visible in the product really exist in nature. Fieldwork may later show that not all of them are relevant for the questions at hand, or that some true differences were not recognized, but the risk of creating artificial patterns is small. Indeed, unclassified but enhanced image prod- ucts have proved especially useful for the monitor- ing of large and unexplored rainforest areas, since they are able to reveal spatial patterns whose ex- istence has previously been unknown (Townshend et al., 1987; Kalliola et al., 1991; Tuomisto et al., 1994; Tuomisto et al., 1995). Consequently, such satellite images can be efficiently used in fieldwork planning: they can help in locating sites that rep- resent formerly uninventoried or otherwise inter- esting vegetation, and they what extent results of fieldwork at any given loca- can also indicate to tion can be extrapolated to other locations. For these reasons, the term "satellite imagery" is used in the following text to mean "enhanced, unclassi- fied satellite imagery." MATERIAL AND METHODS A preliminary interpretation of satellite imagery was used to identify units that were suspected to harbor different kinds of rainforest vegetation in the northern part of Peruvian Amazonia. An attempt was then made to select field study sites so that as many of these units as possible were sampled, while attention was also paid to the adequate geo- graphical distribution of the samples. Initially the satellite image interpretation was based on a Land- sat MSS scene from 1983 centered around the city 1994), but later a more recent TM scene (from 1993, to be of Iquitos (published in Tuomisto et al., published elsewhere) became available, as well as M scenes for adjacent areas. The scale of the im- ages used in the visual interpretation was 1: 250,000 The exact locations of the field study sites were chosen with the help of the satellite imagery so as be both representative of interesting-looking landscape patches (or border zones between patch- es) and practically accessible by roads or navigable rivers. The primary purpose of the study was to document variation within tierra firme (non-inun- dated) forests, so swamps and seasonally inundated areas were excluded from the sampling whenever they were large enough to be identifiable in the satellite images. However, small swamps in depres- sions between adjacent hills and floodplains of small creeks occur throughout tierra firme, anc these were included as a part of the natural varia- tion within the landscape. The present paper will concentrate on docu- menting distribution patterns of pteridophytes. In earlier studies (Tuomisto et al., 1995; Ruokolainen et al., 1997) w the Melastomataceae can be used as indicators of we have found that pteridophytes and more general floristic patterns, because the floristic similarities among sites as measured with either pteridophytes or the Melastomataceae show a very high correlation with the floristic similarities as measured with trees: the correlation between pte- ridophytes and trees can exceed 0.8 (Mantel test, < 0.001; Ruokolainen et al., 1997). This is very practical for large-scale vegetation studies, where it is necessary that the sampling at any one site is floristically representative enough to justify region- al comparisons. Both pteridophytes and the Melas- tomataceae are easy to identify and collect com- pared to trees, because they are smaller in size and include far fewer species. Indeed, the high species diversity of trees makes tree sampling and identi- fication especially laborious, and the number of in- dividuals observed per tree species in any one sam ple plot is often low (Campbell et al., 1986; Balslev et al., 1987; Gentry, 1988; Valencia et al., 1994; Duivenvoorden & Lips, 1993, 1995; Ruokolainen et al., 1997; see also Clark, 1998, this volume). Consequently, chance can have a great impact on the observed floristic composition, and it is difficult Volume 85, Number 1 1998 Tuomisto Satellite Imagery and Biodiversity Patterns Amazon Inventory transects: — Over 5 km long - 1.3 km long * 0.5 km long Figure 2. in Mishana (a) jm Sucusari (b) are discussed in detail in the te to unravel ecological patterns in the tree species distributions. Focusing the inventories on suitable indicator species makes it possible to sample larger areas and more individuals per species in a shorter time, and consequently it becomes more feasible to work out the possible edaphic preferences of each species. It also becomes possible to evaluate with a higher certainty whether two samples actually be- long to the same forest type or not, because each sample contains a higher proportion of the local species pool and the observed similarity patterns are hence more robust. Because the study was aimed at finding out to what extent edaphic conditions influence the dis- tribution of plant species, sampling was done along continuous, finely subdivided transects. This made it possible to directly compare the changes in flo- ristic composition, species abundance patterns, and environmental variables (such as the nutrient con- tent, texture, and drainage of the soil). Several soil samples were taken from each site for chemical analyses in order to quantify the edaphic differ- ences among and within sites. Details on these analyses will be published elsewhere (Ruokolainen et al., 1997, in prep.). In all, 23 transects were sampled (Fig. 2), total- ing almost 83 km or 22 ha. In most cases, the tran- sects were selected within a uniform area of rain- forest as judged from the satellite imagery, but four of the transects were placed so that they crossed Map of the study area in Peruvian Amazonia with I locations of 23 inventory transects. The transects one or several boundaries that were m in the imagery. These were Mishana (marke " in Fig. 2), Trece de Febrero (15 km S of ыы Sucusari (marked with *b" in Fig. 2) and Carbajal (7 20-km transect in the SW corner of the study area). At each of the fieldwork sites, one of four alter- native sampling procedures was followed, depend- ing on the degree of detail required and the time available for studying the site. The field procedures were (1) 500-m-long qualitative transect, (2) 500- m-long quantitative transect, (3) 1300-m-long quantitative transect, and (4) several-km-long semi- quantitative transect. In all procedures, the base- line followed a predetermined compass direction (with allowance made for a 90? angle in four cases). The short transects (less than 2 km long) were 5 m wide, and the longer transects were 2 m wide. In the qualitative transects, a list was obtained of those pteridophyte species that occurred within an estimated 2.5 m on either side of the baseline. The quantitative transects were subdivided into contig- uous 5 m by 5 m subunits, and the corners of each subunit were marked; within each subunit, the in- dividuals of each pteridophyte species were count- ed. The semiquantitative transects were divided into subunits 100 m long, and for each subunit the presence of pteridophytes within an estimated 2 m on the left side of the baseline was recorded. In the first transect (Mishana, marked with “а” in Fig. 2), Annals of the Missouri Botanical Garden no size limit was applied to the pteridophytes, but in the subsequent transects only pteridophytes with at least one leaf longer than 10 cm were taken into account in order to reduce the time spent on look- ing for and identifying tiny plantlets. Pteridophytes with leaves less than 5 cm long were excluded from the Mishana data before analysis for the present paper. Epiphytes and climbers were only included if they had green leaves at a height less than 2 m above ground. Nomenclature of the pteridophyte species is mainly in accordance with the revision of the genus Lindsaea (Kramer, 1957) and Pteridophyta of Peru Tryon & Stolze, 1989-1994). However, Lindsaea lancea (L.) Bedd. var. lancea and L. lancea var. falcata (Dryand.) Rosenst. are in the present paper treated as good species rather than varieties, be- cause (1) they are easy to recognize in the field at any size > 3 cm, (2) they differ ecologically (L. falcata Dryand. grows mainly on decaying wood or litter while L. lancea is mainly terrestrial), and (3) they maintain their distinctness even when occur- ring at the same site and on the same substrate. RESULTS AND DISCUSSION DIVERSITY PATTERNS IN PTERIDOPHYTES The fern genus Lindsaea provides an interesting example of diversity patterns at different spatial scales. The 23 transects (Fig. 2) harbored a total of 11 Lindsaea species, all of which occurred at more than one site. The number of species at any one site ranged between zero and six, and up to nine species were found in the very long transects (Fig. 3A). Transect length obviously contributed to the high Lindsaea species richness of the two longest transects, but among the 0.5-1.3-km-long transects there was no consistent relationship between sam- ple size and number of species found. If only the geographical distribution of the diversity values is observed, it is hard to find any regular pattern in the species richness patterns. A much clearer picture emerges when the soil characteristics of the sites are also taken into ac- count, because these were clearly related to wheth- er a particular species was present or absent at a given site. One of the species thrives on clay soils that are rich in nutrients (Lindsaea phassa K. U. Kramer; Fig. 3B), while two species are restricted to nutrient-poor sandy soils (L. hemiglossa K. U. Kramer and L. tetraptera K. U. Kramer; Fig. 3K, L). One species can be characterized as a generalist (L. divaricata Klotzsch; Fig. 3C), as it can be found almost anywhere save the very poorest and the very richest soils of the region (however, there is some indication that there are actually two closely related forms with different edaphic preferences involved). All the other species are confined to the relatively poor loamy soils, but even among these species there are some interesting differences in distribu- tion patterns. For example, Lindsaea bolivarensis V. Marcano and L. taeniata K. U. Kramer are both relatively frequent at some loamy soil sites, but they rarely occur at the same site (Fig. 3H, I). The general ecological pattern that arises is that there are more Lindsaea species adapted to poor soils than to rich soils, and that the genus is also locally most diverse on relatively poor loamy soils. Indeed, the only site with no Lindsaea species at all has the richest soils of the 23 sites, and the 3 sites that only had one species each (this species was invariably L. phassa) are all among the rich- soll sites. All sites with intermediate to poor soils had more than two species each. It is important to notice here that terms such as and "rich" should be understood so that they only refer to the relative soil fertility of the 23 sites reported here, and that they do not imply "poor" “poorness” or "richness" in a wider context. The present study was conducted in a limited geograph- ical area, and the sampling is far from complete even for that area. Therefore, it is almost certain that the sites discussed here do not represent the full range of existing soil variation in Western Ama- zonia, so no matter whether the sites are here called poor or rich, they might all just become “interme- diate” if more sites were to be included in the com- parison. In some cases it is obvious that internal hetero- geneity within a single transect contributed to the high number of species present. Among the short transects, the most obvious example of this phe- nomenon is the transect close to the village of Mi- shana (marked “а” in Fig. 2). This transect crossed the boundary between intermediate loamy-clayey soll to poor sandy soil at about 600 m from the beginning of the transect (Fig. 4; further details on floristic and soil changes at this edaphic boundary have been published in Tuomisto & Ruokolainen, 1994). A species changed from L. lancea (on loamy soil) to L. divaricata (on sandy soil). Lindsaea falcata was abundant close to the transition zone, with a few individuals at the end of the sandy part. The fa- vored substrate of this species appears to be dead plant material, as it is usually found either on de- caying wood or on microsites with thick litter and t that same point, the dominant Lindsaea humus layers. The same is true for L. guianensis (Aubl.) Dryand., which is most commonly found on decaying tree trunks. This behavior may explain Volume 85, Number 1 uomisto 1998 Satellite Imagery and Biodiversity Patterns 9 26 Lindsaea species divers 377, (total=11 Amazon Lindsaea falcata Lindsaea hemiglossa Lindsaea tetraptera — umber of pinot кч in each of 23 inventory transects. —B-L. N umber of individuals of * 2 et species found in the . + indicates that the species was present but the number of individuals ese was bs recorded. The species are en pum erc in the order of decreasing requirements for soil fertility. why L. guianensis is found at sites with widely dif- bored a total of eight Lindsaea species, the distri- ferent soil characteristics and can co-occur with butions of which are shown in Figure 5. As it such ecologically different species as L. phassa and turned out, transect sections with lower topography L. hemiglossa (compare Fig. ЗЕ with 3B and 3K). tended to have clayey soils, while in the more hilly The fifth species found in Mishana was L. hemi- sections the soils were loamy and poorer in nutri- glossa, which was restricted to the sandy soil at the ents (the satellite image and the details of the soil end of the transect in accordance with its prefer- analyses will be published elsewhere; Ruokolainen ence for sandy soils at other sites (Figs. 3K, 4). et al., in prep.). t a wider scale, spatial variation in species The soil differences were clearly reflected in the composition can be observed along a 43-km-long species distribution patterns. The distribution of transect starting from the river Sucusari (marked — Lindsaea phassa was again unlike those of the other “b” in Fig. 2). The most eye-catching feature of the species: it was relatively frequent in the beginning satellite image covering the area is the alteration of and the end of the transect, but entirely lacking in lighter and darker patches, which corresponds the hilly stretch between km 23 and 38. Lindsaea roughly to the alteration of topographies with lower taeniata, L. lancea, and L. divaricata were also un- and higher hills, respectively. The transect har- common in this area, but showed otherwise more 56 Annals of the Missouri Botanical Garden 1004 Number of individuals in 100 me E Lindsaea lancea ІШ Lindsaea falcata C] Lindsaea divaricata , 804 Bl Lindsaea guianensis pis pm A Lindsaea hemiglossa 61------ loamy soil 40 4 sandy soil 20 4 " A p PA АР о FA а De ША КИША | АШ di 200 400 600 800 1000 1200 Distance along transect (m) Figure 4. Spatial variation in the abundance of the Lindsaea species that were found in the transect in Mishana (for geographic је cation, see Fig. 2). 1204 Topography (m) Lindsaea B present [] absent 40 L. phassa L. taeniata L. lancea L. divaricata L. guianensis L. falcata L. bolivarensis L. sp. 1 0 5 10 15 20 25 30 35 40 km Figure 5. Spatial variation in the occurrence of the Lindsaea species that were found in the transect in Sucusari (for geographic location, see Fig. 2). Volume 85, Number 1 Tuomisto Satellite Imagery and Biodiversity Patterns To raphy (m 120 pography (m) vt T T T Number of Lindsaea species (total 8 spp) Е Number of Thelypteris species (total 11 spp) Number of pteridophyte species (total 130 spp) 0 0 5 10 15 e 6. Spatial variation in the species ci of ha Thelypteris, n pter ee Fig. 2. F T NR in Sucusari (for geographic location n sample units 500 m long, and for overall pteridophyte species richness in vb "500 m scattered distributions, and were often present in subunits where L. phassa was lacking. Тһе remain- ing four species were essentially present where L. phassa was absent, і.е., in the loamy-soil sections of the transect. Lindsaea bolivarensis and an un- identified species of Lindsaea were almost confined to the high hills around km 30. Although the local species richness of the genus Lindsaea tends to be higher on poor soils than on rich soils, this is by no means a universal pattern. For example Thelypteris, another species-rich fern genus, shows the opposite trend. In the Sucusari transect, Thelypteris had no species at all on poor soils, and by far most of the species were found on rich soils (Fig. 6). The diversity pattern of Thelypteris was paral- leled by many other fern genera, with the result that 25 30 35 40 Distance along transect (km) ridophytes in general along the eris, species richness is calculated ‚ 1000 m, and 5000 m long. Lindsaea and Thelyp the overall species richness of pteridophytes was clearly higher in rich-soil sections than in poor-soil sections of the Sucusari transect (Fig. 6). This re- sult was found at several different spatial scales. When subunits of 500 m were analyzed, the num- ber of species per subunit ranged between 9 and 41, which is from less than a tenth to almost a third of the total of 130 species in the transect. Obviously the number of species increases with the size of the observed subunits, but even with subunits as long as 5 km, the number of species never exceeded 76 (just over half of the total). In the least species-rich 5-km subunit the number of species was only 36, i.e., less than in the most species-rich 500-m sub- units. This shows that diversity patterns depend on local site conditions as much as they do on sam- pling scales. Annals of the Missouri Botanical Garden The overall tendency for pteridophytes, that spe- cies diversity is positively correlated with soil fer- tility, explains why species diversity and local el- evation appear negatively correlated in Figure 6. The geological setting at this particular site is sim- ply such that the poorest soils are found at the high- est elevations. Because the Sucusari transect lies entirely within the lowland forest zone (about 100— 200 m above sea level), this pattern cannot be ex- plained by the elevation gradient itself: а vertical distance of 120 m is not big enough to cause such elevation-dependent patterns that are found on mountain slopes. The topography is sufficient, how- ever, to expose sediments of different origins and thereby to give rise to significant spatial variation in soils. ON THE HOMOGENEITY OF THE RAINFOREST Most tierra firme forests are structurally uniform, which is reflected in the paucity of vegetation types recognized for non-inundated areas in Amazonia (e.g., Prance, 1989). Even when this broad category of rainforest is subdivided, attention is usually paid to geomorphology rather than the vegetation itself for the simple reason that geomorphological data are readily available by remote sensing methods, whereas obtaining floristic data requires fieldwork 2 examples, see Malleux, 1975; Encarnación, 1985; Duivenvoorden & Lips, 1993; Tuomisto et al., 1994; INRENA, 1995). en no obvious regional differences have been apparent in tierra firme, researchers have tended to assume that these forests are ecologically uniform at broad spatial scales. Therefore, most theories that have attempted to explain the origin and main- tenance of Amazonian biodiversity have concen- trated on a variety of mainly historical factors. Ex- amples include cyclical changes in climate (Haffer, 1969, 1993; contributions in Prance, 1982; Whit- more & Prance, 1987), intermediate disturbance (Connell, 1978; Salo et al., 1986), random walk (Hubbell & Foster, 1986), and distribution barriers caused by rivers (Hershkovitz, 1968; Salo et al., 1986; Capparella, 1988; Ayres & Clutton-Brock, 1992; Haffer, 1992) ome researchers have stressed the role of edaphic specialization of plant species in promoting beta-diversity (Gentry, 1981; Young & León, 1989; van der Werff, 1992; Kalliola et al., 1993; Tuomisto & Ruokolainen, 1994; Tuomisto & Poulsen, 1996), but others have argued that the evidence is not yet sufficient to distinguish between random dispersal and edaphic influences (Condit, Evidence supporting the ecological differentia- tion model is accumulating, however. The Lindsaea results documented above show that there is sys- tematic floristic variation within the rainforest that can be explained by edaphic specialization of plant species and their differentiated occurrence in dif- ferent habitats. The pronounced variation in pteri- dophyte species richness within the Sucusari tran- sect shows also that alpha-diversity сап vary considerably among adjacent sites in a predictable manner, even in the absence of physical dispersal barriers. Furthermore, the relationships that were found between local diversity and soils in the plant groups dealt with here (Lindsaea, Thelypteris, and the pteridoflora in general) seem also to hold for these groups at the continental scale (Tuomisto & Poulsen, 1996). This underlines the difficulty in ex- plaining these patterns by chance alone (cf. Linhart & Grant, 1996). ESTIMATING SPECIES DIVERSITY If one wishes to use measured alpha-diversities (species diversity within habitat) to yield estimates of gamma-diversity (species diversity at the region- al scale), it is especially important to have a good estimate of beta-diversity (habitat diversity). Re- ported gamma-diversity (known number of species) in western Amazonia is not strikingly high in re- lation to the large area 2. The Amazonian lowlands of Ecuador and Peru together cover an area of almost 600,000 km’, b Шш; 3100 flow- ering plants are known from Amazonian Ecuador Renner et al., 1990) and 7000 from Amazonian Peru (Brako & Zarucchi, 1993). However, the three most species-rich l-ha tree plots in the world are — all situated in this region (Cuyabeno: Valencia et al., 1994; Mishana and Yanamono: Gentry, 1988), and they have about 300 tree species each (> 10 cm DBH). What is the explanation for this discrepancy be- tween spectacularly high alpha-diversity and much more everyday gamma-diversity? Is the forest so homogeneous that species are hyperdispersed, with beta-diversity being very low, or are so many of the existing habitats unknown that a high proportion of the species have remained undiscovered because their habitats have never been inventoried? Some species have certainly escaped discovery, which is obvious from the fact that new plant spe- cies are continuously being described from Ama- zonia. In the present study, 2 probably undescribed species were found among the total of 11 Lindsaea species, which increases the number of known spe- cies by 2296. For several reasons, this may not be a very good estimate of the overall proportion of Volume 85, Number 1 Tuomisto Satellite Imagery and Biodiversity Patterns undescribed plant species. The fern flora of the Iquitos region is already relatively well collected and known, so in less well-known areas the pro- portion of undescribed species may be higher. This assumption is supported by the observation that the sites where the two undescribed species were found have not been included in earlier inventories. Fur- thermore, fern species are generally widespread and relatively well known in comparison with most other plant groups, and therefore the proportion of undescribed species is probably higher among flow- ering plants than ferns. Finally, different genera must hide different proportions of undescribed spe- cies because of differences in their taxonomic com- plexity and the unequal taxonomic attention that has been paid to them, and we do not know whether Lindsaea gives an overestimate or an underestimate of the average among ferns. What the Lindsaea and other fern data (Tuomisto & Poulsen, 1996) do show quite conclusively is that the species are not hyperdispersed; instead, their distributions reflect edaphic conditions. To some degree similar behavior has been shown among trees, but the tree data are less conclusive because tree sample sizes have been too small to represent the local flora well (e.g., Duivenvoorden & Lips, 1993, 1995; Tuomisto et al., 1995; Ruokolainen et al., 1997). This being the situation, it can be asked how representative the existing herbarium collections are of the overall regional flora. Only a few sites have been studied intensively enough to warrant the claim that their floras are well known, and even though species typical of the vegetation types found at these sites would be well represented in herbar- ia, species typical of other vegetation types may be entirely absent. Furthermore, many of the existing data come from general collecting trips, which are concentrated along rivers and roads for obvious rea- sons of accessibility (Renner et al., 1 ‚ SO ri- parian and pioneer species are probably well rep- resented, while species of the forest interior may be much less collected. Other reasons for missing species include preferred sizes of the plants (shrubs are easy to collect whereas lianas are not) and pre- ferred seasons (plants that are not fertile during the collecting trip are ignored). Also, species that have showy flowers and long flowering or fruiting seasons may be collected with a high frequency, while spe- cies with infrequent or inconspicuous flowering these sources of error need to be controlled if reliable estimates of species numbers are to be obtained. It is a general problem in large-scale biodiversity studies in Amazonia and elsewhere that the avail- tend to go unnoticed. A able herbarium data are not presence-absence data, but rather presence-only data. If a species was found and collected at a given site, it is docu- mented as a herbarium specimen. But if a species was found but not collected at the site, no record of it remains. Therefore it is impossible to distin- guish between real absence of a species and ap- parent absence due to non-collection, which has led to serious biases in ras biodiversity centers is problem can only be solved by dise sampling efforts that use study plots or other quantifiable methods that provide comparable data for the different sites. in Amazonia (Nelson et al., HOW MANY KINDS OF RAINFOREST ARE THERE? In general, earlier studies have discussed three habitat types within tierra firme, differentiated by whether the soil is sand, loam, or clay (Tuomisto et al., 1995; Tuomisto & Poulsen, 1996; Ruokolainen et al., 1997). The same basic division is used in the present paper, with some additional variation being recognized within each of the three main ypes. The next question is, how well does this rep- resent the variation found in the region? In other — words, can we assume ” to be a reasonable estimate of habitat diversity, or should we expect to find many more habitats if more sites were in- ventoried? This is a crucial question for biodiver- sity assessments, but an answer cannot be obtained just by field inventories: the huge amount of work involved especially in tree sampling makes it im- possible to establish enough plots to obtain conclu- sive answers by field surveys alone. Satellite imagery can efficiently be used to target field inventories, because it reveals both the exist- ing patterns in the landscape, and the easiest ways to access each of the landscape types. Thereby field sampling can be planned so that the amount of ef- fort invested remains reasonable, while the amount of landscape variation that is covered by field in- ventories is maximized, and unnecessary repetitive sampling within the same landscape type is mini- zed. Satellite images give a clear impression of wide- spread habitat heterogeneity: on the basis of Land- sat TM images we have estimated that many more than a hundred biotopes exist in Peruvian Amazo- nia alone (Tuomisto et al., 1995; see also Kalliola et = 1991; Räsänen et al., 1993; Tuomisto et si . The exact number of vegetation types ca never P Pis d counte (cf. Webb, 1954; Webb et al., 1970; n, 1985), but the impor- tance of meds some сше for the number of habitats and the degree of floristic difference among 60 Annals of the Missouri Botanical Garden them can hardly be overestimated in studies that aim at assessing total biodiversity of a given region (see, for example, the discussion in Campbell et al., 1986). The number of habitats can be estimated from satellite imagery, but the degree of difference between them can only be established by sur- veys. Both components need to be known in order to estimate how high beta-diversity really is. Gamma-diversity is essentially the product of al- pha-diversity and beta-diversity, so it should be ob- vious that it cannot be reliably estimated unless both of these components are satisfactorily known. Given the current world-wide interest in biodiver- sity, it is surprising how little attention the problem of estimating beta-diversity has attracted. t is noteworthy that, in spite of their clear dif- ferences, the 23 fern transects reported in the pres- ent paper are according to satellite imagery situated in the most uniform part of Peruvian Amazonia (PAUT, 1993; Казапеп et al., 1993; Tuomisto et al., 995). The reason that the study was initiated there rather than in one of the more heterogeneous areas is that complete satellite imagery was not available at the time when the work was started, and the northern part of the country was the only area where we knew of any edaphic variation at all in the tierra firme forests: we were attracted there by the famous white-sand forests of Iquitos (Gentry, 981; Encarnación, 1985). Since then, it has grad- ually become obvious that the white sands form only a very small part of the ecological spectrum in the area, and that most of the variation is actu- ally found within the forests on non-sandy soils. Landsat images show both general regional patterns and detailed local patterns, all of which can be postulated to represent ecological and floristic variation in the forest. Obviously only a minute part of the variation has been field verified, but to date we have discov- ered nothing that would contradict this interpreta- tion. It is interesting to note that changes between the biotopes may take place gradually over long distances or more abruptly, and regional variation is found in the relative abundances + | kinds of ecotones (Tuomisto et al., 1 Within Peruvian Amazonia, ifferent 95). In some inundated areas the vegetation 4. are elon- gated in shape and have a uniform general orien- tation, in others they are narrow and aligned ac- cording to the river courses. In tierra firme areas, large smooth-edged patches are typical in northern eru, while smaller and more abrupt patches are frequent in the central to southern parts of the country (Tuomisto et al., 1995). Such differences in landscape structure may have important implica- 1992; Taylor et tions for the biota (Dunning et al., al., 1993), but they have not been paid attention to in either the planning or the interpreting of ecolog- ical, floristic, and biodiversity studies in Amazonia. GEOECOLOGICAL CONSIDERATIONS Western Amazonia has been the scene of a wide variety of geological events during different eras, which has resulted in considerable heterogeneity of terrain at different hierarchical levels (Salo & 1989; PAUT, 1993; 1993). For example, large parts of Peruvian Amazonia ünen, üsünen et al., have been influenced by sea incursions and fluvial dynamics since the late Cretaceous, and therefore the region consists of a mosaic of edaphically and geomorphologically different areas (Salo et al., 1993; Riisiinen et al., 1987, 1992, 1995). Soil characteristics such as nutrient content, texture, and water permeability are determined, 6; Hoorn, among other things, by the geological formations from which the soils are derived, and by the length of time they have been subject to weathering. ecause of the apparent edaphic specificity of many plant species and habitat types, geological formations with special geochemical characteristics and different ages are especially interesting from ecological and biogeographical points of view. Ex- amples include the Pastaza fan with its Holocene (younger than 10,000 years) volcanoclastic material 1990, 1992) and t mation with its marine or brackish licen from the Miocene (Hoorn, 1993; Räsänen et al., 1995). Both of these formations can give rise to soils that (Rüsánen et al., e Pebas for- are chemically unlike anything else in Amazonia and can therefore be expected to harbor edaphi- cally specialized plant endemics. It is noteworthy in this context that the Pastaza fan area has been designated an uninteresting area for biodiversity conservation by a workgroup that selected priority areas on the basis of known en- demism and diversity centers (Workshop 90, 1991). ery few biological specimens have been collected in the Pastaza swamplands, which has resulted in a low number of known species. However, the geo- logical characteristics of the area suggest that it is ecologically unique and should be prioritized in conservation planning (Kalliola et al., 1996). At the very least, the area should be given special atten- tion when biological collection trips are planned in the future. CONCLUSIONS Satellite imagery can be efficiently used in rain- forest studies to recognize different habitats and to map their extent even in areas that are difficult to Volume 85, Number 1 Tuomisto Satellite Imagery and Biodiversity Patterns get to in the field. This provides an unparalleled tool for studies whose aim is to reach regional con- clusions on species diversity. For this purpose, all the habitats that are recognized in satellite imagery have to be field-documented in order to verify to what degree they are floristically distinct, and to quantify species diversity within each of them. Nei- ther local nor regional species diversity can be read directly from satellite imagery, but once the species composition of each of the different habitats has been clarified, diversity estimates for unvisited ar- eas can be obtained by using satellite imagery to correlate them with one of the already field-docu- mented habitat types. Indicator species can be used with great success to facilitate recognizing floristically different habi- tats. Thereby they can also be used to predict dis- tribution and diversity patterns of other plants and animals, but only when it has been clarified by field surveys how these relate to the defined habitats. It is not possible to predict the diversity of such or- ganisms for which this background information is lacking, because different plant and animal groups can show opposing diversity patterns. Only about a dozen vegetation types are usually recognized in Amazonia, and consequently ecolog- ical research results are often generalized as rep- resentative of “the tropical rainforest." Satellite im- age analyses show that the extent of heterogeneity in Peruvian Amazonia is such that extrapolations of field results are not warranted without more de- tailed vegetation mapping, and there obviously is a great need for well-planned work in this field. It is almost ironic that the digital phase in satellite im- age analysis can be reduced to running a spectral enhancement and printing a hardcopy of the result, which can be accomplished by an experienced an- alyst in a few hours. Thereafter, it can easily take an experienced botanist a lifetime to finish the fieldwork needed in order to find out what the dif- ferent color patterns really mean in terms of the diversity of habitats and species. Literature Cited Austin, M. P. 1985. Continuum concept, рм meth- ods and niche theory. Annual Rev. Ecol. Syst. 16: 39— l. Ayres, J. M. & T. H. Clutton- А 1992. River bound- in Ámazonian primates. 537. Balslev, H., J. Luteyn, B. Vllgaard € L. B. Holm-Nielsen. 1987. Composition and structure of adjacent unflooded and floodplain forest in Amazonian Ecuador. Opera Bot. : 37-57. Brako, L. & J. L. Zarucchi. 1993. Catalogue of the Flow- ering Plants and Gymnosperms of Peru. Monogr. Syst. Bot. Missouri Bot. Gard. 45: 1— a D. G., D. C. Daly, С. T. Prance & U. N. Maciel. 6. Quantitative ecological inventory of terra firme and varzea tropical т. оп пе Rio Xingu, Brazilian Amazon. Brittonia 38: 369-39: le A. P. Сы. variation in neotropical birds: Implications for the speciation process. Acta Congr. E Ornith. 19: 1658-1664. Clark, D. 1998. Deciphering landscape mosaics of neotropic А, trees: GIS and systematic sampling provide new views of tropical rainforest diversity. Ann. Missouri ard. 85: 18—33 Condit T 1996. Defining and mapping vegetation pe miga- -diverse tropical forests. Trends Ecol. Evol. Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs. Science 199: 1302-1310. Duivenvoorden, J. F. & J. M. Lips. 1993. ге cology of the Middle Caqueta Basin. Explanatory notes to the maps. Studies on the 2 јео Ш А, Tropenbos Colombia. « 199 A land-ecological study of soils, vegetation, me plant diversity in Colombian Ama- zonia. Tropenbos Series 12, 1—438. Dunning, J. B., B. J. Danielson & H. R. Pulliam. 1992. Ecological processes y ae populations in complex landscapes. Oikos 65: 5. Encarnación, F. 1985. ои а la flora y vegeta- ción de la Amazonia peruana: Estado actual de los es- tudios, medio natural y ensayo de una clave de deter- audi e las formaciones vegetales en la llanura amazónica. Candollea 40: 237-252. Gentry, A. H. 1981. Distributional patterns and an ad- ditional species of the Passiflora vitifolia complex: Am- azonian species diversity due to бере or ated communities. Pl. Syst. Evol. 137: 95- . 1988. Tree species richness of upper сап forests. Proc. Natl. Acad. Sci. U.S.A. 85: 156-159. Haffer, J. 1969. Speciation in Amazonian forest birds. Science 165: "i 1-137. — Jn the “river effect” in some forest birds of southern Amazonia Bol. Mus. Paraense Emilio Goel- di 8: im me's cycle and time's arrow in the his- ш? W қаға Biogeographica (The Hague) 69: 15- Harris, R. 1987. pra re ең Ап Introduc- ig Routledge & Keg w York. Hershkovitz, P. 1968. = 1. ог apr educa of evolutionary 25) іп 1 mammalian tegumentary colors. Evolution 22: 556— Hoorn, C. 1993. Maud incursions and the influence of Andean tectonics on the Miocene depositional history of northwestern Amazonia: Results of a palynostrati- graphic study. Palaeogeogr., Palaeoclimatol., col. 105: 267—309 Hubbell, S. P. & R. B. Foster. 1986. Biology, chance, and history and the structure of tropical rain forest tree com- munities. Pp. 314-329 in J. Diamond & T. J. Case (ed- itors), Community Ecology. Harper and Row, New York. INRENA. 1995. Mapa forestal del Perú 1: 1,000,000. a de Agricultura, Dirección General Forestal, Lim Palaeoe- Kalliola, R., H. Tuomisto & K. Ruokolainen. 1996. Areas mportantes para la conservación de la selva baja per- uana desde el punto de vista geoecológico. Pp. 127— 32 in L. O. Rodriguez (editor), Diversidad Biológica 62 Annals of the Missouri Botanical Garden del Peru: Zonas Prioritarias para su Conservación. GTZ & INRENA, Lima. Linna, M. Puhakka, J. Salo & M. Казапеп. 1993. Mineral nutrients con fluvial sediments in the Peruvian Amazon. Catena 20: 333-349. У ‚Н. Tuomisto € К. Ruoko- linen 1991. The dynamics, distribution and classifi- ation of swamp vegetation in Peruvian Amazonia. Апп. Bot Fenn. 28: 5 239. Kramer, K. U. 19 A revision of the genus Lindsaea in the New World. North- Holland Publishing, Amsterdam. Lillesand, T. M. & R. W. Kiefer. 1994. Remote Sensing and Image Interpretation, 3rd ed. John Wiley & Sons, New York. Linhart, Y. B. & M. C. Grant. 1996. Evolutionary signif- icance of local genetic differentiation in plants. Annual Hev. Ecol. Syst. 27: 2 Mapa forestal del Peru a ex- plicativa). Universidad Nacional Agraria, Lir Mather, P. M. 1987. Computer Processing of ¡E Sensed ен An Introduction. John Wiley & Sons, New 31-277. Nelson, В. W., C. A. Ferreira, M. F. Da Silva & M. I Kawasaki. 1990. Refugia, endemism centers and col- 4. density in Brazilian Amazonia. Nature 345: 714-716. шз id royecto Amazonia de la Universidad de Turku). 3. Mapa geoecológico de selva baja de la Amazonia peruana. PAUT, Helsinki. Prance, G. T. (editor). 1982. Biological Diversification in the Tropics. Columbia Univ. Press, New Yor . 1989. American tropical ten Pp. 99- 132 in H. Lieth & M. J. rger (editors), Ecosystems of the World 14B. Tropical Rain Forest Ecosystems: Biogeo- graphic al and Ecological эша. Elsevier, Amsterdam. Renner, 5. S., Н. Balslev . B. Holm- Nielsen. 1990. Flowering plants of ж еы Ecuador—A checklist. AU Reports 24: 1 fe Run K., A. Linna & H. Tuomisto. 1997. Use of Melastomataceae and ae for revealing phy- togeographic patterns іп Amazonian rain forests. J. Trop. Ecol. 13: 243-256 Räsänen, M., R. Kalliola n M. Puhakka. 1993. Mapa c de la selva me peruana: Каре aciones. Pp. 207-216 іп R. Kalliola, M. Puhakk . Danjoy (edirs) Amazonia Peruana: . ión Ийе Ттор- ical en el Llano Subandino. PAUT and ONERN, Jyväs- kyla, Finland. . J. S. Salo € R. J. Kalliola. 1987. Fluvial per- in the Western Amazon basin: Regulation by и term sub-Andean tectonics. Science 238: 1398— 1401. кеі . M. Linna, J. С. R. Santos & F. R. Negri. 1995. Late Mioc 'ene tidal deposits in the Amazonian foreland basin. Science 269: 386—390. — ———, R. Neller, J. Salo € H. Jungner. 1992. Recent and ancient fluvial de ИТУ Cun ms in > Amazo- nian а 4. eol. 2: 293-300 ——— 555 H. Jungner a 7 Romero 1990. . f the Western Amazon lowland relief: Impact of Andean foreland dynamics. Terra Nova 2: 32 1989. an А | landscape . B. Holm- . Казапеп. patterns in western E Pp. 35 Nielsen, I. C. Nielse als fe . Tropical Forests: Botanical Da Speciation and Diversity. Academic Press, London. Kalliola, 1. Häkkinen, Y. Mäkinen, P. Nie- melä, M. Puhakka & P. D. Coley, 1986. River dynam- ics and the diversity of Amazon lowland forest. Nature 322: des 58. Taylor, P. D., L. Fahring, K. Henein & G. Merriam. 1993. DU is a vital 1. of landscape structure. Oikos 68: 57 a 7: Townshend, Justice & V. Kalb. 1987. Characterization к е ‘ation of South American land cover г using satellite data. Int. J. Remote Sensing 8: 1189-1207 Ty M. & R. G. Stolze. 1989-1994. Pte pne of Peru. Fieldiana, Bot. n.s. 20, 22, 27, 29, 32, Tuomisto, H. & A. D. Polen 1996. Influence E va ic specialization on pteridophyte distribution in neo- tropical rain forests. J. Biogeogr. 23: 283—293. ---« Ruokolainen. . Distribution of Pteri- dophyta and Melastomataceae along an edaphic gradi- ent in an Amazonian rain forest. J. Veg. Sci. 5: 25-34. — A. Linna & R. Kalliola. 1994. Use of digitally proc d d images in studies of tropical rain for- est vegetation. Int. J. Remote Sensing 15: ү 5-1610. ————,K. Ruokolainen. R. Kalliola, A. Linna, W. Danjoy 1995. Dissecting 2. biodi- 269: 63-00. Valencia, R., Н. Balslev & С. y Miño. 1994. High tree alpha-diversity in 121. Ke uador. Biodiversity & 2. 3: 21-28. Webb, . 1954. Is the classification of plant a B either possible or desirable? Bot. Tidsskr. 5 -370. Webb, L. С. Tracey, W. T. Williams € G. М. 1970. Studies in the ш analysis of соп e rain-forest communities. V. А comparison at the prop- erties a es stic and а structural data. J. Ecol. 58: 203-232. Werff, И. van der. 1992. Substrate preference of Laura- ceae s ferns in the Iquitos area, Peru. Candollea 47: 11-2 W bu T. C. & G. T. Prance. 1987. Biogeography and 4 History іп Tropical America. Clarendon Press, Oxford. Workshop m 1991. Biological Priorities for Conservation in Amazonia. Map p Tn ed by Conservation Interna- . oe Pteridophyte species di- versity in the ce entral Peruvian Amazon: хы e of edaphic specialization. Brittonia 41: 388-39: SELECTED PROCEEDINGS FROM THE 1997 MIDWESTERN RARE PLANT CONFERENCE: INTRODUCTION Kayri Havens' The Center for Plant Conservation, headquar- tered at the Missouri Botanical Garden, is a na- tional network of 28 botanical gardens and arbo- reta, each of which is responsible for preserving rare plants within its biogeographical region through protective cultivation and seed banking. The Missouri Botanical Garden is a member of this network, thereby engaged in protecting some of the rarest plant species from Missouri and five nearby states. Аз part of the Garden's commitment to re- gional rare plant conservation, it hosted the first Midwestern Rare Plant Conference in February 1997. The conference will be a biennial event, to be held at one of five regional botanical gardens and arboreta on a rotating basis. The 1997 meeting provided a forum for exchanging research results on rare Midwestern plants, for setting regional plant conservation priorities, and for developing and im- plementing collaborative plant conservation pro- jects in the Midwest. The meeting consisted of two days of scientific paper and poster presentations followed by a one day conservation task force meet- ing organized by the Center for Plant Conservation. Topics covered at the 1997 Midwestern Rare Plant Conference included population biology and genetics, reproductive biology, demography, sys- tematics, autecology, and monitoring and recovery strategies for rare taxa from the upper and lower Midwest as well as the Great Plains. The following collection of papers are a subset of the over 35 invited and contributed presentations given at the meeting. Two of the papers focus on a rare grass, Calamagrostis porteri subsp. insperata. Because the taxon consistently fails to set viable seed, Havens and Holland examined genetic and environmental factors affecting its reproductive failure. In a field study, Bittner and Gibson looked at the taxon's veg- etative performance in relation to a number of en- vironmental variables. Three of the papers focus on the population genetics of rare species. Crawford et al. studied genetic variation in Trifolium stolonifer- um using randomly amplified polymorphic DNA RAPD) markers. In contrast to an earlier allozyme study, they found that even small populations con- tained a number of different genets and are worthy of conservation consideration. Baskauf and Snapp examined isozyme variation in the highly restricted PM cedar glade endemic Astragalus bibullatus. They found little differentiation between populations, suggesting that local adaptation may not be a crit- ical concern when selecting propagules for reintro- duction projects. In the self-incompatible prairie forb Asclepias meadii, Tecic et al. compared allo- zyme variation between sites managed by mowing and by fire. They found less variation in the mowed populations, perhaps because annual summer mow- ing prevents sexual reproduction. In a second paper on Asclepias meadii, Bowles et al. look at habitat differences between populations, discuss how fire and mowing affect population structure, and ex- amine a number of factors that affect outplanting success in restoration projects. Finally, a paper by Clinebell and Bernhardt looks at the pollination ecology of five tallgrass prairie Penstemon species. They found that the assemblage of pollinators pres- ent at a site varied with population size. This may have interesting implications for patterns of gene flow in natural and reintroduced populations. The conference and task force meeting was made possible through the generosity of the following sponsors and co-organizers: Center for Plant Con- servation, Chicago Botanic Garden, Missouri Bo- tanical Garden, Missouri Department of Conserva- tion, The George Gund Foundation, The Holden Arboretum, The Morton Arboretum, The Nebraska Statewide Arboretum, and the U.S. Fish and Wild- life Service, Region 3. ! Manager of Endangered Plant Research, Chicago Botanic Garden, 1000 Lake Cook Road, Glencoe, Illinois 60022, U.S.A. ANN. MissoURI Bor. GARD. 85: 63. 1998. FACTORS AFFECTING REPRODUCTIVE SUCCESS IN A RARE GRASS, CALAMAGROSTIS PORTERI SUBSP. /NSPERATA! Kayri Havens? and Douglas L. Holland? ABSTRACT MR dn n A. Gray subsp. ip dies (Swallen) € 'species of co " by the U.S. Fish and Wildli genotype, pollen source, and Maternal genotype and pollen source (se тА vs light intensity had no significant e the 2000+ florets we C. W. Greene is a rare grass that is currently treated as a fe Service. Ра ен of viable seed had never been observed, and little was ks Sion the factors affecting this reproductive failure. In this study ' we examined the effects of maternal light intensity on caryopsis (the single-seeded fruit РА in grasses) production. . outcross) significantly affected the numb wever, nearly all examined, 4. опе > fully filled, viable caryopsis was found. The conditions under which this r of caryopses initiated, while of the caryopses initiated 42 late in development. In caryopsis was dente suggest future courses of management. Calamagrostis porteri А. Gray subsp. insperata (Swallen) C. Greene is a rare grass currently known from approximately 80 populations in five Mid- western states: Illinois, Indiana, Kentucky, Missou- ri, and Ohio. It was discovered in 1934 in Jackson Co., Ohio, and was originally described as C. in- sperata (Swallen, J. Wash. Acad. Sci. 25: 413. ). It was formerly a Category 2 federal candi- date for listing as endangered or threatened due in part to its extremely patchy distribution throughout its range (Drewry, 1993), and is now, like all former Category 2 candidates, called a “species of con- cern" by the U.S. Fish and Wildlife Service (Drew- ry & Sayers, 1996). It grows primarily in forest T and along edges of upland woods (Spooner et al., 2 porteri subsp. insperata was ге- ported by Greene (1980) to be octoploid (2n — 56), although a related species, Calamagrostis canaden- sis, also 2n = 56, is reported to be tetraploid and has allozyme banding patterns typical of those ex- pected in tetrasomic inheritance (Macdonald & Lieffers, 1991). Diploids (2n — 14) are not known in Calamagrostis, indicating that the genus is quite old and polyploidization occurred early in its evo- lutionary history (Greene, 1980). Although apomix- is is prevalent in the genus, Greene (1984) postu- lated that Calamagrostis porteri subsp. insperata reproduced sexually due to the presence of sexual megagametophytes produced from reduced mega- spores, as well as normal microsporogenesis and pollen formation. However, it has never, to our knowledge, been observed to set viable seed, which probably contributes to its rarity. Several hypothe- ses have been proposed to explain this reproductive failure. First, encroachment of woody vegetation and subsequent shading may inhibit flowering. Flowering tillers are very rare in most populations and are most often seen on plants in relatively sun- ad areas 2. 1984; Bittner & Gibson, 1993). ond, C. 1 subsp. insperata may be self- incompatible мы: e, 1980), and if so, reproduc- tion may be limited by lack of compatible mates. any populations consist of only a few plants, which could be vegetative clones, given that C. por- teri can reproduce asexually via rhizomes (Greene, 1984). In addition, many plants lack viable pollen, as estimated by stainability with cotton blue (Greene, 1980). Information regarding reproduction of many rare or endangered taxa, including C. porteri subsp. in- sperata, is scanty, but vitally important for manage- ment and recovery plans. The purpose of this study was to investigate the effects of light intensity, pol- len source, maternal genotype and pollen viability on reproductive success іп Calamagrostis porteri This NS dank ( alg Greene, S. Ellen M on the project Bade rd manuscript. thor for corresponc address: Chicago Botanic Garden, ence. Missouri Botanical Garden, P.O. Box 299, St. Louis, 1000 Lake Cook Road, Glencoe, Illinois 60022, U.S A study was Viande by funds from the Missouri Botanical Garden and the Center for Plant Conservation. The Ellen Macdonald, George Yatskievych, and Brian Stone for their thoughtful comments Missouri 63166, U.S.A. Present ‚А. * Missouri Botanical Garden, Р.О. Box 299, St. Louis, Missouri 63166, U.S ANN. MISSOURI Bor. GARD. 85: 64-68. 1998. Volume 85, Number 1 1998 Havens & Holland Reproductive Success in Calamagrostis Table 1. Provenance (county and state of collection, voucher number and location, and Missouri Botanical Garden accession number), treatment, and pollen viability for the four genotypes used. Genotype Provenance Treatment Pollen viability 1 Jackson Co., OH, Subpopulation С Sun—“Female” 0% Voucher: Weiland + (МО) Accession # 9017 Shade—“Female” 95% 2 Jackson Co., OH, Subpopulation B Sun— "Female" 9896 Voucher: Weiland "i (MO) Accession # 9017 Shade—"Female" 2096 3 Vinton Co., OH Sun—"Female" 99% Voucher: Rogers 345 (МО) Accession # 872392 Shade—“Female” 98% 4 Texas Co., MO Sun—“Male” 99% Voucher: Summers 3631 (MO) Accession # 902331 subsp. insperata. We examined caryopsis (single- seeded fruits produced in grasses; functionally analogous to seeds in most other species of flow- ering plants) production and viability in two differ- ent light regimes and with three pollination treat- ments: self-pollination, natural crossing, and hand-outcrossed pollinations. MATERIALS AND METHODS Specimens of Calamagrostis porteri subsp. in- sperata were field-collected in 1987 and 1990 from Jackson and Vinton Cos., Ohio, and Texas Co., Mis- souri (Table 1). The plants are maintained in pro- tective cultivation at the Missouri Botanical Garden as part of its cooperative agreement with the Center for Plant Conservation. The Center for Plant Con- servation is an organization dedicated to preserving rare plants in the United States primarily through ex situ cultivation and seed banking. Only four genotypes were included in this ex- periment. These were the only genotypes that flow- ered in the ex situ collection. Since viable seed production has never been observed in natural pop- ulations, growing plants from seed was not an op- tion. Three sets of two genetically identical clones were used as pollen recipients (hereafter “females”; it should be noted that although plants are her- maphroditic, we are referring only to their func- tional gender in this study). The female plants were collected from three different populations or sub- populations in Ohio. The three clumps of tillers were divided and potted separately to produce the pairs of clones. The pollen donor (hereafter *male") was collected in Missouri and is presumably ge- netically distinct from all of the females. All plants were grown in a shaded lath house until flowering culm initiation. Upon flowering, in mid June 1994, one member of each pair of clones was placed in full sun, the other member remaining in the shaded lath house for the duration of the experiment. In- florescence number varied on the female plants from 1 to 65. Two to four of the inflorescences on each female were bagged with nylon stockings prior to anthesis to test for self-compatibility. Approxi- mately half of the remaining inflorescences were hand-pollinated by shaking one of the male's inflo- rescences over the female inflorescence. The remaining inflorescences were not manipulated and allowed to outcross or self-pollinate naturally. Hence, each female plant received three treat- ments, except one plant from genotype #1 that had only one flowering culm, and it received the hand- pollination treatment. Inflorescences were harvest- ed 22 August 1994 after they had turned brown. Approximately 100 florets were randomly selected and removed from each inflorescence. These florets were scored for the presence or absence of cary- opses and, when present, their developmental stage. To assess viability, caryopses were placed in petri dishes on moist filter paper in an incubator at 22°С under long day conditions (16/8 а d monitored for germination for one mont of the caryopses from each treatment were given a cold treatment (12 weeks at 4^C) prior to incuba- tion. All plants in the study were tested for pollen viability using Alexander's stain (Alexander, 1969). Pollen grains that stained dark red were presumed to be viable. SYSTAT 5 computer software (Systat, Inc., SPSS, Chicago, IL) was used for all the statistical analy- ses. А three-way mixed model ANOVA was used to 66 Annals of the Missouri Botanical Garden Micrograph of aborted, non-viable caryop- Figure 1. A partition variance in the number of caryopses ini- tiated into components due to genotype, a random effect, and pollen source and light treatment, both fixed effects. Univariate analyses were done using one-way ANOVAs with post hoc Tukey mean com- parisons after significant differences were found in the multivariate analysis (Dowdy & Wearden, 1983). RESULTS All but one plant produced viable pollen. Viable pollen was not observed in any of the approximately 20 anthers examined in one of the plants of geno- type #1 (sun treatment). A second plant (genotype #2, shade) produced only approximately 20% stain- able pollen. The rest of the plants, including the plant used as a male for hand-pollinations, pro- duced abundant, viable pollen. Pollen stainability for these plants ranged from 95 to 100% (Table 1). Some caryopses were initiated in all plants; how- ever, nearly all of these were shriveled at the basal end and did not germinate (Fig. 1). These caryopses contained endosperm and were easily distinguished from the uniovulate ovaries that did not initiate fruit development. Only one fully filled caryopsis was found, and it germinated after three days on moist filter paper without pretreatment. It was found on the plant that also initiated the most caryopses (ge- notype #2, sun, hand-pollination). Caryopsis initiation varied within and among plants. One genotype (#3) was less successful in Table 2. Results of ANOVA showing the magnitude of variance in number of caryopses initiated attributable to maternal genotype, pollen source, and light treatment. Mean Source d.f. square F-ratio Probability Genotype 2 75123 33.0 P < 0.001 Pollen source 2 5511.2 24.2 P < 0.001 Light treatment || 57.0 0.3 n.s. Error 26 227.4 initiating caryopses across all treatments. There were also significant differences between the pol- len-source treatments. Both hand- and open-polli- nation treatments initiated significantly more cary- opses than the selfing treatment. Light intensity did not have a significant effect on the number of cary- opses initiated (Table 2, Fig. 2). DISCUSSION Reproductive failure in plants can occur at any of several serial stages: flower production, gameto- phyte development, pollination, fertilization, and seed/fruit maturation. In Calamagrostis porteri subsp. insperata, limited success in many of these stages contributes to the extremely low seed set overall. In native populations, flowering tillers are very rare and most often seen in areas of relatively high light (Bittner & Gibson, 1993). While light intensity does not appear to be important at the stages of seed initiation and development, our ex- periment did not address the effect of light intensity on flower production. All plants were grown in the semi-shaded lath house until flowering tillers were initiated; at that point, half of the plants were moved to full sun. Although our plants flowered in the lath house, light intensity in densely shaded woods may be low enough to preclude flowering in many populations. hable gametophyte production is also problem- atic in this taxon. This may be due to meiotic dif- ficulties caused by polyploidy. A shift toward asex- ual reproduction is frequently seen in polyploid species (Levin, 1983). Maternal genotypes varied significantly in the number of caryopses initiated. This probably reflects differences in megagameto- phyte production or viability, rather than pollen or resource limitation. Genotypic differences in cary- opsis initiation were consistent across all pollina- tion treatments including hand-pollinated plants which received abundant pollen, and all plants re- ceived adequate amounts of water and fertilizer. Genetic effects on seed set have been documented in several studies (Mazer et al., 1986; Geber, 1990; Volume 85, Number 1 1998 Havens & Holland 67 Reproductive Success in Calamagrostis FRUITS INITIATED (96) 1 2 3 GENOTYPE 3 Р = o E Ё 1 2 3 SELF OPEN HAND POLLINATION TREATMENT 50 E В < Е = o E 2 tr ш 1 2 LIGHT TREATMENT Figure 2. Mean number of caryopses (fruits) initiated (= SE) for the three maternal genotypes, the three polle treatments, and the two light treatments. Signific dif- ferences eem means are indicated by different letters (P < 0.0: Sultan, 1990). Problems with male gametophyte production were also evident; two of the seven plants produced little or no viable pollen. However, the variation in pollen production was not ex plained by genotype or light treatment. А chromosome counts were not done on our plants, differences in chromosome number between plants would not completely explain the variation seen in gametophyte production for two reasons. First, clones varied in their pollen production, and sec- ond, individuals that failed to produce viable pollen made viable female gametophytes and vice versa. This taxon also has limited success at the repro- ductive stages of fertilization, seed/fruit initiation, and seed/fruit maturation. Plants that potentially received outcrossed pollen, either from open pol- linations or from hand pollinations, were signifi- cantly more successful in initiating caryopses than plants that received only self pollen. Nevertheless, plants that were bagged to prevent outcrossing did initiate some caryopses. Although Greene (1980) postulated the taxon was genetically self-incompat- ible, observing caryopsis initiation in bagged inflo- rescences suggests the grass is self-compatible. Grasses that exhibit self-incompatibility typically have a two-locus, gametophytic system that causes pollen-tube growth to cease in the stigma, either when the tube first touches the stigmatic papillae or in the intercellular spaces of the stigma, prior to fertilization and seed initiation (Richards, 1986). The failure of most self-pollinated ovules to ini- tiate development as well as the extremely high rate of seed abortion in this taxon, even among out- crossed seeds, may be a reflection of a high genetic load. Although recent polyploids are not expected to carry a genetic load (Lande & Schemske, 1985; Hedrick, 1987), the long history of polyploidization in this taxon may have allowed a large genetic load to develop. A similarly high load has been docu- mented i in the tetraploid Vaccinium corymbosum L. (Polygonaceae; Wiens et al., 1989). As in V. cor- ymbosum an . eurekensis, we found embryo-lethal factors expressed in both selfed and outcrossed seeds, indicating dominant mutations may be present. Of the 2000+ florets scored, we found only one fully filled, viable caryopsis. Perhaps most signifi- cantly, this finding demonstrates that sexual repro- duction is possible in this taxon, although it is high- ly unlikely, especially in natural populations. In addition, the conditions under which this caryopsis was produced (pollen from another genotype, full sun) suggest courses of management for this rare 68 Annals of the Missouri Botanical Garden species. А genetic survey would provide valuable information on the amount and pattern of genetic variation present in the taxon. Ап experimental population would allow us to test the effects of cre- ating light gaps and introducing novel genotypes on flowering and seed set in a natural setting. Since seed banking is not currently a viable option for this taxon, ex situ conservation efforts are focusing on tissue culture and cryopreservation of vegetative plant parts, as well as investigating the feasibility of embryo rescue techniques. Literature Cited Alexander, M. P. 1969. Differential staining of aborted and non- Pu pollen. Stain Technol. 44: 117-122 Bittner, R. T. & D. J 1993. Distribution and ecology of бышыгын porteri е insperata (Swal- . Greene in southern Illinois. Report submitted to the Illinois Department of uud n y 5. « S. жшн 98: John Wiley & Sons ‚ New York. Drewry, G. 1993. Plant taxa for listing as endangered or threatened. species: Proposed rules. Fed. Reg. : 51154. апае s for Research. & R. Sayers. 1996. Endangered and threatened wildlife nut plants; Review of plant and animal taxa that are candidates for listing as 2. or threat- ened species. Fed. Reg., 50 CFR Part 17, 96: 4412. Geber, M 1990. The pe of meristem iion in Polygonum arenastrum: Negative genetic « between fecundity and growth. Evolution 44: 799-819. Greene, C. У. 1980. The Systematics of Calamagrostis (Gramineae) in и de America. Unpublished Ph.D. Thesis, Harvard University, Cambridge. 984. Sex Кя and apomictic repro au dE in Сайла ле, т eastern North A ca. Amer. J. 93. Hedrick, P. W. 1987. Се netic d and the mating system in homosporous ferns. Evolution 41: 1282-1289. Krebs, S. L. & J. К. Hancock. 1991. Embryonic genetic load in the Боран blueberry Vaccinium corymbosum (Ericaceae). Amer. J. Bot. 78: 1427-1437. Lande, W. Sc ет Ке. 1985. The evolution of self- (та вани and inbreeding depression in plants. 1 Genetic models. Evolution 39: 24—40. 983. Polyploidy and novelty in flowering plants . Naturalist 122: 1-25. Macdonald, S. E. & V = Lieffers. 1991. Population vari- ation, outcrossing, and colonization of disturbed areas by 22 canadensis Evidence from allozyme analysis. Ame . 78: 1123-1129. Mazer, S. J., A. n Snow 45 M. L. Stanton. 1986. Fertil- ization dynamics and parental effects upon fruit devel- Raphanus E _ nces for lati r. J. Bot. l. 1 Plant BU 5. George en & Unwin. . Қ, D. M., А. :usick, B. Andreas & D. Anderson. 983. је оп Ohi io vascular plants previously consid- m for listing as {еа endangered « pecies. Castanea 50-2 58. Sukan, S 1990. Evolutionary ae alions ж a nolypic plasticity: Genetic diversity for norms o source gradients in 2. Т ria . L. lished Ph. ~ г На da University, 72. Wiens, І)., Nickrent, C. I. Dior C. L. Calvin N. J. 15. 1989. ‘Developmental is and loss ы reproductive capacity in the rare palaeoendemic shrub Dedeckera жже ы Nature 338: 65-67. or threatened action to re- Unpub- MICROHABITAT RELATIONS OF THE RARE REED BENT GRASS, CALAMAGROSTIS PORTERI SUBSP. INSPERATA (POACEAE), WITH IMPLICATIONS FOR ITS CONSERVATION! R. Todd Bittner? and David J. Gibson? ABSTRACT Calamagrostis porteri subsp. insperata was known from only two extant populations in Ohio prior to the late 1980s. Recent searching has documented more than 80 populations of this rare grass in five states. Although more populations exist than previously believed, the habitat requirements for this subspecies are still unknown. To quantify these require- ents, data were collected on photosynthetic photo-flux density (PPFD), air and soil temperature, vapor-pressure deficits (VPD), soil moisture and depth, pH, percent organic matter, and associated species cover from three populations in southern Illinois. Leaf area, predicted total leaf area, and number of tillers per m? were measured to ascertain the relationship between vegetative performance and microenvironmental conditions. Detrended Canonical Correspondence Analysis (DCCA) was used to ordinate the samples using the cover of co-occurring species and the environmental variables. Leaf area, predicted total leaf area, and number E three sites. The vegetative performance n of environmental factors, of which PPFD, VPD. pe soil temperature are extremely influ ential, especially in the spring. The невине а shows that the three populations are distinctly separated, with total predicted leaf area, tiller. density, soil moisture, soil temperature, air temperature, a nd pH being highly correlated to the ordination axes. Habitat modifications dte in dirt d forest canopies detrimentally affect the vegetative performance of this taxon. Conservation biologists, in attempting to evaluate the causes of endangerment to rare plant species in order to aid in their recovery, have employed ecological research to characterize biotic interac- tions and habitat requirements (Schemske et al., 1994). In particular, an understanding of the aut- ecology and natural history of rare species is nec- essary (Brussard, 1991). A number of studies fo- cusing on environmental conditions have led to a better understanding of limiting ecological factors responsible for rarity of endangered species (Gaw- ler et al., 1987; Buchele et al., 1989; Boyd & Hil- ton, 1994; Vivian, 1967). Lack of basic ecological information about the former Federal endangered species candidate (FWS, 1993) Calamagrostis porteri А. Gray subsp. insperata (Swallen Greene (Poaceae) poses a problem to conservation biologists and land man- agers. Аз conservation measures are appropriately planned for this taxon, there is a need to better understand its microhabitat ecology. Calamagrostis porteri subsp. insperata (reed bent grass) was first described from Ohio as grostis insperata (Swallen, 1935). population, three Missouri populations, and one Ar- kansas population (now believed extirpated) were documented in subsequent years. This grass is now known from more than 80 populations in five states: Illinois, Indiana, Kentucky, Missouri, and Ohio (Campbell et al., 1992; Homoya, 19 1993; Bittner, 1995b). Its habitats are extremely varied but are often located on cool, north-facing sandstone bluff edges and tree-fall gaps in dry-me- sic upland forests found in unglaciated areas (Bitt- ner, 1995a). This cool-season, rhizomatous grass rarely flowers and spreads almost exclusively through vegetative growth. It is a tufted perennial that can stand up to 1 m tall. Highly specific flow- ering conditions, self-incompatibility, and poor fruiting success contribute to the reproductive lim- itations of several taxa of Calamagrostis, including 95; Summers, . porteri subsp. insperata (Greene, 19 ! This research was supported by the U.S. Forest Service, the Illinois Department of Natural Resources, and the Department of Plant Biology, Southern Illinois 115. at Carbondale. is Department of Natural Resources, IVC 61348. U. S.A. E. Campus Bld. 11, 815 N. Orlando Smith Road, Oglesby, Illinois 3 Department of Plant Biology, Southern Illinois University at Carbondale, Carbondale, Illinois 62901, U.S.A. ANN. MISSOURI Bor. GARD. 85: 69—80. 1998. 70 Annals of the Missouri Botanical Garden The goals of this study were to determine how microhabitat conditions and ground cover of asso- ciated taxa are related to the vegetative perfor- mance of Calamagrostis porteri subsp. insperata in three of the five known southern Illinois popula- tions (as of 1993), to contribute to the general in- formation about this taxon, and to aid in evaluating its endangered status. MATERIALS AND METHODS STUDY SITES The five Illinois populations of Calamagrostis porteri subsp. insperata are located in Pope County, southern Illinois, at Bell Smith Springs Ecological Area and Lusk Creek Canyon Natural Area (Fig. 1). АП the populations are within the Shawnee Na- tional Forest and are managed by the U.S. Forest Service under prescription 8.2 (to protect, preserve, and enhance the unique scientific, educational, and natural values found within this National Natural USDA, 1992). nly three of the five populations were sizeable Landmark and Ecological Area; enough for the purposes of this research. Two pop- ulations were from Bell Smith Springs [BSS1, Sum- mers 4774 (MO); BSS2, Bittner 347 (ILLS), 348 (SIUC)]. and one population was from Lusk Creek Canyon [LC, Bittner 350 (ILLS ulations were estimated to contain between 4000 and 18,000 tillers. This unglaciated, predominantly forested area lies in the Greater Shawnee Hills Section of the Shawnee Hills Natural Division (Schwegman, 1973). vanian sandstone strata, and the ad: is rug- ged, with many canyons, bluffs well-drained soils have low organic-matter content, moderate Б and moderate available wa- ter capacity (USDA, The study 8. of Calamagrostis porteri )]. These three pop- The bedrock consists of massive Pennsyl- s, and ravines. The subsp. insperata are located predominantly within the dry-mesic, upland forest community. The hab- itat of C. porteri subsp. insperata in these popula- tions is on cool, northwest- and northeast-facing bluff edges and hillsides with high species diversity and many rare, uncommon, and conservative plants (taxa with a high degree of habitat specificity) as associates. Tillers occur in both leaf-litter zones and moss- and lichen-dominated areas within dry- mesic, oak-hickory forests. Within two populations C and an unstudied BSS population), dense mats of Sphagnum species are present throughout large sections. Springs, intermittent streams, or seeps are also common within the study populations. The dominant overstory species association consists of Bell Smith Springs Zn Figure 1. The location of ОН porteri subsp. ша: E at Bell Smith Springs Ecological Area and Lusk Creek Canyon Natural Area in rone Coun- ty, оше гп Illinois. Quercus alba L., Q. rubra L., Q. velutina Lam., Car- ya glabra (Mill. Sweet, and C. ovata (Mill.) Koch (Voigt & Mohlenbrock, 1964) FIELD PROCEDURES In each of the three populations, fifteen 1-m? quadrats were located randomly and accepted if they contained tillers of C. porteri subsp. insperata. An additional four randomly selected 20-m tran- Volume 85, Number 1 1998 Bittner & Gibson Microhabitat Relations of Calamagrostis sects of contiguous 1-m? quadrats were established perpendicular to the bluff edge within the BSSI population (but did not necessarily contain tillers). e number of tillers in each quadrat was sam- pled four times during the growing season of 1993 (Арг, May, July, and September). To calculate mean leaf area per quadrat, the lengths of all the leaves from five randomly chosen tillers, the num- bers of leaves per tiller, and an estimated percent of living area per leaf were recorded twice (May and August). А regression developed from leaf trac- ings of 100 randomly selected leaves was used to predict leaf area from leaf length (R? = 0.94, P < 0.0001). Leaf area was calculated by multiplying the estimated leaf area (predicted from leaf length) by the percent living area per leaf. Mean leaf area per quadrat was calculated by multiplying the num- ber of tillers from the previous month by the mean number of leaves per tiller by the mean leaf area. Predicted total leaf area per quadrat was calcu- lated by multiplying the mean leaf area for the sea- son by the mean number of leaves per tiller by the mean number of tillers for the season. Tiller survivorship was determined at each of the three populations by marking new tillers with plas- tic tags monthly throughout the growing season from April through September in three randomly selected 1-m? quadrats per population. Both new tillers and dead tillers (> 5096 dead area) were recorded each month. determine which factors were related to the microdistribution of C. porteri subsp. insperata, en- vironmental factors were measured twice a day at monthly intervals (April-September) throughout the growing season from 8:00 to 10:00 A.M. an to 4:00 P.M. At the midpoint of each quadrat, light intensity was measured with a line-quantum sensor, model LI-191SA (LICOR, Lincoln, NB) attached to a data logger (LICOR model 1000). The line-quantum sensor averages photosynthetic photon-flux density (PPFD) through 400-700 nm Total PPFD measurements were made in full sunlight through- out the data-collection period and used as a control. Available PPFD within each quadrat was calculat- ed by dividing values for PPFD in each quadrat by control values to give a total light-intensity value. over 1 m. Air, soil, and wet-bulb temperatures were col- lected similarly to PPFD in the individual quadrats, and at 5-m intervals along the transects. The wet- and dry-bulb temperatures were used with baro- metric pressure obtained from the nearby Carbon- dale airport weather station (approximately 65 km NE of the study sites) to calculate vapor-pressure deficit (VPD) values according to Cox (1990). Within each quadrat, morning and afternoon ob- servations of environmental data (PPFD, air, and soil temperature, and vapor pressure deficits) were averaged for each month to obtain six monthly ob- servations. Several transformations were performed to normalize the data according to Sokal and Rohlf (1969). PPFD values were divided by 100 and arc- sine transformed prior to analysis (arcsine / (X * 100)). Vapor pressure deficit values were divided by 10 and arcsine transformed. Soil and air tem- perature values were log-base 10 transformed. Soil moisture was measured by collecting soil samples in May and July from each individual quadrat, and in every fifth quadrat within each tran- sect, following the procedures of Bannister (1986). by 100, arc- sine transformed, and averaged prior to analysis. Soil depth to bedrock was measured in each corner and midpoint for each quadrat with a steel probe, averaged, and log-base 10 transformed prior to analysis. Soil samples, previously dried for soil- moisture measurements, were used to determine the The soil-moisture values were divided pH using a 1:5 soil-water slurry, and organic con- tent using standard procedures (SPAC, 1992). Soil pH was converted to hydrogen-ion concentration for data analysis. The canopy cover of all vascular plant species rooted in each quadrat was estimated in July using a modified Daubenmire scale (Abrams & Hulbert, 1987). Moss and lichen cover was recorded, but identification to species was not made. The mid- point values were used for the species-cover data in an ordination, and for total-cover calculations for each sample. Arcsine-transformed midpoint values were used for statistical analysis. The identity of canopy species immediately above each quadrat was also determined; nomenclature follows Mohlen- brock (1986). ANALYTICAL METHODS The following analyses were performed individ- ually for each of the three populations using SAS, Version 6 (SASI, 1990). Simple and multiple re- gressions were performed to predict tiller densities, mean leaf area, and total predicted leaf area (per- formance) from air temperature, soil temperature, vapor-pressure deficit, PPFD, soil moisture, soil depth, pH, percent organic matter, and vegetation cover (environmental). For the simple and multiple monthly regressions, environmental variables (in- dependent variables) from the previous month and the current month were tested against the perfor- mance of C. porteri subsp. insperata (dependent variables) for the current month. Mean variables 72 Annals of th Missouri Botanical Garden 50 40 S 30 5 20 z ш Q 10 nc 3 = 0 %- * $ E -10 -20 FEB. MAR. APR. MAY JUN. JUL. AUG. SEP. MONTH —# Cumulative New Tiller Mean —0— Tiller Mean —0— Cumulative Dead Tiller Mean Figure 2. Net tiller density and cumulative number of new and dead tillers of Calamagrostis e oris insperata from В551, 8552, and LC from February to September 1993 (mean of nine quadrats from the three sites). were used for simple regressions. The pair of en- vironmental variables with the lowest probabilities (P) for each month and mean were used for the multiple regressions, and for a simple regression of their cross-products, respectively. Separate one-way ANOVAs were used to com- pare tiller presence/absence (dependent variable) at BSS1 for each month and mean using the envi- ronmental variables (independent variables). Simi- larly, environmental data from the previous month and the current month were tested against the per- formance data for the current month. Detrended Canonical Correspondence Analysis (DCCA) was performed on the cover and environ- mental data matrices using CANOCO (ter Braak, 1988), in a manner similar to that of Gibson and Looney (1994), to determine the relationships among the 125 quadrats. The ordination was based upon the 59 quadrats that contained tillers of C. porteri subsp. insperata and from which environ- mental data were collected. The quadrats for which environmental data were not available were not used to construct the ordination axes, but were held "passive" in the analysis and added subsequently, after the axes were constructed on the basis of their cover data. Rare species were downweighted using the rare-species downweighting option (ter Braak, 1988). Reliability of the results was checked and found to be acceptable using Oksanen and Min- chin's (1997) debugged and strict version of DE- CORANA The relationships between the measures of per- formance for C. porteri subsp. insperata (i.e., leaf area, predicted area, and tiller density) and the or- dination axis-scores were determined using Spear- man Rank Correlations (SASI, 1990) RESULTS The highest number of new tillers was produced in April, with March and May yielding the second and third highest number of new tillers, respective- ly (Fig. 2). New tillers were produced each month, but to a lesser extent from June to September. Cu- mulative gains leveled off from May to June after a sharp increase. This plateau was followed by a more moderate increase. Mean tiller density re- mained relatively constant from April to September, with the highest mean in August. No tillers died until June, with the frequency increasing each month thereafter. The mean percentage PPFD decreased markedly from April to May as the forest canopy closed, low- ering the available PPFD on the forest floor to less than 5% (Table 1). Mean monthly temperature of both air and soil increased into the summer, as ex- pected, with a high variation between months. Va- por-pressure deficit monthly means were highest in the late spring and early fall. Mean tiller densities increased slightly from April to May and generally declined into September (Table 1). Tiller density was highly variable among Volume 85, Number 1 Bittner & Gibson 73 1998 Microhabitat Relations of Calamagrostis Table 1. Monthly means and standard deviations for environmental data and tiller densities in three populations (В551 = Bell Smith Springs 1, BSS2 = Bell Smith Springs 2, and LC = Lusk Creek) of Calamagrostis 2. subsp. insperata. (PPFD — Te deficit (mm photosynthetic photo flux density, AirT = air temp. (? = tillers per m?; LC and BSS2 have n = 15; BSSI has C), SoilT = soil temp. (°С), V = 95 — vapor 5 for PPFD ы Tlr and Hg), Tlr = 35 for AirT, Soil T, and VPD). No tiller density data collected for any population in i or August. April May June July August September Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD BSS] PPFD 71.9 118 6.4 5.4 3.3 4.4 27 3.8 2.8 17 2.6 2.1 Air 13.3 13 223 06 269 0.8 298 05 284 07 212 0.4 SoilT 12.8 0.4 14.9 0.7 18.4 0.8 227 09.” 221 0.8 18.8 0.5 VPD 23 0.4 2.5 0.4 2.5 0.4 1.1 0.2 0.7 0.2 2.6 0.3 Tlr 193 206 201 211 — — 211 222 -- — 138 12.6 BSS2 PPFD 53.7 6.9 103 103 2.8 1.7 2.6 2.6 23 27 13 1.1 AirT 17.4 09 20.2 05 296 05 283 0.2 25.6 0.4 207 0.3 SoilT 10.2 0.4 14.7 0.3 18.7 0.4 22.7 0.4 21.8 0.4 17.0 0.6 VPD 24 0.4 2.0 0.2 2.2 0.2 1.2 0.2 1.8 0.2 22 0.2 Tlr 32.0 236 453 351 -- — 414 285 — — 278 180 LC PPFD 41.4 13.0 6.1 3.9 2.5 2.2 3.3 3.0 6.6 22 6.3 5.8 AirT 24.7 1.8 18.3 0.66 284 0.3 30.3 0.2 30.3 04 269 0.3 SoilT 9.8 0.5 13.2 0.3 19.3 0.4 22.0 0.5 22.3 0.4 17.8 0.4 VPD 3.5 0.9 1.6 0.5 2.1 0.2 1.7 0.2 2.0 0.2 3.4 0.3 Tlr 13.1 17.7 15.7 66.1 — — 83.0 67.4 — -- 69.2 53.2 samples within the populations, ranging from O to almost 240 tillers per m?. Mean leaf area declined throughout the year as well, due to leaf senescence. Both leaf area and total predicted leaf area were highly variable within the populations. ive of the nine environmental variables were significantly different between quadrats with and without tillers of C. porteri subsp. insperata at dif- ferent times throughout the year in population BSS1: VPD, air and soil temperature, PPFD, and soil depth (Table 2). Quadrats with tillers present had significantly lower VPD early in the season. Significantly higher soil temperature, air tempera- ture, VPD, PPFD, and deeper soil were observed in quadrats with tillers of C. porteri subsp. insperata during the second half of the growing season Five of the nine environmental variables signif- icantly predicted C. porteri subsp. insperata perfor- mance at В551: УРО, PPFD, pH, and soil organic matter (Table 3). VPD was negatively related to leaf area, and PPFD, pH, and organic matter were pos- itively related to tiller density, leaf area, and total predicted leaf area, respectively. Five of the nine environmental variables at BSS2 significantly predicted C. porteri subsp. insperata performance: PFD, soil temperature, cover, and organic matter (Table 3). Leaf area was nega- tively related to soil temperature, but positively re- lated to plant cover; tiller density was negatively related to VPD, but positively related to PPFD and soil temperature; total predicted leaf area was pos- itively relate D; tiller density for two monthly data combinations was positively related to soil temperature and PPFD combined; and mean leaf area was positively related to total cover and negatively related to soil temperature in combina- tion. Predicted total leaf area was positively related to a simple regression interaction between PPFD and organic matter. The regression predicting tiller density in July from PPFD and soil temperature in June had the highest R? of all the regressions per- formed (0.80). Mean leaf area predicted from cover and soil temperature also had a high variance ac- counted for (К? = 0.74) Five of the nine environmental variables at LC significantly predicted C. porteri subsp. insperata performance: VPD, PPFD, soil moisture, and soil and air temperature (Table 3). Leaf area was neg- atively related to soil temperature and positively related to soil moisture individually, combined, and by their interaction. Tiller density was positively related to VPD and PPFD. Total predicted leaf area was positively related to PPFD and air temperature. Mean soil temperature and moisture predicting leaf area accounted for the most variance (R? = 0.47). А total of 113 species associated with C. porteri subsp. insperata was recorded from the quadrats within the three sites. There were 47 species with a frequency greater than 10% at any site. The woody species (woody vines, seedlings, saplings, 74 Annals of the Missouri Botanical Garden Table 2 absence of Calamagrostis porteri subsp. insperata and environmental variable (Ме accounted (ог (R 2), deg xrees of freedom (d.f.), mean value ANOVA results from Bell Smith Springs (BSS1) listing signific ant relationships between the presence/ es. Given are the probability (P), variance ean), number of occurrences (А), and standard error (s.e.) of each combination of months of measurement. (Environmental variable abbreviations are listed in Table 1. Depth — soil depth (cm), Dep.Var. = dependent performance variable, Ind. Var. = independent environmental variable.) Month Present Absent Dep. Var. Ind. Var. Ind. Var. P Rm d.f. Mean N s.e. Mean ү s.e. May May VPD 0.0122 0.18 1/34 244 28 0.08 2.9] 7 0410 July May VPD 0.0095 0.19 1/34 2.43 27 0.08 2.69 8 0.09 August May VPD 0.0095 0.19 1/34 2.43 7 0.08 2.90 8 0.09 uly July 50111 0.0308 0.13 1/34 2280 27 0.19 22.11 8 ).12 August July Soil’ 0.0308 0.13 1/34 22.90 27 0.19 22.11 9 0.12 August August PPFD 0.0003 0.14 1/90 3.18 06 0.23 1.90 25 0415 August August ArT 0.0306 0413 1/34 2856 27 0.14 27.90 8 015 August August Soill 0.0209 015 1/34 2229 27 0.15 21.00 8 0.08 Augus August VPD 0.0060 021 1/34 0.74 27 0.05 0.54 8 0.06 September August PPFD | 0.0008 012 1/90 317 65 0.23 198 260 (0417 September August AirT 0.0113 018 1/34 2800 26 0.18 27.95 9 0.09 September August Soil’ 0.0488 0.11 1/34 27 26 0.16 21.71 9 0.13 September August VPD 0.0207 0415 1/34 0.74 26 0.05 0.60 9 0.06 September September PPFD 0.0002 0.14 1/90 2.94 65 0.22 163 26 0.51 September September SoilT 0.0004 0.19 1/34 18.94 26 0.10 18.44 9 0.11 ean ean Depth 0.0126 0.07 1/90 28.7: 70 2.06 16.80 21 2.54 were PPFD (г = 0.53) and soil depth (r = —0.43). Tiller density (r — 0.34) and total predicted leaf area (r — 0.33) of C. porteri subsp. insperata were highly correlated (Р — 0.001) to the second axis, while no measures of p mane were signifi- and shrubs) were, in descending order of domi- nance, Parthenocissus quinquefolia (L.) Planch., Os- trya virginiana (Mill.) K. Koch, Toxicodendron rad- icans (L.) Kuntze, Sassafras albidum (Nutt.) Nees, Acer saccharum Marshall, Quercus rubra, Q. alba, Carya glabra, Vaccinium palladum Aiton, Fagus cantly related to the first axis grandifolia Ehrh., Q. velutina, and C. ovata (mean cover > 2% at any site). The herbaceous vascular e species-environmental did biplot shows the environmental variables and species (those with and nonvascular species were, in descending order weightings > 50) ordination scores (Fig. 4). Acer of dominance, mosses, Polystichum acrostichoides saccharum, Luzula multiflora, Очтуа virginiana, (Michx.) Schott, Dichanthelium boscii (Poir.) Gould & L. С. Clark, and Luzula multiflora (Retz.) Lej., (mean cover > 2% at any site). The first, second, and third DCCA ordination axes cumulatively accounted for 21, 39, and 5396 Parthenocissus quinquefolia, Quercus alba, Sassa- fras albidum, Toxicodendron radicans, and mosses were highly weighted in the ordination (weight > 100). The LC samples have a higher air tempera- ture and pH. The BSS2 samples have a higher VPD and soil depth. The BSS1 samples have higher or- ganic matter, total cover, PPFD, soil moisture, and of the variance, respectively (the eigen values were 0.45, 0.41, and 0.30, respectively). Plots from the three sites (BSSI, ‚ LC) were dissimilar to each other and clearly separated when plotted on DCCA axes 1 and 2 (Fig. 3). Only a small overlap is present between the LC and BSS2 samples. The LC samples had a higher tiller density and a greater predicted total leaf area than those from either BSS site. The two variables with the highest correlation 69) soil temperature. DISCUSSION The habitat conducive to supporting populations of Calamagrostis porteri subsp. insperata in the Greater Shawnee Hills Section of the Shawnee Hills to the first axis were soil temperature (r — Natural Division (Schwegman, 1973) of southern Il- < 0.05 for all cor- | relations reported here). The two variables with the highest correlation to the second axis were air tem- perature (г = 0.62) and pH (r = 0.53). The vari- ables with the highest correlation to the third axis linois is comprised of high-quality (Grade A or B; White & Madany, 1978), dry-mesic, hickory forest and bluff edge communities that have and soil moisture (г = 0. upland oak- northeast- or northwest-facing aspects. In determining the microhabitat requirements of Microhabitat Relations of Calamagrostis Bittner & Gibson Volume 85, Number 1 1998 a — его с800'0 96196 « 105 — нән qur шы UBIN — = = = 970 9600 ілу + вәле “pag urs uray = = = = 1Е'0 026070 алаа + воле "рода ивәрү uray — == — m cro 2800°0 оро + воле fea] ued UBIJA — ZE0 782070 11195 - воле eo UBA ure 270 РІ200 опо + LIOS(+/—) воле peo] 9c 0 РЕРО'О алаа + тәү], ure ure = == — = ceo 062070 алаа + EU "ny "дос tee == = == Жап, 98800 алаа + тәү], Anf Апр = == 5% a 150 OSTO'O алаа + тәү, aunf Апр es тз = == 6£'0 Sero аал + EZ зау лау ЭП uonemdoq "m == = — 1е'0 60€0'0 әшейо y Gadd + воле ‘Paid иғәй uray = == — s 2<0 Sc00'0 алаа + воле “pag пиво ura = — — — ОРО 01 100 19407) + воле jes] UBIA ивәр LO 2000'0 LIOS + 21940)(— /4-) воле jes] ©©`0 с200'0 алаа + EZ UBA uray x == == P og'o 6Z£0'0 1195 — воле peo Anf "ny == sess 90 ©©00`0 LIOS + лог, күпү үп 99'0 11000 11105 + аяаа() ЕЛІ 0€'0 ££00'0 алаа + тәү], Апр Апр == == == == 670 12500 1195 + ЕДІ? əunf Апр 08'0 100070 11105 + а4лаа(+) ITLL 790 2000'0 алаа + ЕЛІ? eunf күп ае == — ES ГАТ 20100 ddA - ләү], Хе Апр = = == = со 9900'0 ddA – 19[IUL Хе Хе 2558 uone[ndoq — — — — ЛЕ 2600 оше) + воле "рола ивәр иғәр == x = == 6г0 88100 Hd + воле jes] иғәр UBA n em = == ГО €200'0 алаа + ләү], ny ‘dag p = p pee PZO 86000 ddA = воле Jeo] "ny "ny 1559 uonve[ndoq Y d “IBA PUT ue, за zal d “IBA PUT лед doq “IBA PUT лед doq S9[QEPLIPA S9[qEPLIPA yop, uorssa13o4 o[dnn]y uorssa1da4 o[duitg Саал 19} 9/1 = FP ‘Hd pue oruea() лој 17/1 = p “Add 19} сол = FP :1589 `@1/& = yp suorssoudo1 o[dr[nur əy} 103 “рур = ‘Fp ѕиогѕѕәләл јопрола-зволо pue o[duis əy} 10) 2669 Y ЭП :ѕәгоәйѕ pojeroosse Jo 42400 = 19407) *ju3]uoo 1ojeur oruez1o [IOS = DIUBÍA() *ainjsrour [Ios = OJA[IOS “2 pue [ зојаеј, ur рәт are ѕиоцегләлде əjqenea [ejuouruodtAu^]) 'suoissa18o1 ки pue ІШЕ Əy} ло} рәіѕц әле Y] pue “ZSSg “1558 шоу suoneurquioo щиош ш pypsadsur *dsqns 1210d гуволтошрјољ) jo (Rare jeep [2101 paripald pue ‘rare Jeo| “Хизиәр ләү ал) зопетшојод əy} вицогрола so[qeue4 juopuodoputr jueoyru3rs ayy pue *(—/4-) usis әјешцѕә лојошелра “(2М) 10у pojunoooe aouPLma (4) (| 48404: “E APL 76 Annals of the Missouri Botanical Garden 3 | А + Predicted Агеа м + Tiller Density 2 A A О А А " А 1 a > Р e О ө Хе А п um g п « = ө п d o өп 2 о п B О өг О 4 e п О ca Пп ы о В ° Р -3 — | DCCA І Figure : ЈА Axis I and II plot of quadrats containing Calamagrostis porteri subsp. insperata from sites BSS] (СІ), BSS2 (e. and LC (A). Dependent variables that are significantly (P < 0.05) correlated with each ordination axis are listed with their signs on the appropriate axis. C. porteri subsp. insperata, we found that vegetative performance is related to a complex suite of envi- ronmental factors. All nine environmental variables tested were significantly related at one of the three sites (Table 3). Vapor pressure deficit and PPFD were related to the vegetative performance at all three sites. Additionally, vegetative performance was related to air and soil temperature and soil moisture at LC, cover, soil temperature and organic matter at BSS2, and organic matter and pH at BSS1. No single environmental factor seems to be responsible for vegetative performance. Similarly, no single environmental factor is related to the presence or absence of C. porteri subsp. insperata. Rather, as shown in studies of other taxa (e.g., Gib- son & Good, 1987), it is the relative magnitude and importance of these limiting factors that provide an understanding of the subspecies’ microhabitat re- quirements or realized niche. Although these results show many shared trends between the populations with respect to the envi- ronmental variables as significant predictors of veg- etative performance, the DCCA ordination shows the sites separated when plotted on two axes. It is clear that the ground cover of associated species and the microhabitat are not similar among the three sites (at the local level). At a broader level, this is also supported by a review of the literature that shows a wide range of plant associates (and habitats) for C. porteri subsp. insperata throughout 1994; Campbell et al., Although soil moisture, soil temperature, air tem- perature, and pH were highly correlated with the first two axes in the ordination, only 39% of the variance was accounted for. This demonstrates that more factors are involved with distribution and abundance of the associated spec than were sampled. Total predicted leaf area and tiller density were both significantly related to the second axis. This was due to the much higher tiller densities and more robust leaves present at the LC popula- tion. Leaf senescence contributed greatly to the de- cline in leaf area from May to August at each site. Because C. porteri subsp. insperata grows only on cool, northeast- or northwest-facing slopes, it is possible that there is an upper limiting temperature for growth. Temperature is frequently the main lim- 1980; Barnes et al., 1983), such as C. porteri subsp. insperata. While high soil temperature is related to high per- iting factor for C, taxa (Ode et al., Volume 85, Number 1 Bittner & Gibson 77 1998 Microhabitat Relations of Calamagrostis 2 1.5 Vacpal 1 AIR TEMP. PH Moss Parqui 0.5 T^ erub Hee чн Козсаг - Toxrad VPD Luzmul 5 ORGANIC 9 „гло COVER a 0 ___>РРЕО => SOIL МО Antpl Dicbos т мер, SOIL TEMP. Cunori Cargla Ostvir Fragra — Quealb 4 Acesac -1.5 -1 -0.5 0 0.5 1 1.5 DCCA | Figure 4. DCCA Axis I and П species cover-environmental variable biplot ordination. Environmental variables and species (those with weightings > a coordinates are plotted and labeled on Axes I and II. Acesac = Acer sacch = Antennaria plantaginifolia, Cargla = = Carya glabra, Cunori = Cunila D. dichotomum, Fra ultiflora, E = Очтуа i inn Parqui = Parthenocissus ава, Pola rubra, Roscar — Rosa carolina, Sasalb — Sassafras albidum, Toxrad — Toxi- b — Amelanchier arborea, Antpla origanoides, Dicbos = кемік boscii, Dicdic Juealb — us alba, Queru codendron ien: Vacpal — 2. pallidum formance early in the growing season, it may con- tribute to increased leaf senescence later in the season. Calamagrostis porteri subsp. insperata seems to exhibit microhabitat selection similar to that of Cal- amagrostis canadensis (Michx.) P. Beauv., a boreal forest understory species. Calamagrostis canadensis gra — Fagus grandifolia, Luzmul — Luzula cr — Polystichum acrostichoides, exhibits an opportunistic "guerrilla" strategy for clonal foraging that allows the plants to vegetatively locate and exploit the most favorable microhabitats within the population by expanding rhizomes into conditions with warmer soil and higher light, as demonstrated in a glasshouse experiment (Macdon- ald & Lieffers, 1993). Field observations also doc- Annals of the Missouri Botanical Garden umented that C. canadensis allocates resources to its rhizomes for the invasion of favorable microsites (e.g., treefall gaps). Rhizome expansion and ex- ploitation in C. canadensis is influenced by soil temperature and cover, rather than PPFD alone (Macdonald & Lieffers, 1993). Tillers of C. porteri subsp. insperata at population BSS] were present within the most favorable microhabitats (1.e., quad- rats with higher soil temperature and PPFD; Table 2), as in C. canadensis. Observations of tiller production suggest that there are two distinct periods of production, one in the spring and the second in mid to late summer, with the latter considerably lower. The performance of C. porteri subsp. insperata, especially in the early spring, may be the result of the microenvironmental conditions from the previous year. In C. canadensis, mobilization of rhizome carbohydrate reserves for shoot regrowth is responsible for tiller production in the spring (Hogg & Lieffers, 1991), and it is likely that the same is true for C. porteri subsp. insperata. The second production period may be the result of the photosynthetic output from the current year. The difficulty of understanding the perfor- mance of individual tillers is also compounded by the fact that not all photosynthetic products go into tiller production, but some are allocated for rhizome elongation and foraging (Macdonald & Lieffers, Unfortunately, the present study was unable to determine the conditions favorable for sexual re- production due to the low number of flowering til- lers. Some states, such as Missouri, have seen pro- lific populations (Summers, 1993). No such events have but sporadic flowering episodes in some been noted in Illinois. However, from field obser- vations and 2. from neighboring sam- ples, inferences can be m crohabitat production of flowering tillers. Two sterile flowering e to determine the mi- conditions most conducive to the tillers (inflorescences without reproductive struc- tures) were recorded in 1992, and one in 1993, from the same location (it is unknown if it was the same genet). This location in population 8551, on a bluff edge under а small canopy gap along a hik- ing trail, allowed almost full sunlight to reach the tiller from early morning through midday. Тһе northeast-facing aspect at this location allowed the soil and air to be heated by sunlight early in the day and remain warm throughout the day. In con- trast, samples at BSS2 and LC had northwest-facing aspects, and did not have soil heated directly by sunlight until. midday or the afternoon. Further- more, soil temperature and PPFD were higher ear- lier in the year at BSS1 than at BSS2 or LC. These observations suggest an important role of early—mid growing season soil temperature and light on flow- ering. Late-season environmental factors may be less important for flowering as suggested by Havens and Holland (1998, this issue), who found no effect on sexual reproductive success from differences in late-season light. The flowering-tiller location was also wet for considerable parts of the year because of the presence of an intermittent stream. Environmental data were collected for the entire growing season from a quadrat located less than 1 m from the flowering tiller (1993). This same quad- rat was the DCCA ordination Axis I right endpoint. Axis I was positively correlated with soil tempera- ture (ғ = 0.69) and soil moisture (r = 0.53). These data suggest that there is a very small window of appropriate microhabitat conditions conducive to producing flowering tillers that is dependent on a complex suite of environmental factors including soil temperature, soil moisture, and light. Of course, other environmental and genetic factors not examined may also be related to the ability of this genet to flower. Establishment of new populations of C. porteri subsp. insperata via sexual means would be an un- likely event. It is extremely probable that each pop- ulation consists of a single, fragmented genet. Be- cause C. porteri subsp. insperata is supposedly self-incompatible (Greene, 1980), successful fertil- ization would require simultaneous flowering of two genetically dissimilar pem or genets (over 1 km apart at BSS and 50 m at LC). Many plants lack viable pollen (Greene, 1980) or lack sexual 1993; Bittner, 19952). Even if fertilization occurs, successful seed structures altogether. (Summers, production is still not ensured (Greene, 1980). The remote chance of successful sexua was demonstrated by Havens and Hollanc when they hand-crossed four presumably different genotypes and produced only one seed that ger- minated out of 2000+ fertilized florets examined. Because seed production and suitable habitat are 1 reproduction lland (199 not common, colonization via seed dispersal would be an extremely rare event. Undoubtedly, the infre- quency of this occurrence contributes to the rarity of this taxon. Calamagrostis porteri subsp. insperata maintains its populations primarily by vegetative means, es- pecially in Illinois. The current populations appear to occupy islands of suitable habitat surrounded by ecologically unsuitable habitat. Therefore, it would e unlikely that a population would be able to ex- pand to new sites via vegetative means, except lo- cally through transport of vegetative fragments. Two uncommon to rare associated species were Volume 85, Number 1 1998 Bittner & Gibson Microhabitat Relations of Calamagrostis discovered during the course of this study. At LC and an unsampled BSS population, over 10 m? were carpeted by Sphagnum spp. These areas also con- tained some of the highest densities of C. porteri subsp. insperata in southern Illinois. Another as- sociate, Carex willdenowii Schuhr, is present in three of the populations. It is listed as an Illinois endangered species and is known from only three counties in the southern part of the state (Herkert, 1991). The presence of these species within suit- able habitat might be a good indicator for the pres- ence of C. porteri subsp. insperata in southern Il- linois. A limitation of the present investigation is that populations throughout the range of C. porteri subsp. insperata were not studied. Therefore, we are not able to fully document the variable habitats or climates in which this taxon occurs. Additionally, this research was not exhaustive in examining all possibilities that may have affected the rarity of this taxon. Nevertheless, the limiting factors of specific- ity of habitat, unsuccessful seed production (Greene, 1980; Havens & Holland, 1998), and lim- ited colonization of new habitat are primarily re- sponsible for the few populations present in south- ern Illinois, and are the most likely reasons why C. porteri subsp. insperata is rare. e long-term survival of C. porteri subsp. in- sperata in southern Illinois is threatened in several ways. Natural disturbances, such as treefalls, may not create enough favorable microhabitats within extant populations to maintain viable populations Коо Increased forest succession and clo- ure of the tree canopy caused by fire suppression Hero likely affected er taxon adversely (Bittner, 1995a; Ambrose et al., ditionally, several populations are изван ынай їп агеа (80—120 m?), which increases the chances of destruction of a population by a single stochastic event. These actions could eventually degrade or extirpate the current populations. Since it is extremely unlikely that new populations would be produced by either vegetative or sexual means, it is necessary that the existing populations be properly protected and managed to ensure their long-term survival Future research is needed to determine the ef- fects of management on vegetative and sexual per- formance. Summers (1993; in Ambrose et al., 1994) has suggested that canopy gaps should be created within and around the populations to determine how the populations respond to such management. Only a small percentage of the canopy (10-25%) should be removed at first. Summers stated that too much disturbance (e.g., timber cuts) leads to rapid growth from nearby species that outcompete and choke out C. porteri subsp. insperata. He further added that prescribed burning should be conducted on an experimental basis and suggested that flow- ering events may be related to mild disturbance. Literature Cited Abrams, M. . C. Hulbert. 1987. Effect of topo- graphic пейіш q fire on species composition in tall- grass prairie in northeast Kansas. Amer. Midl. Natural- ist 117: 442-445. Ambrose, D. M., W. R. . M. R. Репѕкаг & D. W. Schuen. 1994. хн ruso Abstract for Cal- amagrostis porteri subsp. insperata, Reed Bent Grass. The Nature Conservancy. Arlin Bannister, P. 5. Pp. 73- 144 in D. Moore & S. B. Cha Methods i in Plant Ecology. Blac I Scientific, Oxford. England. Barnes, P. W., L. L. Tieszen & D. J. Ode. 1983. Distri- bution, production, ia ин of C,- and C,-domi- nated communities in a mixed prairie. iud J. Bot. 61: 741-751. Bittner, R. Т. 1995а. Distribution and Microhabitat Re- lations of a Rare Grass, Calamagrostis Porteri subsp. nsperata, in Southern Illinois. Unpublished Master's Thesis, Southern Illinois University, Carbondale. 995b. New populations of rare species in Illi- nois. sup 14: 59-60. Boyd, R. . D. Hilton. 1994. Ecologic studies of the ac species Clematis socialis Kral. Castanea 59: 31-40. ion C. J. G. ter. 1988. CANOCO—A FORTRAN Pro- m for Canonical Community Ordination by „Ки [Detrended] [Canonical] Corre ‘spondence Analysis, and Redundancy Anal- ton, Virginia )X— Baskin & C. C. Raskin, 1989. Ecol- ogy of the endangered species Solidago shortüi. 1. Ge- у, TETRI and physical habitat. Bull. Torrey Bot. Club 116: 344-355. Campbell, J. J. N et A. Bonney, J. D. Kiser, L. W. Korn- man, J кост герог, L. E. Medley & A. C. Risk 1992. ына Inventory of Endangered, Threat- ened, Sensitive, and Rare Species, Daniel Boone Na- "mr d Morehead Ranger District. U.S. Forest ер 3 C. Y. 1990. Laboratory Manual of General Ecology. 6th ed. Wm. С. Brown, Dubuque, lowa. FWS (Fish and Wildlife Service). 1993. 50 CFR Part 17, Plant Taxa for Listing as Endangered or Threatened Species; Notice of Review. Part IV. Fed. Reg. 51144—51190 Gawler, 5 D r & E. S. Menges. 1987. En- vironmental factors affecting establishment and ip of Pedicularis furbishiae, a rare endemic of the St. John dis Valley, Maine. Bull. Torrey Bot. Club ne 280— T им D. J. & R. E. Good. 1987. The seedling habitat of Pinus echinata and Melampyrum lineare in oak-pine forest of the New Jersey Pinelands. Oikos 49: 91—100. & P. B. Looney. 199 Vegetation Remy of е д on Peridido Kx Florida. J. Coastal Res. 10: 133-14 80 Annals of the Missouri Botanical Garden Greene, C. W. 1980. The Systematics of Calamagrostis (Gramineae) їп Eastern North America. Unpublished Ph.D. Thesis, ‘Harvard University, Cambridge, Massa- 1998. Factors affecting re- .. porteri . 85: 64—68. ‘husetts Havens, K. & D. L. Holland. productive success in a rare grass, subsp. insperata. Ann. Missouri Bot Ga Herkert, J. R. (editor). 1991. Endangered ind . ecies of Illinois: Status and Distribution, Volum Plants. Illinois Endangered Species Protection Board. pow 1 Hc V. J. Lieffers. 1991. The relationship be- serves grostis canadensis. Canad. Homoya, 1995. New vascular plant records 5 Indi- ana. Indiana Academy of Science, 111" Annual meet- ing, Nov. 2-3, 1 ct.] 5 V. J. Lieffers. 1993. Rhizome plas- ticity and clonal foraging o ias el pura in i d to habitat heterogeneity. J. Ecol. 81: 769— Mohlenbrock. R. H. 1986. Guide to the Vascular Flora of P Southern Illinois Univ. Press, Carbondale. Ode, D. J., L. L. Tieszen & . Lerman. 1980. The ке contribution of С, and C, plant species to pri- mary production in a mixed prairie. Ecology 61: 1304— 1311. Oksanen, J. & P. R. Minchin. 1997. Instability of ordi- nation results under changes in inipit data order: Ex- сі. € —454. . Ve 1990. SAS Lansuase: Refer- Cary, North Carolina. edie SASI (SAS Institute duc ). ence, Version 6, Ist ed. Schemske, D. E., B. C. Husband, M. H. Ruckelshaus, C. Goodwill, I M Ги & J. С. Bishop. 1994. Evalu- to the conservation of rare and endan- 5с 1. ond iM Plan for the Illinois ature Prese s System, Part 2. The Natural Divisions o eis Illinois 1. Preserves Commission, Rock- Sokal, R. R. & F. J. Rohlf. po Biometry. W. H. Free- man, San Francisco, Califor SPAC (Soil and Plant ie Council, Inc.). 1992. Handbook on Reference Methods for Soil Analysis. Soil and Plant A ae Council, Athens, Georgia 93 Summers, B. 1 otanical Survey for | | and State Listed Plant Species in Oregon and Shannon Counties, Eleven Point € o Mark Ту National Forest. U.S. Forest $ Swallen, J. R. 1935. States and Mexico SDA. ash. Acad. Sci. 25: 413-414. U 975. So : Survey of Pope. Hardin, and Massac Counties, Шпо —— 1992. Final Supplemental Environmental Im- act Statement: Amended Land and Resource Mar anage- ment Plan, Shawnee National Forest. USDA Forest Ser- vice Eastern Region Vivian, V. E. 1967. Shortia galacifolia: Из life history and Me requirements. Bull. Torrey Bot. Club 94: 369-387 Voigt, J. W. & R. H. Mohlenbrock. 1964. Plant Com- munities of Southern Illinois. Southern Illinois Univ. Press, “| е J. & бод Мадапу. 1978. Classification of Nat- Am in Illinois. Pp. 310—405 in Illinois Nabaril ree Inventory Technical Report GENETIC VARIATION IN RUNNING BUFFALO CLOVER (TRIFOLIUM STOLONIFERUM: FABACEAE) USING RANDOM AMPLIFIED POLYMORPHIC DNA MARKERS (RAPDs)! Daniel Ј. Crawford?, Elizabeth J. Esselmam?, Jennifer L. Windus*, and Carol S. bin? ABSTRACT Trifolium stoloniferum Muhl. ex Eaton (Fabaceae), a perennial, stoloniferous herb commonly known as running buffalo throughout its known geographic distribution t (U.S.A) | using Random Ampli эи! is currently res шінде to five states and is federally ong populations of T. stoloniferum fied Polymorphid: DNAs (RAPDs) as markers. The average within-population banding similarity values for > ш. from 21 populations are high, ranging from “ 920 to 0. е 5.2 - 0.952). Th 0.902 (me e mean banding similar s for 884), implying that much of the өлмей resides among populations in this sp alo differences abe average similarities within and between patches at the population level, suggesting substructu comparisons between populations range from 0.856 ecies. There are ring within populations. A large proportion of plants sampled within populations have different banding patterns, indicating that populations do not consist of one to several genets perpetuated pe ы. stolons. Our results agree with mes in showing relatively low leve ls of dive iversity; certain ones are identical to each other, a and m ow that even the smallest populations have a high proportion of different genets and thus are worthy of further оаа ii for conservation It is generally accepted that the long-term via- bility of a species is correlated with the levels of genetic variation it maintains (Vrijenhoek, 1994) because species with low levels of diversity may lack the ability to adapt to new and changing en- vironments (Godt et al., 1996). It follows, therefore, that successful strategies for the maintenance of rare species must include an understanding of the levels and distribution of genetic diversity. Population biologists have described levels of genetic variation in plant and animal populations primarily by utilizing allozymes (Hamrick & Godt, 1990). Enzyme electrophoresis has several advan- tages over other readily available methods for as- sessing genetic variation in most plant species (Hamrick et al., 1991). Allozymes exhibit Men- delian inheritance and thus are not subject to en- vironmental effects on the phenotype as may occur for quantitative morphological or physiological traits. In addition, allozymes are inherited as co- dominants, allowing the identification of alleles and loci; the allelic frequency dat y be used to calculate diversity at different елігі lev- els such as within and between populations, geo- graphical areas, etc. (Nei, 1973). Although protein electrophoresis has several advantages, one short- coming is that only a relatively small number of loci can be surveyed and only those encoding pro- teins are sampled. Also, rare plant species may have little or no detectable allozyme variation (Crawford et al., 1994; Lesica et al., 1988; Soltis et al., 1992) so that no assessment can be made ! We thank Marjorie Becus, Paul J. Harmon, Tom Bloom, and Ethyl Hickey for their help in the collection of leaf material for this study. This work was supported by a grant from Region 3 с ће U.S. Fish & Wildlife Service and the Ohio Department of Natural Resources, Division of Natural Areas and Prese ? Department of Plant Biology, The Ohio State University, Columbus, Ohio 43210, ? Department of Biological Sciences, Southern Illinois University at Edwardsville, Édwardssille Illinois 62026, U.S.A. * Division of Natural Areas “ Preserves, Ohio Department of Natural Resources, 1889 Fo 1, Columbus, Ohio 43224, U.S untain Square Court F- > Department of Pediatrics, um of Pennsylvania Health System, Philadelphia, Pennsylvania 19104, U.S.A. ANN. Missouni Bor. Савр. 85: 81—89. 1998. 82 Annals of the Missouri Botanical Garden of how genetic diversity is apportioned within the species. Random amplified polymorphic DNAs (RAPDs) have gained popularity as molecular markers in re- cent years, with Hadrys et al. (1992) providing an early perspective for the use of RAPDs in molec- ular ecology. RAPDs have both advantages and lim- itations relative to allozymes for the study of vari- ation in rare Advantages include the availability of 151 of the 10-base primers APD bands with PCR. This allows for the sampling of many more loci than is used to generate the possible with allozymes, and parts of the genome in addition to those encoding soluble enzymes can art & Porter, 1995; Whitkus et al., 1994). Use of additional primers will, in most be examined (Stew instances, eventually reveal variation within popu- lations and/or species of plants in which allozyme diversity is lacking or extremely low (Brauner et al., 1992; Crawford, 1997; Crawford et al., 1994: Meyer et al., d Rieseberg € Gerber, 1995; Rieseberg et al., The 1. a 1% fall into the two broad categories of reproducibility and interpretation of banding patterns on the one hand and analysis of the data on the other. The question of reproducing the same banding pattern for a given plant and primer in the present study will be discussed later. It is usually assumed that co-migrating bands in gels are homologous (represent the same segment of 2. but there have been few studies to doc- ument . The work of Rieseberg (1996) found that an " the RAPD bands were homologous for different species of Helianthus. While the level of error involved in assuming homology for co-mi- grating RAPD bands from different plants is not known, available data suggest that the assumption is valid in most cases for populations of the same species or closely ee 2. ies (Lannér et al., 1996; Stammers et al., 1 Because RAPDs a inherited primarily as 1. the genetic pasie of the banding patterns cannot be inferred, as is usually possible with allozymes, and this precludes the routine use of gene diversity statistics (Nei, f it is possible to assume from other infor- mation (such as allozymes) that populations are in Hardy-Weinberg equilibrium, then one can infer al- lelic frequencies from the absence of a given band in individuals (which are assumed to be homozy- gous recessive). However, rare plants often have lit- tle or no allozyme diversity within populations, and it is not possible to make assumptions about equi- librium within the very population where RAPDs represent an alternative to allozymes for assessing diversity. Various methods have been used for an- alyzing RAPD banding patterns, and several will be used in this study. Trifolium stoloniferum Muhl. ex Eaton (Fabaceae: Papilionoideae), commonly known as running buf- falo clover, is a rare, highly stoloniferous, perennial = . The species is known historically from nine states in the midwestern United States and was thought to be extinct from 1940 until the early 1980s, when it was rediscovered in West Virginia (Brooks, 1983; Campbell et al., 1988; Homoya et al., 1989; Cusick, 1989). Subsequently, additional populations were discovered in Indiana, Kentucky, Ohio, West Virginia, and Missouri. Since then run- ning buffalo clover has been listed as an endan- gered species by the U.S. Fish and Wildlife Service FWS, 1987). Recently, a number of populations have been found, particularly in Kentucky and —. West Virginia. Most populations are small (10—200 plants), but several populations are known with more than 1000 rooted crowns. Past genetic studies of T. stoloniferum using al- lozymes suggested low genetic diversity in this spe- cies (Hickey et al., 1991; Hickey & Vincent, 1992). These studies also indicated that smaller popula- tions have lower levels of diversity than larger ones, the majority of the diversity occurs among popu- lations, and that gene flow between populations is limited, even between those separated by short dis- lances. The purpose of this study was to assess genetic variation within and among T. stoloniferum popu- lations throughout their known geographic distri- bution using RAPD markers. Specific questions ex- amined include: (1) How much RAPD diversity exists in the species? (2) How is the diversity in the species distributed? (3) Do populations of T. stoloniferum consist of more than one genotype genet)? (4) Do larger populations contain more ge- netic variation than smaller ones? (5) Is there ge- netic substructuring, or a spatial distribution of dif- ferent genotypes within and among populations? METHODS POPULATIONS EXAMINED NA was extracted from the leaves of 390 in- dividual plants from 21 populations in Ohio, Ken- tucky, West Virginia, Missouri, and Indiana. State and county locations for populations are shown in Figure 1 and Table 1. Estimated population sizes are indicated in Table 1. One to two leaves were collected from each plant in April 1994 through June 1996. In larger populations, leaves were col- lected along a randomly placed transect, while leaves were collected throughout the smaller pop- Volume 85, Number 1 1998 Crawford et al. 83 RAPD Diversity in Trifolium stoloniferum INDIANA Dearborn MISSOURI Madison and Texas KE OHIO Clermont and Hamilton — 4 00М WEST VIRGINIA Randolph and Tucker NTUCKY Madison and Woodford 870W Map of state and county locations (in black) for populations of Trifolium stoloniferum examined in this stud Count names are given below each state name. State and county locations for each population are listed in у. 8 pop Table ulations. Twelve of the 21 populations sampled were examined for evidence of substructuring, ei- ther by comparing overall similarities within and between subpopulations or by determining the presence of unique RAPD markers between sub- populations. Voucher specimens for each popula- tion (e.g., Crawford et al. 1601-1621) are deposited at OS. DNA EXTRACTIONS Total DNA was extracted from 0.07—0.09 g nem weight) of mature leaf tissue following a minipre extraction technique from Doyle and Doyle (1987). Leaf tissue was ground in 0.7 ml of 2 x CTAB isolation buffer [100mM Tris-HCl, pH 8, 1.4 M NaCl, 20 mM EDTA, 2% hexadecyltrimethylam- monium bromide (CTAB), 1% Nabisulfite, 0.296 2- mercaptoethanol]. The ground leaf tissue was in- cubated at 60°С for 30—60 min., extracted once with chloroform-isoamyl alcohol (24:1), fuged at high speed for 2 min., and the supernatant was collected. Nucleic acids were precipitated by adding 0.46 ml isopropanol, recovered by high- speed centrifugation for 2 min., washed in 0.8 ml 76% ЕЮН/0.01 M NH,OAc, and resuspended in 0.1 ml 10 mM NH,OAc/25mM EDTA. centri- RAPD PROTOCOL AND ANALYSIS The protocol for amplifications is a modification of the procedure of Williams et al. (1990). Reac- tions were performed in volumes of 25 Ш contain- ing 1.5 mM MgCl, 100 uM each of dATP, dCTP, dGTP, and dTTP, 0.2 ИМ primer, 1.5 ul genomic DNA (an amount that gave reliable amplifications) and 0.75 unit of Taq polymerase. Amplifications were performed in a Perkin Elmer Cetus DNA Thermal Cycler programmed for 40 cycles of 1 min. at 94°С, 2 min. at 35°C, and 2 min. at 72°C. Am- plifications were performed with five primers, A-3, A-4, A-7, А-8, and А-9 from Operon Technologies. Amoke products were analyzed by electro- phoresis in a 1.5% agarose gel in 1 X TBE (tris- PR EDTA) buffer, detected by staining with ethidium bromide, and photographed on a transilluminator. A 100 base-pair ladder of DNA fragments (BRL) was included on each gel as a size reference. Amplification products from individuals of each population and each primer were initially electro- phoresed in separate gels. One to three individuals that (either singly or collectively) contained all bands in a population were identified from these gels. These individuals were then amplified a sec- ond time and run together on a second gel along with individuals from every other population that had been judged to contain the same bands. This was done to ensure that the presumed similarities and differences in bands between populations were reproducible. RAPD marker similarity was calculated by de- Missouri Botanical Garden Annals of the 84 [с6'0 2990 "16'0–Р18'0 001 6 6 0001-08 ydjopuey ‘uny suqof teddy 0<60 0690 <160-С180 001 Ol Ol 00€ ydjopury морлис 6<60 9680 966 0-SL8°0 280 €I SI OSE-SEE udjopuey “Ss лэлецс 6V6'0 28870 0160-9790 560 €l ТІ OOS [70001 yd¡opuey ‘uny dure?) yooy 096'0 t680 666'0-0€8'0 98'0 él ТІ OSI yd¡opuey *urejunojy. Yo y 786'0 668'0 8560-0790 со S 6 ScI yd¡opuey “үріп 7960 6880 026 0-9С<80 S8'0 LI 0c 0001—0€S€ ләҹә, OUJ 176'0 9C8'0 9880-<1980 001 9c 9c 000€ yd¡opuey *qouy цәполгу вал 1594 себ'о 188'0 0c6'07018'0 860 ср 9p 0052 ПОЦІШЕН `ѕпецэту ttOO 2280 126'0-£08'0 LLO 0c 9c 006 поттшең ALMAN 6t6 0 988'0 716'07Р28'0 02'0 L 01 0c зиошләүо) “ро, 69860 ¿060 9T6'0-1€8'0 0€'0 SI 0€ 00€ пош uey 0€6'0 ¿680 9T6'0-€c8'0 72'0 11 ес OSI uoj[rueg *uaa44) ssa13uo-) on) SC€6'0 7990 8680-8790 160 OI 11 OOS SEX9] `әрцѕәџц, CC6'0 098'0 2€£6'0-v08'0 76'0 SI 9I 00€ UOSIPRIA чәцӯрәрү LInossmy 656'0 0680 9260-<6/70 560 ТІ SI Su p19JpooA "Ou ү Үс60 с690 T16'0-8v8'0 98'0 81 Ic OSI uosIpep “ло Аәшд лода у с?60 9690 666 0-€F8'0 180 сс LG 00c uosIpe[y 9945 ләп) 156'0 €68'0 €£08'0-€£€6'0 790 6 ТІ 001 uosipe[y “espu-y 0c6'0 €280 906'0-P08'0 0071 1с 1с 561 UOSIPR| 189104 әй”) Амопупом ¿60 8980 с16'07018'0 560 ТІ SI DOSZ шофіғә( ‘AJLA uopprH епетриј uone[ndod suorje[ndod suorje[ndod лао е po[duies sjue[d jo вод Ајопов uonye¡ndod (вимоло ројоол pa¡dues suonejndod 24) шцим лао е ҷим цим әп|рл хушеришиѕ — .i2quinu/sad4jouas рәмәз40 Jed po[dures Jo 19quinu) 2218 pue suoneoo[ Ajunoo/a]elq Аудејиш5 КЕ ә2рдәлР Jo әЗиР p2A12sqo JO лодиту вәлвә[ Jo лодшпу поцвпдой payewysy IBRIIAY IBRIIAY jo 19GUINN грерпјош os[e зле uorje[ndod yore 10у (/961 *әзооу Y pueus[[;) зопјел хәрш Хивәлір ord&10u23 pue sodAjouas ролловдо jo лодшти au[ "штагћиојој“ што jo suonejndod Zuowe pue шуим suosueduioo ушеш лозмлеш 7 зведоле pue 'po[dures sjenprarpur jo 1oquinu *suorjeoy поцепао “| a1qe AL 140] 10] 4] y gel | np ! uen M (Id VH tarpui - [ uot d `1 ?Iq*L Volume 85, Number 1 1998 Crawford et al. 85 RAPD Diversity in Trifolium stoloniferum termining overall band similarity among all indi- viduals and then calculating mean similarity for plants within and between populations. A second approach was to ascertain the number of different discrete banding patterns (multilocus genotypes found within populations, and to assess the number and frequency of variable bands within and be- tween populations. Overall RAPD similarity within and between populations was calculated using a program written by Vera Ford, University of California, Davis (pers. comm.). Pair-wise individual similarity values for all plants were averaged for a measure of within- population similarity and again for between-popu- lation comparisons. Only bands that were scorable (1 or 0) in both individuals were used in this com- putation. Bands not clear or unscorable between ыгар" individuals were not included. The calculation for pair-wise individual similar- ity is: (2 8 x B) ” (B, + B) Where В. = the number of bands scored 1 for both individuals i and j. the number of bands scored 1 for individual i and likewise for j. The calculations and assumptions for average with- in and between-population or group similarity are as follows: Assume the first population consists of individuals numbered n, to n,, and the second population con- sists of individuals numbered m, to m,. Average pair-wise similarity of individuals in the first population is given by: У 5, for all i and j with n € i Sj Vol. 1. Kluwer Academic Publishers, Netherlan Williams, T A. R. Kubelik, K. Livak, J. A. Rafalski & S. ingey. . DNA poly morphisms amplified by arbi- trary primers are useful as genetic markers. Nucl. Acids Res. 18: 6531—6535. POPULATION GENETICS OF THE CEDAR-GLADE ENDEMIC ASTRAGALUS BIBULLATUS (FABACEAE) USING ISOZYMES' Carol J. Baskauf? and Sharon Snapp* ABSTRACT The rare cedar-glade endemic Astragalus шиш (Fabaceae) has low levels of genetic p red pall ed i^ id populations. Within-population means across 15 25.6% polymorphic loci, and 0.06 150 zyme loci resulted in estimates « alleles per loc 064 for observed ha terozygosity. Populations are genetically very 4. d » OW Ж and genetic identity values ranging from 0.981 to 1.000. Sites where this species na turally occurs should be protecte ad, but, considering the low levels of population differentiation, the source of e ар or seeds needed to establish new populations may not be the most critical concern. Astragalus bibullatus Barneby & E. L. Bridges (Fa- baceae) is a rare plant endemic to the limestone (“ce- dar”) glades of middle Tennessee’s Central Basin. It is a perennial that overwinters as a rosette, flowers in April and May, and ri Baskin, 1989). irme 4... was described as a new species in 1987, w ns fruits in June (Baskin & the Great Plains taxon A. crassicarpus Nutt. var. crassicarpus considered to be its closest relative (Barneby € Bridges, 1987). Known from only a few sites, A. m n is federally listed as endangered (FWS, 1 Conservation biologists are eo concerned about levels of genetic variability present in rare species. Many authors have pointed out that species with little genetic variability would have limited evolutionary potential under heterogeneous or changing environ- ments (e.g., Frankel, 1970, 1974; Franklin, 1980; tonovics, 1984; Lande & Barrowclough, 1987; Huen- neke, 1991). Compared with more geographically widespread species, rare and localized species often (but not always) have low levels of genetic ir (Hamrick & Godt, 1990; Hamrick et al., 1991; ron, 1987, 1991). Such low levels of genetic vari- ability could be the result of inbreeding and/or ran- dom genetic drift in small populations (chronically small, or small due to founder events or other genetic bottlenecks), or perhaps adaptation to a narrow set of environmental conditions. en estimating the genetic diversity of a spe- cies, population genetic structure can be examined to evaluate the level at which most variability occurs whether at the level of the individual, the popula- ~ tion, or the entire species), and the genetic similarity of populations can be estimated. Such analyses can help in management decisions for rare species. For example, population *C" of A. bibullatus is located on private property the owner plans to bulldoze, so state conservation officials hoped to transplant most of these individuals to an A. bibullatus site on pro- tected public land. Despite being protected, such a location could be unsuitable if the *C" population were genetically дине distinct. from the resident plants. Possible negative effects could include re- duced genetic diversity via local selection and ran- dom genetic drift, and poor growth of the transplants if genetic differences involve unique adaptations to ocal environmental conditions. Such possibilities raise the question of how genetically similar the А. bibullatus populations are. This study examines the population genetic structure of A. bibullatus, using isozymes to esti- mate the genetic variability of this narrow endemic and the genetic similarity of its populations. ! We thank the Tennessee Natural Heritage jg сре especially A. She and collecting leaves from one population, and | he Пред i in locating plants at one various ways. D. E. McCauley. Í the manuscript. This research was supported by a grant to € and Conservation and the Center of Excellence * Austin Peay State University, С larksville, Memphis, Tennessee 38163, U.S.: ANN. Missourt Bor. Christie for work on Tm map. The N site and permitte id pollen collection t Then Steven J. Bas Baskin, and an anonymous re vie Tennessee 37044. U.S.A. GARD. 85: 90-96. 1998. . for showing us Astragalus 4. sites Nature Conservancy of ' see skauf facilitate d » pier in er ы helpful criticism of ear 1 J. Baskauf from the Tennessee Department of оа nl versions of > for Field Biology at Austin Peay Sir U niversity. ? Department of Biology, Austin Peay State University, Clar sville, Tennessee 37044, U.S.A. Present. Address: University of Tennessee, Volume 85, Number 1 1998 Baskauf & Snapp 91 Population Genetics of Astragalus bibullatus Kilometers Relative positions of 2. of Astrag- Es [DN “W” represents both WO and WS. The unlabeled population was not sampled (see pon Popu- lation С (not show ni is about 20 km southwest of this area. Specific details about population locations have been omitted to protect ИШ» endangered species. MATERIALS AND METHODS COLLECTION OF SAMPLES Populations of Astragalus bibullatus were sam- pled during the summer of 1995. Leaves were col- lected and refrigerated in moist reclosable bags. Those collected in the field were stored on ice for a few hours before they were refrigerated. Spatially separated clusters of plants are referred to here as “populations” and labeled specifically by letters (A, C, D, V, WO, and WS). Population C is relatively isolated, being about 20 km from the other populations; on the other hand, WO and WS are only about 70 m apart. All populations then known for the species were sampled except one— a possibly artificially established colony with per- haps 60—100 plants (FWS, 1991) to which the landowner refused access. At the time of sampling, population C existed almost entirely as potted plants in the greenhouse because individuals had been dug up for transplanting. Figure 1 shows the relative positions of all populations except for C, with *W" representing both WO and WS Virtually all plants present were sampled for C, D, and V. Individuals were sampled haphazardly from A, WO, and WS, with a large fraction of the plants present included in the sampling. The ho- lotype for Astragalus bibullatus was collected from population А and is deposited in VDB (see Barneby & Bridges, 1987). ELECTROPHORESIS Electrophoresis procedures generally followed Werth (1985). Leaves were homogenized on ice in the simple extraction buffer to which 10% polyvi- nylpyrrolidone and 0.696 mercaptoethanol had been added immediately before grinding. Crude ho- mogenate was then adsorbed onto filter paper wicks and loaded onto 12% starch gels. Various individ- uals were used as marker genotypes on gels throughout this study. Four buffer systems (three continuous and one discontinuous) were used to visualize 12 enzyme systems: (1) tris-borate EDTA, pH 8, for alcohol dehydro- genase (ADH) (1.1.1.1), aldolase (ALD) (4.1.2.13), and glyceraldehyde-3-phosphate de- hydrogenase (NAD-dependent form) (G3PDH) (1.2.1.12); (2) tris-citrate, pH 8, for isocitrate dehydrogenase (NADP-dependent form) (IDH) (1.1.1.42), mal- ate dehydrogenase (MDH) (1.1.1.37), and phos- phoglucoisomerase (PGI) (5.3.1.9); (3) histidine-citrate, pH 5.7 (Wendel & Weeden, 1989), for menadione reductase (MNR) (1.6.99.-), phosphoglucomutase (PGM) (5.4.2.2), phospho- gluconate dehydrogenase (PGD) (1.1.1.44); and (4) the discontinuous system from Ridgeway et al. (1970), pH 8.1, for aspartate aminotransferase (AAT) (2.6.1.1), leucine aminopeptidase (LAP) (3.4.11.1), and triose-phosphate isomerase (ТРІ) (5.3.1.1). Staining protocols generally followed Wendel and Weeden (1989). Other staining solutions are de- scribed in Werth (1985) (IDH), Moran and Hopper (1983) (MNR), Soltis et al. (1983) (G3PDH), and Baskauf (1993) (LAP, MDH, PGM). Loci and alleles were numbered from the electrophoretically fastest to the slowest. ANALYSIS Allele frequencies, measures of genetic variabil- ity, and Nei’s (1978) unbiased genetic identity were calculated using BIOSYS-1 (Swofford & Selander, 1989); x? goodness-of-fit tests of genotype frequen- cies for deviations from Hardy-Weinberg expecta- tions (using the Levene correction for small sam- ples) and x? contingency tests to examine the independence of allele frequencies among popula- tions were performed. Hierarchical cluster analysis (UPGMA) (Sneath & Sokal, 1973) was used to group populations by genetic similarity using Nei’s genetic identity. Wrights (1978) F-statistics (Fs and Ку) for evaluating within vs. among population 92 Annals of the Missouri Botanical Garden Table 1. Allele frequencies and sample size (N) for polymorphic loci in Astragalus bibullatus. Population Locus/allele A C D V WO WS PGM-1 (N) 32 28 21 16 22 18 1 0.203 0.000 0.024. 0.000 0.045 0.028 2 0.016 0.107 0.048 0.063 0.045 0.139 3 0.781 0.893 0.929 0.938 0.909 0.833 PGM-2 (М) 32 28 21 16 22 18 1 0.859 0.839 0.929 0.875 0.364 0.833 2 0.063 0.036 0.000 0.000 0.091 0.111 3 0.078 0.125 0.071 0.125 0.545 0.056 ADH-1 (N) 30 24 16 8 21 15 l 0.367 0.188 0.250 0.063 0.262 0.233 2 0.633 0.813 0.750 0.938 0.73 0.767 PGD-2 (N) 32 21 21 15 22 16 1 0.984. 1.000 0.976 0.967 0.841 0.938 2 0.016 0.000 0.024. 0.033 0.159 0.063 variability were calculated according to Weir and Observed heterozygosity (H,) for these loci ranges Cockerham's (1984) procedures, which correct for effects of sample size and provide a weighting sys- tem for multiple alleles at a locus. t-tests were used to determine whether the value of an F-statistic dif- fers significantly from zero. RESULTS Fifteen putative loci were considered to have been resolved, coding for only 10 of the enzyme systems. This is because interpretation was at least partially unclear for MDH, PGI, PGM, PGD, and TPL, usually due to the presence of a larger number of invariant bands than could be accounted for by the typical number of loci found in diploid plant species. Crossing studies are not helpful in such a case involving invariant loci, and comparisons of banding patterns of leaf tissue versus soaked pollen (Weeden & Gottlieb, 1980) did not aid interpreta- tion. Liston (1992) reported duplication of certain isozyme loci (PGI, PI, perhaps MDH) for some Astragalus taxa, ы it is possible that there may be several cases of gene duplication in А. bi- bullatus as well. Of the 15 loci resolved, 11 appear to be invariant for this species (ALD, LAP, AAT, IDH, PGD-3, MDH-3, the two G3PDH loci, and all three MNR loci). Allele frequencies for the four polymorphic loci are given in e 1. Astragalus bibullatus does show some genetic variability for soluble enzymes, but at a low level (Table 2). Within populations, 2096 to 2796 of the loci included in this analysis are polymorphic (P). The mean number of alleles per locus (A) is 1.4. from 0.038 (for V) to 0.099 (for WO), with a mean of 0.064. Species level estimates are similar, with А = 1.4 and P = 27%. These estimates of isozyme variability may be overestimates, considering that some unknown number of clearly invariant loci were excluded from the analysis. As a whole, the populations are somewhat dif- ferentiated from one another at three of the four variable loci, as indicated by the significant (Р < .01) x? contingency tests of allele frequencies (Ta- ble 3). Further analysis revealed that despite being separated by only about 70 m, WO and WS show highly significant differences (Р < 0.001) in allele frequencies at PGM-2. In fact, WO appears to be genetically the most distinctive population in the species. Nonetheless, the populations of A. bibullatus are all very similar genetically. Genetic identity values among these populations are consistently high, ranging from 0.981 to 1.000 (Table 4, Fig. 2). An Е, of 0.089 (Table 5) indicates that less than 10% of the total genetic variability of the species is the result of differences among populations, and in fact the jackknifed mean Fy, does not differ signifi- cantly from zero (P > 0.05). Therefore, most vari- ability in this species is due to genetic heteroge- neity within populations rather than genetic differentiation among populations. ;enotype frequencies for variable loci do not de- viate significantly from the Hardy-Weinberg expec- tations within populations; thus expected hetero- zygosity values (Н,) are values (H,) for this species (Table 2). This situation very close to observed Volume 85, Number 1 Baskauf & Snapp 93 Population Genetics of Astragalus bibullatus Table 2. Genetic variability* at 15 loci for Astragalus bibullatus. Population N A P H, H, A 29.3 1.4 26.7 0.064 0.074 (0.7) (0.2) (0.034) (0.040) C 26.1 1.3 20.0 0.061 0.053 (0.7) (0.2) (0.033) (0.029) D 20.4 1.3 26.7 0.056 0.047 (0.4) (0.2) (0.034) (0.027) V 15.8 1.3 26.7 0.038 0.036 (0.8) (0.1) (0.019) (0.018) WO 21.3 1.4 26.7 0.099 0.095 (0.4) (0.2) (0.049) (0.047) WS 14.0 1.4 26.7 0.068 0.072 (1.0) (0.2) (0.032) (0.034) Mean (all populations) 1.4 25.6 0.064. 0.063 an sample size per locus М), mean number of alleles per locus (А), percentage of loci polymorphic (P), observed ( heterozygosity (H,), expected heterozygosity (H,) as an unbiased estimate (Nei, 1978). Standard errors are indicated in parenthes is reflected in the fact that F, values are close to zero and the jackknifed mean does not differ sig- nificantly from zero (Table 5). These data suggest that A. bibullatus may be primarily an outcrossing species; however, this species’ mating system has not been studied. DISCUSSION Although not completely lacking in genetic di- versity at isozyme loci, the narrow endemic Astrag- alus bibullatus has low levels of variability. This is true for each population and for the species as a whole. In a compilation of plant isozyme studies, Hamrick and Godt (1990) reported population level means of A = 1.72, P = 43.0%, and H, = for 85 studies of widespread species, as opposed to A = 1.39, P = 26.3%, and H, = 0.063 for 100 studies of narrowly endemic species. Thus means estimated for the rare A. bibullatus (A = 1.4, P = 25.6%, and H, = 0.063) are comparable to those given for endemics in general. Low levels of vari- ability at isozyme loci also have been reported for able 3. Independence of allele frequencies for pop- ulations of Astragalus bibullatus: X contingency analyses. #o Locus alleles DF x P PGM-1 3 10 36.2 0000 PGM-2 3 10 67.705 0.00000 ADH-1 2 5 8.312 0.13986 PGD-2 2 5 16.539 0.00546 some western species of Astragalus with restricted geographic ranges (Karron, 1991; Liston, 1992). O the other hand, Travis et al. (1996) found 220 vari- able AFLP markers and substantial differentiation among populations for the rare Astragalus cremno- phylax Barneby var. cremnophylax. These data are not directly comparable to isozyme data, however, and it is not known what levels of diversity or pop- ulation differentiation would be detected by an iso- zyme survey of this taxon. А few other species endemic or nearly endemic to the limestone glades of Tennessee have been as- sayed for variability at isozyme loci. Levels of ge- netic variability estimated for Echinacea tennes- seensis (Beadle) Small (Asteraceae), another federally listed endangered species, were similar to those of A. bibullatus: A = 1.3, P = 23.0%, and H, = 0.071 (Baskauf et al., 1994). A much less rare congener, Astragalus tennesseensis A. Gray, had opi high estimates, with A ТІ, P = 3.1%, and Н, = 0.148 blade fioi Wiltshire, able 4. Genetic identities (Nei, 1978): pairwise com- parisons for populations of Astragalus bibullatus Popula- tion A C D V WO WS A жжжжж C 0.997 xe D 0.998 1.000 ***** V 0.992 1.000 0.999 ***** WO 0.981 0.985 0.981 0.982 ***** WS 0.999 1.000 1.000 0.999 0.984 жж 94 Annals of the Missouri Botanical Garden Similarity" 0.95 0.96 0.97 0.98 0.99 1.00 | | | | | | | | | | — А C 4 D L WS C v wo | | | | | | | | "221 0.95 0.96 0.97 0.98 0.99 1.00 Similarity' "Nei's (1978) unbiased genetic identity Fig is Мера (1978) unbiased genetic identity. 1994). However, Dalea foliosa (A. Gray) Barneby, another legume federally listed as endangered, has much lower estimates, with А = 1.15, P = 13.8%, and H, — 0.045 (calculated from Wiltshire, 1994). Population sizes of Astragalus bibullatus appear to be quite variable among years and have been extremely small at times (Somers & Gunn, 1990; A ), a factor that could contribute to low levels of genetic variability. For example, the “A” population was reported to consist of only a couple plants in 1979, but had increased to 171 plants by Table 5. F-statistics for polymorphic loci in Astraga- lus bibullatus. Locus F Fip PGM-1 —0.108 0.041 PGM-2 0.041 0.194 ADH-1 — 0.029 0.019 PGD-2 0.090 0.053 Mean* —0.009 NS 0.089 NS (0.029) (0.060) * Means jackknifed over polymorphic loci (Weir $ Cockerham, 1984), with standard errors indicated in pa- rentheses. Neither mean differs significantly from zero (NS, P > 0.05) re 2. Populations of / Astragalus bibullatus clustered according to genetic similarity. The similarity measure used 1988 after the site had been cleared of woody veg- etation. Such dramatic population fluctuations or extinction and recolonization events, even when rare, can greatly decrease effective population sizes and thus genetic variability (Wright, 1940; Nei et al., 1975; Lande & Barrowclough, 1987; McCauley, 1993). served for some species showing no genetic vari- 1988; Waller 7), as well as some Astragalus species Such population crashes have been ob- ability at isozyme loci (Lesica et al., et al., showing very low levels of variability (e.g., A. clar- ianus Jepson; Liston, 1992 "Genetic bottlenecks” resulting from population crashes are not the only factor that could affect genetic variability in A. bibullatus. Even at the best of times populations of this species are not large, and the smaller a population the more quickly ran- dom genetic drift is likely to erode variability. On the other hand, this plant is a perennial that prob- ably has a long-term seed bank like many of its congeners (e.g., А. tennesseensis; Baskin & Baskin, 9). Both of these features would favor the re- tention of genetic variability within the species. The fine-scale differentiation observed between "populations" WO and WS, which are separated by only 70 m, was unexpected considering the great Volume 85, Number 1 998 Baskauf & Snapp 95 Population Genetics of Astragalus bibullatus similarity among populations as a whole for this species. WO is also the population that displays the highest levels of heterozygosity. Of all populations, WO occurs in the most open habitat—a regularly mowed area along a private lane. The WS plants, on the other hand, grow in one of the most shaded spots among trees. Our isozyme data suggest that limited gene flow occurs between these two popu- lations despite their close proximity. Gene flow could be restricted due to pollinator behavior, or could be ineffective due to differential selection pressures. The most immediate threat to survival for A. bi- bullatus appears to be lack of protected habitat, with all populations but two occurring on privately owned land. The plight of the “С” site is a clear indication of this threat. WO could be a particularly good population to try to protect, given that it is genetically the most distinctive (indicated by ge- netic identity values) and the most variable (indi- cated by heterozygosity estimates); nevertheless, all of the populations are genetically quite similar. Ex- tinction because of environmental stochasticity is a risk for any highly localized species limited to a few populations (Lande, 1988; Simberloff, 1988); thus the establishment of new populations of this species is advisable. А seed storage program is al- ready in progress (K. Havens, pers. comm.). The low level of population differentiation observed for A. bibullatus suggests that the origin of seed used in establishing new populations probably is not a critical consideration. Similarly, these data provide no evidence of major genetic differences that might make inadvisable the transplanting of individuals from С to the V population. Overall, it appears that protection of natural pop- ulations and the establishment of new populations are high priorities in alleviating the threat of ex- tinction for this rare species. In addition, further research is needed. Little is known about the life cycle and ecology of Astragalus bibullatus, and any management plans would benefit from this type of information. Furthermore, it would be interesting to know how the genetic variability of this cedar-glade endemic compares with that of its widespread prai- rie relative, А. crassicarpus var. crassicarpus, and such a comparison is planned. Literature Cited а J. 1984. Genetic variation within popula- s. Pp. 229-241 in R. Dirzo & J. Sarukhan (editors), rspectives on Plant Population Ecology. Sinauer, Sun- Aarand, xem husetts. Barneby, R. C. & E. L. Bridges. 1987. A new species of | BR ү ина eae) from Tennessee's Central Basin. Brittonia 39: 358-363. Baskauf, C. J. 1993. раар, Population Genetics and bm. of a Rare Videspread Species of Echinacea (Asteraceae). Ph. р, pon ic Vander- bilt 4. Nashville, Tennessee. D. E. McCauley & W. G. Eickmeier. 1994. Ge- netic 4. оҒ а таге and a widespread species of Echinacea а Evolution 48: 180-188. Baskin, J. M. & C. C. Baskin. 1989. Cedar glade endem- ics in Tennessee, m a review of their autecology. J. I ee Acad. Sci. 64: 63-74. тет J. A 083. Extinction, survival, and genetic variation. Pp. 125—163 in C. M. Schonewald-Cox et al. (editors), Genetics and Conservation. Benjamin/Cum- mings, Menlo - California Bradshaw, A. D. 1984. Ecologic Ж —À of genetic variation between гоо: Pp. 213-228 in R. Dirzo & J. Sarukhan (editors), Perspectives on hu ас tion Ecology. Sinauer, Sunderland, 5. FWS (U.S. Fish & Wildlife Service). Su pe RP and threatened wildlife and plants: an bibulla- tus (Guthrie's ground-plum) determined to be endan- gered. Fed. pner 56: 48748-48751. Frankel, O. H. 1970. Variation, the essence of life. Sir William Macleay Memorial Lecture. Proc. Linn. Soc. Lond. 95: 158-169. 1974. Genetic conservation: Our evolutionary ibebendibility. Genetics 78: 53-65. Franklin, I. R. 1980. Evolutionary change in small pop- ulations. Pp. 135-149 3. Soulé & B. A. Wilcox (editors), Conservation pe Ап Exululiomus- Eco- pen Perspective. Sinauer, Sunderland, Massachu- е 5. Hamrick J. L. & M. J. W. Godt. 1990. All di ant species. Pp. 43—63 in A. H. D. pun et tj amd Plant Population Genetics, Breeding, and Ge- netic Resources. Sinauer, Sunderland, Massac и etts. urawski & M = 19: 91. бање ора been species traits and E diversity: Implications s for conservation biology. Pp. 75- in D. A. Falk & K. E. Holsinger (editors), Genetics and Conservation of Rare Plants. Oxford Univ. Press, New York. Huenneke, L. F. 1991. Ecological implications of genetic variation in plant populations. Pp. 31—44 in D. A. Falk & K. E. Holsinger (editors), Genetics and Conservation of Rare дан Oxford Univ. Press, New Үог Karron, J. 1987. A comparison of levels of genetic Pn and self-compatibility in dp do pod еы and widespread plant congeners. Evol. Ecol. 7-58 . 199]. Patterns of genetic E and breeding systems in rare plant species. Pp. 87-98 in D. A. Falk . E. Holsinger (editors), Genetics and nas d of Базе Plants. Oxford Univ. Press, New У Lande, R. 1988. Genetics $ demography in » biological conservation. Science 241: 1455-1460. & С. Е ашн ы. 1987. Effective popula- lion size, genetic variation, and their use in population management. Pp. 87-124 in M. E. Soulé (editor), Viable эке for 2. Cambridge Univ. Press, Cambridge Lesica, P. К. Е. boss F. W. Allendorf & D. E. Bilder- back. 1988. Lack of genetic diversity within and among populations of an ташты! plant, Howellia aquatilis. Conservation Biol. 2: 275-282. Aston, A. 1992. Isozyme ted of Astragalus sect. Annals of the Missouri Botanical Garden Leptocarpi subsect. Californici (Fabaceae). Syst. Bot. 17: 367-379 McCauley, D. E. 1993. Genetic consequences of extinc- tion and recolonization in ae habitats. Pp. 217— 233 in P. M. Kareiva et al. ors), Biotic Interactions anig Global Change. Sinauer, Sunderland. Massachu- setts. “а G. Е. & S. D. Hopper. 1983. Genetic diversity and the insular нЕ structure of the rare granite rock species, Eucalyptis caesia Benth. Austral. J. Bot. 3l: чое 978. Estimation of average heterozygosity and ма үз from a small number of individuals. Genetics 89: 583—590 4. & R. Chakraborty. 1975. The bo Џепеск effect в genetic variability in 4... Evolution 29: 1- Ridgeway, С. Ј., S. p Sherburne & R. D. Lewis. 1970. Polymorphisms in the ce of т herring. ans. Amer. 4. бос. 99: 147— Simberloff, D. The aes т «үн апа community 1. to conservation science. Annual Rev. Ecol. Syst. Sneath, P. H. A. onomy. Freeman, San Francisco. Soltis, D. E., C. H. Haufler, D. C. Darrow & G. V. Gastony. 1983. Starch gel electrophoresis of ferns: А compilation of grinding buffers, gel and electrode buffers, and stain- ing schedules. Amer. Fern J. a 9-27 Somers, P. & S. C. Gunn. 1990. Status теріні e bibullatus Биле & a 11. report, U.S Fish € Wildlife Serv Soulé, M. E. 1980. Thres holds for survival: Maintaining fitness and evolutionary potential. Pp. 151-169 in M. 1973. Numerical Tax- E. Soulé & B. A. Wilcox (editors), Conservation Biology: An Evolutionary-Ecological Perspective. Sinauer, Sun- derland, Massachusetts Swofford, D. L. & R. K. Жашайын 1989. BIOSYS-1, Computer Program for E Analysis of Allelic Variation in Genetics. Release 1.7. User's Manual. University of Illinois, Urbana, Illinois Travis, .. J. Masc ый & P. Keim. 1996. An analysis of g пеге variation іп Astragalus cremnophylax var. ral Иа a critically endangered plant, using AFLP markers. Molec. Ecol. 5: 735-745. Waller, D. M., D. M. O'Malley & S. C. Gawler. 1987. Genetic variation in t reme endemic Pedicularis furbishiae (Scrophulariaceae). Conservation Biol. 1: 335-340. Weeden, N. F. & L. D. Gottlieb. 1980. Isolation of cyto- plasmic enzymes from pollen. Pl. Physiol. 66: 400—403. ir, B. S. & € Estimating F- statistics for the analysis of population structure. Evo- lution 38: 2. Wendel, J F. Weeden. 1989. Visualization and interpretation > me isozymes. Pp. 6—45 in D. E. Sol- tis & P. S. Soltis (editors), 2. in Plant 175 е , Diosc orides Tres Portland, Oregon. T 985. 2. ап Pru laboratory a field m => Wiltshire, B. 1994 sessme de “of niei “diversity in Astragalus tennesseensis and the federal endangered Dal foliosa. Unpublished 1. и Southern Illinois Universi Carbondale, Illino Wright, S. 1940. Breeding structure of p opulations in re- lation to L- 2. Amer. Naturalist 74: 232-248. 1978. Variability Within and Among Natural Populations, Vol. 4. Evolution and the Genetics of Pop- ulations. Univ. Chicago Press, Chicago. GENETIC VARIABILITY IN Diane L. Tecic?^, Jenny L. McBride’, THE FEDERAL THREATENED 7 » Boules' qn Daniel L MEAD’S MILKWEED, ASCLEPIAS MEADII TORREY (ASCLEPIADACEAE), AS DETERMINED BY ALLOZYME ELECTROPHORESIS! ÅBSTRACT Mo st populations of the federal threatened Mead’s milkweed, Asclepias meadii Torr. (Asclepiadaceae), occur primarily issouri, where annual summer mowi remaining small populations in Illinois, Iowa, pur аши уан Missuri p d observed heterozygosity of 0.158. other milkweed species. More than half of the total 4 depauperate populations or for restoring new populations; however, sampling from о species” genetic diversity. Such sources would include the large fire-managed populations. Empirical data are needed to determine the popu ulation- -genetic consequences of long-distance crosses and introductions that are apparently needed to restore viable populations of this species in the eastern part of its range. The restoration of declining species enters a realm of conservation biology in which environ- mental, demographic, and genetic factors limit the survival and growth of populations (Shaffer, 1981; Wilcox & Murphy, 1985; Gilpin & Soulé, 1986). When once widespread self-incompatible or out- crossing plant species have reduced population sizes, forced inbreeding can result in either total reproductive failure or lower reproductive fitness with increased homozygosity and де оа potential (Menges, 1991; Schaal et al., 1991; Wel- 94). Thus, restored 4. of сме must be large enough to withstand loss from envi- ronmental or demographic events, and populations of either self-incompatible or outcrossing species must also contain sufficient numbers of genetically | c ~ ! Funding for this project was provided primarily by the U.S. Forest Service and the U.S. Fish & Wildlife Service. e thank representatives from many agencies for ; assistance and additional funding that facilitated the plant-tissue ч кун and data processing. These include: Beth Shimp and Larry Striteh, U.S. Forest Service; John Schwegman Illinois Department of Conservation; Paul McKenzie, U.S. Fish & Wildlife Service; Dean Roosa, Fort Dodge Community College; John eie lowa Department of Natural Resources; Bill Pusateri, lowa Department of Transportation; Bill atson, lowa Chapter of The Nature Conservancy; Do Conservation; Paul | Қ Мона enisi of Nat Е. 7 m; Hayworth, and Barbara Schaal for extremely helpful comments on 2. on Kurz, Tim Smith, and ural Resources; Craig Fre ? Department of Plant Biology, Southern Illinois University, Carbondale, Illinois 62901-6509, U.S.A. * Current address: Horseshoe Lake State Park, 3321 Highway 11, Granite City, Illinois 62040, U.S.A. ^ The Morton Arboretum, Lisle, Illinois 60532, U.S.A. ? Corresponding au ANN. MISSOURI Bor. GARD. 85: 97-109. 1998. Annals of the Missouri Botanical Garden = A SCONS => О ш z Цом О 17 > © О өө О 19 16 "e. о YO ШЕК е URE 1,2,3 > 10 Mr ANS ee Qs • e 18 8,9 e g 11 12 13 14 15 Site Number, Name & State 1 Rockefeller, KS 6 Jack's, KS 11 Hinton Creek, KS 16 Helton, MO Dog Leg, KS 7 Osawatomie, KS 12 Cook Meadow, MO 17 Woodside, IA 3 French Creek, KS 8 Garnett, KS 13 Niawathe, MO 18 Saline, IL 4 High, KS 9 Sunset, KS 14 Wah-kon-tah, MO 19 Ford, IL 5 Colyer, KS 10 Fowler Hill, KS 15 Weimer Hill, MO 2. |l. Distribution by county of Asclepias meadii. Closed circles are counties with extant populations; open s are о from which populations have been extirpated. Allozyme study-site locations are numbered corre- le жен чн to Tab different individuals to allow reproduction, avoid inbreeding, and have evolutionary potential (Les et al., 1991; DeMauro, 1993, 1994) Understanding and resolving genetic problems is an important need in the conservation and recovery of declining plant species (Asins & Carbonnell, 1987; Barrett & Kohn, 1991; Falk & Holsinger, 1991; Amos & Hoelzel, 1992; Fenster & Dudash, 1994). This need is exemplified for the federal threatened (Harrison, 1988) Mead's milkweed, As- clepias meadii Torr. (Asclepiadaceae). This species is an obligate outcrossing, long-lived, late-succes- sional perennial that is restricted to virgin tallgrass prairies, prairie haymeadows, and glades (Betz, 1989). The plant is pollinated by small bumblebees (Bombus) and miner bees (Anthophora) and, like all milkweeds, its pollen is dispersed in pollinia and its seeds are wind-dispersed from follicles (Betz & Lamp, 1992; Betz et al., 1994). Plants have an un- derground rootstock that produces multiple ramets, and from which rhizomes up to 1 m long have been observed in the field and in potted plants (M. L. Bowles, pers. obs. Asclepias indi formerly ranged from Kansas eastward through Missouri and Illinois to northwest Indiana, and north into southern Iowa and north- west Wisconsin (Fig. 1). The origin of A. meadii is speculative. Gleason (1922) indicated that it was derived from the more eastern A. amplexicaulis Sm., and Fernald (1925) suggested it survived gla- Volume 85, Number 1 1998 Tecic et al. 99 Genetic Variability ciation in the driftless area of Wisconsin and ad- jacent states. Woodson (1954) and Noamesi and Il- tis (1957) proposed a more likely post-glacial migration from the Ozark uplands similar to that proposed for Silene regia, a prairie species whose highest genetic diversity is in the Missouri Ozarks (Dolan, 1994) When A. meadii was discovered in 1843, it may not have been rare. Owing to habitat destruction and fragmentation, this species is presently known from a total of just 31 counties in parts of its former range in Illinois, Iowa, Missouri, and Kansas, the latter of which contains the largest number of pop- ulations. In Iowa, small populations, each with rel- atively few plants, are found in prairie remnants in four counties. Of the two known Illinois *popula- tions," one is represented by only a from a railroad prairie in Ford Co. The other occurs in the Shawnee Hills, Saline Co. This population comprises four colonies fragmented by forest en- single plant croachment among four small glades, one of whic is isolated from the others by more than a kilometer. The largest colony has 17 ramets, and the others have only a few ramets. In the southwestern and west—central part of Missouri, А. meadii occurs in small populations within remnant prairies. It is also locally abundant in large glades of the St. Francis mountain system of the (Ozark Plateau in Iron and Reynolds Cos., Missour Although A. meadii is s localis abundant in Kan- sas and Missouri, most suitable habitats were con- verted to native prairie haymeadows nearly a cen- (Fitch & Hall, 1978). This cultural ractice removes developing seed pods (follicles), thereby eliminating pollen-mediat- ed gene flow and sexual reproduction (McGregor, 1977; Betz, 1989). In such populations, reproduc- tion is limited to vegetative spread of rhizomes. One exception is Kansas University's Rockefeller Prai- rie, Jefferson Co., from annual haymowing to a two- or three-year burning rotation in about 1956 (Fitch & Kettle, 1988). Because A. meadii is a long-lived perennial (Betz, 1989), it can survive decades of haymowing and reproductive failure, and may increase ramet (but not genet) numbers by rhizomatous spread. In Kansas, which was converted a seven-year demographic study of 140 ramets in unmowed railroad prairies, Betz (1989) annually re- moved all mature seed pods and found a 9.396 de- cline in ramet numbers, suggesting that some sex- for long-term maintenance of natural populations. This life-his- ual reproduction is required tory strategy correlates with low levels of follicle production reported for most milkweeds (Wyatt, 1976; Wyatt & Broyles, 1994). For example, al- though 7796 of 140 ramets studied by Betz (1989) flowered annually, less than six follicles matured annually. The glade populations of A. meadii in the Ozark Mountains of Missouri have escaped conversion to agriculture and haymowing but, as in Illinois, ap- pear to be fragmented by forest encroachment re- sulting from fire suppression (Guyette & McGinnes, 1982; Guyette & Cutter, 1991; Ladd, 1991). The greatest concentration of these plants occurs in Iron Co. at Weimer Hill and in adjacent Reynolds Co. on Proffit Mountain, both of which are fire-managed natural areas. The Weimer Hill population contains 100 or more plants distributed among numerous glade openings (each less than 1 ha in area) that occur along a kilometer of bluff line habitat. As with most milkweeds, A. meadii appears to be self-incompatible, as selfed garden plants do not produce follicles (M. L. Bowles, pers. obs.). Incom- patibility in Asclepias appears to involve a late-act- ing ovarian system, and is not thoroughly under- stood (Seavey & Bawa, 1986; Shannon & Wyatt, 1986; Broyles & Wyatt, 1991; Kahn & Morse, 1991; Karron, 1991; Broyles & Wyatt, 1993a; Wyatt & Broyles, 1994). This life-history charac- teristic, combined with low effective population sizes (in some cases so low as to prevent sexual reproduction), impacts directly upon federal recov- ery planning to recover or restore sustainable pop- ulations of A. meadii (M. L. Bowles, pers. obs.). Furthermore, it is critical to assess the amount of genetic variation among populations to determine how genetic differentiation should be managed in recovery or restoration operations. This information may be especially important for restorations, where limited habitat size may require metapopulation management in which translocation of genetic ma- terial is required to maintain a high level of genetic diversity among populations (Bowles et al., 1998). e purpose of this research was to determine the amount and distribution of genetic variation within and among A. meadii populations, and to compare the effects of management (burning versus mowing) on this variation. Allozymes provide a rel- atively rapid means to assess neutral genetic vari- ation within and among populations of rare, endan- gered and/or threatened plant species (Waller et al., 1987; Lesica et al., 1988; Nickrent & Wiens, 1989; Billington, 1991; Hickey et al., 1991; Schaal et al., 1991; Godt & Hamrick, 1996a, 1996b), and mul- tilocus genotypic diversity within populations (Sipes & Wolf, 1997). Detection of genetic diversity by DNA-based methods such as RAPDs (random am- plified polymorphic DNA) is gaining use, especially when allozyme variability is low (e.g., Rieseberg et 100 Annals of the Missouri Botanical Garden Table 1. Mead's milkweed study sites sampled for allozymes (see Fig. 1 for site locations). Population sizes obtained from Kansas and Missouri Natural Heritage Program census data or M. L. Bowles (pers. obs.). Population size Population Population approx. no number name State County Habitat or management indiv. l. Rockefeller Kansas Jefferson former ћаутеадом“ 200 2. Dog Leg Kansas Jefferson former haymeadow 20 3. French Creek Kansas Jefferson haymeadow « 25 4. High Kansas Leavenworth haymeadow « 25 5. Colyer Kansas Douglas haymeadow « 25 6. Jack's Kan Jouglas haymeadow > 300 7. Osawatomie Kansas Miami roadside « 10 8. Garnett Kansas Anderson aymeadow > 100 9. Sunset Kansas Anderson haymeadow > 100 10. Fowler Hill Kansas ranklin haymeadow « 25 11. Hinton Creek Kansas Bourbon former haymeadow 400 12. Cook Meadow Missouri Barton rotation" « 25 13. Niawathe Missouri Dade rotation‘ < 300 14. Wah-kon-tah Missouri St. Clair rotation* < 20 15. Weimer Hill Missouri Iron ades' 100 16. Helton Missouri Harrison preserve/burned <5 17. Woodside lowa Adair haymeadow (mowed late) < 30 18. Saline Illinois Saline glades 40 19. Ford Illinois Ford railroad prairie «5 * Burned every 2 or 3 years since 19 e Former hayme 1990. * Burn/hay/rest since ca. 1 Metapopulation managed E: ар al., 1989; Rieseberg & Gerber, 1995). However, when the resolution of this genetic variation is fine- scale, such as genotypic variation within popula- tions, allozyme variation may provide a more con- servative system that is useful across a broader range of genetic diversity within and among popu- lations (Swensen et al., 1995). Allozymes have al- eady proven useful in population-level studies of relatively common (Broyles & Wyatt, ; 1993b; Foré Guttman, 1996) as well as rare milkweeds (Edwards & Wyatt, 1994). E MATERIALS AND METHODS POPULATION SAMPLING samples were collected in 1992 from indi- vidual ramets of Asclepias meadii from 19 popula- tions throughout the extant range of the species; supplemental collections were made from the Wei- mer Hill study site in 1993 (Table 1, Fig. 1). When possible, at least 20 individuals were sampled from each population. However, this number was some- times not achieved because of the limited number of individuals located during this study. This was despite previous censuses of larger populations at some large sites, and no doubt was due to difficul- eadow, prey when ed flower since ca. 98( ties in locating small vegetative plants when flow- ering individuals were absent (Alexander et al., 1997). Sampling was minimally destructive, and consisted of removing one or two leaves from the lower nodes of the stem. Samples were individually numbered and labeled by population name, sealed in a plastic bag, and stored on ice until returned to the lab. Samples that were not extracted imme- diately were quickly frozen with liquid nitrogen, then maintained at —75°C until extraction. ISOZYME ELECTROPHORESIS Enzymes were extracted by using a Polytron ho- mogenizer (Brinkman Industries, Westbury, NY) to grind the leaf tissue in ca. 2.0 ml of *microbuffer" pH 7.5 (Werth, 1985) supplemented with 596 (w/ vol) polyvinylpyrrolidone (MW 40,000) and 0.1% 2-mercaptoethanol. For extremely small samples, hand-held glass homogenizers were used with vary- ing amounts of extraction buffer (less than 1.0 ml). Samples were kept on ice throughout the extraction process. Cellular debris was pelleted via centrifu- gation at 10,000 rpm for 10 minutes, and the su- pernatant was poured into a microcentrifuge tube and stored at — 75°C. Volume 85, Number 1 1998 Tecic et al. 101 Genetic Variability For electrophoresis, the extract was absorbed onto 6 X mm Whatman #3 filter paper wicks and loaded into 13% starch gels (Starch Art, Smith- ville, Texas) for typical horizontal starch gel elec- trophoresis (Wendel & Weeden, 1989). Ten enzyme systems representing 12 putative loci were assayed using three gel electrode buffer systems. Alcohol dehydrogenase (Adh-2), isocitrate dehydrogenase (Idh-1), glutamate dehydrogenase (Gdh-2), and shi- kimate dehydrogenase (Skdh) were resolved using the Tris-citrate pH 7.5 buffer (Soltis et al., 1983). Aspartate aminotransferase (Aat-2, -3), glucose phosphate isomerase (Gpi-2), malic enzyme (Me-1), menadione reductase (Mnr), phosphoglucomutase (Pgm-1), and triose phosphate isomerase (Tpi-1, - were resolved using a lithium hydroxide buffer (Ridgeway et al., 1970). Finally, malate dehydro- genase (Mdh-2) and 6-phosphogluconate dehydro- genase (6-Pgd) were resolved with a pH 6.0 histi- dine citrate (Olmstead, 1989). Enzyme-staining tocols were essentially as reported in Soltis and ds (1989). Shikimate dehydrogenase showed ac- tivity but was not resolved for all populations; hence it was excluded from analysis. Menadione reductase (Mnr-2) showed high levels of variability within and between populations, but resolution was N ~ inadequate; hence this system was also excluded from analysis. GEL SCORING AND ANALYSIS Gel banding patterns were recorded photograph- ically and genotypes were inferred based upon knowledge of enzyme subunit composition and the number of loci per enzyme system commonly seen in other plants (Gottlieb & Weeden, 1981; Weeden & Wendel, 1989). Enzyme patterns previously doc- umented for other species of Asclepias were also consulted. Allelic isozymes were measured and re- corded as relative mobilities using the most com- mon allele as the standard (relative mobility of 100). When more than one locus appeared for an enzyme system, the most anodal one was designat- ed “locus one.” Relative ыыы numbers were converted to al- phabetic genotypes prior to entry into BIOSYS-1 (Swofford & Selander, 1981). This program was used to calculate genetic variability measures by population, such as mean sample size per locus (V), mean number of alleles per locus (A), percentage of polymorphic loci (P), direct-count heterozygosity (Н), expected heterozygosity given Hardy-Wein- berg equilibrium (Н,), and chord and arc genetic distances between populations (Cavalli-Sforza & Edwards, 1967). To analyze partitioning of genetic variability within and among populations, the fixa- tion index (Wright, 1978) was used. Because of its small sample size, the subdivided Saline Co., Illi- nois, population was analyzed as a single popula- tion, as well as the larger fragmented population at Weimer Hill, Missouri. We also used hierarchical F-statistics (Wright, 1978) to analyze genetic vari- ation within and among seven subdivided popula- This allowed comparison to hierarchical analysis of patch sub- division of Asclepias verticillata L. made at a similar landscape scale in southern Ohio (Foré & Guttman, 996). The PCORD (McCune, 1993) software pro- gram was used to analyze population samples by Prin- cipal Component Analysis (PCA), using allele frequen- cies in a variance-covariance matrix, and to calculate allele diversity (H^ for each population using the Shan- non diversity index where H’ = —X plog p, where p, — allele frequency at each locus. Because Asclepias meadii has the potential to form more than one aerial shoot (ramet) from its rhizome system, the true number of genets per pop- ulation could not be determined at the time of sam- pling. Following analyses of the electrophoretic data, it was determined that for some populations, individual ramets represented the same genet. This was determined by (1) the presence of identical multiple-locus genotypes from the isozyme analy- ses, (2) identical RAPD patterns on some collec- tions (D. Hayworth and B. Schaal, pers. comm.), and (3) ramet proximity obtained from field maps. Such ramet genotypes were then “collapsed” into a single genotype and, for analysis purposes, as- sumed to be parts of the same genet. This process results in a conservative estimate of the actual number of genets because different genets could have the same multiple-locus genotype. To assess tions sampled at Weimer Hill. the relative abundance of ramets and genets within each population, the percentage of all ramets as- sociated with each genotype was determined, and their mean and standard error calculated for each population. The average percent ramets/genet is also calculated as the inverse of the number of ge- notypes multiplied by 100. RESULTS Twelve polymorphic gene loci were identified corresponding to 10 enzyme systems for Asclepias meadii (data matrix available upon request). Al- though no monomorphic loci were found, two of the polymorphic loci (Adh-2 and Me-1) had one allele that was essentially fixed and a second allele that was scored only once. There was an average of 1.5 alleles per locus (Table 2), with 42 alleles occurring | Garden issouri Botanica Annals of the M 102 ОСОО + ISTO 750'0 + 6510 L0'O + OLS соғ CT COL + 8'0t ‘ps + чвзи 191'0 2910 OVS 1 1791 001 1 | p104 “OL ТОГО 2600 сос VI OSS OOL + 00'0c © Ol 3ut[PS 781 6t UO 2910 LG VI Cee 8с + 2991 9 РА SPISp9oN Al 2910 0с<<70 997% €I OSS 001 I с чоцон “OL ГО Тего C 81 0757 сс + 0rt€ 16 ВР ПІН PWM “CT 21270 9070 08'c 81 t'8€ ср9 + 7991 9 8 Че-поз- UEM “pl 1610 ТРГО С 81 070€ SOL + OSZI 8 сс эшеме “El сего SETO 897 €I OSS 121 + 00'05 © 9I морвә YOO) “EI 6210 GLTO сб LI 686 LY + ITH 6 SI 991) чоң “TT с60'0 29Г0 09%< СІ 2791 001 1 Т ШЕН 49[^94 701 TECO 8120 Le Uc 051 ee + IUIT © Ol SUNS "6 OT UO 6ЕГО 89'c 81 070€ 61 + OSZI 8 Ic Huw ^g 90c'0 POLO LEG 91 EEE OES? + 00'0€ [4 € 9IUIO]EMES() "L 9c UO SLUO 89'c 9'I Тр 696 + 0007 5 81 syf 79 29го 18ГО ele STI 070€ ci + ITI 6 él 1олј05 “€ 8с00 8200 сес ГІ £8 9'£c + 00'0€ с € эшеід фан cp 7200 €90'0 OLS el Gee voc + ELE EE Е 6 19915 YU 76 SZUO YOTO LG cl Liv LEL + 00766 v L 391 301 72 100 polo 08'c 81 Є`ӨС те 999 SI Ic Top ?x»oH °H "Н Н "V (%)d ‘ps x Joues ѕәадјоиәх) војашес uone[ndoq лошел 2 ивәрү "кивовАтолојоц po1oadxo = ^g :ХАизовАлолодоц paA1asqo = “y :xoput Хишәлір попивус = „у :snoo[ orgdaouisqod ләй so[oj[e jo лодшти = "y ‘foj orydiowájod шәләй = (%)4 :suonepndod рәәмҹуш spray 10] вгоцацеје Хизләлїр әцәцә=у 76 әде, Volume 85, Number 1 1998 Tecic et al. 103 Genetic Variability across all 12 loci. Ten alleles at ten different loci were abundant, averaging over 90% frequency among all populations. There were 25 rare alleles (less than 1046 frequency), representing all but the Mdh-2 locus. Rare alleles were distributed widely among populations. Only two sites, High Prairie and Fowler Hill, did not have any of the 25 rare alleles but these sites had sample sizes of only three and four, respectively. Eleven populations had one or more of 15 alleles that were unique to single populations. The Saline Co. population had the highest number, representing the Got-2, Рет-1, and Tpi-1 loci. None of the other three disjunct populations in eastern Illinois, eastern Missouri, or Iowa had unique alleles. The mean percentage of polymorphic loci across all populations was 40.896, with 0.159 observed and 0.151 expected mean heterozygosity (Table 2). Percent polymorphic loci was positively correlated with sample size (r? = 0.432, P = 0.0022) and reached 7596 at Weimer Hill (n = 48) and Sunset haymeadow (n = 10). The lowest values (16.7% occurred in sites with single plants, but Saline Co. (п = 10) had only 2596 polymorphic loci. Observed and expected heterozygosity were not correlated with either sample size (r? = 0.026, Р = 0.507; г — 0.0046, P — 0.783, respectively) or number of genotypes (r? = 0.0002, P = 0.954; г = 0.0664, P — 0.281, respectively). Among all A. meadii pop- ulations, 123 multiple-locus genotypes were iden- tified among the 237 “collapsed” samples, with 79 (64.296) restricted to single populations (Table 2). Genotypes and sample size were positively corre- lated (2 = 0.8016, P < 0.0001), with the highest numbers at Weimer Hill (27 genotypes) and Rocke- feller (15 genotypes). The mean allele diversity (H^) was 2.70, and values ranged from 2.52 at High Prairie to 2.80 at the Rockefeller and Wa-kon-tah prairies (Table 2). These values were not correlated with sample size (г? = 0.092, P = 0.206). Although the highest Н” (2.80) occurred at Rockefeller Prai- rie, five haymeadows had higher H'' values than the fire-managed Weimer Hill (H' — 2.72). Results of genetic-distance analyses (data not shown) indicated that all populations clustered at a chord distance (Cavalli-Sforza & Edwards, 1967) of 0.32 or less, and that clustering did not conform to expectations based on geographic proximity. The Aia, greatest interpopulational distances also involved those populations with the smallest sample sizes, and hence these relationships may be artifactual. There was also no clear geographic pattern among populations with PCA of allozyme frequencies (re- sults not shown). The first PCA axis accounted for 35.1896 of the variation and was most highly cor- related (i.e., factor loadings > 0.4) with variation in two alleles at the pgm-1 locus. The second axis accounted for 57.696 cumulative variation and was most highly correlated with variation in one allele at the pgi-2 locus. The third axis accounted for 77.9% cumulative variation and was also highly correlated with variation in one allele at the pgi-2 locus. Three additional axes brought cumulative variation to 94.9%, with the strongest correlation in four alleles at the gdh-2 and got-3 loci. Wright’s fixation index provides information on the degree of fixation of individuals relative to their specific population (Ғ,) and to all populations (F,), and the differentiation among all populations rela- tive to complete fixation (Ғ.,). An Fsr value of 0.0 indicates that all variance resides within popula- tions whereas a value of 1.0 shows that all variance is between populations (i.e., no alleles are shared among populations). The F,, value for А. meadii was 0.261, which shows that about 7446 of the ge- netic variation sampled resides within any one pop- ulation. Hierarchical F-statistical analysis of the subdivided Weimer Hill provided information on the proportion of variance explained by the inter- actions of subpopulations to the total population. An F, value of 0.355 for subpopulations to total indicates that about 6596 of the variance in the Weimer Hill population of А. meadii occurs within its subpopulations. The type of management regime in effect at each population 1s shown in Table 1. Fire-managed pop- ulations contained more genotypes and a greater proportion of ramets with different genotypes than haymeadows (Tables 1, 2). For example, in thirteen haymeadows, the average percentage of all ramets for each genotype ranged from 11.1196 (at three sites) to 100%, with an average of 30.98%. In con- trast, the average percentage of ramets per genotype was 6.66% and 3.70% in the fire-managed Rocke- feller and Weimer Hill prairies, respectively. Al- though these average values are positively corre- lated with sample size (г = 0.3464, P = 0.008), they are meaningful when they represent total pop- ulation samples. For those populations with ten or more samples, a plot of the ramet/genet ratio and the number of genotypes graphically illustrates the effects of burning versus mowing (Fig. 2). The burned Rockefeller and Weimer Hill sites had high numbers of genotypes and low ramet/genet per- centages, whereas the mowed haymeadows had higher mean ramet/genet percentages and lower numbers of genotypes. DISCUSSION Given that several other species of Asclepias have been examined using allozymes, it is informative to 104 Annals of the Missouri Botanical Garden 50 50 [ ] %Ramet/Genet =—@— Мо. Genotypes = 3 40 |--|- 40 8 c 2 Ф = (5 о һе с Mes Ф Ф 30 ----30 © Е “- а о Е Ф д Ё с 20 - ---- 20 5 © z = < I Ана Бі 10 'w—-——e———e-..- - -- -] ---- 10 ~ e = he ~ ы ы р =] o Y т 5 5 $ б 5 = » _3 o = c c с © о xo ~ -— о > б о 6 Ф Ф о о G > = 88 Е x e t O S ӛз D 8 = z > с Dr ін > < > Burned Mowed Figure 2. Effect of management methods on genetic structure of Mead's milkweed populations. Only those popu: lations with allozyme sample sizes gre percent ramets per genet and multiple-locus genotypes. place the A. meadii results in context by comparing genetic diversity statistics among these species (Ta- ble 3). Although the objectives and methodologies differed somewhat in each study, similar patterns or trends are apparent. The mean number of alleles per locus and percentage of polymorphic loci for Asclepias texana Heller, a rare species, and A. per- ennis Walt., a widespread sister species, are re- markably similar to each other and to those of A. meadii. Heterozygosity levels in А. meadii were not as high as in A. exaltata L. or A. verticillata, two widespread North American species, but higher than in A. texana and A. perennis. These data in- dicate that genetic diversity in milkweeds may be less affected by effective population size and num- ber than in other plants, likely because of their obligate-outcrossing mode of reproduction (see be- low). Despite severe reductions in population size, А. meadii has retained a comparatively high level of genetic diversity, thus making it an excellent candidate for restoration activities. Because levels of heterozygosity and mean allele diversity were not correlated with sample size or number of genotypes, eater than or equal to 12 are shown. See text and Table 2 for calc ulations of mea these estimates are likely accurate reflections of the actual genetic diversity in A. meadii populations. In contrast, the percentage of polymorphic loci (P) was significantly correlated with sample size, pos- sibly because sample size exceeded 20 plants for only 4 of the 19 populations studied. However, samples for many of the fragmented populations in Iowa, Illinois, and northern Missouri often repre- sented the entire population, and thus these results are meaningful. The partitioning of genetic variation in А. meadii, as measured by F-statistical analyses, showed that the greater proportion of genetic variation (7496) is within populations. However, this species partitions more than two to three times as much variation be- A. texana, А. per- ennis, or А. exaltata (Table 3). Similarly, the hier- archical partitioning of 6546 of the genetic diversity within population subdivisions at Weimer Hill was much less than the 97% found within patches of A. verticillata (Foré & Guttman, 1996). These com- parisons indicate that A. meadii maintains more than a moderate amount (2696) of genetic variation tween populations as compared to Volume 85, Number 1 Tecic et al. 105 Genetic Variability Table 3. Comparison of milkweed genetic-diversity statistics based on allozyme studies: A, number of alleles per polymorphic locus; % unique alleles = percentage of alleles proportion of genetic expected heterozygosity, F,r (also reported as С) = observed heterozygosity, H, — о percent polymorphic loci, Н restricted to single populations; Р(%) variation among populations. % unique Total Total samples Asclepias Reference Р(%) 40.8 loci pops. species 0.159 0.151 0.261 this paper 35.7 1.5 A. meadii Edwards & Wyatt (1994) Edwards & Wyatt (1994) Broyles & Wyatt (19932) Foré & Guttman (1996) ~ “2 е^ А. texana 0.082 0.055 942 0.182 0.093 0.214 64.5 75.5“ 2.17 10.5 0.197 0.033" 459 A. verticillata * Only polymorphic enzyme systems (of 20 examined) were used in this analysis; Р would be lower if loci were considered monomorphic that had the most common allele at a » Based on F,, (patch-total) hierarchical analysis (Wright, 1978). within its populations. However, this variation is only slightly higher than the expected level (20%) or widespread, outcrossing species reported by Hamrick and Godt (1990) and follows from the life- history characteristics of milkweeds in general (i.e., wind-dispersed seeds and durable pollinia carried by insects). Thus, any natural A. meadii population that meets a minimum size might be expected to comprise much of the genetic diversity occurring across the range of the species, but many genotypes may be distributed among populations. Despite the modern rarity of А. meadii, its populations maintain higher levels of genetic variation than naturally rare or endemic Esp plants such as Pedicularis furbishiae S. s. (Waller et al., 1987), Howellia aquatilis A. le (фа et al., 1988), and iw ium stoloniferum Eaton (Hickey et al., 1991). outcrossing breeding system (Schoen & cd 1991) and occasional pollen- or = mediated gene flow among populations (Wyatt & Broyles, 1994), along with the great longevity (Betz, 1989) and increased survival of heterotic individuals (Schaal & Levin, 1976; Mitton & Pierce, 1980; Le- dig, 1986), may contribute to the maintenance of genetic diversity within fragmented populations of A. meadii. Although generally considered neutral, allozyme frequencies may be associated with different soil characteristics (Heywood & Levin, 1985). of (1) large genetic distances among populations and (2) a geographic pattern to genetic variation is somewhat surprising because А. meadii is widely distributed over areas where soil conditions range from acid and nutrient-poor in the south to calcar- eous and nutrient-rich in the north (Bowles et al., 1998). However, our samples did not adequately cover the northeastern part of the former range of this species. Asclepias meadii populations in ungla- ciated Missouri or Kansas could represent points of origin for all А. meadii populations. Given their geographic proximity and habitat similarity, glade populations in southern Illinois also might be most closely related to those in Iron Co., Missouri. How- ever, allozymes provide little positive information to address biogeographic hypotheses relating to pre-glaciation refugia, lineages and migration routes, Pe 5270 and current distribution pat- terns for А. т. Burning on mowing management practices ap- pear to have different effects on the genetic struc- ture of A. meadii populations, with high numbers of genotypes and low ramet/genet percentages in burned sites, and fewer genotypes with higher ra- met/genet percentages in haymeadows. Although these results may correlate with sample size, sim- 106 Annals of the Missouri Botanical Garden ilar ramet/genet percentages were found with APDs (D. Hayworth & B. Schaal, pers. comm.), and higher ramet densities with lower percentages of flowering ramets also occur in haymeadows 1998). This evidence suggests that haymowing, which prevents sexual reproduction, promotes clonal spread of certain genotypes but at- trition of others. Although sexual reproduction is (Bowles et al., arrested by the mechanical removal of flowering stems, if such “pruning” does not remove all pho- tosynthetic tissues from ramets, or misses smaller ramets, vegetative growth may continue throughout the growing season. If above-ground tissue is lost, it is also possible that mowing may stimulate lateral buds to form new shoots, thereby increasing rhi- zome branching. New growth may also be enhanced by reallocation of resources that might have gone toward sexual reproduction (Bowles et al., 1998). The net effect would be an increase in the number of vegetative shoots (ramets) that subsequently arise from the underground root system. Over time, sto- chastic factors would result in the successive loss of genotypes that do not spread vegetatively, since they are prevented from undergoing sexual repro- duction. The consequences of mowing on the functional dynamics of milkweed populations are apparent. Because of the self-incompatibility of Asclepias meadii, its populations may be sensitive to mini- mum numbers and spatial patterns of genotypes within populations. Even though populations have high levels of heterozygosity, they could still have limited reproductive capacity if they have low num- bers of genotypes. A greater percentage of similar genotypes in haymeadows would inflate population numbers, while effective population size (М,) re- mained relatively small. Even if large numbers of genotypes are present in haymeadows, the potential for crossing among genetically identical individuals and for consequent reproductive failure is in- creased in these populations because most local movement of pollinators will be within clones (Pleasants, 1991). vents sexual reproduction in haymeadows, regard- less of their genetic structure. The maintenance of most genetic diversity within populations has important implications for restora- tion and management of Asclepias meadii. This structuring of genetic diversity may be selectively Finally, repeated mowing pre- advantageous by maximizing the number of differ- ent compatibility types, thereby avoiding inbreed- ing among related individuals. Although plants with different multiple-locus allozyme genotypes are ge- netically distinct, they may still share identical al- leles at a compatibility locus, thus preventing sex- 1991). Such genetic processes may limit reproduction in extremely ual reproduction (Les et al., small A. meadii populations, such as those in Saline Co., Illinois. Demographic factors such as physical ЙЕСІНІ of population subdivisions and non-syn- chronous flowering of compatible plants may pre- vent pollen transfer among these populations. Al- though five different genotypes were detected among the four Saline Co. populations (Table 2), such genetic, demographic, and stochastic factors are apparently preventing sexual reproduction, since it has not occurred in this population. Finally, reproduction in small milkweed populations may also be limited by lack of appropriate pollinators, even if sexually compatible plants are present. Such factors may be in operation in the subdivided Weimer Hill population, where hierarchical F-sta- tistics found a lower percentage of heterozygosity among population subdivisions (6596) than that found among all milkweed populations (7496 However, the Weimer Hill population regularly pro- duces о seeds via natural pollination (Bowles et al., 1998). он intervention, small, fragmented, clonal populations of Asclepias meadii, such as those in the eastern part of the range or possibly those in small haymeadows, appear to have relatively low opportunities for sexual reproduction and therefore high extinction probabilities. Because of the high proportion of rare or unique alleles within popula- tions, a large reservoir of genetic variation would be lost with each extinction. To achieve greater vi- ability and evolutionary potential in small popula- tions, genotype diversity should be maximized through either pollen flow or the introduction of ad- ditional, genetically different plants. An argument that local genotypes will be lost through genetic swamping is not relevant, since they would almost certainly be lost otherwise through attrition and lack of sexual reproduction, and once genes are introduced, local habitat selection can act on novel combinations. The same consideration applies to the restoration of new milkweed populations, re- quiring the establishment of many genetically dif- ferent individuals to maximize reproduction, evo- lutionary potential, and high population growth rates. The questions then arise: How many geno- types must be introduced to restore populational viability, and what seed sources should be used? Approximately 30 allozyme genotypes were detect- ed at Weimer Hill, yet it is not presently clear that this represents the minimum number that should be used for restoration purposes. If the primary re- covery objective for А. meadii is to maximize levels of genetic diversity and numbers of genotypes with- Volume 85, Number 1 1998 Tecic et al. 107 Genetic Variability in populations, then replicating the diversity found in natural viable populations is a logical paradigm. If, as suggested earlier, there has been little selec- tion for genetic change across the range of A. mea- dii, populations in unglaciated Missouri or Kansas would thus be suitable sources of dni material for population restoration in any part of this spe- Multiple populations “lso might be used to collectively maintain higher levels of ge- kd cies range. netic diversity by management as a metapopulation through periodic transfer of pollen or seed among sites to maintain high levels of diversity and avoid negative effects of inbreeding. nother concern is that long-distance crossing for population recovery could potentially disrupt lo- cally co-adapted gene complexes and cause out- breeding depression in naturally evolving popula- However, heterosis resulting from such crosses might outweigh any deleterious conse- quences (Fenster & Dudash, 1994), and such com- plexes might be regularly broken in outcrossing though out- breeding depression may occur in milkweed spe- cies (Wyatt, 1976), no evidence has been found for optimal outcrossing because of their usually large neighborhood sizes (Broyles & Wyatt, 1991; Wyatt & Broyles, 1994). No differences were seen in seed tions. species with large neighborhoods. A germination percentages among natural and geo- graphically distant crosses of A. meadii (Bowles et al., ). This study also showed that seedlings from distant crosses were competitively superior to seedlings derived from natural crosses, suggesting that heterosis may counteract negative effects of outbreeding. Clearly, these experiments must be ex- tended to the field to ascertain long-term fitness components relevant to restoration efforts. CONCLUSIONS Conservation efforts aimed at increasing popu- lation size and stability of Asclepias meadii can be augmented with genetic data. Most genetic varia- tion in А. meadii is contained within populations, and genetic analyses do not provide conclusive ev- idence for geographic patterning of genetic varia- tion among А. meadii populations. Regardless of genetic differences that may occur among popula- tions, the high level of allozyme diversity within extant populations indicates that restoration should attempt to maximize genetic diversity. If different genotypes are correlated with the outcrossing breeding system of milkweeds, then the number of genetically different individuals in restorations must also be maximized. The fragmented eastern populations of A. meadii are apparently too small to allow population recovery or restoration from their in situ genotypes, and hence supplemental propagule sources must be identified among the re- maining Missouri and Kansas populations for effec- tive restoration. This selection process should bal- nce the maintenance of potential genetic differences across the range of the species against for maximizing genetic diversity in res- torations. Experimentation is needed to evaluate the genetic consequences of population restoration of А. meadii, especially the effects of long-distance the nee crossing among geographically different popula- tions and the long-distance movement of genotypes. Literature Cited Alexander, H. M., N. S. Slade & W. D. Kettle. 1997. Application of mark- fone n to estimation of the population size of plants. Ecology 78: 1230-1237. A 1992. Applications of molec- ular genetic techniques to conservatio small rae Biol. Conservation 61: 133-1 Asins, M. J. & E. A. Carbonnell. 1987. Concepts in- ar in measuring genetic Fundit and its impor- tance in 2. of plant genetic resources. Evol. Tre nds РІ. : 51—61. E. ar R. Kohn. 1991. Genetic and evo- 2 El 1: = Ф = uu "d с e = Ф Y | = > vation of Rare Plants. Oxford Univ. Pre ess, New Yo E Betz, R. F. . Ecology of Mead's milkweed (Asclepias meadii Шы Pp. 187-191 in T. B. Bragg & J. Stub- bendieck (editors), Proceedings of the Eleventh North American Prairie Conference. Univ. Nebraska at Lin- coin. ; Н. F. Lamp. 1992. Flower, pod, and seed pro- duction in кер вресїев of ew eeds (Asclepias). Pp. 25-30 іп D. D. Smith . Jacobs (editors), ise ERE a Vanishing Me. y erbe the welfth North Ameri Prairie Conference. Univ. . гп М“ Cedar R. D. Struven, 4 г Wall & F. B. Heitler. 1994. Insect pollinators of 12 milkweed (Asclepias Pp. 45-60 in В. С. Wickett, P. D. Le & P. Pratt (editors), Proceedings of the Thirteenth North American Prairie Conference. Department of Parks & reation, Windsor, Ontario, Billington. Н. L. 1991. Effect a ок size on ge- үү dere in a dioecious conifer. Conservation Biol. : 115-1 Bowles. M. J. L. McBride & R. F. Bet an- agement = restoration ecology of the pire thirest- ened Mead’s milkweed, Asclepias meadii еее: сеае). Ann. Missouri Bot. Сага. = 2r Broyles, S. nity ie sis in a natural popula of езе «кайи, Multiple pater- nity, functional gender, and the “pollen-donation hy- pothesis." Evolution 44: 1454-1468. & ective pollen dispersal in a natural polio of ¡Ari exaltata: The influence of pollinator behavior, genetic induite and mating success. Amer. Naturalist 138: 1239—1249 ———— а —— 199 3a. The consequences of self-pol- 108 Annals of the Missouri Botanical Garden lination in 42. 172 а self-incompatible milk- weed. Amer. J. prs 41-44. | сть. Allozyme diversity and ge- netic structure in 1. Appalachian жылы of poke milkweed, Asclepias exaltata. Syst. Bot. 18: 18— 30. Cavalli-Sforza, L. L. ©. Edwards. 1967. Phylo- genetic analysis: Models and estimation procedures. ee 21: 550- ПеМашо, М. М. Ted pe of breeding system to rarity in the Lakeside ipe ‚ menoxys acaulis var. p лш» Biol. 550. 994. Development and implementation of a re- covery program for the federal threatened Lakeside dai- sy (Hymenoxys acaulis var. glabra). Pp. 208—321 in M. L. Bowles & C. J. Whelan (editors), Restoration of En- dangered Species: Conceptual Issues, Planning ae Im- plementation Cambridge Univ. Press, Cambridg K. Dolan, R. W. 1994. Patterns of isozyme variation in г lation to population size, isolation, and 2. history in royal catchfly (Silene regia: Caryophyllaceae). Amer. J. Bot. 8l: 065 ‹ 972. Бан, A. L. & R. Wyatt. 1994. Population genetics of the rare Asclepias texana and its Е read sister spe- les, А. perennis. Syst. Bot. 19: 291-307 A. & K. E. Moliner 1991. cie Con- servation of Rare Plants. Oxford Univ. Press, New York. M. R. Dudas; M Genetic чы ~ tion. Pp. tors), 11 of Endange nd Species Issues, Planning and нашы Cambridge Univ. Press, Cambridge, U. 1925. Persistence of d in m iated . Mem. Amer. Acad. Arts 15 Fernald, M. L. areas of boreal America -242. Fitch, H. S. & E. R. Hall. 1978. A 20-year record of succession on reseeded fields of tallgrass prairie on the m kefeller c tract. Museum of Natural His- ry, Univ. Kansas, Lawrence. & W. г Kettle. 1988. Kansas ecological reserves (Univ. of Kansas natural areas). Trans. Kansas Acad. 1: 6 S. 1. Guttman. 1996. Spatial and temporal genetic structure of Asclepias verticillata (whorled milk- weed) among prairie iue thes in a forested landscape. Car ser J. Bot. 74: 1289-1297. Gilpin, M M. E. Soulé. 1986. Minimum viable En Processes of species extinction. Pp. 19—34 in M. E. Soulé (editor), io Biology. Cam- bridge Du Tres Cambridge, Massachusetts. G га Н.А. The vegetational гу X the Mid- dle West. fm 2. end Pes Godt, M rick. 1996a. Genetic al and morphological ш. in Liatris helleri (As- threatened pu species. Biodiversity & Ganee 4 5: 461-47 & Ф ы pom m. Genetic structure of two en- dangered рис hor plants, Sarracenia е sii and d сета oreophila (Sarraceniaceae). . J. Bot )16-1023. Gottlieb, L. D. & N. F. Weeden. 1981. Correlation be- tween subcellular location and phosphoglucose isom- erase variability. Evolution 35: 1019-1022. Guyette, R. & B. E. Cutter. e ring analysis of a post 2. savanna in the Missouri Ozarks. Nat. Area J. 11: 93- & E McGinnes. 1982. Fire history of an Ozark зой n in Missouri. Trans. Missouri Acad. Sci. НА, J. L. 1. К Godt. 1990. Allozyme diversity in plant ms Pp. 43 np in A. H rown, M. T. Clegg. A. L. Kahler & B. S. Weir (editors), Plant Pop- ulation Genetics, 22 and Genetic Resources. Sinauer, a Massachusetts. Harrison, W. 1988. Endangered and threatened wild- life and 1. Determination of threatened status for Asclepias meadii (Mead's milkweed). Fed. Reg. 53: 33982- Fe d J. 5. & D. А. Levin. 1985. а be- allozyme frequencies and soils characterisites in Gaillardia pulchella (Compositae). eris 39: 1076- 1086. Hickey. А. Vincent & 5. 1. Сиштап. 1991. ко: ariation in running buffalo с 2. volorum. Јана 2. Biol. 5: 309-316. Kahn, A. P. & D. H. Morse. . Pollinium gemination and 2. па 24. іп self- and cross-polli- nated common milkweed Asclepias syriaca. = г. Midl. Naturalist 126: 61—71. Karron, J. D. 1991. Patterns | genetic variation and iw systems in rare plant species. 87-98 in ‚ А. Falk & К. ы Holsinger (editors), 2 "m Cierto of Rare Plants. Oxford Univ. Press, N Ladd. D. 1991. .. Е the role x fire in Mis- souri oak n Pp. ) in G. V. Ebinger . Wi helm im Proc еч ef ‘the Oak Woods. . Workshop. Eastern lllinois Univ., Charleston, Illinois. Ledig, F. T. 1986 in outbreeding plants. Pp. 77-104 in M itor), Conservative Biology. Sinauer, Sunderland, Mas- = т Heterozygosity, heterosis, and fitness Soulé (ed- sachusetts ез, D. H., J. A. Reinartz & E. | Esselman. 1991. Се- netic consequences of rarity in Aster furcatus (Astera- ceae), a threatened, self-inc esi plant. Evolution 45: 1641-1650. Lesica, P., R. Е. Leary, К. W. Allendorf & D. E. Bilder- back. 1988. L: т af genetic diversity within and among populations of an жеш plant, Howellia aquatilis. Conservation Biol. 2: 282. cCune, B. . Multivariate an Maus on the PCORD system, August, 1993 Version. Oregon State Univ., Cor- vallis, Oregon. | 1977. Rare native vascular plants of . Techn. Publ. State Biol. Surv. Kansas 5: А Menges. | Е. 5. 1991. The application of 2” rs population theory to plants. Pp. 45—61 in D. K. E. Holsinger (editors), Genetics and үка» of Rare a 2. Univ. Press, New Mitton, J. . Pierce. ы 2... іп ваша! populations. Genet- ics 95: 1043-1054. Nickrent, D. L. & D. Wiens. 1989. Genetic diversity in the rare California shrub 227 eurekensis (Poly- gonaceae). Syst. Bot. Noamesi, G. K. & H. H. Пп. gs i ani rm of the flora of Ж шай No. 40. Asclepiadaceae— Milkweed family. Trans. Wisconsin Ао: Sci. 46: 107 n. A The "distribution of 114. Olmstead, R. 1989. Phylogeny, phenotypic evolution, and biogeography of the Scutellaria angustifolia complex Volume 85, Number 1 1998 Tecic et al. Genetic Variability 109 (Lamiaceae): Inference from morphological and molec- 0—338. 99]. Evidence for short- Mas csi dis- persal of pollinia i іп Asclepias syriaca L. Funct. Ecol. 5: 75-82. Ridgeway, С. J., S. W. Sherburne € R. D. Lewis. 1970. Polymorphisms in the esterases of Atlantic herring. Trans. Amer. Fish. Soc. 99: 147-151. Rieseberg, L. H. & D. Gerber. 1995. Hybridization in the Catalina Island mountain mahogan ок tras- kiae): RAPD wk свита Biol. € à 5. Zona, L. Aberbom & T. D. Martin. 1. Ну- Бейше in qe pues endemic, Catalina mahogany. Conservation Biol. 3: 52—58. Schaal, B. A. & D. ^ Levin. 1976. The demographic genetics of Liatris cylindracea Michx. (Compositae). Amer. Natu ralist 110: 19 1- 7 | J. Leverich & S H. Rogstad. 199]. Com- айан of methods for as plant conservation E & K. E. виа genetic variation in Pp. 1123-134 in D. A. Falk er (editors), Genetics and Conservation of Rare Plants. Oxford 2. Press, New York. Schoen, D. J. & A. H. D. Bro 1991. Intraspecific vari- ation in population pe diversity and effective popu- lation size correlates with t н вуеш in plants. Proc. Natl. Acad. Sci. USA. 88: 449 Зеауеу, S. К. & K. S. Bawa. 1986. Late-acting self-in- compatibility in angiosperms. Bot. Rev. (Lancaster) 52: 195-2 Shaffer, M. L 1981. Minimum population sizes for spe- cies conservation. BioScience 31: 131—134. Shannon, T. R. & R. Wyatt. 1986. Pollen раар. of Asclepias exaltata: Effects of flower age, dry ime and pollen source. Syst. Bot. 11: 322-325 Sipes, S. D. & P. G. Wolf. 1997, Clonal structure and patterns of allozyme diversity i in the rare endemic Cy- cladenia humilis var. jonesii (Apocynaceae). Amer. J. Bot. 84: 401-4409. Soltis, D. E. & P. S. Soltis. 1989. rins in Plant Bi- ology. Dioscorides Press, Portland, Or ‚ C. H. Haufler, D. C. Darrow & G. J. Gastony. 1983. Starch gel electrophoresis of ferns: А compilation of grinding buffers, gel and electrophoretic buffers and 2 Amer. Fern |. 73: 9-27. С. J. Allan, М. Howe, М. J. Elisens, S. A. Junak 5 L. H. Rieseberg. 1995. Genetic analysis of the endangered island endemic Malacothamnus fas- ciculatus (Nutt.) Greene var. nesioticus (Rob.) Kearn. Malvaceae). кошы Biol. 9: 404—4 “А D. L. & R. B. Selander. 1981. BIOSYS-1: А ORTRAN program p the comprehensive analysis of electrophoretic data in population genetics and system- atics. J. Heredity ч 281-283. Waller, D. . O'Malley & S. C. Gawler. 1987. Genetic ia in the extreme endemic Pedicularis 22. (Scrophulariaceae). Conservation Biol. =. РБ "s б & J. F. Wendel. 1989. 21 ~ plant isozymes. a 46-72 in D. E. Soltis «СР. S. Soltis (ed- . Isozymes in Plant Biology. E had regon. Weller, S. G. 994. The po» of rarity to plant Urge biology. Pp. 298-321 in M. L. Bowles & C. J. Whelan (editors), Restoration of Endangered Spe cies: E Issues, Planning and Imp cu. Cambridge Md Press, Cambridge, U.K. Wendel, J. F. . M. Weeden. 1989. Visualization and cao 2 plant isozymes. Pp. 5—45 in D. E. Sol- & Р. S. Soltis (editors), Isozymes in Plant Biology. Dioscorides Press, Portland, Oregon. Werth, C. R. 1985. Implementing an isozyme laboratory at a field Ron. Virginia J. Sci. 36: 53-76. po B. A. & D. D. Murphy. 1985. Conservation strat- : The effects - fragmentation on extinction. Amer. Naturalist 125: 8 7A Woodson, R. E. 1. The North American species of Asclepias L. Ann. Missouri Bot. Gard. 34: 353-432. Wright, S. 1978. Variability Within and ке Natural Populations. Univ. Chicago Press, Chicag Wyatt, К. 1976. Pollination and fruit-set A ión A reappraisal. Amer. J. Bot. 63: 845-851. ----« . B. Broyles. 1994. Ecology and evolution of reproduction i in milkweeds. Annual Rev. Ecol. Syst. 25: 23441. MANAGEMENT AND M. L. Bowles, J. L. McBride, and R. F. RESTORATION ECOLOGY OF 2° THE FEDERAL THREATENED MEAD'S MILKWEED, ASCLEPIAS MEADII (ASCLEPIADACEAE)! ABSTRACT The federal threatened Asclepias meadii Torr. (Asclepiadaceae) is a perennial, self-inc ompatible prairie о impe riled by habitat destruction and population fragmentation. Many large E persist in prairie haymeadows in Kansas and Missouri despite removal of seed poc y by annual summer mowing. Only a few small populations remain in Illinois, lowa, and northern Missouri. Recovery these small populations and restoration of new populations are recovery objectives for this species. This study was ene ted to determine habitat differences among are о а how а = and fire management affect the structure of A. meadii populations, and to test the effects of differ management treatments on restoration of this species. Soils analysis showed a onse puedes with өөн ин он оп E nutrient- -poor soils, and еи окы dne on calcareous, nutrient-rich soils. Milkweed rame densities were lower in fire-managed prairies than in haymeadows; but burned sites had 68% flowering ramets while haymeadows had oniy 19% flowering ramets. This suggests that burning has selected for greater resource allocation toward sexual reproduction, while annual hay mowing has selected for greater resource allocation toward clonal spread. The Morton Arboretum is conducting experimental restoration a Asclepias meadii in the eastern part of its range, an objective of Federal Recovery Planning. In greenhouse and garden 1... Wii tition from oats as antly reduced seedling growth, with greater growth in artificially s 52 seedlings. At seven restoration sites in Illinois and northern Indiana, significant variation in milkweed pun пайоп; survivor BR a growth was caused ds weather, differences оа sites, and site management. Seedlings were vulnerable to drought, with greater survivorship when rainfall was 200% of normal. Planted j jave ‘nile milkweeds "reater survivorship than seedlings, and less sensitivity to d ен а oni growth and survivorship also occurred in al than in unburned plots at three sites, but not in all life-stages. Propagated Ap from Missouri seed sources were larger than Kansas plants in the garden, but not in the field. ыле work is needed to determine if restored populations can become viable, and if there are negative effects of crossing and 1. 'ating genotypes. Managing for viable populations of endangered history stages (Fenster & Dudash, 1994; Weller, species requires knowledge of their life-history and 1994; Pavlik, 1994, 1996; Guerrant, 1996). This habitat requirements, reproductive biology, and the paper examines factors that affect the management demographic, genetic, and ecological traits that апа restoration ecology of Mead's milkweed (Ascle- make them vulnerable to extinction processes (Gil- pias meadii Torr., Asclepiadaceae), a federal threat- іп & Soulé, 1986; Lande, 1988; Menges, 1986, ened (Harrison, 1988) plant essentially restricted 1991, 1992). Similar information is roued to re- to the virgin tallgrass prairies of the midwestern cover and restore new populations of these species, ^ United States (Betz, 1989). To better understand especially how site-management and stochastic en- how to manage and restore this species, we com- vironmental processes affect population growth pare habitat characteristics across its range, ex- based on their genetic attributes, reproductive amine the effects of management on population characteristics, and survivorship at different life structure, and use experimental propagation and ! We thank the Illinois Department 9r Conservation, Illinois Conservation Foundation, Indiana Division of Nature Preserves, U.S. Fish & Wildlife Service, U.S. Forest Service, and the Chevron Corporation for funding our restoration work on Mead's milkweed. We also ‘han the many agencies and peupie who provided extremely helpful field assistance and support for se кше. We are grateful to John Bacone, Marcy DeMauro, Craig Freeman, Cloyce Hedge, Craig Johnson, Mike Jones, Dave Ketzner, Amelia Orton- Palme m. Tom Post Dave Mauger, John Schwegman, Beth Shimp. Larry Stritch, and Paul Tessene for permission, fundi ield assistance, or application of site-management treatments. We also thank the Illinois Nature Preserves Commission, . of the Grand Prairie, and Natural Areas uardians for permission to work at Pellville and Munson prairies, and Dean Kettle, University of Kansas, for A. meadii data ion the Rockefeller Prairie. Finally, we thank Mich E Klonowski for graphic zm assistance, Chris Whelan for statistic ‘al advice, and Steve Broyles, Kay Havens, and Don Les for extremely dr ? The Morton Arboretum, 4100 Illinois Route 53, Lisle, ш: 60532-1293, ANN. Missouni Вот. Garb. 85: 110-125. 1998. Volume 85, Number 1 Bowles et al. 111 998 Management and Restoration Ecology = ir SCONS 5% > IO|W 6,7 1 O 4 » 0 О ө О 2 5 e o YO ОЛ s N О UI RII 14,15 —— 11,12 eO —5 13 e/o? К/А N 5А 5 еее О ө 3 ее 8,9 e e e 10 16 17 20 19 18 Site Number, Name & State 1 Palatine, IL 6 Cummings, IA 11 Colyer, KS 16 Cook Meadow, MO 2 £ 7 Martensdale, IA 12 Jack's, KS 17 Niawathe, MO 3 Saline, IL 8 Garnett, KS 13 Flower Hill, KS 18 Weimer Hill, MO 4 Woodside, IA 9 Sunset, KS 14 French Creek, KS 19 Paint Brush, MO 5 Flaherty, IA 10 Hinton Creek, KS 15 Rockefeller, KS 20 Wah-kon-Ta, MO Figure l. Distribution by county of Asclepias meadii. Closed circles are counties with extant populations; open circles are counties from which populations have been extirpated. Study site locations are numbered corresponding to Table restoration to understand the effects of environmen- tal variation and site management on population establishment. DISTRIBUTION AND STATUS The range of Asclepias теади follows the tall- grass prairie, extending eastward from Kansas through Missouri, Iowa, and Illinois to southwestern Wisconsin and northwestern Indiana (Fig. 1). Be- cause of conversion of tallgrass prairie to agricul- ture, А. meadii has been reduced to about 150 pop- ulations, primarily in Kansas and western Missouri native haymeadows. These haymeadows are usually summer-mowed, which removes milkweed pods (follicles), preventing seed dispersal and sexual re- 1977; Betz, 1989). Hay mowing has occurred almost annually for a century (Fitch & Hall, 1978), yet many haymeadows con- tain hundreds of milkweed ramets. One former hay- meadow, the Rockefeller Prairie, Jefferson Co., Kansas, has been fire-managed since the 1950s and production (McGregor, may contain 200 or more plants (Alexander et al., 1997). Less than 20 former haymeadow populations have been preserved as public prairies in Missouri since the mid 1980s, and only one site contains a large population (Smith, 1997). A metapopulation of A. meadii occurs across a complex of non-mowe igneous glades in southeast Iron and Reynolds Cos., groups has more than 100 plants and occurs at the Missouri. The largest of these population 112 Annals of the Missouri Botanical Garden Figure 2. Flowering umbel and follicle of Mead's milkweed (Asclepias теаай). жы; with permission from Erigenia (Journal of the Illinois Native Plant Society); drawing by Nancy Hart Stieber. fire-managed Weimer Hill site in Iron Co. East- ward, small colonies occur at two sites in northern Missouri, six Iowa sites, and five sites in Illinois; populations are extirpated from Wisconsin and In- diana (Betz, 1989; M. L. Bowles, pers. obs.). BIOLOGY Asclepias meadii (Fig. 2) is a long-lived rhizo- matous perennial herb. As in many prairie plants, dormant season fire appears to stimulate flowering (Betz, 1989; H. M. Alexander, pers. comm.). Mature plants usually have 6-12 paired leaves and a single terminal umbel with about 12 flowers and usually produce a single narrow pod (per plant) with about 60 seeds (Betz, 1989). Flowers within an umbel are open for about 5—6 days, and flowering occurs for about 10—12 days within populations. Plants flower as early as late May in the south through mid to late June in the north, depending upon yearly grow- ing season conditions. Pollinia are most frequently removed by miner bees (Anthophora sp.), or by small bumblebees (Bombus sp.) (Betz et al., 1994). In a seven-year study, 7796 of over 100 A. meadii ramets flowered annually, but less than 6.496 ma- tured pods, averaging 61 seeds/pod and 47.696 seed germination (Betz, 1989). This correlates with low levels of pod production reported for most milk- weeds (Wyatt, 1976), in which reproduction is reg- ulated by their breeding system and by resource allocation (Willson & Price, 1980). Most milkweed species are self-incompatible, requiring crosses be- tween genetically different individuals to produce viable seeds (Kephart, 1981; Shannon & Wyatt, 1986; Kahn & Morse, 1991; Broyles & Wyatt, Volume 85, Number 1 1998 Bowles et al. Management and Restoration Ecology 113 199], 1993; Wyatt & Broyles, 1994). The apparent longevity of А. meadii and its restriction to virgin prairies suggests that it is a late-successional spe- cies characterized by poor colonizing but good com- petitive abilities. As a result, seedling establish- ment may be infrequent but is probably required for long-term population maintenance and neces- sary for population establishment. As yet, little ex- perimental information is available about seedling ecology (Betz, 1989). sclepias meadii is genetically diverse, with 7496 of its allozyme diversity maintained within popu- lations or metapopulations and no geographic ge- netic pattern among populations (Tecic et al., 1998). Allozyme samples from the Rockefeller and Weimer Hill populations have found 15 or more genotypes per site, with small ramet: genet ratios, while haymeadows and small populations are high- ly clonal, with fewer genotypes and large ramet: genet ratios (Tecic et al., 1998). Formerly wide- spread species with outcrossing breeding systems become vulnerable to extinction because of lowered reproductive potential in fragmented populations (Schaal et al., 1991; Weller, 1994; Les et al., 1991; DeMauro, 1993). Such is the case for A. meadii. Because of its self-incompatible breeding system, small fragmented populations that are reduced to single clones, such as in Illinois, no longer produce seeds and are vulnerable to stochastic extinction processes. Viable restored populations of self-in- compatible species should contain high levels of genetic diversity, which will enhance outcrossing and seed production while lowering inbreeding (DeMauro, 1993). For A. meadii, this may require infusion of large numbers of different genotypes from across the range of the species. Such efforts would maximize evolutionary potential and de- crease inbreeding, but could alter historic lineages and produce outbreeding depression if co-adapted gene complexes exist and are disrupted (Fenster & Dudash, 1994). This issue is often contentious for restoration ecologists and will only be resolved with case-by-case experimentation among different plant groups (Bowles & Whelan, 1994). Ex-siTU CONSERVATION To facilitate recovery of Asclepias meadii, The Morton Arboretum has assembled a genetically di- verse garden population and nursery to provide a propagule source for population restoration and re- search. The garden environment consists of 1 X 2- m elevated beds filled to 0.3 m with wood chips, in which potted milkweeds are propagated. The pots allow isolation of plants and removal for arti- ficial cross-pollination. Seed sources have included extant populations and herbarium specimens rep- resenting western Missouri and Kansas (Bowles et al., 1993a). Important seed sources have included the Rockefeller Prairie and Weimer Hill, which have been supplemented by pollen crosses from fragmented eastern populations in southern (Saline Co.) and central (Ford Co.) Illinois, northern Mis- souri (Harrison Co.), and southern Iowa (Adair Co.). y 1996, the garden population contained 58 adult plants representing 28 different genotypes. OBJECTIVES Our objectives were to analyze ecological factors affecting the distribution, population structure, and restoration of Asclepias meadii. In this paper, we first examine the distribution of A. meadii in rela- tion to soil characteristics, which, based on the lack of a strong geographic genetic pattern in this spe- cies, might be expected to show little variation. We also compare its population structure under re- gimes of hay mowing and burning, the primary management alternatives for maintaining and re- storing prairie. If mowing removes live biomass and prevents sexual reproduction but not vegetative spread, we would expect spatial population struc- ture to correspond to genetic differences found be- tween mowed and burned populations. We then use greenhouse and garden experiments to compare germination and growth among different seed sources and to test seedling growth under different competition and moisture conditions. To further ex- amine factors affecting restoration potential of this species, we compare survivorship and growth of seeds and juvenile milkweeds from different sources planted into native prairie habitats under experimental burning treatments and stochastic cli- matic variation. METHODS STUDY AREAS Studies were conducted at 1 former and 19 cur- rent Asclepias meadii stations in Kansas, Missouri, Iowa, and Illinois (Table 1). We collected soil sam- ples during 1992 and 1993 from 18 sites, and pop- ulation data during the 1992 flowering period from 10 sites. These included the fire-managed Rocke- feller Prairie and Weimer Hill, both of which were spring-burned in 1992, five annually mowed pri- vate Kansas prairie haymeadows, and three Mis- souri former haymeadows that were protected in the mid 1980s. Two of the Missouri sites are now man- aged in hay-burn-rest rotations, while flowering 114 Annals of the Missouri Botanical Garden e 1. Me ad's milkweed (Asclepias meadii) study sites sampled for soils (5), ab plant size (P). ' and for population structure and sites were sampled for soils in Ford and Saline counties. See Figure 1 for site locations. Site number & name State & county Samples Management history 1) Palatine Illinois/Cook 5 railroad dina ини destroyed) 2) Ford Illinois/Ford S (2) ailroad prairie (two samples) 3) Salin Illinois/Saline S (2) A алк (two «Ж 4) Woodside lowa/Adair 5 haymeadow (mowed in Sep.) 5) Flaherty lowa/Clarke 5 preserve 6) Cummings lowa/Warren 5 preserve 7) Martensdale lowa/Warren S preserve 8) Garnett Kansas/Anderson P haymeadow 9) Sunset Kansas/Anderson S.P haymeadow 10) Hinton Creek Kansas/Bourbon S.P haymeadow 11) Colyer Kansas/Douglas S.P J haymeadow 12) Jack's Kansas/Douglas 5 haymeadow 13) Fowler Hill Kansas/Franklin 5 haymeadow 14) French Creek Kansas/Jefferson S.P haymeadow 15) Rockefeller Kansas/Jefferson S.P former haymeadow (burned every 1—3 yrs. since 1956) 16) Cook Meadow Missouri/Barton S.P Паутеадом (unmowed when milkweeds flower since ~ 17) Niawathe Missouri/Dade S.P former haymeadow (burn/hay/rest rotation since — 1980) 18) Weimer Hill Missouri/Iron S.P — glade/metapopulation (burned) 19) Paint Brush Missouri/Pettis 5 former haymeadow (burn/hay/rest rotation since — 1980) 20) Wah-kon-Tah Missouri/St. Clair P former haymeadow (burn/hay/rest rotation since — 1980) milkweeds are left unmowed in the third site. Greenhouse and garden studies were conducted at the Morton Arboretum, Lisle, Illinois. Restoration planting experiments were conducted at seven prai- rie habitats in northern Illinois and northeastern Indiana (Table 2). None of these sites contain native А. meadii populations, but all lie within the former range of the species. All sites are protected and managed by prescribed burning. They provide an among-site drainage gradient from dry-mesic to me- sic, and a successional gradient from early- to late- successional vegetation (e.g., Betz, 1989; Betz & Lamp, 1989; Bliss & Cox, 1964) SOIL SAMPLING AND ANALYSIS Composite soil collections were made from each site by pooling multiple A-horizon samples from milkweed habitat into a 2-quart plastic bag that was transported to the Morton Arboretum Soils Lab, Lisle, Illinois. Soils were refrigerated until shipped for analysis by the A & L Great Lakes Laboratories, Fort Wayne, Indiana. Samples were tested for pH, milli-equivalents cation exchange capacity (CEC), percent organic matter (£6 OM), and parts per mil- lion (ppm) phosphorous (P), potassium (K), mag- nesium (Mg), and calcium (CA), following methods of Page et al. (1982). Sample sites were ordinated by their soils data on PCORD software (McCune, 1993) using the Bray-Curtis technique (Beals, 1984) with a variance-regression endpoint selection and Euclidean distance measure. Samples also were clustered by their soils data on PCORD using Ward's method with a Euclidean distance measure (Sneath & Sokal, 1973). Sample means and stan- dard deviations were calculated for geographically similar groups. SAMPLING AND ANALYSIS OF NATURAL POPULATIONS With one exception, all Asclepias meadii sites were sampled from stratified random transects through milkweed populations during their flower- ing period, when plants can be most easily located (Alexander et al., ment of plots along transects rarely encountered 1997). Because random place- milkweeds, flowering plants were randomly select- ed as centers of non-overlapping 10-m? circular plots, in which the numbers of flowering and non- flowering plants were counted. This maximized sampling of flowering plants across sites, and also helped focus plot placement on the clonal ramet clusters of this species. For Rockefeller Prairie, density data were obtained by placing а 10-m? grid over a 2-m”-resolution grid map of milkweed clone locations. At this scale, the comparatively small ra- met clusters at Rockefeller (Alexander et al., 1997) were easily quantified within sampling plots. All Volume 85, Number 1 1998 Bowles et al. 115 . and Restoration Ecology Location, site characteristics, and number of experimental Mead's milkweed (Asclepias meadii) plantings made at Mead's milkweed restoration sites. Table 2. Number planted (1994—1996) Tubers Genotypes Seeds Management treatment Drainage/successional stage Size 16 ha state County, Site name 117 Burned-unburned 1994 Lake Co., Indiana Will Co., Biesecker (Cook) Vermont Cemetery Pellville Cemetery Hickory Creek dry-mesic/late-succ. mesic/late-succ. Illinois 140 Burned-unburned 1995 mesic/late-succ. ord Co., Illinois Will Co., Illinois DuPage Co., Illinois dry-mesic/early-succ. 103 Burned 1994—1995 dry-mesic/mid-succ. 20 ha Schulenberg Prairie N dry-mesic/late-succ. a 47 ha Munson Cemetery | Unburned 1995 mesic/mid-succ. DuPage Co.. Illinois W. Chicago Prairie ramets sampled were quantified by their reproduc- tive status and size. The extensive vegetative spread of ramets in haymeadows prevented quan- tifying reproductive status at the genet level. For each flowering ramet, the number of flowers per el was recorded. For each plant, the area (length X width) of one randomly chosen leaf from the largest pair of leaves was recorded. This leaf- area was transformed to a plant leaf-area index by multiplying the measured leaf-area for each plant times the number of leaves on the plants and Duncan's multiple range tests (Steele & Torrie, 1960) were used to compare morphological and population structural differences between fire-managed and mowed populations. The burned Rockefeller and Weimer Hill sites were re- tained as separate treatments because they occupy ecologically different habitats. To attain similar, but not equal, sample sizes, data were pooled within the Missouri former haymeadows, and within the Kansas current haymeadows. Separate tests were made of the null hypotheses that there were no sig- nificant differences between these four groups for One-way ANOV. mean ramet plot densities, mean percent flowering ramet densities, mean umbel size, mean leaf-area indices of flowering and nonflowering plants, and mean total leaf-area of all A. meadii ramets per plot. GREENHOUSE AND GARDEN EXPERIMENTS АП seeds used in propagation studies were moist-stratified in Petri dishes at 5?C for 4 months efore planting in 10-cm-deep flats filled with a mixture of equal parts standard greenhouse soil and prairie loam (Betz, 1989). Seeds were planted in mid May and germinated within 10 days. Flats were retained in the greenhouse until seedlings had de- veloped one pair of leaves, after which they were transferred outside into full sun, which is required for optimum growth (Betz, 1989). Seedling perfor- mance was quantified by percent germination for seed pods and seed sources, while performance of older plants was quantified by leaf-area indices af- ter their first year of growth. A one-way AN was used to compare mean percent seed germina- tion of pods from naturally pollinated Rockefeller and Weimer Hill populations against pods derived from garden crosses using geographically distant seed sources. To assess the effects of grass competition on moisture availability and milkweed seedling growth, we established a drainage gradient among 12 seedling flats (40 seeds/flat), randomly chosen for placement either above (n — 4 flats), midway (n 116 Annals of the Missouri Botanical Garden = 4), or below (n = 4) the level of a 30-cm-deep ] X 2-m sand bed kept moist by watering. Flats placed below the sand level were expected to have the poorest drainage and highest moisture levels because of capillary movement from the sand bed; elevated flats were expected to have greatest drain- age. For each drainage position, two flats were planted with seeds (pooled from multiple pods) ob- tained by long-distance out-crossing of pollen from natural populations to our garden plants, and two were planted with naturally pollinated seeds pooled from Rockefeller and Weimer Hill pods. After seed- ling germination, half of the flats in each moisture/ seed source treatment were randomly selected and sown with annual oats (Avena sativa L., Gramineae). This oat species has similar photosynthetic re- sponses to native prairie grasses such as Sorghas- trum nutans (L.) Nash (Fay € Knapp, 1993), and its annual growth habit can rapidly develop a fi- brous root system that can compete with forb seed- lings for water and other resources. The flats were watered periodically to sustain seedling growth, al- lowing surficial drying between watering periods. We measured the moisture gradient during the growing season by inserting an electrical conduc- tivity moisture meter probe into each flat 24 hrs. after watering. Analog readings ranged from 1 (dry) to 8 (wet). After the milkweed seedlings and oats had become dormant, all seedlings were excavated and their tubers weighed. Effects of the drainage and oats on mean moisture level readings were test- ed with a fixed model two-factorial ANOVA (Sokal & Rohlf, 1981). Effects of the moisture gradient, oats, and seed sources on mean milkweed tuber weight were measured in a mixed model three-level к ANOVA (Sokal & Rohlf, the hierarchical effects of oats, moisture gradient, and seed source by flat. 1981) comparing FIELD ESTABLISHMENT EXPERIMENTS Experimental plantings of seeds and seedlings were conducted at the seven study sites between 1994 and 1996. Our strategy was to maximize num- bers of genetically different individuals within sites so as to increase potential for compatible outcross- ing and seed production. This resulted in planting of 1232 seeds and 630 juvenile plants representing | ilkweed seeds and juveniles were planted in early May prior to the breaking of prairie plant dormancy. Clusters of five seeds were planted 1 cm deep in a 1-dm? area in which the soil was loosened with a hand trowel. One-year-old dormant milkweed tubers were re- moved from flats, weighed, and packed with sphag- 53 different genotypes (Table num in plastic bags for translocation. They were planted upright with buds about 3 cm below the soil surface, in 10-cm-deep incisions made in the prairie with a tile spade. АП plantings were watered immediately. Seeds and tubers were planted within 1-m? plots placed along permanently marked stratified random transects. Usually two seedling clusters and four tubers were planted per plot, and were placed with- in interstitial patches between bunchgrasses. The transects crossed burned and non-burned sections of Biesecker in 1994, Pellville in 1995, and Mun- son in 1996, allowing comparison of these manage- ment treatments on growth and survivorship. Seed- lings were monitored for first-year survivorship, while juvenile plants that emerged from tubers were monitored for survivorship and measured for leaf- area indices. Percent survivorship was determined over time for annual cohorts and compared between seed and tuber plantings. Survivorship was com- pared between burn and nonburn treatments using Chi-square analysis of 2 X 2 contingency tables, and mean leaf-area indices were compared between the same treatments using t-tests. We also com- pared differences in mean leaf-area indices among planting sites and seed sources using one-way AN- s. А two-way VA was used to compare leaf-area indices between garden and field sites planted with Rockefeller and Weimer Hill seed sources using seeds pooled from multiple pods. RESULTS SOIL CHARACTERISTICS Soil samples ordinated and clustered along a chemical- and nutrient-concentration gradient cor- responding to their geographic distribution (Fig. 3). Missouri and southern Illinois sites had the lowest scores on the first axis, Kansas sites were inter- mediate, and Iowa and northern Illinois sites had the highest scores. All soil variables were positively correlated with the first axis, with ‚ and ppm Mg, K, and Ca having correlations greater than 0.5. In general, Missouri and southern Illinois sites are acid and nutrient-poor, Kansas sites are interme- diate, and Iowa and northern Illinois sites are cal- careous and nutrient-rich (Table 3). Missouri and southern Illinois had comparatively low mean p Zo OM, СЕС, and ppm Ca, but comparatively high ppm P. Kansas samples were generally intermediate but variable, with comparatively high mean pH, low mean % OM, intermediate СЕС, and extremely low ppm Р. Iowa and northern Illinois had the highest mean values for all soil variables. Volume 85, Number 1 998 Bowles et al. Management and Restoration Ecology 117 Ахїз 2 Axis 1 Figure 3. Bray-Curtis ordination of Mead's milkweed (Asclepias meadii) habitats by soils characteristics. Axis 1 represents a geographic gradient with increasing рН, % OM, CEC, and nutrient concentrations (see Table 3). Cir- cles are clusters кек by Ward's method. Number codes: northern Illinois (1—2), ere Illinois (3), Тома (4-7). Kansas (9-1 5. Missouri (16—19). See Table 1 for site names and Figure 1 for site locations. SITE-MANAGEMENT EFFECTS ON POPULATION STRUCTURE Asclepias meadii ramet densities were signifi- cantly higher in Kansas haymeadows than in Rockefeller Prairie, and were intermediate in the former Missouri haymeadows and Weimer Hill (Ta- ble 4). At Rockefeller and Weimer Hill, 68% of all ramets were flowering, while only 18.696 of all plants flowered in haymeadows. On a plot basis, the percentage of flowering ramets averaged under 3296 in Kansas and former Missouri haymeadows, but more than 60% at Weimer Hill and more than 80% at Rockefeller (Table 4). As a result, the density of flowering plants, but not nonflowering plants, was similar across all study sites and there was no sig- nificant difference in mean plot leaf-area across all sites (Table 4). However, flowering ramets were larger than nonflowering ramets across all sites, and Rockefeller plants were larger than haymeadow plants in flowering and nonflowering groups (Table 4). Flowering plant umbels were larger in burned habitats, averaging about 12 flowers, and had about 10 or fewer flowers in Kansas and former Missouri haymeadows (Table 4). GREENHOUSE AND GARDEN EXPERIMENTS Between 1993 and 1996, seed production among our garden plants averaged 56 seeds/pod, which Geographic differences in mean (+ s.e.) soil chemistry values and nutrient concentrations of habitats supporting extant or former Mead's milkweed (Asclepias meadii) Table 3. populations. See Figure 1 and Table 1 for location and description of sample areas. Sample CEC pH ppmP ppmK ppmMg ppmCa % OM size Community Region N. Illinois 412.86 + 108.85 2992.9 + 921.7 229 + 87.88 75.43 + 41.69 6.71 + 3.27 6.67 + 0.82 6.47 + 0.77 20.64 + 3.27 11.09 + 4.00 5.54 + 0.96 Prairie Prairie/Haymeadow 2200 + 754.43 255 = 38.19 1.71 = 1.25 15.27 + 3.48 Kansas 33 + 530.49 758 116.67 + 54.74 23.43 + 73.0 6.67 + 1.25 6.30 + 2.24 5.05 + 1.97 Haymeadow/glade & S. Illinois 118 Annals of the Missouri Botanical Garden Site management effects on mean ramet and mean percent flowering ramet density per 10-m? plot, mean umbel size, mean leaf-area index, and average plot leaf-area of Mead's milkweed (Asclepias meadii) in burned prairies (Rockefeller and Weimer Hill), former haymeadows (Missouri), and current haymeadows (Kansas). Similar lower case letters indicate similar means across variables with Duncan's multiple range test at Р = 0.05. Table 4. Test statistic and probability Weimer Hill Missouri prairies Kansas haymeadows 3.625ађ = 0.84 Rockefeller 2.525a + 0.31 Variable 6.15b + 0.91 F — 5.16, P = 0.0024 Mean ramet n=8 67.26a + 9.12 40 80.07a + 4.87 density + s.e. 19.71, P < 0.0001 F= 5.40 34 = 10.09bed + 0.52 26.05b = 9.0 31.29b n Mean % flowering = 15 9.27cd + 0.65 n 40 11.89abc = 0.35 n ramet density + s.e F — 4.68, P — 0.0038 1.15 zum 1 12.40аЬ Mean flowers 45 142.89b + 12.4 n 162.33 ab 3 per umbel = s.e. 54.71. P « 0.0001 Е = 13 4. 137.52b + 2 = 27.75 1 92.365 + 17.76 + 8.03 196.09a Mean leaf-area index n= 45 e ~ of flowering plants + s.e. F = 15.89, P < 0.0001 + 10.85 138.54a Mean leaf-area index of ~ s.e. + nonflowering plants 551.70 + 106.85 Mean plot 605.42 + 118.18 33 243.8] + 56.34 1 — De n n =8 n= leaf-area + s.e. was similar to the 60 seeds/pod average found by Betz (1989) from 1965 to 1971 for native plants in Missouri and Kansas railroad prairies. Under greenhouse conditions, 74.3% of 1665 seeds ger- minated. Also with greenhouse propagation, there was no significant difference (F = 0.78, 0.4679) in mean percent germination per pod among our wild-collected seed from Weimer Hill (n = 11 pods, x = 60.73 + 10.3), Rockefeller Prairie (n = 11 pods, x = 71.12 + 8.4), and pods derived from garden crosses among pea different seed sources (n = 17 pods, x = 74.65 + 6.4). Four life-stage classes could be sonido in garden- propagated plants: first-year seedlings (< 15 cm high, < 5 cm long linear leaves), second-year ju- veniles (> 15 cm high, cm lanceolate leaves), and flowering or nonflowering adults (> 30 cm high, > 1 cm broad lanceolate-sagittate leaves with cordate bases). In the field, ramets of adults can revert to juvenile form in successive years (Betz, 1989). The presence of oats significantly reduced soil moisture levels in milkweed flats, reversing the re- lationship between moisture level and drainage se- quence (Table 5). Overall moisture levels were higher in flats without oats, where the highest drainage position had the lowest mean moisture level. With oats, the moisture gradient was re- versed, with a higher mean moisture level at the highest drainage position. Although drainage per tion did not affect tuber weight (F — 0.9095), the presence of oats significantly heel the overall mean weight of milkweed tubers, with differences between seed sources (Table 5). Oats reduced the mean weight of tubers from the natural Rockefeller and Weimer Hill seed sources, but not the mean weight of tubers produced by garden out- crosses among geographically distant seed sources. FIELD ESTABLISHMENT EXPERIMENTS By the end of 1996, 332 seedlings and 290 ju- veniles representing 46 genetically different indi- viduals had been established at the seven study sites. Significant variation in the germination, sur- vivorship, and growth of these plants was caused by weather, fire, sources. Although greenhouse seed propagation rates were stable in 1994—1996, seedling survivorship in the field was 1196 or less in 1994 and in 1995, but increased to more than 40% in 1996 (Fig. 4). This difference in survivorship corresponded to below- normal May-July rainfall їп 1994—1995 and an above-normal May-July rainfall (20096 of normal) and differences among seed Volume 85, Number 1 1998 Bowles et al. Management and Restoration Ecology 119 Table 5. population crosses vs. artificial long-distance outcrosses) on mean weights of Mead's milkwee Experimental effects of artificial drainage gradient position and pum e or absence of oats (Avena sativa) on mean moisture levels (conductivity), and effects of presence or absence of o ats and seed source (natural within- d (Asclep ias meadit) tubers. Drainage gradient position: high = greatest elevation and ieu mid — ppro elevation pi drainage, low = lowest elevation and е Oats' level: F = 2.39, P = 0.0931; effect on moisture lev oats X drainage interaction: F = 9.03, P = : F = 75.58, P < 0.0001; drainage effect on moisture ht: F = dis p^ effect on tuber weight 5.656, P = 0.035; i source effect on tuber weight: F — 4.75, P — 0.0001. Mean moisture level (conductivity) = s.e. Mean tuber weight + s.e. in grams Gradient position Seed source high mic low within-population long-distance n — 40 n — 40 n — 40 n = 90 n — 90 Oats present 3.25 + 0.19 2.40 + 0.14 2.54 + 0.18 0.219 = 0.010 0.272 = 0.011 Oats absent 4.62 + 0.19 5.28 + 0,22 5.83 + 0.20 0.338 = 0.0125 0.326 = 0.013 in 1996, which enhanced seed germination and seedling survival. Seedling cohort survivorship over time was low, dropping to about 10% after two or three growing seasons; a few seeds delayed germi- nation until the second growing season, causing the upward deflection of the 1 in 1996 (Fig. 5). Survivorship of planted tubers was higher, with 5096 or more of these cohorts aliv 994. survivorship curve after two or three growing seasons (Fig. 5). Differ- ences among four planting sites also significantly affected sizes of three-year-old plants, as measured by leaf-area index (F = 21.53, P < 0 ). Plants with the largest mean leaf-area occurred in the Ver- mont Cemetery and averaged twice the size of plants at Biesecker Prairie. Plants at the Schulen- berg and Hickory Creek sites were intermediate in ze. Prescribed burning had either positive or neutral effects on seedling and tuber survivorship and leaf- area in all three study sites (Table 6). Two years RELATIONSHIP BETWEEN PRECIPITATION AND MILKWEED SEEDLING SURVIVORSHIP % seedling survivorship 1994 C] Greenhouse 1200 % of normal May—July precip. 1996 — Precipitation Figure 4. Temporal к between percent of normal May-July dinem ы and first-year survivorship of and field-germinated greenhouse- : ead's milkweed (Asclepias meadii) seedling . Precipitation levels are northeastern Illinois summaries from ihe National Oceanic and Atmospheric тея 120 Annals of the Missouri Botanical Garden SURVIVORSHIP OF SEEDLINGS FROM SPRING— PLANTED MEAD'S MILKWEED SEEDS 100 Nus на — Е “а. o “з. = “ы о "Sul a EN о EL £ — 2 4n ов IU 6 o ^N o © 22) 0 1. 2 3 Year æ= '94 cohort =- "95 cohort -*- '96 cohort SURVIVORSHIP ОҒ SPRING—PLANTED ONE- YEAR OLD MEAD'S MILKWEED TUBERS ТОО o t 2 a о с > ж гу 5 ao м о © 3 0 1 2 3 Year -e- '94 cohort =- '95 cohort -*- '96 cohort Cohort survivorship curves for spring-planted Mead's milkweed (Asclepias meadii) seedlings (upper) and tubers (lower) at seven northeastern Illinois and Indiana restoration sites. after the burn at Biesecker Prairie, seedling sur- vivorship did not differ significantly between treat- ments, but tuber survivorship and mean leaf-area index were greater in the burn treatment. One year after the burn at Pellville, seedling survivorship and tuber leaf-area did not differ significantly, but tuber survivorship was greater in the burn treat- ment. At Munson, only seedling survivorship was significantly greater in the year of the burn. The overall lower survivorship of tubers in 1996 may have been caused by excessive handling and stor- age. After these plants were excavated from flats and weighed, they had to be held in cold storage for about a month before planting due to a late cold spring. In the garden versus field comparison of perfor- mance between Rockefeller and Weimer Hill seed sources, garden-grown plants had significantly larg- er n leaf-area indices than field plants (F = 363. 78. P « 0.0001). After three years, most field plants had reached the size of second-year juvenile garden plants, while garden-grown plants had at- tained an adult size. Averaged across both planting treatments, Weimer Hill plants were larger than Rockefeller plants (F = 85.09, P < 0.0001). There was also a significant site X seed source interaction Volume 85, Number 1 1998 Bowles et al. 121 Management and Restoration Ecology Table 6. Effects of prescribed burning on Mead's milkweed (Asclepias meadii) seedling and tuber ме апа mean leaf-area indices in three restoration plantings. Survivorship P values аге based on Chi Leaf-area index P values are based on Student t-tests. gency tables. Site/planting date Biesecker/1994 Pellville/1995 Munson/1996 Variable Burned (94) Unburned Burned (95) Unburned Burned (96) Unburned Seedling (N) 18 15 30 25 49 50 survivorship (1996) 17% (P < 0.825) 13% 13% (P = 0.844) 8% 59% (Р = 0.057) 38% Tuber (N) 41 39 14 15 39 40 survivorship (1996) 83% (P < 0.001) 51% 86% (Р = 0.013) 33% 41% (Р = 0.216) 58% Leaf-area (/V) 4l 39 14 15 39 40 index = s.e. 25.51 (P = 0.006) 14.08 16.54 (P — 0.636) 14.59 4.42 (P — 0.467) 5.14 (1996) + 223 + 209 + 1,74 +291 +064 + 0.68 (F — 83.28, P « 0.0001); Weimer Hill plants had more than twice the leaf-area of Rockefeller plants in garden, but not in field, habitat. Overall tuber weight also had a significant (P < 0.0001) but small (r? = 0.18) positive correlation with leaf-area index for field-planted milkweeds in 1996. Exces- sive handling and storage may have reduced this correlation. DISCUSSION IMPLICATIONS OF HABITAT DIFFERENCES AND GENETIC ARIATION Despite the apparent lack of allozyme differen- tiation across its range (Tecic et al., 1998), Ascle- pias meadii occupies a strong geographic gradient in soil characterisitics, which have been found to affect allozyme frequencies in some species (e.g., Heywood & Levin, 1985). Our initial success in restoring plants fd Missouri and Kansas seed sources into the nutrient-rich soils of northern Il- linois also suggests that these soil differences may not be critical to A. meadii. It is unknown if size differences in garden habitat between plants from these seed sources reflect an important fitness com- onent, or simply phenotypic variation associated with genetically diverse plants under noncompeti- tive conditions. Certainly, they were not expressed in the field, where competitive stress should make differences that reflect fitness more apparent. In Pitcher’s thistle (Cirsium pitcheri Torr., Composi- tae), a monomorphic species with little geographic allozyme variation, plants from Wisconsin and In- diana seed sources differed in seedling morphology in the greenhouse, and in subsequent survivorship and growth when planted in an Illinois restoration (Bowles et al., 1993b; Bowles & McBride, 1996). When plants from geographically distant seed sources are integrated in restorations and cross-pol- linate, the potential exists for disruption of natu- rally evolved lineages and outbreeding depression caused by breaking up locally co-adapted gene complexes (Fenster & Dudash, 1994). The breeding system, pollen packaging as pollinia, and strong- flying pollinators of milkweeds may contribute to usually large neighborhood sizes (Broyles & Wyatt, 1991; Wyatt & Broyles, 1994), which could select against deleterious effects of outcrossing. This may vary among milkweed species. For example, Wyatt (1976) found greater percent fruit set from within- population crosses of А. tuberosa in comparison to between-population crosses across a wide geo- graphic region, suggesting outbreeding depression. However, our failure to find differences in percent seed germination between natural and geographi- cally distant crosses suggests that outbreeding may not be a critical factor for А. meadii at this early life-stage. In our garden experiment, seedlings from distant crosses developed larger tubers than seed- lings from natural populations when grown in com- petition with oats, and the larger plants correlated with greater plant size when outplanted the follow- ing year. This apparent heterosis effect could out- weigh deleterious consequences of long-distance crosses (Fenster & Dudash, 1994). Additional stud- ies are needed to assess the survivorship and growth of seedlings and backcrosses from geo- graphically distant sources to determine if popula- tion viability is negatively affected. In addition to total lack of reproduction, small populations of self-incompatible plants can undergo rapid increases in inbreeding, which could be det- rimental to population growth in restorations 122 Annals of the Missouri Botanical Garden (DeMauro, 1993). For A. meadii, inbreeding de- pression could be a serious problem, but we have only circumstantial evidence. For example, our three-year 7596 greenhouse germination rate was significantly greater (Х? = 280.6, P — 0.001) than the 47.696 germination found by Betz (1989) for 2429 wild-collected seeds. This lower germination have been caused by inbreeding in the isolated linear railroad prairie populations that Betz stud- ied. In contrast, our seed collections were either from populations known to be genetically diverse or from controlled garden crosses. Although in- breeding may reduce seed production, inbreeding- induced differences in fitness in plant species may be expressed under stressful field conditions and at different life stages depending upon their breeding systems (Dudash, 1990; Fenster & Dudash, 1994; Car & Dudash, 1996). Further етене апа field studies are needed to assess inbreeding effects in А. meadii. HABITAT MANAGEMENT EFFECTS Prescribed burning and mowing appear to have different effects on population structure of Asclepias meadii. The greater leaf-area of flowering milk- weeds at Rockefeller and the larger umbel sizes and greater percentage of flowering plants in burned sites suggest that these plants are placing more resources into the potential for sexual repro- duction than are plants in haymeadows. For ex- ample, although the smaller ramet and flower sizes of haymeadow plants could reflect stress from sum- mer mowing, the greater ramet densities but similar average plot leaf-area in Kansas haymeadows sug- gest a reallocation of resources into vegetative spread due to lack of sexual reproduction. However, this vegetative spread appears to accompany the loss or attrition of genetically different individuals, which would limit sexual reproduc pe in this self- incompatible species (Tecic et al., 1998). In com- cross-pollination among sexually compatible plants. The Missouri populations on former haymeadows had a comparatively low, although not significantly different, mean plot leaf-area, which does not strongly support our resource reallocation hypoth- eses. Population structure at these sites could be responding to the novel effect of reduced mowing frequency, which may contribute to loss of geneti- cally similar ramets once annual mowing was stopped. Fire is а natural factor responsible for mainte- nance of prairie, with varying effects on individual plant species (Collins & Glenn, 1988; Collins & Gibson, 1990; Evans et al., 1989; Collins & Wal- lace, 1990). Asclepias meadii appears to be fire- adapted. Betz (1989) found 77.1% flowering stems in annually burned prairies in railroad rights-of- ways, and greater flowering occurs in years of pre- scribed burns at Rockefeller Prairie (H. M. Alex- ander, pers. comm.). This is reinforced by our find- ing of increased milkweed juvenile growth and survivorship in burned tracts. Because of the low annual fruit production and seed production in this species (Betz, 1989; Alexander et al., 1997), fire may be critical for long-term population mainte- nance and could accelerate restored population growth. POPULATION DEMOGRAPHY The slow growth of restored milkweeds and lack of sexual reproduction after three years extremely limits demographic interpretations during this pe- riod. Seedling survivorship was essentially less than 10% for the first two cohorts, and no seedling plants attained the sizes reached by second-year juveniles in garden plots. The 40% seedling sur- vivorship in 1996 should allow greater second-year survivorship of this cohort, but development into a juvenile or reproductive state may require many years under field conditions. There also have been few transitions from juvenile to reproductive states. For example, 596 of all plants flowered in 1995, but fewer plants flowered in 1996, and only two plants flowered in both years. Also, none of these plants have produced seed pods, either due to lack of pollination or compatible crosses, or to inability to allocate enough resources to produce pods. Guerrant (1996) suggested that a restoration strategy of using outplanted juveniles rather than seeds would increase survivorship and population growth. Our preliminary data also indicate that ini- tiating a milkweed population with planted juve- niles can reduce mortality rates and accelerate de- velopment of larger population sizes. However, garden propagation increases chances of recruiting less fit plants that might not survive selective pres- sures that operate at the seedling stage. Planting of seeds is required to help assess if plants can ac- tually complete their life-cycles, but this necessi- longer restoration process with higher mor- tality rates and greater seedling vulnerability. tates a ENVIRONMENTAL EFFECTS ON DEMOGRAPHY Demographic monitoring of the trends of restored population growth can be enhanced by resolution of critical factors affecting reproduction, survivor- Volume 85, Number 1 Bowles et al. Management and Restoration Ecology 123 ship, and growth (Pavlik, 1994). Three environ- mental factors—moisture levels, competition, and site variation—appear to have important effects on population establishment and growth of Asclepias ro (1994) found higher seedling re- cruitment of raa dis acaulis (Pursh) K. L. Par- ker var. glabra (A. Gray) K. L. Parker (Compositae) uring years without summer drought. We also meadii. De found that growing season rainfall strongly influ- enced the fate of field-planted seeds, with survi- vorship exceeding 4096 only in 1996, when rainfall exceeded 20096 of normal. This suggests that re- cruitment into natural populations is uncommon and rainfall-dependent, providing a selective re- quirement for longevity of adult plants. As a result, lack of growing-season rainfall could negatively im- pact demographic processes when restoring popu- lations from seed. Competition from grasses, with their fibrous root systems, can negatively affect forb survivorship and growth 2. 1986; Louda et al., 1990; Hook et al., significantly reduced =“ кае шой. apparently through competition for moisture. This indicates s expected, oats that competition from existing prairie grasses should slow plant growth and delay transition of cohorts into adult stages. Asclepias meadii may be adapted to the bunch-grass structure of late-suc- cessional prairie by an ability to establish in patch- es between grasses, and by using its longevity to persist in this competitive but stable environment. The highly significant difference between garden and field effects on milkweed growth also indicates the high level of competition in field habitats. Greenhouse-propagated plants often flower and may produce pods in three years. But, as indicated, few field-planted milkweeds have flowered, and seed- lings from planted seeds still resembled first-year greenhouse seedlings after three years. However, significant differences in growth between different sites indicate that a wide range of conditions af- fecting plant growth exist in field habitats. For ex- ample, the greater size of plants restored at Vermont Cemetery may correlate with its lower landscape position and greater moisture retention compared with other, more well-drained restoration sites. Our results also suggest that burning should ac- celerate demographic processes by increasing seed- ling survivorship and growth of plants. However, weather cycles such as drought may override pos- itive effects of fire, as different precipitation levels clearly affected seedling establishment over time. Although optimum seedling growth for A. meadii is in full sun (Betz, 1989; Bowles, pers. obs.), seedling survival requires adequate moisture that could be retained longer in soils under unburned vegetation. These factors also would be affected by site drain- age, exposure, and soil water-holding capacity. Burned mesic habitat may have optimum germi- nation but strong late-season grass competition. Dry-mesic habitat may have less competition but stronger moisture requirements for seedling estab- lishment. Because weather is unpredictable, exper- imental burn and non-burn treatments appear nec- essary for milkweed establishment when supplemental watering is unavailable. Summer mowing may have some benefit in restoring А. mea- dii if it increases ramet numbers, possibly through decreased competition from warm-season grasses. However, we have no data on the effects of mowing on seedling recruitment, and long-term repeated mowing appears to cause genetic attrition (Tecic et al., 1998 SUMMARY AND CONCLUSIONS Pavlik (1996) identified proximal (completion of life-cycle, cohort replacement, and population in- crease) and distal (attainment of Minimum Viable Population) restoration objectives for plants. Res- toration of Asclepias meadii is clearly in the prox- imal stage, and the time scale for even short-term success of this late-successional species is un- known. These small populations remain vulnerable to impacts from stochastic demographic or environ- mental events that could eliminate all or a large proportion of their plants (Menges, 1991, 1992), and their effective population size (N,) is controlled by their outcrossing mating system, which requires crossing with different genotypes. А realistic short- term goal should be establishment of the number of genotypes present in natural populations. Allo- zyme sampling found 27 genotypes in the Weimer Hill population, and 15 genotypes at Rockefeller Tecic et al., 1998), and over 200 genetically dif- ferent plants actually may be present at Rockefeller (Alexander et al., 1997). To restore large numbers of genotypes will require mixing geographically dif- ferent seed sources. Such crossing has the potential for disrupting locally adapted lineages and causing outbreeding depression, but the unusually large neighborhood sizes of milkweed species may buffer them from these genetic effects. Although our pre- liminary success with planting of Missouri and Kansas milkweed seedlings into northern Illinois habitats supports this hypothesis, additional work is needed to understand long-term consequences. Additional constraints on the development of large milkweed populations are related to problems of size or scale (White, 1996). Asclepias meadii is so uncommon within natural habitats that it is rare- жт- 124 Annals of the Missouri Botanical Garden ly encountered by random sampling, and its ap- parent requirement of late-successional vegetation currently limits restoration to small sites, because no large sites exist in the eastern part of its range. Restoration of larger high-quality prairies is also not a short-term process, and has not been attained even after 20 years (Schramm, 1992). As a result, habitat size may regulate population growth by lim- iting effective population size and reproductive po- tential, and by enhancing inbreeding effects. De- velopment of an inordinately large or dense population within a small area could result in den- sity-dependent disease or insect infestations that would have disastrous effects on populations. For example, severe damage, and possibly mortality, to A. meadii can be caused by milkweed cerambycid beetles (Tetraopes sp.) and by milkweed weevils (Rhyssematus sp.). The adult Tetraopes feed on leaves and flowers, while their larvae feed on roots. Rhyssematus grubs feed on milkweed pith, which may weaken the stem, while adults may topple the terminal umbel, thereby preventing seed produc- tion (Betz, 1989; R. F. Betz, pers. obs.). Managing fragmented A. meadii restorations as metapopulations may help resolve the population size dilemma. The transferring of genetic material among sites could maintain a high level of genetic variation across restorations (Lacy, 1987, 1994; Te- cic et al., 1998). This would help provide demo- graphic stability by enhancing outcrossing potential while avoiding problems otherwise associated with high densities of small populations. Literature Cited Alexander, H. ade & W. D. Kettle. 1997. Application of n mar rien -rec E ы to estimation of the E size of plants. Ecology 78: 1230-1237 Beals, E. W. 1984. Bray-Curtis ordination: An effective Е for analy sis of multivariate ecological data. Ad- s Ecol. Res. 14: 1—55 en %. F. 1989, Ecology of Mead’s duds (Asclepias meadii Torrey). Pp. 187-191 in T. B. Bragg & J. Stub- bendieck (editors), Proc 1. of the Eleventh North American Prairie Conference. Univ. Nebraska at Lin- coln. . F. Lamp. 1989. Speci les волио об | old seller silt loam cemetery prairies. Pp. 33-39 in T. B Bragg & J. Stubbendieck (editors), Proceedings Е the Eleventh 4. American Prairie Conference. Univ. Хе- braska at Lincoln. . К. n" J. E. Wall & F. B. Heitler. 1994. Insect pollinators of 12 illnd (Asclepias) species. Pp. 45—60 in Proceedings of the Th 5 North American Prairie Conference. R. G. Wicket D. Lew- is, А. Woodliffe & P. Pratt (editors), о. of Parks y: Ren теапоп, Windsor, Ontario, Canada. Bliss, L. . Cox. 1964. Plant community struc- ture nd oil variation cue a northern Indiana prairie. >-128 Amer. Midl. Naturalist 72: 5 = © Bowles, M. L. & J. McBride. 1996. .. (Cir- sium pitcheri) . Pp. 423-431 in D. Falk, C. I. Millar & M. Olwell (editors), Restoring Di- versity: Strategies for ‘Reintroduction of Endangered Plants. Island Press, Washington, D.C. ; С. d (editors). "1994. Restoration of s: Conceptual Issues, Planning, and Cable Univ. Press, Cambridge, Endangered S ШЕ о. U.K z & M. M. DeMauro. 1993a. Propaga- tion oir rare E from 1. seed collections: Im- plications for species | and herbarium тап- agement. Restoration Ecol. 2. Flakne, K. McEac wn N. Pavlovic. 1993b. Recovery planning and ени ыу of the federally threatened Pitcher thistle (Cirsium pitcheri) in Illinois. Natural i^ as J. 13: 164—176. Broyles, S. B. & R. Wyatt. 1991. Effective pollen dis- persal in a d population of Asclepias exaltata: The influence of 2. behavior, genetic | and mating success. Amer. Naturalist 138: 123€ ) 1993. The consequence ina elf-pol- lination in Asclepias анан а зе incompatible milk. weed. ; -44. Carr, D. E. 1996. Inbreeding depres- sion in two species of Mimulus (Sc rophulariaceae) with .. mating systems. Amer. J. Bot. 83: 586-593. Collins, S. L. & D. J. Gibson. 1990. Effects of fire on Wallace PN North American Tallgrass Prairie. Univ. Okla- homa fes das —— & S. М. С en 1988. Disturbance and commu- nity facea in North Americ an же ies. Pp. 131-143 in de During, M. Werger & J. Willems (editora, Diver- and жм in Plant Communities SPB Academic Publishing The Hague, T erlands. L. L. Wallace men 1990 1 North American Tallgrass Prairie. Univ. Oklahoma Press, Nor- man. DeMauro, M. M. 1993. ЛЕ of breeding system to rarity in the Lakeside menoxys acaulis var. En Conservation Biol. 7. 54 1994. Development and 1. ntation of a re- covery program for the federal threatened Lakeside dai- y Vomenoxys acaulis var. glabra). Pp. 298-321 in M. L Bowles . Whelan (editors), Restoration of En- dangered Species: Conceptual Issues, Planning, and Шана Cambridge Univ. Press, Cambridge, M. R. 1990. Relative fitness of selfed and out- eny in a self-compatible, protandrous spe- cies, po a angularis 1 2 еге) іп Іһгее еп- vironments. Evolution 44: 1129 "ane Evans, E. W., J. M. Briggs, E. J. Finck, D. J. Gibson, S. W. James, D. W. Kaufman & T. R. 1 1989. Is fire а 1. in grasslands? Рр. 159-160 іп T. B. Bragg & J. Stubbendieck (editors), Proceedings of the Eleventh North American Prairie Conference. Univ. braska at Lincoln. Fay P. A. & . Knapp. 1993. Photosynthetic and stomatal respondas of Avena sativa (Poaceae) to a vari- able light environment. Amer. J. Bot. 80: 1369-1373. Fenster, C. B. . Dudash. 1994. Genetic consid- erations for rs population restoration and conserva- tion. Pp. 34—6 L 'helan (edi- tors), o ог Endangered Spas ies: Conceptual 4. Bowles & € Volume 85, Number 1 Bowles et al. 125 Management and Restoration Ecology Issues, Planning, and Implementation. Cambridge Univ. 1978. A 20-year record of succession on reseeded fields of 2. ка on the Roc 'kefeller Experimental Tract. Publ. . Nat. Hist 15. E. Soulé. 1986. Minimum viable p з ses of species extinction. Pp. 19—34 n M. E. Soulé (editor), Conservation Biology: The Sci- ence of Scarcity and Diversity. Sinauer, Sunderland, tts Guerrant, E. O., Jr. 1996. Designing populations: De- mographic, genetic, and horticultural dimensions. Pp 171-207 in D. А. Falk, C. I. Millar & M. Olwell (edi- tors), Restoring Diversity: Strategies for Reintroduction of ОН Plants. Island Press, Washington, Gurevitch, J. . Competition and the local distribu- tion of the grass Зура neomexicana. Ecology 67 97. Harrison, №. Е. 1988. Endangered and threatened wild- life and plants; Determination of threatened status for Asclepias теади (Mead's milkweed). Fed. Reg. 53: 33982-33994. Heywood, J. S. & D. A. Levin. en . be- tween allozyme frequencies and s eristics in LU pulchella (Compositae). eres 39: 1076— 10 Hook, т B., W. К. Lauenroth & I. C. Burke. 1994. Spatial patterns of roots in a semiarid grassland: Abundance of canopy openings and regeneration gaps. J. Ecol. 82: 485—494. Kahn, A. P. & D. Н. Morse. 1991. Pollinium germination and putative m penetration in self- and cross- (ov nated common milkweed Asclepias зупаса. Amer. Midl. Naturalist 126 1-71. Ke 5 1981 reeding systems in Asclepias їп- carnata, A. syriaca, Fut verticillata. Amer. J. Bot. 68: . С. 1987. Loss of genetic diversity from managed populations: пе effects of drift, mutation, im- migration, se Ea population subdivision. Con- servation Biol. 1: 14: 158 1 Managing — 5. in captive populations of animals. Pp. 63-89 in M. L. Bowles Š C. J. Whelan (e е Ба, a paños cies: Conceptual Issu lanning, a Implementation Cambridge Univ. Press, Cam ridge. | nde, Н. 1988. Genetics “ ЕЗ in biological conservation. p ience 1: 1455-1460. Les, D. H., J. A inhaits & ч J. Esselman. 1991. Ge- netic consequences of rarity in Aster furcatus (Astera- ceae), a E» “= d, мо dant f plant. Evolution 45: 1641- Louda, S. ^ is M. PM vin & S. K. Collinge. 1990. Pre- dispersal seed predation, ~ seed predation and competition in the t of seedlings of a native ques in ud hills pr prairie. wies Midl. Natu- ralist 124: 105-113. McCune, B. г Multivariate analysis on the PC-ORD system. Oregon State University, Corvalis. „ 1977. Rare native vascular plants of nsas. Techn. Publ. State Biol. Surv. Kansas 5: 1-44. 5 1986. Predicting the future of rare plant Корова Demographic monitoring and modelling. . Areas ч Ө: 13—25. The application of minimum viable pop- он ied to plants. Pp. 45—61 in D. A. Falk & К. E. Holsinger (editors), Genetics and Conservation of Rare Plants. Oxford Univ. Press, New York. 992. Stochastic modeling of extinction in plant барий oe Pp. 253-275 in P. L. Fiedler & S. K. Jain (editors), мы mis: Biology: The Theory and Prac- ure Conservation. Chapman & Hall, New York. Page, A. L., R. H. Miller & D. R. Keeny (editors). 1982. Methods of Soil Analysis, Part 2: Chemical and Micro- bial Properties, 2nd ed. American Society of Agronomy & Soil Science Society of America, Madison, Wiscon- sin. Pavlik, B. M. 1994. Demographic monitoring and the re- covery of endangered dum populations. Pp. 322-350 in M. L. Bowles Whelan (editors), Restoration of aic: Species: ое Issues, Plannin and Implementation. Cambridge Univ. Press, Cam- bridge, U.K 1996. T and measuring success. Pp. 127-15 55 in D. A. Falk, C. I. Millar & M. Olwell (edi- tors), bu ring Dude Strategies for Reintroduction of wes Plants. Island Press, Washington, D.C Schaal, B. A., Leverich & S. H. Rogstad. 1991. comparison ab methods for ме ар Lara на in plant 23-134 in D. A. = & K.E. ie A ditors), (dente and Conser- ation of Rare P Oxford Univ. Press, New 5с ian P. 1992. Prairie restoration: A — ci year An ege on establishment and management. Pp. 69— 177 in D. D. Smith & C. A. Jacobs жүн Proceed- ings of the Twelfth North dog Prairie Conference. niv. Northern lowa, Cedar Fa Shannon, T. R. 8 R. Wyatt. 1986. Pollen MN f Asclepias exaltata: Effects of flower Ed drying t 4. E source. Syst. Bot. 11: 322— mith, T. 1997. Management Guidelines o») Masts mile E 1. meadii Torrey ex А. Gray). Missouri De- ent of Conservation, Тон City, Missouri. Sneath, P. H. A. и la к Sokal. 1973. Numerical Tax- onomy. W. H. Fr n, San Fra Sokal, R. R. & TE “Rohlf. 1981. man, New Yor Steele, R. G. D. & J. H. 1960. Principles and Procedures of Statistics. hs ыы Hill, New . McBride, M. L. Bowles & D. L. `6 enetic variability in the federal iei ened Mead’s milkweed, Asclepias meadii Torrey (Ascle- piadaceae) as determined by allozyme electrophoresis. Ann. uva Bot. Gard. 85: 97-109. Weller, S. The re ШЕП of rarity to plant repro- ductive vis Pp. 90-117 in M. L. Bowles & C. Whelan (editors), 2. ‘of Endangered Species Conceptual Issues, Planning, and d ale eis книг Univ. Press, Cambridge White, P. 5. 1996. em and biological poles ік nj: 1 D. A. Falk, C. I. Millar € M. Diversity: 172. for Re- introduction of рааны Plants. Island Press, Wash- ington, D.C. Willson, M. F. & P. W. Price. 1980. Resource limitation of fruit and seed production in some Asclepias species. Canad. J. Bot. 58: 2229-2233. Wyatt, R. 1976. Pollination and fruit-set in Asclepias: A reappraisal. Amer. J. Bot. 6 5l. ----45 Broyles. Ecology and evolution of reproduction in milkweeds. bool Rev. Ecol. Syst. 25: 423-441. ncisco Biometry. W. H. Free- THE POLLINATION ECOLOGY OF FIVE SPECIES OF PENSTEMON (SCROPHULARIACEAE) IN THE TALLGRASS PRAIRIE! Richard R. Clinebell II? and Peter Bernhardt? ABSTRACT The floral ecology of Penstemon cobaea Nutt. var. cobaea, P. cobaea var. purpureus Pennell, P. digitalis Nutt. ex Sims, P. ај ышы Nutt., P. pallidus * sit linc ‚ Kansas, . Flowers e floral sinus. The tubular, white flow and Missouri. All five species show protandry, but the ver, the онд, pollination system Small, and P. tubaeflorus Nutt. was studied by sampling populations at nine prairie acad stigma lies only 2 mm away w that only P. digitalis sets species exhibit a horizontal enis of the corolla and of Penstemon spp. correlate with corolla rs of P. 2. appear to be pollinated by a c шаның of diurnal Lepidoptera and some native bees favoring a ded deposition of pollen on mouthparts and upper thorac The four remaining species and P. cobaea visited infrequently by the r sylvanicus subsp. pennsylvanicus forage primarily on may ор spp.. Megachile brevis, and anthophorids (alone hamata and prac terminalis) carry dorsal stigmas while the Penstemon pollen because they contact anthers and s ve gullet- or bell-shaped corollas ornamented with vi species appear to be alinae primarily by polylectic/polyphagic bees (including six Bombus spp.), w rare кан мазр, Pseudomasaris occidentalis те large, gullet flowers of P. olet-purple blote hes с or lines. The: ese = "2 — =) c reduce Large-bodied Bombus y forage exclusively ectar. [n co bodied members of the Anthophoridae (Ceratina). Colletidae (ола spp.). Halic tidae (Augochlorella, аа La- sioglossum), and Мерас 'hilidae (Hoplitis and Озта) forage actively for Penstemon pollen encouraging repeated, v ywers. le queens a contact with the sexual organs of the f (especia ally Р di Bombus workers were Ф = е ted entral re more prevalent at large 2. йн учы е le on 4. sites. The importance of small bees as pollinators appeared to vary indirectly with o population size. Penstemon (Scrophulariaceae: Cheloneae) is a North American genus of about 270 species (Wolfe et al., 1997) distributed from Alaska to Guatemala. Within the Great Plains, Freeman (1981) recog- nized 22 species of Penstemon in two subgenera and five sections. The Upper Mississippi Valley supports two additional species, Penstemon arkan- sanus Pennell and P. hirsutus (L.) Willd., suggesting that 24 Penstemon species are native to midwestern American prairies. Despite the species richness of Penstemon in orth America, analyses comparing life-histories within this genus lag far behind classical (Pennell, 1935; Keck, 1938) and molecular (Wolfe & Elisens, 1993) taxonomies. We lack significant literature on breeding systems in Penstemon compared to other scrophulariaceous genera (e.g., Pedicularis; Macior, 1982) distributed through the Northern Hemisphere . References to the pollination bi- “~ Катрпу, ology of Penstemon by Реппе! (1935) were derived primarily from predictions that were based on floral morphology. In contrast, what literature does exist on Penstemon pollination often shows a lack of con- sensus regarding the efficiency of different bee taxa as true pollen vectors. Field studies of Penstemon pollination began with Robertson (1892, 1929), who noted p in "P. laevigatus" (= P. digitalis) and “Р. pube cens" (— P. pallidus) in Illinois and collected a ial of 20 different bee species in their flowers. Rob- ' This work is part of the first author's doctoral dissertation being prepared и in the Dept. of Biology, Saint Louis niversity, under the direction of P. B Federation of Missouri (Bell 2 Sc holarship), t (Litzsinger Road Ecology Center), the Missouri Departm St. Louis office tables and. graphics. Stanley r e Sawyer provided statis ernhardt. Financial e for fie Idwo ansas ent t of Conservation, of the Nature Conservancy. Donald Hardin p 22. in the field and in the urinis of tical advic he Conservation = Е Е 8; - = а: = e] a s Michener and colleagues identified the Ну- menoptera A referred other insect specimens to appropriate nois We thank Paul Wilson and an anonymous reviewer for ful comments and criticisms. 2 раи of Biology, Saint Louis University, 3507 Laclede Ауе., St. Louis, Missouri 63103, U.S.A. ANN. Missouni BOT. GARD. 85: 126-136. 1998. Volume 85, Number 1 1998 Clinebell & Bernhardt Pollination Ecology of Penstemon 127 ertson was convinced that pollination was effected exclusively by long-tongued bees probing for nectar because small-bodied, short-tongued bees were ex- cluded from the floral throat by the ornamented staminode. Clements and Long (1923) collected flo- ral foragers on P. gracilis, in the Colorado foothills, concluding that Osmia and Hoplitis spp. (Mega- chilidae) were the most common floral visitors. Straw (1956) insisted that the pseudomasarid wasp- pollinated Penstemon spectabilis Thurb. was derived from a stabilized hybrid between bee-pollinated P. grinnellii Eastw. and hummingbird-pollinated P. centranthifolius (Benth.) Benth. This has since been discredited by Grant (1994; see also Wolfe & Eli- sens, 1993). Crosswhite and Crosswhite (1966) studied insect-pollination of P. gracilis and P. pal- lidus in southern Wisconsin and northern Illinois. Unlike Robertson, they emphasized the role of small, solitary bees, especially Osmia and Hoplitis spp. Crosswhite (1965) provided the only informa- tion on compatibility systems in the genus, con- cluding that P. pallidus was self-compatible. Law- son et al. (1989) analyzed the pollen loads of bees collected on Penstemon haydenii S. Wats. in Ne- braska, noting an abundance of Hoplitis and Osmia, but did not comment on which bees regularly con- tacted the receptive stigmas. Here we report the results of four years of field studies on six Penstemon taxa in five species: Pen- stemon cobaea var. cobaea, P. cobaea var. purpureus, P. digitalis, P. grandiflorus, P. pallidus, and P. tu- baeflorus. Continued studies of pollination systems of this genus in the American Midwest are needed for two, overlapping reasons. First, further analyses and descriptions of pollination mechanisms in Pen- stemon will make it possible to map pollinator shifts onto a phylogenetic tree (see Armbruster, 1993; Goldblatt et al., 1995). Second, Penstemon spp. are native to prairies that are reduced and much en- dangered habitats through the American Midwest. Basic information on pollen dispersal in prairie Penstemon spp. should contribute ultimately to con- servation policies and restoration projects. MATERIALS AND METHODS The taxonomy of Penstemon species follows Yat- skievych and Turner (1990). Floral phenologies are based on censuses (as specified below) at each study site for one to four flowering seasons. POPULATIONS Penstemon cobaea Nutt. var. cobaea (pollinators collected 24 May-22 June 1994-1997, N — 116 Sixteen populations of this taxon on or near the Konza Prairie (Kansas) have been studied, most of them small (< 30 blooming stems). The Konza Prairie Research Natural Area is an 8616-acre pre- serve managed by Kansas State University for the Nature Conservancy. Two large populations (7 500 blooming stems) were studied: (1) in 1995 in White Pasture near the northeastern corner of the Konza in an area considered more floristically rich in tall- grass prairie forbs than much of the rest of the site; and (2) in Wright Prairie, adjacent to the part of Konza known as the Texas Hog Pasture, for which access was provided by Valerie Wright. Penstemon cobaea var. purpureus Pennell (28 May-4 June 1995-1996, N — 15). Three popula- tions of approximately 100 flowering shoots each were monitored in roadside glades in Christian, Ozark, and Taney Cos., Missouri. All three sites are listed on the Missouri Department of Conservation Rare Plant Inventory. Penstemon digitalis Nutt. ex Sims (30 May-5 July 1994—1996, N — 325). This species was stud- ied on four sites, two tallgrass prairie restorations in eastern Missouri and two tallgrass prairie relicts in western Missouri: (1) Litzsinger Road Ecology Center of the Missouri Botanical Garden, St. Louis Co., Missouri. This wet mesic prairie restoration was seeded in 1989 and still contains а prepon- derance of weedy, early successional prairie spe- cies. (2) Shaw Arboretum Experimental Prairie of the Missouri Botanical Garden, Franklin Co., Mis- souri. This site is about 25 years old and is more diverse than Litzsinger, but not so diverse as the natural communities. (3) Paint Brush Prairie, Pettis Co., Missouri. The colony studied is in a relatively impoverished part of the site, but is adjacent to floristically rich prairie. There are notable differ- ences, discussed below, between the floral foragers here as compared to the restorations. (4) Hi-Lone- some Prairie, Benton Co., Missouri. The study area here is a Penstemon monoculture embedded in common lenticular sedges (Carex spp.). Penstemon grandiflorus Nutt. (21-30 May 1996— 1997, N — 44). This species was studied in a pop- ulation of about 50 blooming plants on and near an ungrazed prairie haymeadow on the Poole Ranch in Geary Co., Kansas, about 2.8 miles south of In- terstate 70 оп Kansas Hwy. 177. This species is extremely rare on the Konza Prairie proper. Penstemon pallidus Small (15-24 May 1996- 1997, N — 71). Two sites were used for this taxon in both 1996 and 1997: (1) Fults Hill Prairie, Mon- roe Co., Illinois. This area is perched atop the Mis- sissippi River bluffs and is considered the best sur- viving undisturbed loess hill prairie along the Illinois bluffs. (2) Paint Brush Prairie, Pettis Co., 128 Annals of the Missouri Botanical Garden Table 1. Pollen loads of foragers collected on Penstemon spp. Penstemon — Penstemon + Other species No pollen only — other species only pollen Total Large Gullet-Corolla Species Penstemon cobaea var. cobaea Bees Augochlorella striata (Provancher) F 2 0 0 0 2 Bombus nevadensis auricomus (Robt.) Q 4 l 0 0 5 Bombus nevadensis auricomus W 2 0 0 0 2 Bombus pennsylvanicus (DeGeer) Q 19 8 2 1 30 Bombus pennsylvanicus 1 0 0 0 1 Ceratina strenua о 0 1 0 0 1 Halictus ligatus 1 0 0 0 1 Hoplitis pilosifrons (Cresson) F 2] 2] 3 5 50 Hoplitis pilosifrons M 0 0 0 1 1 asioglossum (= Dialictus) spp. F 6 2 0 ] 9 Synhalonia hamata (Bradley) F 1 0 0 0 1 Others Bombylius 4 0 0 3 7 Euphoria BL (Fab.) 0 2 0 4 6 У = 116 Penstemon cobaea r. purpureus Bees Bombus pennsylvanicus Q 0 0 0 1 l Hoplitis pilosifrons F 1 2 0 1 4 Lasioglossum imitatum (Smith) К 1 0 0 0 1 Osmia distincta (Cresson) F 1 1 0 0 2 Synhalonia rosae Robertson F 1 0 0 0 1 Xylocopa virginica L. К 0 1 1 1 3 Others Вотђућих 0 0 0 1 1 Рата“ sp. F 1 0 0 1 2 У = 15 Penstemon grandiflorus ees Augochlorella persimilis (Viereck) F 1 0 0 0 1 Augochlorella striata 11 2 1 4 18 Bombus griseocollis (DeGeer) Q 0 1 0 0 1 Bombus pennsylvanicus Q 5 3 0 0 8 Hoplitis n 3 0 0 0 3 Lasioglossum (— 5... spp. F 6 7 0 1 14 Meg achile | brevis Say M 1 0 0 0 1 X e virginica M 1 0 0 0 1 У = 44 Small Gullet-Corolla Species Penstemon digitalis Bees Anthophora terminalis ;resson F 6 0 0 1 7 Augochlorella striata F 3 1 0 3 7 Bombus bimaculatus Cresson Q 2 0 1 2 5 ombus bimaculatus W 24 8 6 3 4l Bombus bimaculatus M 5 0 0 0 5 Bombus fraternus (Smith) W 8 1 1 0 10 Volume 85, Number 1 Clinebell & Bernhardt 129 1998 Pollination Ecology of Penstemon Table 1. Continued. Penstemon Репѕіетоп + Other species о pollen only other species only pollen Total Bombus griseocollis Q 0 1 1 2 4 Bombus griseocollis W 3 5 0 1 9 Bombus impatiens Cresson W 9 1 1 0 11 Bombus nevadensi 28 2 6 9 45 Bombus nevadensis W 5 5 4 2 16 Bombus pennsylvanicus Q 3 3 1 2 9 Bombus A W Т 6 1 2 16 Ceratin 7 5 0 5 17 Hop, pon F 6 2 0 1 9 Hylaeus spp. F 2 0 0 2 4 еня (= Dialictus) spp. Е 36 5 3 12 56 Osmia spp. F 11 0 0 1 12 Synhalonia hamata F 22 13 2 2 39 Synhalonia hamata M 1 0 0 0 1 Others Pseudomasaris s 1 0 0 0 1 Pterourus troilus troilus L. F 1 0 0 0 1 У = 325 Penstemon pallidus Bees Apis mellifera L. W 1 0 0 0 1 Augochlorella striata F 0 0 1 1 2 Bombus bimaculatus Q 2 0 0 1 3 Bombus bimaculatus W 1 1 0 0 2 Bombus nevadensis auricomus Q 0 0 0 2 2 mbus pennsylvanicus Q 0 0 0 2 2 Ceratina spp. F 2 0 0 Б 7 Hoplitis pilosifrons F 13 1 0 6 20 Hoplitis producta (Cresson) F 1 0 0 0 1 Lasioglossum (= Dialictus) sp. К 1 1 0 0 2 Osmia spp. F 15 3 0 6 24 Synhalonia rosae M 3 0 0 2 5 У = 71 Tubular-Corolla species Penstemon tubaeflorus Bees Anthophora abrupta vx F 0 0 0 2 2 Anthophora abrupta 1 0 0 2 3 Anthophora ursina т Е 1 0 0 0 1 Augochlorella striata Е 0 0 0 1 1 bus pennsylvanicus ©) 3 0 0 0 3 Osmia distincta 2 0 0 2 Others Pterourus troilus troilus F 1 0 0 z 3 Pterourus troilus troilus M 0 0 0 1 1 У = 16 22 = 587 К = Female, M = male, ©) = queen, W = worker. 130 Annals of the Missouri Botanical Garden Missouri. This Osage Plains prairie is particularly rich in late spring forbs which co-flower with pen- stemons. Penstemon tubaeflorus Nutt. (4—5 jme 1996, N — ]7) This species was studied at Long Bald, a dolomite glade in Caney Mountain Conservation Area, Ozark Co., Missouri. This is a large and spec- tacular glade surrounded by dry oak woods, and it contains a wealt including the Ozark endemic Echinacea paradoxa (Norton) Brit- of wildflowers, ton. Taken together, the study areas of this project represent a diversity of Penstemon habitats across Alth ough we have pooled the floral forager lists in this paper (Table the tallgrass prairie biome. 1) by Penstemon taxa to conserve space, we em- phasize the value of repeated samples of the same study areas over several years in evaluating the fi- delity of floral foragers to specific penstemons. The sites themselves are, for the most part, well- known nature preserves in Illinois, Kansas, and Missouri, for which there is a large literature avail- able (e.g., Evers, 1955; Freeman & Hulbert, 1985; Ochs, 1993; Toney, undated; Solecki et al., 1986 These references contain much information on site history, location, ecology, and floristic composition. Floral fragrance. cies were placed in clean, glass, stoppered vials for Whole flowers of each spe- periods up to two hours, following Buchmann et al. (1978). At the end of the two-hour period, the vials were uncorked and smelled. Bagging experiments. То determine the poten- tial role of self-compatibility and mechanical self- pollination (autogamy) in the absence of insect vis- itation, inflorescences were bagged during the flow- ering season. Nylon stockings or wood and mesh exclusion cages were placed over ten flowering shoots, P. cobaea var. purpureus, six of P. digitalis and three of P. grandiflorus for the length of individual flow- ering periods. Infructescences were examined for the production of fruit and seed within two months following the withering of the last flower on the shoot Field observations record- ed foraging behavior of insects on, and within, Pen- stemon flowers. This included noting when foragers collected nectar and/or pollen and whether insects contacted anthers and stigmas while foraging. Pollinator analyses. he collection and analyses of floral foragers on Penstemon flowers represent the harvest of 35—5 insects/site each year. As these Penstemon рори- lations represent protected species on nature pre- in bud, of P. cobaea var. cobaea, twelve of serves, overcollection of prospective pollinators may lower seedset. Otherwise, the protocol for net- ting, killing, and processing specimens followed Bernhardt (1990a, b) and Bernhardt and Weston (1996). Floral foragers were collected on penstemon blossoms and killed in jars poisoned with ethyl ac- etate. To verify the presence of Penstemon pollen, each euthanized insect was placed on a separate, clean glass slide and bathed in a few drops of absolute ethanol. At this time additional masses of pollen packed onto scopae and corbiculae were teased or scraped off with a dissecting needle. When the eth- anol had evaporated, the pollen film left on the glass slide was stained with two to three drops of Calberla's fluid (Ogden et al., 1974). Because in- sects were sacrificed in a communal killing jar, there is the danger of pollen contamination as bod- ies of different insects contact each other (see Bern- hardt & Weston, 1996). Therefore, we excluded pollen counts of less than 50 grains on insects since the vast majority of floral foragers were extremely hairy and liable to pick up loose grains due to static electricity. Pollen loads were counted on each slide as follows: a total of 200 pollen grains were counted ` there were more than 150 grains of Penstemon pollen in the count (> 75%), the load was classified as a “ Pen- stemon Pollen Only Load." If more than 150 grains for each insect in increments of five. of non-penstemon pollen was present the load was called an *Other Species Only Load." If more than 50 grains of both Penstemon and at least one other species were present, this was If less than 200 grains were present on the slide, the load non-Penstemon called a *Penstemon + Other Species Load." was scored as a "No Pollen Loa Insect length was recorded by measuring six pinned specimens of each species, gender, and/or caste. The body length of each specimen was mea- sured from the labrum to the apex of the abdomen (Bernhardt & Weston, 1996). Insect specimens were identified by C. D. Mich- ener, R. Brooks, and colleagues at Snow Entomo- logical Museum, University of Kansas, and by Arduser of the Missouri Department of Conserva- tion. Vouchers are deposited at the Snow Museum. Bee genera follow Michener et al. (1994). RESULTS Floral phenology. Midwestern Penstemon spp. are found in bloom from early May until early July (Fig. 1). Flowering seasons overlap broadly between species. Penstemon pallidus is the first species to bloom in spring, while the two varieties of P. cobaea Volume 85, Number 1 1998 Clinebell & Bernhardt Pollination Ecology of Penstemon 131 Penstemon pallidus | P. grandiflorus | P. cobaea var. purpureus | | P. tubaeflorus H P. cobaea var. cobaea — — P. digitalis EM ананан. КЕ IE 15 20 25 30 5 10 15 20 26 30 Figure 1. Floral phenology of Penstemon spp. (1994—1997). are usually the last to begin flowering. Penstemon digitalis has the longest flowering period. Bagging experiments. In both varieties of P. co- baea, no fruits were produced in the absence of visiting insects. In P. grandiflorus approximately 50% of the flowers produced full-sized capsules to- tally devoid of seed (parthenocarpy). In P. digitalis, the vast majority of flowers produced seed-contain- ing capsules. Floral presentation. АП Penstemon spp. are protandrous with the stigma located within 2 mm of the two pairs of fertile anthers. All Penstemon spp. present their flowers horizontally with respect to the peduncle Flowers of all Penstemon spp. produce a barely discernible fragrance reminiscent of ripe cante- loupe. Penstemon tubaeflorus is the only species studied with flowers that are pure, translucent- shiny white to the human eye. Flowers of P. gran- diflorus appear pink-mauve with faintly discernible nectar guides. Penstemon digitalis and P. pallidus have white corollas with the floral throat streaked with narrow, light purple lines. Penstemon cobaea var. cobaea has pink or white corollas, while corol- las of variety purpureus are deep purple (although we have records of red or white morphs in Taney Co., Missouri). Corollas of both varieties of P. co- baea have deep purple blotches and/or broad lines on the throat. The corollas of P tubaeflorus (sinus 4-6 mm wide) have the most narrow floral tubes. The co- rollas of both varieties of P. cobaea (sinus 18-25 mm wide) and P. grandiflorus (1: mm wide) form large, bilabiate gullets. In contrast, the corol- las of P. digitalis (sinus 8-12 mm) and Р pallidus (4—7 mm) form small gullets. Floral foragers. Approximately 600 insects were caught foraging on Penstemon flowers. Penste- mon tubaeflorus was the only species that was vis- ited consistently by the butterfly Pterourus troilus, but captures do not reflect the density of field ob- servations (Table 1). The head and proboscis of this butterfly contacts stigmas and dehisced anthers while it probes for nectar, with pollen loads depos- ited on the proximal portion of the proboscis (Table 2). Bees greater than 10 mm in length cling to the petal lobes while probing for nectar, acquiring dor- sal depositions of pollen on their heads (Tables 1, 2) In all remaining species pollination appeared to be dominated by Hymenoptera, with both varieties of P. cobaea also visited by beeflies (Bombylius spp.) that carried Penstemon pollen while contact- ing the stigmas (Table 2). Bumblebee flower-beetles (Euphoria sepulchralis, Scarabaeideae) collected on > cobaea var. cobaea carried loads of Penstemon pollen, but they consumed basal floral organs and we were unable to determine whether they con- tacted stigmas (Table 2). Penstemon digitalis and P. cobaea var. purpureus were visited by females of Pseudomasaris occiden- talis (Vespidae). These wasps contacted the stigmas while swallowing pollen and/or foraging for nectar at the base of the corolla tubes, so pollen deposition was both ventral and dorsal. The foraging behavior of bees was determined by physical size. Bees greater than 10 mm in length (Anthophora, Bombus, Megachile, and Synhalonia 132 Annals of the Missouri Botanical Garden Table 2. Size and behavior of floral foragers on Penstemon spp. _ )bserved Observed — Contacted anthers = Number X = Length of foraging foraging and stigma while Floral forager caught body* (mm) for pollen for nectar foraging Coleoptera Euphoria sepulchralis 6 12.5 m t = Diptera Bombylius spp. 8 12.0 - + + Hymenoptera Bees Anthophora spp. F 9 12.1 Е + + Anthophora spp. М 3 13.3 - t + Apis mellifera W 1 13.0 Е + + Augochlorella spp. Е 31 7.2 + — + Bombus queens 118 Е + + B. bimaculatus 8 18.5 B. griseocollis 5 22.8 B. nevadensis 52 27.8 B. pennsylvanicus 53 21.0 Bombus workers 98 = + + B. bimaculatus 43 15.3 B. fraternus 10 — B. griseocollis 9 19.5 B. impatiens 11 12.6 B. nevadensis 9 18.7 B. pennsylvanicus 16 22.5 Bombus males B. bimaculatus 5 = + + Ceratina spp. 25 6.1 + = + Halictus ligatus Ё 1 8.5 + = + Hoplitis pilosifrons Ё 86 Tel + - T Hoplitis pilosifrons M 1 8.0 ? ? ES oplitis prod. 1 8.5 + - + Hylaeus spp. F 4 5.0 + - + Lasioglossum (7 Dialictus) i F 82 6.0 + - + Megachile сш 1 11.0 - + + Озтїа sp 40 8.4 + + Synha а T F 40 15.3 = + + Synhalonia hamata M 1 12.5 = + + Synhalonia rosae F 1 14.2 = + + Synhalonia rosae M 5 13.7 a + + Xylocopa virginica Ё 4 21.4 ? ? ? Xylocopa virginica M 1 22.5 ? ? ? Wasps Pseudomasaris sp. F 3 19.0 + + + Lepidoptera Pterourus troilus troilus F 4 225 == + +? Pterourus troilus troilus M 1 22.0 — + +/? * Measurement refers to body length measured from the clypeus (excluding the length of the proboscis) to the tip of the abdomen. F = female, M = male, Q = queen, W = worker. Volume 85, Number 1 1998 Clinebell & Bernhardt Pollination Ecology of Penstemon 133 spp.) entered the floral tube and probed for nectar, receiving dorsal depositions of pollen from de- hisced anthers while contacting the stigmas (Table 2). These bees did not collect Penstemon pollen by scraping anthers with their legs. Although Xylocopa virginica was over 10 mm in length, bees of this species mostly obtained nectar from Penstemon flowers by piercing the base of the corolla tube without entering the floral throat. Observations of queen and worker Bombus spp. showed that these large bees typically foraged first on the lowest open flower on an inflorescence, and then climbed up the inflorescence following the spiral of open corollas and visiting flowers in both phases of protandry. АП the large bee species were observed visiting the flowering shoots of several genets in succession. In contrast, bees less than 10 mm in length (Au- gochlorella, Ceratina, Hoplitis, Hylaeus, Lasioglos- sum, and Osmia spp.) were not observed to probe for nectar. These bees were observed clinging up- side-down and collecting pollen from dehiscent an- thers. Penstemon pollen was deposited ventrally on these insects, and bees contacted stigmas while for- aging for pollen. These small bees did not appear to discriminate between flowers in the young (de- hiscent) versus old (empty) anther phases of pro- tandry in Penstemon flowers based on corolla fea- tures. These bees were observed to enter the floral tube, cling to the old, empty anthers and contact the stigmas for a few seconds, but then they exited the flowers without attempting to scrape old an- thers. These small bees were observed visiting sev- eral flowering shoots in succession before leaving the site. Approximately 5796 of all bees captured on Pen- stemon spp. carried pure loads of Penstemon pollen in their corbiculae or scopae (Table 1). An addi- tional 2096 carried Penstemon pollen mixed with the pollen of wm co-flowering species. Mixed pol- len loads included both nectar-producing (Baptisia spp., Delphinium зрр., Onosmodium sp., Pycnan- themum sp., Rubus sp., and Teucrium sp.) and nec- tarless (Rosa spp., Schrankia nuttallii) species. Pol- linaria of several Asclepias spp. were found attached exclusively to the first pair of legs. ome queen bumblebees appeared to restrict their foraging to selected Penstemon spp. Bombus pennsylvanicus was most abundant on the large-gul- let species: P. cobaea and P. grandiflorus. Bombus nevadensis was found primarily on the small, gullet- bell-shaped corollas: P. digitalis and P. pallidus. No morphometric difference could be found between B. pennsylvanicus and B. nevadensis queens. Bum- blebees were seldom observed or collected in flow- ers of populations of P. cobaea var. cobaea that con- sisted of < 50 flowering shoots (Konza Prairie sites 1995 and 1997). The density of bumblebees belonging to the neu- ter, worker caste was highly skewed in the data set. Ninety percent of all workers of six Bombus spp. Table 1) were restricted to the flowers of P. digi- talis at the Litzsinger Road Ecology Center, a re- stored site. The remaining 10% were distributed among all remaining study sites for all Penstemon spp. Although the flowering period of P. digitalis at the Litzsinger Road site overlapped broadly with P. digitalis at the two remaining sites, the presence of worker-caste bumblebees on virgin prairies and on old restored prairies was only a fraction of the total catch of workers of Bombus species at a five-year restoration site. he density and diversity of bees less than 10 mm long differed among the bee-pollinated Penste- mon spp. (Table 1). Over 50% of all bees collected on P. cobaea var. cobaea were Hoplitis pilosifrons (Megachilidae). In contrast, approximately 4596 of bees on Р. grandiflorus were Augochlorella striata (Halictidae). Most of the small-bodied bees col- lected on P. digitalis were Lasioglossum spp. (Hal- ictidae). Osmia spp. (Megachilidae) and Hoplitis pi- losifrons were equally represented on P. pallidus. Note that most of these bee taxa were collected in smaller proportions on Penstemon flowers other than the modal Penstemon species (listed above) at each site (Table 1). These smaller bees were ob- served to visit more than one open flower on an inflorescence, and to visit several flowering shoots in succession. Small bees were rarely observed and collected on flowers of P. cobaea var. cobaea in pop- ulations of > 500 flowering shoots (Konza Prairie sites 1995 and 1997). These results on the foraging preferences of bumblebees versus small, solitary bees on P. cobaea populations were highly signifi- cant by chi square (T = 37.33, P < 0.001). Sta- tistical procedures follow Conover (1980). Remem- bering that the square root of the chi square test statistic is approximately the absolute value of the test statistic of the normal distribution, we can es- timate that there is an extremely low probability, on the order of one occurrence in a million, that these results are due to chance alone (Stanley Saw- yer, pers. comm.). We also emphasize that we do not have a high enough level of sample replication in our data set to use parametric statistics, which would distinguish species effects from site effects, and acknowledge that we thus cannot make defin- itive comments here on this issue. “~ DISCUSSION Pollen dispersal varies among prairie Penstemon species in the American Midwest. The putative 134 Annals of the Missouri Botanical Garden trend toward butterfly pollination in Penstemon tu- baeflorus is reflected by the constriction of the floral tube. Otherwise, there appeared to be few floral characters indicative of classic psychophily in this species (Barth, 1985; Proctor et al., 1996). For ex- ample, flowers were not held erect and lacked the characteristic pigmentation associated with “but- This suggests that this pollination system is recent and that this species may be de- terfly flowers." rived from a bee-pollinated ancestor. In the four remaining Penstemon spp., bee visi- tation was dominant, involving a broad diversity of potential pollinators representing four families of Apoideae. Furthermore, bee pollination in mid- western Penstemon spp. appeared to be a two-tiered system. Pollination by bees in most angiosperm sys- tems is based on either passive or active contact (sensu Bernhardt, 1996) with the anthers. Pollina- tion by passive contact implies that the bee first contacts dehiscent anthers and receptive stigmas while foraging for nectar, or other rewards, toward the base of the flower. Pollination by active contact suggests that the bee contacted the receptive stigma while foraging specifically for pollen. Both modes of pollination occurred in four Penstemon зрр., but each pollination mode was performed by two dif- ferent sets of bees. Large-bodied, long-tongued bees effected passive pollination, while small-bod- ied bees, with tongues of varying lengths, effected active pollination. Since Penstemon flowers were protandrous, this implies that active pollination re- quired automimicry (sensu Dafni, 1993) to encour- age cross-pollination. Small bees foraging for pollen were unable to discriminate initially between fresh, dehisced anthers and old, empty anthers, suggest- ing that pollen collection in Penstemon flowers rep- resents a form of partial reinforcement for smaller bees. We have found no previous citations of pollina- tion of Penstemon spp. by bumblebee flower-beetles (Euphoria sepulchralis). These greenish scarabs are not hairy, but became dusted with Penstemon pollen in the process of feeding on nectar and floral organs while residing inside the base of the corolla tube for long periods of time. It is reasonable to suggest that they played a minor role in the pollination of Penstemon spp. on the Konza Prairie Research Nat- ural Area. Likewise, uncommon pseudomasarid wasps probably contributed to the pollination of P. digitalis and P. cobaea var. purpureus. The litera- ture suggests that pseudomasarids are important pollinators of some Penstemon spp. native to the west coast of North America (Straw, 1956). Three patterns of interactions between Penste- mon flowers and bees were documented. First, the foraging specificity of queens of Bombus nevadensis to P. digitalis and P. pallidus versus that of B. penn- sylvanicus to P. cobaea and P. grandiflorus provides an interesting dichotomy. The morphometric differ- ences between the two Bombus spp. with respect to glossa length and body parameters are minor, ac- cording to taxonomic authorities (Medler, 1962). In both species, the combined length of the glossa and prementum is about 12.5 mm. Based on our col- lections, it would appear that B. nevadensis pre- ferred Penstemon corollas with small gullets, while B. pennsylvanicus preferred corollas with large gul- lets and much expanded sinuses. (In contrast, both B. griseocollis queens, which rarely visit penste- mons, and B. fraternus queens, which did not visit penstemons at all in this study, have a combined glossa plus prementum length of about 10 mm.) However, statistical analyses of the foraging pref- erences of queens of B. nevadensis and B. pennsyl- vanicus would require additional data sets from a far greater number of sites. In fact, we acknowledge that the above-mentioned relationship between cer- tain Bombus queens and Penstemon spp. may ac- tually reflect a combination of overlapping phenol- ogies and biogeographies. econd, bumblebee queens far outnumbered neuter workers on Penstemon flowers. This was to be anticipated considering the well known, annual life-cycle of Bombus spp. in the Northern Hemi- sphere, since Bombus queens always forage for pol- len in spring until they raise a significant retinue of workers (Heinrich, 1979). What was not expect- ed was the relative abundance of workers on res- toration sites (V = versus the relative absence of workers (N — 7) on both original, tallgrass prai- ries and true glade sites. We were unable to deter- mine whether Bombus workers simply avoided Pen- stemon populations in “virgin” prairies and glades or whether Bombus colonies matured more rapidly within, or adjacent to, restored sites. Whatever the case, this pattern, if continuous, may have direct implications for future conservationists who empha- size the restoration of original pollination systems. Third, collections of small-bodied, solitary bees indicated. selective foraging preferences that cor- related negatively with the size of Penstemon pop- ulations. This occurred exclusively in P. cobaea var. cobaea within sites on the Konza Prairie Research Natural Area. Over four seasons, 59 solitary bees were collected on populations with less than 50 flowering stems, but only four specimens were ever collected on populations with more than 500 flow- ering stems. These massively flowering populations of P. cobaea appeared to be pollinated primarily by Bombus queens. Why did solitary bees avoid the Volume 85, Number 1 1998 Clinebell & Bernhardt Pollination Ecology of Penstemon 135 greater resources offered by a much higher density of Penstemon flowers? Our only explanation is that Bombus queens outcompeted solitary bees for ac- cess to Penstemon corollas in larger populations, even though they did not compete for the same re- source. Bernhardt (1993) mentioned older reports that female xylocopines (Anthophoridae) recog- nized and avoided flowers of Passiflora spp. that had been visited and marked with a glandular se- cretion by another female of the same species. Per- haps solitary bees recognized and avoided Penste- mon flowers that had been visited and marked by Bombus spp. The comparative disinterest that Bom- bus queens showed to smaller populations of P. co- baea may have reflected both economic and ener- getic considerations. As bumblebees are strong fliers and trapline foragers (Heinrich, 1979; Bern- hardt & Montalvo, 1979) they may have avoided smaller populations of flowering plants because such populations offered weaker visual/olfactory displays during peak flowering periods. ile a traplining forager would seem most appropriate for the pollination of smaller, isolated populations, their visitation contradicts bumblebee economics (Heinrich, 1979). vide inadequate rewards for both the bees and the Small flowering populations pro- rvae ple by the foraging queens (Bern- ът и et m (1996) argued that generalized pol- lination is the rule, not the exception, in most an- giosperm species. Floral generalization becomes se- lectively advantageous when pollinator populations fluctuate, so that pollinator shifts can occur in an- giosperm populations. e do not accept that bee pollination in all mid- western Penstemon spp. must also reflect fluctua- tions in generalized trends favoring either large or small bees. We interpret bee pollination in four Penstemon spp. as two specialized syndromes con- current within the same flowers. А coexisting, two- tiered syndrome has been selected for pollination by both large-bodied, long-tongued, nectar-foraging bees and small-bodied, pollen foragers. Note that all Bombus spp. and most of the solitary bees (ex- cluding Osmia s. str.) collected in this study are polylectic taxa in grasslands (Bernhardt, 1990а, b). Note also that the same bee species may adopt ei- ther active or passive pollen collection on different, coblooming flowers (Bernhardt, 1996). With two bee-pollination syndromes operating in the same Penstemon flower, either syndrome can encourage outcrossing, regardless of demographics, in any Penstemon population. It seems naive to pre- sume that dichotomous modes of pollination in the same flower usually represent shifts in generalized patterns. For example, the older literature has long suggested that a two-tiered system has existed in some species pollinated by both birds and bees. Both birds and bees contribute to pollination in one species because two different modes of attractants and rewards overlap within the same flower (Grant & Grant, 1968; Breedlove, 1969; Macior, 1975; Grant, 1976; Schemske, 1978). If bees and ho- meothermic vertebrates can pollinate the same flower, we should also be able to postulate two dif- ferent groups of bees pollinating the same flower for different rewards. Frankly, a generalized mode of pollination must always be implied when field studies fail to incorporate basic analyses of pollen loads and observations of foragers contacting re- ceptive stigmas. Literature Cited логоа W. S. 1993. Evolution of plant pollination eses and tests with the е пе otropical vine 0- Insects апа Flowers "The Biology of E -m George Allen & Unwin, Princeton, New бөлке P. 1990а. Pollination ecology of Oxalis vio- lacea (Oxalidaceae) following a controlled grass fire. Pl. Syst. Ev a da -155. . Anthecology of Schrankia nuttallii (Mi- iR чи on thie tallgrass prairie. Pl. Syst. Evol. 170: 55. 1993. Natural Affairs; A Botanist Looks at the Attachments Between Plants and People. Villard, New ork. 1996. Anther M in animal pollination. Pp. 192-220 i in W. С. D'Arc R. C. Keating (editors), The Anther; Form, Function = Phylogeny. Cambridge Univ. Press, Cambridge, U.K. & E. A. "iei cid 1979. The pollination of Echeandia macrocarpa (Liliaceae). Brittonia 31: 64—71. — & P. Weston. 1996. The pollination ecology of Persoonia (Proteaceae) in eastern Australia. Telopea 6: 5-803. Breedlove, D. E. 1969. The systematics of Fuchsia Sec- tion Encliandra Wis ад Univ. California Press, Berkeley and a ngeles. тене SE La, ^B Jones & L. J. va 1978. Vi- bratile inus on of Solanum douglasii and S. xanti (Solanaceae) in southern California. Wasmann J. Biol. 35: 1-25. Clements, F. E. & F. Long. 1923. 2” роШпа- tion. Publ. npe Inst. Wash. 336: 1-274. Conover, W. J. 1980. Practical Nonparametric Statistics, 2nd ed. John Wiley and Sons, New York Crosswhite, F. S. 1965. Revision of Penstemon (Sect. Pen- stemon) Series Graciles with a Synopsis of the Genus. Unpublished Master's Thesis, University of Wisconsin, Madison. — — — & С. D. Crosswhite. 1966. Insect pollinators of Penstemon series Graciles (Scrophulariaceae) with notes on Osmia and other Megachilidae. Amer. Midl. Natu- 67. . Pollination Ecology; A Practical Ap- маб Oxford Univ. Press, Oxford, О.К. 136 Annals of the Missouri Botanical Garden Evers, R. A. 1955. Hill prairies of Illinois. Bull. Illinois The Bee Genera of North and Central America (Hy- Nat. Hist. Surv. 26; 409—410. menoptera: Apoidea). ao Institution Press, Freeman, C. C. 1981. A Biosystematic 4 of the Ge- 1. D.C., and Londor reat Plains. Ochs, 993. An Ecological сев ey of the Litzsinger nus Penstemon (Scrophulariaceae) in t Unpublished Masters Thesis, Kansas Bus University, an, 2. . Hulbert. 1985. Ап annotated list of the a i of Konza Prairie Research Natural Area. s. Kansas Acad. Sci. 88: 84-115. Goldblatt, P., J. C. Manning & P. Bernhardt. 1995. Pol- lination biology of 2. subgenus Lapeirousia (Iridaceae) in southern Africa: Floral divergence and adaptation for long-tongued fly pollination. Ann. Mis- souri Bot. Gard. 82: 517—534. Grant, K. A. & V. Grant. 1968. Hummingbirds and Their Flowers. 4. Univ. Press, New York and London. Grant, V. 1976. Isolation Бегей Aquilegia formosa and A. 2. А reply and reconsideration. Evolution 30: 625—628 1994. Modes and origins of mechanical and ethological isolation in angiosperms. Proc. Natl. Acad. Sci. U.S.A. 91: 3-10. Heinrich, B. 1979. Bumblebee Economics. Harvard Univ. Press, Cambridge and London. Kampny, C. M. 1995. Pollination and floral diversity in Scrophulariaceae. Bot. Rev 22. 350-366. Keck, D. D. 1938. Studies bi Penstem I. The section Aurator. Bull. Torrey Bot. Club 65: m awson, pedino & T. L. Griswold. 1989. Pollen collec ion and oiher insect visitors to 2. һауаепі S. Wats. Pp. 233-235 іп T. B. Stubbentlece ico Proceedings of ag: Fie North Americ p ips Conference. Macior, L. W. The pollination of Delphinium tri- corne 1. ala ede). Amer. J. Bot. 10: 1009-1016. 1982. Plant community and pollinator dynamics in the ee of pollination mechanisms in did laris (Scrophulariaceae). Pp. 29—45 in J. A. Armstrong, J. M. Powell & A. J. Richards (editors), Pollination and Evolution, Royal Botanic Gardens, Sydney, NSW, Aus Medler, Í T. 1962. Morphometric studies on bumblebees. Ann. Entomol. e r. 55: 212-218. Michener, C. D., R. J. McGinley & B. N. Danforth. 1994. n Ecology Center. ion Department, Missouri Botanical Garden, St. Louis. Ogden, E. C., G. S. Raynor, J. V. Hayes, D. M. Lewis & H. Haines. 1974. Manual for жеты A bee Pollen. Hafner Press, New York. Pennell, F. W. 1935. бо In Scrophulariaceae of eastern temperate о 27 Acad. Nat. Sci. Phil- adelphia, pL 1: 196-273. o & A. ins 1996. The Natural History Timber Press, Portland, Oregon 92. Flowers and ao Asclepiadaceae Acad. . Louis. 5: Proctor, M., P. of нен Robertson, C. to Sc a eae. Trans. 069—598 1929. NS and Insects. C. Robertson, Car- lines: Шіп Schemske, D. W 1978. Evolution of reproductive char- acteristics in /mpatiens (Balsaminiaceae): The signifi- cance of cleistogamy and chasmogamy. Ecology 59: 596-61 А Solecki, M ‚ E. A. Cook & P. S. Haverland. 1986. а ин 2. of Three Missouri Tall- grass Prairies with Reference to Past Management. Con- servation Commission of the State of Missouri, Jefferson ity. Straw, R. M. 1956. 2” isolation in Penstemon. Amer. Naturalist 90: 46— Toney, T. E. ж кет — Prairies of Missouri. Mis- souri Department of Conservation, Jefferson City. Waser, N. M., L. Chittka, M. n Price, N. M. Williams & J. Ollerton. 1996. Generalization in л pollination systems and why it matters. Ecology 77: 1043-1060. Wolfe, A. D Elisens. 1993. Diploid hybrid spe- ciation in Penstemon revisited. Amer. J. Bot. 80: 108. 1094 ‚ М. J. Elisens, L. E. Wat gs C. W. dePamphilis. 1997. Using restriction-site variation of PCR- Р cpDNA genes for pre analysis o e do neae (Se Edd eae). Ame ot. 64. Yatskievych, . Turner. 1990; бил ine dica ги 2” Сте Syst. Bot. Missouri Bot. Саг A COMBINED CLADISTIC Owi I. Nandi?, Mark W. Chase? and ANALYSIS OF ANGIOSPERMS 74% K Endress? USING rbcL AND NON-MOLECULAR DATA SETS! ABSTRACT А combined analysis of 162 extant angiosperm taxa for which rbcL 2” -data and/or an appreciable amount of non-molecular information is available was calculated. A non-molecular tree, an rbcL tree, and a combined tree are presented. Only the rbcL and the combined data set show large numbers of preces. with as percentages greater than 50%, whereas the non-molecular trees show only eleven clades of this kind; this seems due to the number of missing cells in the non- -molecular matrix. We tried to identity non-molecular characters (including biochemical) M ¡here clades turned out to be new when compared to o oupings fond in the non-molecular analysis that parallel the rbc а, topologies include a grade tala Шеше, жыны and Amborellaceae (magnoliid II); а clade containing Magn oliales, Laurales, Aristolochianae, and monocots (magnoliid I); a hamamelid group; subgroups gn asteri ls (e.g., a s.l. clade; a partial Malpighiales grade containing Quiinaceae, Linale Flacourtiaceae s. str., and Ochnaceae; and some smaller clades, similar to the corresponding groups found in rbcL is ide Шы. Aristolochianae-monocots; Hydrangeaceae-Cornales; Lecythidaceae—Scyto- etalaceae; Pittosporaceae—Araliales; Geissolomataceae-Stachyuraceae; Connaraceae—Oxalidaceae). Capparales s.l. and ihe nitrogen-fixing clade, two novel milesi lar clades, are only found in the rbcL and the combined trees. Cistaceae shown to share important characters with Malvales s.l. and are consistently found within this clade. These e in a publi ogeny. Considerable progress has been made in na a morphological/ chemical data set that parallels the broad coverage of angiosperms seen in DNA studie New opportunities for the study of seed-plant phylogeny have opened due to the continued de- velopment of computer hardware and software. In addition, gene sequencing has become reasonably fast, and large nucleotide data matrices have been produced (e.g., Chase et al., 1993; Savolainen et al., 1996; Soltis et al., 1997b). These studies have stimulated even more molecular work on macrosys- tematics, including the addition of more “critical” taxa to the data matrices, comparison with results from other gene sequences, and the combination of nucleotide with non-molecular data matrices, as has been undertaken in this study. Other examples of broadly sampled combined nucleotide and non- molecular studies are those of Doyle et al. (1994) and Chase et al. (199 The non-molecular investigations of this study originated from the question of the position of Cis- taceae within eudicots. Cistaceae have been in- cluded in Violales (Takhtajan, 1966; Cronquist, 981) and Malvales (Dahlgren, 1980), yet the most natural (i.e., phylogenetic) position has remained a matter of debate (Thorne, 1983, 1992). Thus, it was a major objective of our non-molecular study to identify the accurate position of Cistaceae and their allies (Bixaceae and Cochlospermaceae) within the eudicots. Most families that have commonly been allied with Malvales or Violales are included in ' We thank Thomas Baumann, Roland Eberwein, Helena Eklund, Mary Endress, Else Marie Friis, Robert эру Anton Igersheim, Alexander Kocyan, Reto Nyffeler, Markus Кеш, Rolf Rutishauser, Harald Schneider, Jürg Schón berger, Edwin Urmi, and Christian Wagner for their criticism and help, and Phil Ackery, Victor Albert, Abelardo ord, Aparicio, Pieter Baas, Harv Donald Les, Richard Olmstead, and Valentina Ukraintseva for useful communications. We g Jauch, Shuqing He, Rosemarie Siegrist, Annette Nandi-Koch, and the many and dads in Kenneth Cameron, Michael di Jam Hoot, Lola Lledó, буша Morton, Yin- inis 16 Vincent viam = for Zoll rude 107, 8008 Zürich. өлке dom Bil Alverson, Diane Bowman, Anette de Bruijn, Walter Judd, Douglas Soltis, and an us ? [nstitute of Systematic Botany, Тын of Züric rvey Ballard, Gilles Dutartre, James Farris, Jeffrey Harborne, Mark оп Larry Ни! atefully Sanat at Urs rbcL seque en, and Susan Swensen Na s are also due to cally examining the m * Royal Botanic Gardens, Kew, Richmond, Surrey, TWO 3DS, United King ANN. MISSOURI Bor. GARD. 85: 137-212. 1998. 138 Annals of the Missouri Botanical Garden the present study. Many more taxa were added to evaluate the robustness and position of the two or- ders within eudicots and to compare the trees ob- tained with other studies (Hufford, 1992; Olmstead et al., 1992; Albert et al., 1992; Chase et al., 1993). Monophyly of the taxa used has also been eval- uated by comparing molecular results with macro- Berg, 1977). This was done to determine which terminal taxa should be used in the non-molecular sampling. In the non-molecular matrix we make use of those clades found in the rbcL "pied that are compat- systematic studies (e.g., Urticales: ible with widely accepted groups. Thus in some cases we have sampled individual families, whereas in others we have used orders (e.g., Gentianales, Annonales, etc.), superorders (e.g., Faganae, Aristo- lochianae), or larger groups (e.g., monocotyledons) that no recent research (molecular or non-molecu- lar) has indicated are other than monophyletic. Fla- courtiaceae s.l. were split into two groups, one wit cyanogenic glycosides (e.g., Kiggelariaceae) and one without major caveat for the non-molecular matrix is that we often used single character-states for poly- morphic taxa. These assignments are based on as- sumptions of character polarity, which could result in mistaken interpretations of character evolution. We accept that in some specific cases mistakes may have been made, but we felt that some simplifica- tions were required to deal with such large taxo- nomic units. However, the character-state assign- ments were carried out using a consistent approach (see Appendix 5). Because the independent trees are often highly similar, we gain confidence that the historical signal present in the non-molecular data has not been grossly distorted by this method of character-state assignment. We hope that further progress in non-molecular investigations will obvi- ate the need for such a procedure in future studies. We are certain that this approach can be greatly improved upon. Analysis of large, non-molecular matrices is not without precedent in plant system- atics (e.g., Donoghue & Doyle, 1989). Working with such large non-molecular matrices could h desirable effects (і.е., not finding the shortest trees or all islands of trees, cf. Maddison, 1991). These large matrices must nonetheless be much less con- founding than matrices using exemplar taxa for groups that are not monophyletic. We were interested in finding a large set of non- molecular characters that would contain phyloge- netic information. We tried to characterize larger ave un- taxonomic groupings, especially new ones, by non- molecular synapomorphies, as produced by Mac- Clade 3.04 (cf. Maddison & Maddison, 1992). We wanted to see in which way the non-molecular data set changed or confirmed the topology of the rbcL tree, and vice versa, when both data sets are com- bined (we agree that the inclusion of rbcL results in delimiting the terminal taxa and in looking for taxa with more ancestral characters within the larg- er of these terminal taxa makes it impossible to claim that both data sets are totally independent). We also examined by simple comparison whether the different samplings of taxa in the rbcL analysis of Chase et al. (1993) and that of the present study affect the topology of the rbcL tree. Finally, we were also interested in the stability of the topologies ob- tained after applying the parsimony jackknife pro- gram (Farris et al., 1997) and bootstrapping (Fel- senstein, 1985). MATERIAL AND METHODS GENERAL METHODS The matrices were analyzed using PAUP 3.1.1 (Swofford, 1993). The shortest trees were collected and swapped on to completion, keeping in this case all additional trees found at this shortest branch length. In the Results and Discussion sections, we will mostly use the same terms for the larger an- giosperm clades used by Chase et al. (1993: part B of figs. 1-15) to facilitate comparisons. Ceratophyllum was specified as the outgroup in agreement with the results produced with rbcL (Chase et al., 1993). We simply used Ceratophyllum as the outgroup to avoid the issue of where in the angiosperms the root should be placed. This topic will be discussed in other papers; we view it as too complex an issue to be dealt with adequately here. The use of non-angiosperm outgroups (Gnetales) for the non-molecular matrix is difficult. Important morphological structures cannot be adequately ad- dressed in terms of their homology at present. Be- cause we used Ceratophyllum as the default out- group, we will not concern ourselves with the evaluation of its position. We were interested only in examining general patterns within the angio- sperms for both rbcL and non-molecular data. All matrices are available on diskette or by e-mail (m.chase@rbgkew.org.uk) from the second author (please provide a single high-density diskette). n each case, the products of the initial searches were sets of trees with equally weighted characters. We intended to use the jackknife procedure of Farris et al. (1997) but found that the number of missing cells in these matrices makes this method unsuitable because it found no support for any groupings in the combined matrices [J. Farr comm., reports that missing data signific mb lower s, pers. Volume 85, Number 1 8 Nandi et al. 1 Combined Cladistic Analysis of Angiosperms jackknife values; we have also investigated this em- pirically in another study (Fay et al., 1997)]. We therefore used the bootstrap consistently for an eas- ier comparison. For trees illustrated in this paper, branch lengths are shown above the branches (ACCTRAN optimization, Swofford, 1993), and all branches not present in the strict consensus trees are indicated by an arrow. Bootstrap support for supported groups is indicated below the branches. NON-MOLECULAR DATA MATRIX (APPENDIX 1) A selection of 161 angiosperm taxa was scored for 252 characters (Appendix 3); 151 taxa were ul- timately included in the non-molecular search (115 families, 32 orders, and 4 supraordinal taxa, mainly in the sense of Takhtajan, 1987; Appendix 2). Data were taken from selected synoptic literature, from primary literature (especially in dilleniids sensu Cronquist, 1981) and from original observations by the first author [leaf venation and dentation (studied in the Herbaria of Zürich, Geneva, and Vienna and in a number of botanical gardens) and observations from anatomical sections or SEM micrographs in Cistaceae, Cochlospermaceae, Bixaceae, Diptero- carpaceae, Sarcolaenaceae, Sphaerosepalaceae, and Berberidopsidaceae]. We had also at hand the extensive anatomical slide collection of the third author. Uncertain cases with regard to the presence or absence of oxalate crystals and to ovule anatomy were resolved by careful observation of selected slides (e.g., oxalate crystals seem to be absent from Amborella; see also Metcalfe, 1987). We used the characters that we considered to contain the most significant phylogenetic informa- tion. Floral developmental information could only be scored with two characters (characters 223 and 224 in the matrix) due to the complexity of com- Characters grouped into the following classes: 1. Serology (16 paring developmental data. were characters); 2. Chemical compounds (88 charac- ters); 3. Characters at cellular level (22 characters); 4. Embryology (18 characters); 5. Seed anatomy (21 characters); 6. Stem morphology and anatomy (24 characters); 7. Leaf characters (17 characters); 8. Floral and fruiting characters (46 characters). The procedure of assigning character-states to taxa is documented in more detail in Appendix 5, but in general the hypothesized plesiomorphic state was used if more than one state occurred within a ter- minal taxon. In a few cases, paleobotanical infor- mation was also included (e.g., Magnoliaceae, Pla- ‘eae; Crane, 1989; Crane et al., 1991; Пусћег & Crane, tanaceae, Buxac 993; Drinnan et al., 1984). Most of the characters are mutually independent. Other characters were chosen as hierarchical sets e.g., characters 8—10, 80—85, 97—100, 200-202, or 250-251), character pairs (223-224), or char- acter triplets (108-110, 142-144). 2... with the molecular data set was reache nly allowing four character-states (“А,” “С,” “С, > "i “А” can be een to “0,” *C" to “1,” *G" to “2,” and “Т” t . Characters with more than four states were diee up into character sets (233— 235), but few characters required such modifica- ~ ion. Of the 252 characters, 207 are binary (in 186 of these simple presence/absence coding is involved) are multistate characters. The common strategy of character-state assignment described in Appendix 5 was to find a basal pattern for each taxon. With this procedure, we tried to reduce char- acter radiation due to the evolutionary processes within the terminal taxon. In characters with states that have low probability to be evolved, due to their complexity, presence of a state was favored in cod- ing over absence of a state (e.g., presence of phloem м-. stratification; 169; presence of bixoid exotegmen in the chalazal region; 159; presence of salicoid leaf dentation; 201). This implied that the character was coded as being present if at least one representative of the terminal taxon is known to exhibit the char- acter-state. By analogy, for dithetic characters, in which both states are more or less equally likely to have evolved, the character was coded as polymor- phic if both character-states occur in a terminal taxon (e.g., successive or simultaneous microspo- rogenesis; 128). More specific rules, which were rarely applied to cover more complex hypotheses of character evolution, are given in Appendix 5. These exceptions were applied restrictively, since the number of assumptions prior to analysis of data should be low. Appendix 5 also lists characters for which the putative ancestral character-state was searched by scoring the character-state in putative- ly basal members of the terminal taxon (e.g., Ul- maceae in Urticales, Ceratonia in Fabaceae, Ery- throspermum in Kiggelariaceae). iochemical characters were scored as absent, present, or “?” (unknown or uncertain) in certain terminal taxa by considering the extent of knowl- edge of the biochemical substance classes in ques- tion (in the specific taxon). In wood anatomy, the broadly acknowledged evolutionary trends (see e.g., Carlquist, 1988a: chapter 11) from ancestral to de- rived character-states were used strictly to find character-states of terminal taxa [i.e., uni-or biser- iate circular or scalariform vessel side-wall pitting was preferred over opposite, and opposite over al- 140 Annals of the Missouri Botanical Garden ternate (184); vessellessness was preferred over presence of vessels (integrated in 185); scalariform perforation plates were preferred over mixed sca- lariform and simple plates, and the latter over sim- ple plates (185)]. l characters were scored as unordered. Am- biguous characters for which no priority rule was applied (see Appendix 5) were coded as polymor- phic. Assignment of character-states was compli- cated by the fact that often only the presence (and not the absence) of a character-state is noted in the literature. If one or more thorough published stud- ies of a character class in a certain terminal are available, and a given character-state was not de- scribed, it was coded as being absent. This rule was applied to, for example, presence or absence of exotegmic palisades or exotegmic longitudinal fi- bers in seeds (157). Some extrinsic characters such as hostplant and paleobotanical data were not in- clude evaluated from the trees produced. in the matrix, but their optimization was Information on taxon circumscription, characters and character-states (including four data errors that were detected after all analyses were completed), character definitions, procedures of character-state assignment, and sources are given in Appendices 2-6. The four errors were examined for effects by initiating searches on the trees found; we found no additional or shorter trees, so we assumed that com- plete new searches were not warranted. In Search I, 100 random-addition replicates were run using TBR branch swapping (Swofford, 1993), but keeping only 10 trees per step (TBR = tree bisection-reconnection). The first tree of the short- est tree set obtained was swapped to completion (1.e., MULPARS turned on; Swofford, 1993). The steepest descent option (Swofford, 1993) was not used. This island that was found (sensu Maddison, 1991) contained 136 trees of length 3546, consis- tency index (CI) = 0.09, and retention index (RI) — 0.41. In Search II, 1000 random-addition replicates were run using SPR algorithm and keeping only 1 tree per step (ӘРЕ = subtree pruning-regrafting; Swofford, 1993). The shortest tree was swapped to completion with MULPARS turned on. The steepest descent option was not used. The resulting island obtained with this method contained 8 trees of length 3545, CI = 0.09, and RI = 0.41 These two searches were very slow (much slower than the rbcL and combined searches), and we sus- pected that shorter trees could be produced by an- other strategy, described below. In Search Ш, the taxa were divided into three groups. Group I contained taxa 1 to 37 (presumed magnolids, lower eudicots, and caryophyllids). Group II contained taxa 38 to 106 and taxa 149, 150, 161 (presumed rosids). Group III contained taxa 107 to 148 (presumed asterids). First, group II was processed. One hundred random-addition replicates were done using the TBR algorithm and keeping maximally 10 trees per step. The topology of the strict consensus of the shortest tree set (con- taining two most-parsimonious trees) was defined as a constraint framework for the following step. The undred random- addition replicates were done with TBR swapping and keeping maximally 10 trees per step. After this, taxa of group I were added. One the constraints were omitted and all taxa of the first tree of the shortest tree set obtained were allowed BR algo- rithm. The resulting tree set was defined as a con- straint for the following step, with the taxa of group III added. One hundred random-addition replicates to swap freely to completion using the were done with TBR swapping but keeping maxi- mally 10 trees per step. The constraints were omit- ted again, and all taxa of the first tree of the shortest tree set obtained were allowed to swap freely to completion using the TBR algorithm. The trees ob- tained were 3544 steps long. This tree set was re- weighted based on the rescaled consistency index with maximal weight of 10. Twenty steps of length reduction were done with this new weight set using the TBR algorithm. After this, all characters were again weighted equally. The trees obtained in the last procedure were swapped to completion. More than 2200 trees of length 3541 were obtained. The search was stopped due to memory constraints. The first 50 trees of the obtained tree set were reweight- ed based on the rescaled consistency index with maximal weight of 100. Twenty steps of length re- duction were done with this new weight set using the TBR algorithm. Afterward all characters were again weighted equally. The trees obtained in the e l. As evidence in support of the use of successive weighting, we used MacClade 3.06 (M dison, 1992) to plot how many steps we the shortest trees found with the non-molecula re contributed by informative characters in each data matrix. In pa was changing 60 times ar data only. —B( — Maddison & Mad- art B, for e data were 2. on the op). The non- mu ‘ular data optimized on one of dele >). The non-molecular n жаз on a tree = = from the combined analysis. —C (bottom). Plot of the rbcL data | onto the combined t Volume 85, Number 1 Nandi et al. 141 998 Combined Cladistic Analysis of Angiosperms 28 non-molecular data on non-molecular tree сда 4 ini 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 у non-molecular data on combined tree = “о 30 31 32 33 so rbcL data on combined tree Ё 3 E 2 12345 6 7 8 9 1011 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 number of steps 142 Annals of the Missouri Botanical Garden last procedure were swapped to completion. The steepest descent option was also not used in Search III. An ultimate tree set of 17 trees with a length of 3539, CI = 0.09, and RI = 0.41 was obtained; the first tree of this set is illustrated in Figure Arrows indicate groups not found in all 17 trees. This is obviously only one island of many that exist for this data set, but it is the shortest tree length that we were able to obtain, and in spite of the unorthodoxy of the procedure, it produced far short- er trees than any "standard" method (i.e., with rep- licates of random taxon-addition). The three taxon- groups in Search III were formed by comparison with rbcL topologies. This introduces some bias in the non-molecular trees, but the application of these three groups was responsible for finding the shortest trees. MATRICES FOR rbcL AND COMBINED DATA The techniques involved in collecting our rbcL data have been previously published (Chase et al., 993; Chase et al., ecause each rbcL se- quence represents a specific single plant (Appendix 7), which we assume can represent из family or other higher taxon, we used a single rbcL sequence to represent each of the terminals scored in the non-molecular matrix. In general, we selected as the rbcL representative a species that was not es- pecially sequence-divergent within its group. Many of the sequences included in the present study are previously unpublished. We analyzed this new rbcL matrix to be certain that we could obtain results similar to those of other published rbcL topologies. Taxa for which we still have no molecular data are marked with “$” in Figure 4 A, B Problems in amalgamating nucleotide and non- molecular data sets are discussed in Chase et al. (1995). The main problem is that non-molecular characters were scored for higher taxa whereas each rbcL sequence represents a single plant. In our experience, this technique does not appear to produce spurious results (several such studies have been published and more are in progress in the laboratory of M. Chase; Chase et al., 1995; Gadek et al., 1996; Morton et al., 1997). The trees ob- tained with more taxa, thus spanning the diver- gence levels present within a family, do not produce wildly different patterns, nor does substitution of one species in a family for another greatly affect the position of the family (provided the family is monophyletic). This fact is obvious when one com- pares the patterns found for multiple members of a family in the 1993 rbcL tree (Chase et al.) with the position of that family in the present analysis, in which only one taxon was included. Furthermore, there should be no expected correspondence be- tween a morphologically plesiomorphic taxon and plesiomorphic molecular characters, so taxon se- lection based on presumed *basalness" is not an important consideration. However, if faced with a choice between species that are highly divergent and others that are only slightly divergent, then use of one of the latter is helpful in avoiding spurious placements of the family We excluded the first 27 base positions at the 5' end of rbcL, leaving a maximum of 1401 basepairs (bp) of data for each species (some were less than this, although. попе substantially less than. 1300 yp). Of these 1401 sites, 785 (56%) were variable and only 562 (4096) were potentially informative. We used 1000 replicates of random-taxon entries and the TBR-swapping algorithm; only five trees were retained per step, which reduces the amount of time spent swapping on trees from suboptimal tree islands (Maddison, 1991). Although all trees shown were produced by successive weighting, we have shown them with their Fitch branch lengths Fitch, 1971; i.e., characters with equal weights, character-states unordered) in Figures 3 and 4. In all but the strictly non-molecular matrix, we em- ployed successive weighting to down weight or eliminate the effects of characters that changed ex- cessively (Farris, 1969; Carpenter, 1988, suggested that successive weighting should be used merely to select а subset of the trees found with equal weights). To illustrate the reasons why we favor the use of successive weighting, we plotted the number of steps vs. the number of characters in MacClade 3.06 (Maddison & Maddison, 1992) for both the non-molecular and the rbcL matrices (both on the combined tree and on the shortest non-molecular — trees). Once excessively homoplasious characters were down weighted, it was logical not to use those characters in estimating internal support. Hence relative weights were employed in the bootstrap- ping procedure (5000 replicates for each matrix) except for the non-molecular matrix, which was evaluated with equal weights. The rescaled consis- tency index (RC; Swofford, 1993) was used to cal- culate the successive weights (with a base weight of 1000) based on the best fit in any tree for each character, and in bootstrapping characters were sampled with equal probability rather than having the frequencies depend on the weights. We used a “fast” bootstrapping procedure in which a minimal amount of NNI swapping was used (the fastest and most superficial of the PAUP swapping algorithms; we permitted only 20 trees to be retained at each step). This procedure obviates the need to swap Volume 85, Number 1 Nandi et al. 143 Combined Cladistic Analysis of Angiosperms extensively, significantly shortening the time to car- ry out bootstrapping. Well supported groups are present in the starting trees (due to the quick dis- tance-based calculations that PAUP and other par- simony programs use to generate a tree upon which swapping is then carried out) and do not need any swapping to be identified. If extensive swapping is required to “find” groups, then they are obviously weakly supported or unsupported; groups with in- termediate levels of support necessitate at least some swapping to be effectively evaluated, hence the limited use of NNI swapping. We expect that the use of successive weighting will in many cases, as here, find trees for which the Fitch lengths (equal weighting) are longer than for the shortest trees found with Fitch parsimony. This is due to the fact that when highly homoplasious characters are down weighted, more consistent characters (those with higher relative weights) will be optimized more parsimoniously, thus forcing more changes into al- ready highly homoplasious characters because such actions actually reduce the weighted tree length. Some characters in these matrices do change ex- cessively often (see Results below), and thus it seems logical to us that once we have eliminated the effects of highly variable characters in the tree search procedure, these weights should be em- ployed as well to evaluate internal support. Char- acters that change as often as 40—60 times should be eliminated from consideration; it seems obvious to us that these characters are not useful at this taxonomic level. In the interests of retaining a rea- sonable lack of a priori sifting of characters, we kept all characters until the initial patterns ob- tained indicated a lack of appropriateness of some data (such winnowing is of course not possible with А data unless one resorts to whole-category weights, and we do not find any evidence that such weighting schemes are appropriate; Chase et al., 1995, found that third codon positions in rbcL were better phylogenetic data than first or second posi- RESULTS AMOUNTS OF HOMOPLASY The numbers of times each character of the non- molecular or of the molecular matrix changed on different trees are illustrated in Figure 1. As esti- mated on the combined tree found with successive weighting, some characters in the non-molecular matrix were changing up to 60 times (Fig. 1B); i the rbcL matrix, fewer sites were changing as fre- quently, although one site did change 57 times (Fig. 1C). For the non-molecular data, 26.0% of the characters changed five times or less (Fig. 1B; ver- sus 30.3% on the shortest non-molecular tree, Fig. 1A), and 26.4% of the rbcL characters fell into this same category. Examples of non-molecular char- acters that changed frequently are distributed among different character types; pollen: polar pol- len diameter (131) changed 45 times, and sexine texture (135) changed 45 times; seed anatomy: ovu- lar or seed vascular bundles (145) changed 45 times, and embryo size (163) changed 45 times; wood anatomy: wood parenchyma (174) changed 46 times; fruits: seed to carpel number (249) changed 45 times. None of the serological or chemical char- acters changed more than 28 times. Both molecular and non-molecular data had nearly the same per- centage of reasonably non-homoplasious charac- ters, but many of the non-molecular characters pos- sessed only two alternate states; therefore when these characters change two or more times homo- plasy is involved, whereas base positions in DNA sequences can change up to three times without hus this comparison of percent- cussed here. It should be noted that with successive weighting of nucleotides based on the those changing three times uniquely (e.g., from A to C, А to С, and A to T) retain the same weight as those changing only once, whereas binary characters that tions). change three times will be drastically down weight- — Figure 2. One of seventeen equally most- ae trees derived from the non-molecular matrix in Search ІП found with equally d characters. T hese found in all seventeen tre with an arro es have 3539 steps with CI = 0.09 and RI = 0.41. Branches not w. Numbers above the branches are the numbers of estimated re markec substitutions (ACCTRAN е канд Underlined жекте below branches аге bootstrap values; branches without an underlined number i bootstra with Ceratophyllacea are derived. Wi Mostly rosid, derived clade. rap percentages of less than 5096. —A (left). as the outgroup. Magnoliids and hamamelids form thin subio ‘ots, rosids form a grade in which the asterids s caryophyllids are embedde *Triaperturate taxa embedded within uniaperaturate grade. t Glucosinolate-producing taxa. First-brancl hing portion of the tree, arranged a grade, out of which most of the eudicots d. —B (right). 144 Annals of the Missouri Botanical Garden _ rosids Gunneraceae Eucommiales Pittosporaceae III ae М asterid I, II & IV N \ e N | 5 asterids 9 rosids 9 pu Coriaraceae = pha liaceae г ephalotaceae T rossosomataceae T aeoniaceae Trochodendrales Proteaceae lower hamamelids T orellaceae Chloranthaceae Ceratophyllaceae Volume 85, Number 1 Nandi et al. 145 1998 Combined Cladistic Analysis of Angiosperms Dioncophyllaceae Áncistroc adaceae lumbaginaceae Droseraceae тасш сш hvllid | elaceae Podostemaceae caryopny S 5.1. immondsiaceae Balanophoraceae 14 29 22. "E n 1 7-1,4 |Б ЕЙ ma] 11 Cistaceae Bixaceae Malvales s. 1. ae | aerosepalaceae usiaceae onnetiaceae 8 17 b al Я рее rosids 17 оис arales 4 ас Ec s. Str. " 1ggelariaceae i ¡Slicese LII Celastrales s. str. ересек орао й Lr Linales 5, Е. " Qu inaceae pu Ene Hamamelidales 146 Annals of the Missouri Botanical Garden ed. The main point here is that successive weight- ing is based on the rescaled consistency index. This permits the dissection of patterns of change more accurately than merely eliminating base positions that change excessively. Weighting with the RC is also much more appropriate for DNA sequences than simple weighting with CI. Furthermore, merely eliminating all characters that change more than a certain arbitrarily set number of times (e.g., more than ten times) will eliminate some characters with multiple states (e.g., nucleotides) that retain a great deal of signal. NON-MOLECULAR DATA MATRIX AND TREES In the construction of the non-molecular matrix, some original observations, mainly in Malvales and Violales, placement of Cistaceae applying cladistic methods are convincingly supported by specialized synapo- were made. The observations on the morphies. А seed with a specialized structure in the chalazal region (an exotegmic palisade layer curved inward at the chalaza, and with a hypostase plug fitting into this dome-shaped curvature), was found in the seeds of several taxa (159). We termed this chalaza structure a bixoid chalaza (Nandi, 1998a). The occurrence of this chalaza type was known for Bixaceae, Cochlospermaceae, and Cis- taceae. We found it also in Pakaraimaeoideae and Monotoideae (Dipterocarpaceae) and in Sarcolaen- aceae (the character-state for Sarcolaenaceae was not included in the data matrix because it was found after the processing of the matrix). In Fla- courtiaceae, salicoid leaf dentation (201, definition see Appendix 4) was found in twelve more genera not previously known to exhibit this condition: Dis- someria, Byrsanthus, Calantica, Carriera, Flacour- tia (only some species), Homalium, Ludia (not well developed), Oncoba, Poliothyrsis, Scolopia, Trimer- ia, and Xylosma. One of the 17 most-parsimonious trees (the first one found during the search) from Search III of the non-molecular analysis is shown in Figure 2. It has a length of 3539 steps, CI = 0.09, and RI = 0.41. Branches not found in all 17 trees are indicated by solid arrows. Judging from the bootstrap results, in- ternal support for this topology is weak; only eleven groupings rec eived poolsteap Bupport of 9001 Or 4 greater: Myri dons (56%), ,,.... (59%), СІ ah mo || FA ¿452 < 12 a /Penta- phylacaceae/Marcgraviaceae (59%), Clethraceae/ Eric - (79%), bs оре in cud нөгө (51%), Camy (58%), .. эрди (75%), Carica- ceae/Passiflorales (5296), Connaraceae/Oxalidaceae (68%), Flacourtiaceae s. str./Kiggelariaceae (81%), and Akaniaceae/Bretschneideraceae (72%) Dilleniids sensu Takhtajan (1966) and Cronquist (1981) were not recovered in this analysis, and we thus treat the eudicots as being composed of ran- unculids, hamamelids, caryophyllids, rosids, and asterids; we use the narrower oo (1.e., as- terid I, rosid IL, etc.) as neces New groupings that are similar t to those obtained from rbcL studies are a grade containing Illiciales, Austrobaileyaceae, and Amborellaceae (magnoliid II), a clade containing Magnoliales, Laurales, An- nonales, Aristolochianae, and monocots (magnoliids D. a lower hamamelid group, a number of sub- groups of asterids (e.g., a similar asterid III clade containing Scytopetalaceae, Lecythidaceae, Sapo- taceae, Ebenaceae, Theaceae, Primulales, Styra- caceae, and Ericales), an expanded caryophyllid group, a Malvales s.l. group, a partial Malpighiales Marcgraviaceae, Actinidiaceae, Clethra- ceae, i» containing Quiinaceae, Linales s. str., Pas- orales, Violaceae, Kiggelariaceae, Flacourtiaceae s. str, and Ochnaceae, as well as some smaller clades (Hydrangeaceae-Cornales, Lecythidaceae— Scytopetalaceae; Pittosporaceae—Araliales; Geisso- lomataceae-Stachyuraceae; Connaraceae-Oxalida- ceae). Chloranthaceae appear consistently as an isolated family. The uniaperturate magnoliids plus the monocots form a grade and not a clade, although a large portion group. The early- branching taxa include Chloranthaceae, Amborella- ceae, Winterales, and Illiciales/Austrobaileyaceae; if the root belongs elsewhere, then Ceratophyllaceae would be a member of a group with these taxa. Cer- tain triaperturate groups (i.e., eudicots) also fall into of them do form a monophyletic this grade; these include most of the “lower” hama- melids, plus Ranunculidae, Nelumbonaceae, Berber- 2A). The rosids also form a grade that gives rise on the one hand to idopsidaceae, and Eupteleaceae (Fig. : the asterids and on the other hand to the expanded caryophyllids (Caryophyllidae в.1.). The composition of Caryophyllida le in the number of groups never associated previously as a whole in any traditional classification with Caryophyllales (Fig. 2B). These include Dioncophyllaceae, Ancistrocladaceae, ae sl. is remark: Droseraceae, Nepenthaceae, Tamaricaceae, Franken- laceae, Asteropeiaceae, Podostemaceae, Simmondsi- aceae, and Balanophoraceae. An expanded Malvales complex (Malvales s.l.) is present among the rosids (Fig. 2B), but many other groupings within the rosids found in rbcL trees are not evident in the non-molecular trees. In particu- lar, the smaller groupings of rosids (rosids I, II, Ш) Volume 85, Number 1 1998 Nandi e 1 a НА Analysis of Angiosperms are not evident, nor are the clades composed of glucosinolate-producing (Rodman et al., о nitrogen-fixing families (Soltis et al., 199 e composition of the expanded asterid assem- ~ blage contains many of the same groupings seen in the rbcL trees, in particular the asterid III grouping composed of Ericales, Ebenales, Primulales, and some Theales. In addition, several other taxa are also present here that would not be expected, either on the basis of molecular studies or previous tax- onomies. These include Gunneraceae, Sabiaceae, Santalales, Oncothecaceae, Aextoxicaceae, Bruni- aceae, and Rhabdodendraceae. rbcL TREES The Fitch search of the rbcL matrix produced more сан ps trees of 6057 steps, CI — 0.22, an — 0.43 (the search was discontinued at 5000 trees because the memory was becoming too low; all 5000 were swapped on to completion). Sev- eral characters changed more than 40 times (Fig. 1C). After down weighting by the use of successive weighting only nine trees were found. These nine ad 6091 Fitch steps, CI = 0.22, and RI = 0.42 ке length of 531,123 steps, CI = 0.62, RI ranches not found in all nine weighted trees are marked with arrows (Fig. ЗА). The first tree found at this length is illustrated in Figure 3 (ACCTRAN optimization). These trees are in gen- eral agreement with the results of Chase et al. (1993). The magnoliids form an unsupported mono- phyletic clade (bootstrap of less than 50%) that 1s sister to all eudicots, which are strongly supported (9796). Within the magnoliids, Laurales are sister to the monocotyledons; this is different from either of the two searches presented in Chase et al. (1993), but the present study uses different taxa, and in all cases this grouping is unsupported in the present rbcL tree (bootstrap of less than 5096). The relationship between Annonales, Magnoliales, and Myristicaceae has some bootstrap support (63%). Weak support (5696) is also shown for the associ- ation of the strongly supported pairs Nymphae- aceae/Amborellaceae (9296) and Illiciales/Austro- baileyaceae (98%). Ranunculidae/Eupteleaceae (supported at 86%) are sister to the rest of the eudicots. The lower ha- mamelids form a grade between Ranunculidae/ Eupteleaceae and asterids/caryophyllids/rosids. Among hamamelids, Buxaceae are strongly sup- ported (8246) as sister to Didymelaceae; Sabiaceae and Proteaceae are weakly supported (53%) as a clade. The clade of Gunneraceae/Myrothamnaceae is also supported to a similar degree (6396). Ber- beridopsidaceae/Aextoxicaceae form a pair without internal support, which is the sister group to the rest of the asterids. The only newly added family that falls into the asterid I and П пае is боловао. Within the asterids, Gent are weakly supported (67%), Dipsacales (59%) to Pittosporaceae/Araliales, a pair which has high support (95 analysis of rbcL does not recover exactly the same relationships within the asterid I and II groups as in Chase et al. (1993), but the sampling is much more sparse here. The asterid IV group of Cornales/ as is a relationship of 6). In general, this Hydrangeaceae is also recovered, but is not sup- ported by the bootstrap. The asterid III grouping is weakly supported (52%), and there are additional families comprising this group that we 1993). These include Pellicieraceae, Tetrameris- taceae, and Marcgraviaceae, which are strongly supported (100%) in a clade including Balsami- naceae; the (87%) as sister families. Lecythidaceae and Scy- topetalaceae are also strongly supported (99%) as sister families, but other recent research (Morton et al., 1997) demonstrated that the latter is embedded in the former. Diapensiaceae are weakly supported (65%) as the sister of Sarraceniaceae, and the pairs Ericales/Actinidiaceae and Clethraceae/Cyrillaceae are also weakly supported (58% and 64%, respec- tively). The expanded caryophyllid clade first identified in Albert et al. (1992) and further investigated in Williams et al. (1994) received weak internal sup- port in this analysis (69%). Additional newly iden- tified members of this clade include Tamaricaceae/ re not covered in Chase et al. first two are also strongly supported => Figure 3. Опе of nine токар лв most- pem rbcL trees found with successive We ond Branches not pon in all nine E marked with an These trees have 6091 steps (Fitch len ual weights) with CI — and RI Numbers above de pieds bs are the numbers of estimated иеэ (ACCTRAN 1. а 4. ‘rs below branches are bootstrap values; branches withou t). First-branching portion of the tree, arranged with Ceratophyllaceae as P es of less than 50%. —A (left percentage derlined number had bootstrap outgroup. Magnoliids form a en ш is sister to the eudicots. Within eudicots, ranunculids ry hamamelids form grade in which the asterids are UU панк гага bul rosids, see Fig. 3B). — Note that the glucosinolate and neta dun families form cla B (right). Rosid dle 148 Annals of the Missouri Botanical Garden rosids asterid Ш caryophyllids s. 1. asterids L4! Euc ucom + Hydrangeaceae ЕГ y |састасеае СД asterid I & П РРР asterid IV гы Berberidopsidaceae [24 Ceratophyllaceae Volume 85, Number 1 Nandi e 1 еы dude Analysis of Angiosperms 6 Irvingiaceae лг Ё Malpighiales 3 16 Scyphostegiaceae —— Flacourtiaceae s. str. 43 Erythroxylaceae — "T Lacistemataceae Z Rhaninaceae rosid I i EE. orynocarpaceae N, E Cucurbitalés fixing clade 2 Fagan lastrales s. str. meri PA Plagiopteridaccac Els p ea ygophiyllaceae rameriaceae Dipterocarpaceae Cistaceae Malvales S. Str. Malvales s. l. acea 8 pi aerpsepalaceae | Thyme wa s. 1. 100 78 a 14 100 17 E Ruta ES 66 4 21 Leitneriaceae 100 21 m Vapindales s. = s. 21. s My Js = Myrtales s.l. dI 35 — Саррага!ев 12 31 Salvadoraceae FOS) 9 1781 24 Caricaceae eT сеге Capparales s. 1. 12 5 Bretschneideraceae 100 17 é p || 5 rossosomataceae 1з (621 - gtachyurace eaceae LLL Ж solomatace elianthaceae TE E idales rosid III | L2» Vitace 150 Annals of the Missouri Botanical Garden Frankeniaceae (supported at 8696), Asteropeiaceae, Simmondsiaceae, Rhabdodendraceae, and Dionco- phyllaceae/Ancistrocladaceae (supported at 60%). Polygonaceae/Plumbaginaceae are strongly sup- ported (91%). Within the rosid clade (Fig. 3B), the same three major groups as in Chase et al. (1993) were recov- ered, but only rosid I (5596) has any bootstrap sup- port, and rosid III is a grade (Hamamelidales pair with Vitaceae). The rosid I group includes several newly sequenced families: Caryocaraceae, Clusi- aceae, Corynocarpaceae, Dichapetalaceae, Elaeag- naceae, Flacourtiaceae s. str., Kiggelariaceae (the cyanogenic glycoside-producing genera of Flacour- tiaceae s.l), Lacistemataceae, Medusagynaceae, Salicaceae, Scy- phostegiaceae, and Surianaceae. Within the rosid I clade, Plagiopteridaceae are strongly supported (100%) as the sister lic of Celastrales s. str. (with more ee igs the former are embedded within the ase, unpub- Plagiopteridaceae, Quiinaceae, mnaceae are strongly (83%). The cunonialean clade (61%) comprises Ox- alidaceae, Connaraceae, Eucryphiaceae/Cunoniales (98%), Cephalotaceae, and Tremandraceae/Elaeo- carpaceae (91%); all but the first two listed have moderate support as a clade (81%). ithin the moderately supported Malpighiales clade (76%), Flacourtiaceae s. str. are also strongly supported as a clade (100%; with increased sampling the first two families are embedded within the last; Chase et al., 1996). Chrysobalanaceae/Dichapetalaceae/ e eae have mo erate DAE support (8596), a ceae a weak support (68%). he composition of the rosid II group is more or less like that in Chase et al. (1993), except that it includes Myrtales/Vochysiaceae and leaves out Ge- raniaceae, which appear in rosid III instead. There Salicaceae, Scyphostegiaceae, and e is no internal support for this clade, but it is re- covered in all most-parsimonious trees. Several new families (since Chase et al., 1993) are represented: Bixaceae, Cistaceae, Geissolomataceae, Salvadora- ceae, Sphaerosepalaceae, Staphyleaceae, Stachyu- raceae, and Thymelaeaceae. С n /<% || IQ ~ Geissolomataceae is supported at 60% bootstrap level, and within this clade, a subclade of the last three is strongly supported (95%). The mustard-oil clade has moderate ai (7 as ane within it ыу supported (100%; the last two at 82%) ge ~ lari /Salvad /Capparales has moder- ate bootstraps (78%; the last two at 100%). Vochy- siaceae yrtales is strongly supported at 100%, and this pair has moderate support (76%) as the sister of Sapindales/Rutales/Leitneriaceae (100%) plus Malvales s.l., comprised of Dipterocarpaceae/ Cistaceae (100%), Malvales s. Sphaerosepalaceae, and Thymelaeaceae. str., Bixaceae, s mentioned above, the rosid III group forms a grade and contains in addition to those families identified in Chase et al. (1993), Geraniaceae and Vitaceae. With more sampling, Geraniaceae are placed near Crossosomataceae. Dilleniaceae, Me- lianthaceae, and Santalales are not clearly associ- ated with any other lineage. COMBINED TREES Analysis of the combined matrix with equal weights produced only 40 trees of 10,183 steps, CI = 0.16, RI = 0.39. As with rbcL alone, many char- acters changed excessively and so we employed successive weighting, which produced a single tree (Fig. 4) with the length of 10,271 Fitch steps, CI = 0.16, and RI = 0.38 (weighted length 631,329, CI = 0.56, RI = 0.63). In general, this topology is like that for rbcL, but there are a number of differ- ences. The differences of the combined tree from the non-molecular trees are more substantial, as are the differences between the rbcL and the non-mo- lecular trees. The major differences of the com- bined from the rbcL trees are as follows: the mag- noliids form a grade rather than a clade; the ranunculids are sister to one of the clades, Platan- aceae/Nelumbonaceae, that make up the hamame- lid grade; the caryophyllids are sister to a combined rosid/asterid clade, in which these are monophy- letic sister groups; Malvales s. str. are the sister to Bixaceae, Cistaceae, and Dipterocarpaceae in the combined tree, whereas they are placed between Bixaceae and Cistaceae/Dipterocarpaceae in the rbcL trees; the Bixales group with the taxa having a bixoid chalaza in their seeds (Nandi, 1998a; 159) thus appears as monophyletic in our combined tree (Sarcolaenaceae, for which no rbcL sequence was available, do have a bixoid chalaza; this character was found too late to be included in the matrix; if this character-state could be coded, Sarcolaenaceae would probably appear in the Bixales group as well); Fabaceae are placed outside the nitrogen-fix- ing clade in the combined tree; Clusiaceae are placed in a clade with Caryocaraceae, Elatinaceae (only non-molecular data), and Bonnetiaceae (only non-molecular data) in the combined tree, whereas they appear as the sister group of Euphorbiales in Volume 85, Number 1 1998 Nandi et al. 151 Al ado Cladistic Analysis of Angiosperms the rbcL trees; Violaceae are found in a group that is sister to a large clade containing, e.g., Flacour- tiaceae s. str. and Euphorbiales, whereas they are placed differently in the rbcL trees; Kiggelariaceae are sister to Flacourtiaceae s. str., Scyphostegi- aceae, and Salicaceae in the combined tree, where- as they are more distant from Flacourtiaceae s. str. in the rbeL tre The addition of the non-molecular data to the rbcL matrix greatly reduced the number of trees obtained; in the case of the Fitch analysis it dropped from more than 5000 (at which point the memory was exhausted) to only 40, and in the weighted analysis from nine to only one. If there were agreement between patterns in the molecular and non-molecular data, then an effect of increased support might be observed in the combined anal- ysis. This is partly the case, but the amount of miss- ing data in the non-molecular analysis makes this assessment difficult; there are many exceptions no- ticed by comparing Figures 3 and 4. For example, support for an expanded Nymphaeales (including Amborellaceae, Austrobaileyaceae, and llliciales) decreases slightly (from 5696 to 5396), but the sup- port for the two pairs, Nymphaeaceae/Amborella- ceae and ОА Коне забез, decreases markedly, from 9296 and 9896 to 86% and 7596, respectively. Citing all such cases is not a worth- while endeavor at this stage of investigation. At the least, it can be stated that the addition of the non- molecular data does not drastically alter the pattern obtained with rbcL alone, nor does it greatly de- crease bootstrap support. TAXA IN THE rbcL TREES FOR WHICH INSUFFICIENT NON-MOLECULAR DATA ARE AVAILABLE А number of small families have been included in the non-molecular matrix, but little information is available for them. The presence of such taxa can destabilize results and produce lower levels of internal support. Most of these taxa received strong support for placement in the rbcL trees, and they seem relatively securely placed in the combined tree. We point out these taxa here to draw attention to them: Tetrameristaceae and Pellicieraceae are found near Marcgraviaceae in the rbcL and com- bined trees (Figs. 3A, 4A) and are also sister groups tending to be placed in asterids in the non- molecular trees (Fig. 2A); Corynocarpaceae falls in the clade formed by Coriariaceae and Cucurbitales (including Datiscaceae and Begoniaceae) in the rbcL and combined trees; Leitneriaceae have a sta- ble position near Rutales/Sapindales; Huaceae are placed near Celastrales s. str.; and Lacistemataceae and Scyphostegiaceae are found near Flacourti- aceae s. str. DISCUSSION Certain caveats must be proffered before further consideration of the results of these analyses. To overcome the disadvantages of high taxon number and large amounts of missing data (which always slows the process of finding shorter trees; pers. obs.), we regrouped the taxa into three subgroups ПРЕ III). These corresponded to what we pre- med were magnoliids, lower eudicots, and cary- sint (group I), rosids (group II), and asterids up Ш). This led to the advantage of shorter and more ы computation and ultimately yielded E trees up to six steps shorter than those found with the other two search strategies. This method is somewhat biased in presuming three major groups, but the final unconstrained swapping and reweight- ng procedures should a for the biases Mbs introduced. With an RI of 0.41, this matrix is highly likely to be subject to у ој (Maddison, 1), and this appeared to cause problems for standard types of search strategies. This is likely true also for the rbcL and combined matrices, al- though these were clearly more consistent in find- ing reasonably similar tree lengths in each of the random replicates of taxon-entry order. The methods used for coding of the non-molec- ular data can be improved. Assessments of char- acter polarity before analysis are assumption-laden. Coding only a single character that is assumed to be the plesiomorphic state for cases in which poly- morphisms occur could result in spurious place- Figure 4. (Fitch length; i.e., estimated changes equal weights) with CI ЕЕЗ The single most-parsimonious combined tree found with successive weighting. The tree has 10,271 steps = 0.16 and RI = 0.38. Numbers above the branches are the numbers of (ACCTRAN optimization). Underlined numbers below bran ches are bootstrap values; branches with- out an underlined number had bootstrap percentages of less than 50%. —A (left). First-branching portion of the tree, arranged with Ceratophyllaceae as E outgroup. Magnoliids form a grade composed of tw a the eudicots. Within eudicots, ranunculids and hamamelids form a grade. The family outside the main 2. clade (Fabaceae). r to = неби у (for rosids, see Fig. amilies form clades. *Taxa for wo major subclades (magnolid 4B). —B (right). Rosid clade. Note that the glucosi- which rbcL sequences were unavailable. +Nitrogen-fixing 152 Annals of the Missouri Botanical Garden Scytopetalaceae ЖЕ T e ythidaceae asterid III asterids - Gentianales 12 Scrophulariales м asterid IV caryophyllids s. 1. 4 lower hamamelids ranunculids 55 N | N magnoliids Chloranthaceae С ~ eratophyllaceae Volume 85, Number 1 Nandi et al. 998 Combined Cladistic Analysis of Angiosperms 23 г Passiflorales 10 Linales - Euphorbiale aes Es Chr гузо soba РВЕ Dic арш ар сеае 50 Два ЛЕРДЕ Aries Kiggela elar ceae L Апасов Malpighiales rosid I alot eae Stasbur епасеае pected e$ Copiado i tie =- Cunonisles 6.1. Cochlospermaceae§ ; Malvales 5. str. Malvales s. 1. Sphaerosepalaceae Rutales, ; 3 1 Leitneriaceae Sapindales s. 1. 2L Sapindales 4] "rri Vochysiaceae Myrtales s. 1 Capparales Salvadoraceae мо L 9 т Crassula А A | Saxifragales S. str. rosid III niaceae 16 г2 Dille 2° Vitaceae Hamamelidales 154 Annals of the Missouri Botanical Garden ments of some taxa. Ап example of an assumed apomorphy can be seen in the results for Lilianae in Chase et al. (1995), in which the combined rbcL and morphology trees indicated that an inferior ova- ry was the plesiomorphic condition. This is the op- posite conclusion one would reach based upon gen- eralized character trends in angiosperms, and such conclusions could result in spurious assessments of relationships. Moreover, coding terminals as poly- morphic can also produce erroneous topologies (Nixon & Davis, 1991). Adding terminals would be a solution, but it would involve unmanageable ma- trix dimensions and the need for more specific data on variation within larger clades. For example, if a taxon B is deeply nested within a large taxon A, it would be difficult to detect this relationship with our data. Taxon B would most likely attach to a subgroup of taxon A (which in our matrix may be absent). This would mean that large taxa have to be split up. An example of this problem is our use of Rutales, Sapindales, and Leitneriaceae. With more sampling within the two orders (Gadek et al., 1996), Leitneriaceae is embedded within Simarou- baceae of Sapindales. In our trees, it appears as sister to Sapindales. Despite these caveats, the ap- proach used here is made stronger by the inclusion of many more characters than taxa. As long as most characters are accurately scored, the general re- sults should contain useful and new information, and the “phylogenetic signal” should not be overly distorted. The non-molecular matrix often deals with large taxonomic units composed of many families. The results are thus meaningful only if these taxa are monophyletic. We used higher-order taxa when the results of Chase et al. (1993) and many published studies (Appendix 6) coincided with the traditonal circumscription of these groups. In the rbcL and combined analyses, these groups were represented by only a single sequence of a representative spe- cies. The effects of using exemplars is discussed in Sytsma and Baum (1996), but the results do not differ significantly from other rbcL studies using more than single representatives. A different approach would have been to use species as terminals, preferably the same species as covered by the rbcL database. This approach, however, would have the disadvantage that not all character fields would be investigated for the spe- cies or even the genus in question. Moreover, it seems most likely that the coverage of all angio- sperms with exemplar species would require a sam- ple of more terminal taxa than in this study. This again would necessitate more phylogenetically in- formative characters. This species-terminal ap- proach, at the present time, is impractical and could not be effected; there simply are not enough species studied for all these characters. We intend this study to be an example of the direction that we think phylogenetic studies should be taking. We will be most gratifie searchers take our matrices and improve upon if other re- them. The literature is voluminous. We have surely missed a number of papers, but these matrices are now there to be completed. The gaps will be ob- vious to those who are interested. The missing cells need to be filled in, and we can see that if they are, there is hope for improvement. For those taxa on which we have focused most and incorporated more of the relevant literature (e.g., Malvales s.l.), the non-molecular (Fig. 2B), the molecular (Fig. 3B), and combined trees (Fig. 4B) are all highly congruent. The non-molecular results for Cochlo- spermaceae, Bixaceae, Cistaceae, Dipterocarpa- ceae, and Sarcolaenaceae also demonstrate that the search for characters has to consider a wide array of subject areas. The final caveat concerns the use of successive weighting to “improve” the matrices (Farris, 1969). Some readers will wonder how this procedure has "distorted" the results produced by equal weight- ing, the results of whic data sets contain characters that are excessively "noisy," and these can be detected by an exami- nation of their consistency on any of the trees (Fig. 1). This is evidence that, although these characters may be useful at some hierarchical level, they are not useful at the broad scale being studied here. A priori one cannot and should not make this sort of decision; it is simply too assumption-laden. When the initial results from an analysis indicate that cer- tain characters are relatively more inconsistent than others, then the effects of the former should be less- ened and those of the latter enhanced (i.e., made more consistent). The effect of successive weighting we have not shown. All is never vastly different from that of equal weight- ing; in the great majority of cases, successive weighting merely identifies а subset of the trees found with equal weights as optimal (i.e., those that favor the more consistent characters). This is not the case with any of the trees found here, but both the rbcL and combined results have nearly the same Fitch length as the most-parsimonious Fitch tree (the weighted trees are only 0.5696 and 0.8696 longer than the Fitch trees for rbcL alone and the combined matrices, respectively; the CI and RI for the Fitch and weighted trees are nearly identical and only differ at the third decimal point levels, CI = 0.217 versus 0.215 and RI = 0.428 versus 0.424 for rbcL, and CI = 0.160 versus 0.159 and RI = Volume 85, Number 1 1998 Nandi et al. 155 Combined Cladistic Analysis of Angiosperms 0.390 versus 0.384 for the combined analysis). We attempted to use successive weighting on the non- molecular data, but, like the search protocol itself, this procedure would have occupied many months of computer time and was therefore abandoned. (A) TREES AND GENERAL PATTERNS No previous cladistic analysis of the angiosperms has used as many higher-level taxa as this, includ- ing Chase et al. (1993) and Soltis et al. (1997b), which both used more species but fewer families. Of course, many of the families are subsumed in these trees by higher-order taxa (1.е., monocotyle- dons, Faganae, Urticales). This process of selecting terminals did not have a great effect on topology for the rbcL-only analysis, which deviates only slightly from that seen in Chase et al. (1993), and our results also do not differ drastically from those produced by 185 rDNA either (Soltis et al., 1997b). Several more divergent families are differently placed, which could be due to the overall sparser sampling permitting branch attractions to occur. These families (relative to Chase et al., 1993) are: Geraniaceae, in rosid II near Crossosoma before, here in rosid III (Fig. 3B); Vitaceae, in an isolated position with Dilleniaceae before, here with Ha- mamelidales; Krameriaceae/Zygophyllaceae, near Rosaceae in rosid I before, here in an isolated po- sition as sister to the rosid I clade; and Fabaceae, which in the rbcL trees falls into the nitrogen-fixing clade (Fig. 3B) but in the combined tree is sister to Zygophyllaceae/Krameriaceae. everal taxa occupy isolated positions in the rbcL and combined trees, and these would appear to be critical for understanding the patterns observed in the largest clades (i.e., rosids, asterids, and cary- ophyllids). These include Aextoxicaceae, Berberi- dopsidaceae, Dilleniaceae, Gunneraceae, Myro- thamnaceae, Vitaceae, and Santalales. These taxa have shifted positions in every published large rbcL analysis, but they always come out as the sister taxa of the largest clades of eudicots. Within the aster- ids, Aquifoliaceae, Eucommiales, and Icacinaceae perform similarly; among rosids the Celastrales s. str./Plagiopteridaceae, Huaceae, Krameriaceae/Zy- gophyllaceae, Melianthaceae, and Crossosomataceae/ Qi, 1 IQ || 1 ji X || 4 / I clades are likewise unstable. Their positions in the non-molecular trees are generally different from their positions in the rbcL and combined trees. Aextoxicaceae, Berberidopsidaceae, Dilleniaceae, Gunneraceae, Myrothamnaceae, Vitaceae, and San- talales, those taxa that fall as sister groups of the asterids, caryophyllids, and rosids in Figures 3 and 4, are embedded among the magnoliids or included in the rosid groups that fall apart from the main rosid clade in the non-molecular trees (Fig. 2A). These taxa have a large number of plesiomorphic traits. For example, Berberidopsis (Berberidopsida- ceae) has an undifferentiated perianth, plesio- morphic wood (presence of mostly solitary vessels with scalariform perforation plates and opposite side-wall pitting, absence of septate fibers), and tri- colpate pollen (Miller, 1975; Lemke, 1988). The rbcL data contain significantly greater phy- logenetic information than the non-molecular data in this broad study (e.g., they delimit more groups with greater levels of internal support). In part, this must be ascribed to the structure of the non-mo- lecular matrix, containing many empty cells and also a larger number of polymorphisms. Moreover, it has become obvious that all larger clades of an- giosperms can only be characterized by few non- molecular traits (see part b of Discussion). This re- sults in a matrix that seems to yield only slightly longer trees using standard methods (i.e., no com- partmentalization). In phylogenetics, it has been underestimated that the more “signal” (i.e., the less randomness) is contained in a data matrix, the eas- ier it is to find optimal trees. The 1993 rbcL tree was obtained in a relatively short search (Chase et al., 1993); trees only five steps shorter were found by Rice et al. (1995) after many more months of search on more than one computer. The only dif- ferences between these minimally shorter trees and the trees found in 1993 concern groupings that are weakly supported regardless of their positions. Nothing more of significance has been obtained ex- cept a huge outlay of computing time and personal effort; the 1993 tree contained all of the strongly supported groupings, and represents well the phy- logenetic signal present in rbcL data. It should be accepted that with large searches for which exact solutions are impossible (such as this and the other large angiosperm matrices) excessive swapping over several months is not reasonable; effort is bet- ter spent in finding additional data. When all groups are strongly supported, then finding the op- timal solution will be easy and the trees accurate (Soltis et al., in press). Even after many additional months of search on the 1993 rbcL matrix, we can- not say that anything new was learned. The most that was achieved was the observation that many groups, especially those with long branches, were unstable. Of course, we performed bootstrap anal- yses here, and this makes the general weakness of the rbcL tree evident. Unpublished analyses of atpB for nearly 300 seed plants take even less time than rbcL and contain even more groups with strong sup- 156 Annals of the Missouri Botanical Garden port. Soltis et al. (1997b) presented 185 rDNA data for 232 seed plants. The authors reported that more time is required for 185 than for rbcL alone or rbcL- 18S combined searches, and again the major prob- lem with large data sets is not just their size, but also the degree of randomness and missing cells that they contain. The large number of question marks and lack of support in the non-molecular matrix are serious obstacles to rapid search. Like- wise, they do not permit the use of the jackknife (J. Farris, pers. comm.), which is a fast and accu- rate method of finding groups with strong internal support, regardless of the size of the matrix (Farris et al., 1997). Some authors have suggested that hybridization or other forms of horizontal gene transfer cou have a major effect on higher level studies within the angiosperms and could be expected to create conflicts between data categories (Syvanen et al., 1989; Syvanen, 1994). Others did not give hybrid- ization a major role at higher levels (Chase et al., 1993). We do not deny that high levels of parallel- isms exist among angiosperms, but we find the ex- planation of widespread horizontal gene transfer as the cause (Syvanen, 1994) unappealing and not conducive to further investigation. Studies of nu- clear 185 rDNA (Soltis et al., 1997b) and plastid atpB (Savolainen et al., 1996) find results highly congruent with those of rbcL. In particular, the con- gruent topologies found with plastid genome se- atpB) as well as with nuclear genome sequences (18s rDNA) argue against hy- quences (rbcL an bridization being a major problem in higher level plant systematics. Reticulate evolution, dating to a time when hybridization was still possible between now distant lines, appears to have only minor ef- fects on macrosystematic patterns (for discussion of effects leading to parallelisms, see also Kubitzki et al., 1991 (B) NON-MOLECULAR CHARACTERS OF TAXON GROUPS DISCUSSED ON THE BASIS OF THE COMBINED DATA TREE (FIG. 4 A, B) We argue that the trees with the greatest under- lying data are the most appropriate to discuss; thus, unless specifically stated, we will discuss only the combined tree from Figure 4. We focus on a series of characters that appear to contribute to the to- pology obtained in the combined tree. This is not meant to be an exhaustive examination of these top- . We intend instead to illustrate some of the ub in the non-molecular data that agree with ће distribution of variation in the rbcL matrix. Characters described are synapomorphies as yield- ed by MacClade 3.04 on the combined tree, unless stated otherwise. Other characters that are widely represented within a clade may represent synapo- morphies if the topologies are only slightly rear- ranged; since many of these branches are weakly supported, discussing these characters as either synapomorphies or plesiomorphies seems prema- ture and potentially misleading. Therefore we dis- cuss many characters as simply being widespread or frequent within clades; many of these will even- tually be demonstrated to be synapomorphies. Due to the large number of missing cells and low levels of internal support with present data, it seems most prudent to consider only their relative frequencies or tendencies of occurrence rather than to frame this discussion as an investigation of synapomor- phies. Magnoliidae. The strict dichotomy of the leaf parts in Ceratophyllum is unusual in angiosperms, even if compared with other water plants showing the Hippuris syndrome of leaf architecture (cf. also Cook, 1978; Rutishauser & Sattler, 1987). The in- florescence is a spike with the flowers frequently arranged in two orthostichies (Raynal-Roques, 1981). This inflorescence type shows some similar- ities to the decussate spikes in Chloranthus and could reflect an old pattern. Also the flowers in Ceratophyllum are unisexual (Endress, 1994b), and this could be plesiomorphic for angiosperms or apo- morphic as a result of adaptation to an aquatic aut- ecology. Chloranthaceae occupy an isolated and perhaps early-diverging position (see also Nixon et al., 1994). This is concordant with the fact that the oldest fossils known at present that are clearly at- tributable to angiosperms are Chloranthaceae-like. Chloranthoid pollen was described from the Valan- ginian of Israel (Brenner, 1996). Hedyosmum-like flowers are known from the Valanginian or Hauter- ivian of Portugal (Friis et al., 1994; Crane et al., 1995; E. M. Friis, pers. comm.) and are thus even older than the Ceratophyllum-like horned fruits found from an Aptian locality (Dilcher, 1989). The fact that distinct Aptian fossil material has been found that appears to combine characters of Chlo- ranthaceae, Piperales, and Circaeasteraceae (Ran- 1995) indicates that early angiosperms exhibited a suite of traits that are now unculidae; Crane et al., only known to occur individually within distinct terminal clades of extant angiosperms. The decus- sate arrangement of the flowers in spicate inflores- cences in Chloranthus and the Late Cretaceous Chloranthistemon (Endress, 1987; Eklund et al., 1997) is paralleled by the decussate inflorescences Volume 85, Number 1 1998 Nandi et al. 157 Combined Cladistic Analysis of Angiosperms of Ephedra (Hufford, 1996). The comparison of branched male structures in Gnetales and Chloran- thaceae is problematical because of unclear ho- mologies (Endress, 1987; Friis & Endress, 1990; Doyle, 1994, 1996). For comparison of Chlorantha- ceae with Gnetales see also Taylor and Hickey (1996) and critical discussion by Doyle (1996) and Endress and Igersheim (1997). Also the highest di- versity of pollen aperture types within an angio- sperm family seems to occur in Chloranthaceae (not expressed in the characters used for this analysis; see, e.g., Erdtman, 1952). Sesquiterpenes, as y-ele- mene, can serve to indicate relationships of Chlo- ranthaceae to other angiosperm families. At pres- ent, y-elemene is known only from Chloranthaceae, Piperaceae, and Aristolochiaceae (Hegnauer, 1962-1994). The germacrene acoragermacrene ос- curs only in Chloranthaceae and monocots (Heg- nauer, 1962-1994). These two compounds seem to indicate an evolutionary relationship of Chlorantha- ceae to Aristolochianae—monocots or are a relict of previously more widespread traits. Amborella also occurs in an isolated position in our non-molecular trees (Fig. 2A). Amborella was found as the sister group to the rest of angiosperms in a subset of the 18S rDNA trees (Soltis et al., 1997b), but in a clade supported by the jackknife along with Illiciales, Austrobaileyaceae, and Nym- phaeales in the combined analysis of rbcL and 18S in Soltis et al. (19972). Probably ancestral or erratic characters of Amborella include the presence of S- type plastids in the sieve-tubes (107), uniaperturate in addition to inaperturate pollen grains (129; Sampson, 1993), minute embryos (163), scanty wood parenchyma (174), no fibers (not coded), tra- cheids (177), wood rays of Kribs heterogeneous type I (179), circular tracheid side-wall pitting (similar to some Gnetales; 184), no vessels (not coded; probably plesiomorphic), no discontinuous calyx-corolla transgression (210), practically or- thotropous ovules (246), and stipitate fruits (239) (Metcalfe & Chalk, 1950; Behnke, 1981; Cron- quist, 1981; Takahashi, 1985; Carlquist, 1988a; Endress, 1994c). Brenner (1990, 1996) reported that angiospermous, inaperturate pollen which may have evolved into a Clavatipollenites pollen-type, are present in the Valanginian and ains, Hauterivian of Israel. Judging from these paleobo- tanical finds, one may take into consideration whether the inaperturate pollen grains found in Ceratophyllum, Ascarina (Todzia, 1993), Amborella, Trimenia papuana (see Sampson & Endress, 1984), and many Laurales (Gomortegaceae, Hernandi- aceae, Lauraceae, Monimiaceae except Atherosper- matoideae) are reductions or represent an old, con- served, character-state. Neglect of the presence of inaperturate pollen in the above-mentioned mag- noliid taxa based on the assumption that the ina- perturate condition does not represent the basal pollen type could result in different topologies at the base of the tree. All taxa of magnoliids and early-branching eu- dicots included in this analysis have ovary-to-car- pel length ratios greater than 1:2 (i.e., with short or absent styles; 236). The formation of long styles in relation to the whole carpel thus seems to be an apomorphic tendency in basal angiosperms. А me- sotesta (middle layer of outer integument in the seed; 154) with sclerified cells is present in many magnoliids: Chloranthaceae (Chloranthus spp.), Nymphaeaceae, Austrobaileyaceae, Illiciales, Ar- istolochianae (Aristolochia spp.), Myristicaceae (Horsfieldia, Myristica), Annonales, Magnoliales (Corner, 1976; Endress, 1980; Takhtajan, 1988). A mechanical layer in the mesotesta is also found in some early-branching eudicots (Eupteleaceae; Bux- aceae: Sarcococca; and Hamamelidaceae; Corner, The clade formed by Laurales, Aristolochianae, monocots, Myristicaceae, Annonales, and Magno- liales (magnoliid I clade; Fig. 4A) shows a frequent occurrence of the phenylpropane asarone (41). Asa- rone is known from Lauraceae (Sassafras), Pipera- ceae (Piper), Aristolochiaceae (Asarum), Annona- ceae, and Magnoliaceae (Magnolia) (Gildemeister & Hoffmann, 1956; Hegnauer, 1962-1994; Sethi et al., 1976; Keller, 1982). Outside of this clade Heg- nauer (1962-1994) cited only three families of an- giosperms that produce asarone. The same clade contains the only plant taxa that Hegnauer (1962- 1994) and Harborne and Baxter (1993) found to produce the lignans galbacin (57) and veraguensin (59). The neolignan licarin (58), though described from Krameria (Dominguez et al., 1992), is also predominantly found in this magnoliid I clade (Gottlieb et al., 1988, stated that neolignans have their center of diversification in the magnolialean families). Galbacin, a tetrahydrofuranoid lignanoid, occurs in Lauraceae (Persea), Aristolochiaceae (Ar- istolochia), Myristicaceae (Knema, Virola), and Hi- mantandraceae (Galbulimima) (Hegnauer, 1962- 1994; Harborne & Baxter, 1993). Veraguensin, also a tetrahydrofuranoid lignanoid, is known to occur in Trimeniaceae (Trimenia), Lauraceae (Ocotea), Saururaceae (Saururus), Myristicaceae (Virola), and Magnoliaceae (Magnolia) (Harborne & Baxter, 1993). Licarin has been found in Lauraceae (Li- caria), Aristolochiaceae (Aristolochia), Myristica- ceae (Myristica), апа Magnoliaceae (Hegnauer, 158 Annals of the Missouri Botanical Garden 1962-1994; Ionescu et al., 1977; Le Quesne et al., 1980; Harborne & Baxter, 1993). The alkaloid liriodenine (83) is known only in the magnoliid I clade, as well as in Ranunculidae and Nelumbonaceae; the last-mentioned taxa fall into the sister group of the remaining eudicots. As with two of the three lignanoids mentioned above, liriodenine is not known from any families outside of these clades, most significantly not from the mag- noliid subclade containing Winterales, Nymphae- aceae, Amborellaceae, Austrobaileyaceae, and Il- liciales (Hegnauer, 1962- ; Harborne & Baxter, 1993), hereafter the magnoliid II clade (Fig. 4A). The magnoliid I clade further has sieve-tube plastids of the P-type (107) in a majority of families (Behnke, 1981), whereas all members of the mag- noliid II clade except Canellaceae (here in Winter- ales) have S-type plastids. monocots are further linked by the common pres- ence of crystal sand (in Piperaceae, Metcalfe & Chalk, 1989, and Araceae, but not in Acorus, So- lereder & Meyer, 1928; Franceschi & Horner, 1980; Seubert, 1993; 115), of a dispersed vascular system (in Piperaceae and monocots, but not in Saururaceae; 167), and of frequent trimery in peri- anth (212-214), androecium (221), and gynoecium (233-235). Aristolochianae and monocots also cluster on the basis of the widespread occurrence of two stamen whorls (not coded). More similarities, perhaps as the result of common ancestry, are enu- merated by Burger (1977) and Dahlgren and Clif- ford (1982). All magnoliid I families except mono- cotyledons/Aristolochianae stratified phloem (169) and wedge-shaped phloem rays (Met- calfe & Chalk, 1950; Cronquist, 1981; Carlquist, 1988a; 170 Both the non-molecular and the combined trees show Chloranthaceae as an isolated family apart from the main magnoliid clades. Also equally iso- Aristolochianae and share a lated in all trees are Amborellaceae, Austrobai- leyaceae, and Illiciales (magnoliid II clade), sepa- rated from the more typical magnoliid I clade, in which the monocots are sister to Piperales/Lacto- ridaceae/Aristolochianae (Fig. Eudicots. Eudicots are held together by their triaperturate pollen grains (129), which most likely evolved in parallel in Illiciales (Erdtman, 1952; Doyle et al., 1990; Qiu et al., 1993). Many early-branching eudicots have represen- tatives with tricolpate pollen grains; these are cited here, as in Chase et al. (1993), as the ranunculids and lower hamamelids (the latter a grade composed of several small clades). These taxa are nearly all relatively small and could be considered remnants of previously more widespread and numerous ar- chaic lineages. In our scheme, these lineages would include Berberidopsidaceae, Nelumbonaceae, Pla- tanaceae, Ranunculidae, Proteaceae, Gunneraceae, Myrothamnaceae, and Trochodendrales. Vitaceae and Aextoxicaceae appear to be related also to these, but exhibit some more advanced characters, such as tricolporate pollen, which is more predom- inant in derived eudicot lineages (see, e.g., Erdt- man, 1952). In Nelumbo, both tricolpate and mono- sulcate pollen are reported (Kuprianova, 1979; Blackmore et al., 5; coded only as tricolpate in the matrix because we became aware of the occur- rence of monosulcate pollen in Nelumbo only after analysis). The sister group of the rest of eudicots consists of Nelumbonaceae, Platanaceae, Euptele- aceae, and Ranunculidae. A number of these fam- ilies have some members with palmately veined leaves or leaves with no dominant single primary vein (i.e., caeasteraceae, Menispermaceae, Lardizabalaceae, Cir- Ranunculaceae, Berberidaceae, Nelumbonaceae, and Platanaceae; 198). The leaves of Kingdonia and Circaeaster are particularly inter- esting for their dichotomously branching venation, which is rare in angiosperms (for the conditions in Kingdonia, see Foster & Arnott, 1960; morphoge- netic interpretations by Hagemann, 1970, and Ha- gemann & Gleissberg, 1996). Foster and Arnott (1960) hypothesized that the dichotomous venation pattern in Kingdonia represents an ancestral char- acter-state. Imprint leaf fossils from the Early Cre- taceous of Madagascar have been found that show characters similar to extant Circaeaster (O. Appert, pers. comm.). Circaeasteraceae and Kingdoniaceae were placed in the ranunculalean clade by Oxel- man and Lidén (1995; here including Trochoden- dron) based on an analysis of 288 rRNA. They were also given family rank (as members of a distinct order in the Ranunculidae) by Takhtajan (1997). Proteaceae and Sabiaceae are linked by the com- mon presence of wedge-shaped phloem rays (Met- calfe & Chalk, 1950; 170) and of a nectary disk (Haber, 1959, 1961, 1966; van Beusekom, 1971; 231), a rare character in the early-branching an- giosperms. Buxaceae and Didymelaceae share а simple, bract-like perianth (possibly plesiomorphic; 209) and encyclocytic stomata (195), the latter a rare character-state present in only eleven taxa of our analysis (see also Metcalfe & Chalk, 1950, 1988, 1989). Aextoxicaceae and Berberidopsida- ceae are also linked by this same character (Cron- quist, 1981; Baas, 1984). Ellagic acid is not only absent from the magno- lids, with the exception of Nymphaeales (Ambor- ellaceae have not been sampled), but also from the Volume 85, Number 1 1998 Nandi et al. 159 Combined Cladistic Analysis of Angiosperms first-branching eudicots, Ranunculidae, Euptele- aceae, Platanaceae, Nelumbonaceae, Proteaceae, Sabiaceae, Buxaceae, and Trochodendrales. Gallic acid (70) shows a similar distribution (Hegnauer, 1962-1994; Gibbs, 1974). The morphological data set does not establish the sister-group position of Ranunculidae to the re- maining eudicots but places them nested in the magnoliid I clade (Fig. 2A). Perhaps a better knowledge of the biochemistry of some basal eu- dicots (Eupteleaceae, Platanaceae, Sabiaceae, Di- dymelaceae) would cause a somewhat modified placement of Ranunculidae. The non-molecular trees also do not consistently separate Nelumbo from magnoliid I (Fig. 2A). In all trees Trochoden- dron, Tetracentron, Proteaceae, Sabiaceae, Buxa- ceae, and Didymelaceae are in an isolated position (cf. also Drinnan et al., 1994). Berberidopsis, a di- typic Australian- Chilean disjunct genus, shows no close relationships to Flacourtiaceae in either data set. А distinct position of Berberidopsis within the core-Flacourtiaceae s.l. was already indicated by Keating (1975) on the basis of pollen morphology and by Miller (1975) on the basis of wood anatomy. The rbcL and combined analyses place Berberidop- sidaceae and Aextoxicaceae (also from Chile) as sister to the asterids (Figs. 3A, 4A), and the non- molecular analysis places them with the magnoliid I clade (Fig. 2A). Dilleniaceae and Vitaceae have never been con- sidered closely related, but they share oxalate raph- ides (Metcalfe & Chalk, 1950; 113), an endotesta containing radially elongate cells (156), and a tra- cheidal exotegmen (Corner, 1976; 157). Caryophyllidae s.l. Albert et al. (1992) found an unexpected grouping of Droseraceae and Ne- penthaceae with Caryophyllales; the latter have been considered to have no particularly close rel- atives, other than perhaps Plumbaginaceae and 1981). This clade ap- pears in all trees, even non-molecular, with a re- Polygonaceae (Cronquist, markably similar composition (Figs. 2B, 3A, 4A). Most taxa of the clade formed by Rhabdodendra- ceae (1), Caryophyllales (2), Tamaricaceae (3), Frankeniaceae (4), Asteropeiaceae (5), Nepentha- ceae (6), Droseraceae (7), Dioncophyllaceae (8), Ancistrocladaceae (9), Simmondsiaceae (10), Plum- baginaceae (11), and Polygonaceae (12), here termed as caryophyllids, have some taxa with tri- colpate or polycolpate (stephanocolpate) pollen grains (2, 3, 4, 5, 7, 8, 9, 10, 11, 12; Erdtman, 1952; Cronquist, 1981; 129). Many (2, 5, 6, 7, 8, 9. 10, 11, 12) also have spinuliferous or punctite- gillate pollen sexine (Erdtman, 1952; 135). The similarity of pollen grains of some Polygonaceae and some Caryophyllales was noted by Erdtman (1952). Likewise the resemblance of pollen of Dro- seraceae and Nepenthaceae is noteworthy (e.g., Erdtman, 1952; Basak & Subramanyam, 1966; Takahashi & Sohma, 1982). Anomalous secondary growth seems to be particularly well represented in the caryophyllids, occurring in Rhabdodendraceae, Caryophyllales, Frankeniaceae, Dioncophyllaceae, Simmondsiaceae, and Plumbaginaceae (Carlquist, 1988a). Similarly, interxylary phloem occurs in sev- eral taxa: Rhabdodendraceae (Record, 1933), Car- yophyllales, Simmondsiaceae (Bailey, 1980), Plum- baginaceae, and Polygonaceae. A character-state that was coded as present in only seven taxa out- side the extended caryophyllids is the presence of maximally biseriate wood rays, displayed in Fran- keniaceae, Asteropeiaceae, Dioncophyllaceae, An- cistrocladaceae, Droseraceae, and Simmondsiaceae (Metcalfe € Chalk, 1950; Carlquist, 1988a; Carl- quist & Wilson, 1995). The caryophyllids, except for Rhabdodendra- ceae, are further characterized by the presence of only alternate intervessel pitting (secondary xylem present in 2, 3, , 7, 8, 9, 10, 11, 12; 184). The exclusive occurrence of the alkaloid ancistrocla- dine in Amaranthaceae (Arora & Metha, 1981; 85), Dioncophyllaceae, nauer, 1962-1994) also suggests a degree of relat- edness. All caryophyllid families for which infor- mation was available (3, 4, 6, 7, 9, 11, 12) have an endosperm provided with starch grains (161); only and Ancistrocladaceae (Heg- ten other taxa in the matrix share this condition. Tamaricaceae were previously put into the “Nelk- engruppe,” roughly corresponding to modern con- cepts of Caryophyllales, by Hallier (e.g., 1914). A tendency linking Tamaricaceae and Frankeniaceae is the presence of exotestal cells with convex sur- faces, being represented as papillae in Frankeni- aceae or as hairs in Tamaricaceae. Netolitzky (1926) mentioned that the chalazal hair tuft in Ta- maricineae is first developed as papillae. Corner (1976) also postulated a link of Frankeniaceae to Tamaricaceae through exotestal morphology. More- over, Tamaricaceae as well as Frankeniaceae have appendages on the ventral side of their petals. Airy Shaw (1951) suggested a close affinity of Drosera- ceae, Nepenthaceae, Ancistrocladaceae, and Dion- cophyllaceae. Schmid (1964) added new evidence for this grouping. The latter alignment, containing three carnivorous families (Droseraceae, Nepentha- ceae, and Dioncophyllaceae) with different trapping systems, has been supported by Hegnauer (1962- 1994) on biochemical grounds. The presence of the naphthoquinones plumba- 160 Annals of the Missouri Botanical Garden gine (99), droserone (100), and related 1,4-naph- thoquinones is another link between Nepenthaceae, Droseraceae, Ancistroclad , Di phyll and Plumbaginaceae (Hegnauer, 1962-1994; Zenk et al., 1969; Durand & Zenk, 1974; Lavault & Bru- neton, 1980; Williams et al., 1994); these com- pounds are otherwise known to be accumulated only by several species of Iridaceae and Ebenaceae (Hegnauer, 1962-1994). It is mainly the coincidence of these trends in chemistry and pollen morphology that places the families mentioned above into the expanded cary- ophyllid clade in the non-molecular tree. Thus, it is the coding of the presence of variably exhibited specialized traits that is responsible for the pres- ence of the caryophyllid clade in nearly the same composition as in the rbcL trees. Rhabdodendra- ceae, which fall into the asterids in the non-molec- ular trees (Fig. 2A), presumably do so because they have unitegmic ovules (see below); with the weak support (69%) present in the rbcL data for the ex- panded caryophyllids, Rhabdodendraceae move into this clade in the combined tree. One of the remarkable aspects of the caryophyl- lid clade is the diversity of life history strategies that is found among these taxa. Many of these taxa are adapted to either xeric or saline conditions, and some (i.e., Plumbaginaceae, Frankeniaceae, and Tamaricaceae) have multicellular glands that ex- crete salt (Hill & Hill, 1976; character not coded), whereas others such as Droseraceae, have similar glands that produce mucilage and enzymes used to trap and digest insects (Juniper et al., 1989). A similar Caryophyllidae s.l. was also inferred from 185 rDNA data (Soltis et al., 1997b). Com- paring the present non-molecular, rbcL, and com- bined trees, the caryophyllids appear in no consis- tent position with respect to the rosids or asterids. Future combined studies may establish the inter- relationships of these clades. Asteridae s.l. The larger asterid clade found with rbcL (Olmstead et al., 1992, 1993; Chase et al., 1993; Savolainen et al., 1994; Soltis et al., 1997b) has been remarkably consistent in composi- tion as well as in the general patterns of relation- ships. This same grouping is present in the non- molecular trees (Fig. 2A), except that some unex- pected taxa have additionally been placed here (i.e., Gunneraceae, Sabiaceae, Rhabdodendraceae, and Santalales), presumably because these are highly autapomorphic (e.g., Gunneraceae) or they have unitegmic ovules like asterids (e.g., Sabi- aceae, Rhabdodendraceae). The absence of these groups from the asterids with rbcL analysis can be interpreted as meaning that the distribution of uni- tegmic ovules shows some degree of homoplasy. The presence of unitegmic ovules is a consistent character-state in most asterid clades. In a clade corresponding to asterid I, II, and IV of Chase et al. (1993), most taxa have unitegmic ovules (Hy- drangeaceae, Cornales, Oncothecaceae, Sphenos- temonaceae-Aquifoliaceae, Icacinaceae, Eucom- miales, Dipsacales, Campanulales, Solanales, Gentianales, Scrophulariales, Escalloniaceae, Pit- tosporaceae, Araliales, Menyanthaceae, and Loa- saceae; 137). In addition to characters correlated to some degree with unitegmic and tenuinucellar ovules (137, 139; the correlation including the presence of an integumentary tapetum and endo- sperm haustoria), the asterids also have a higher percentage of taxa with united sepals (215), and especially with united petals (217), than rosids; this is much more evident than in the more restricted definition of asterids by either Cronquist (1981) or Takhtajan (1987). Caricaceae are the only rosid family in this analysis in which all genera have united petals (best developed in the male flowers). In addition to the tendency for the union of perianth whorls, asterids show a higher degree of haploste- mony than rosids, a character that is perhaps func- tionally linked to the more synorganized perianth/ androecium. The core asterids (sensu Cronquist, 1981), Solanales, Campanulales, Gentianales, and Scrophulariales, are held together by alternate ves- sel side-wall pitting (184), simple vessel perfora- tions (185), and rounded vessel transverse section (Metcalfe & Chalk, 1950; 187). Loasaceae and Hydrangeaceae both show the presence of deutzioside (Bliss et al., 1968; Uesato et al., 1986), an iridoid compound known only from these two families (Hegnauer, 1962-1994). Other rbcL studies (Soltis et al., 1995a) demonstrated that these two families are sister taxa (but an rbcL se- quence for Loasaceae was unavailable for the pres- ent study). Iridoid compounds occur in 19 taxa in our matrix, 16 of which belong to the extended as- terids. Light-colored, obdurate, protruding, non- glandular leaf teeth characterize different taxa of Hydrangeaceae (character not coded; O. Nandi, pers. obs.). The investigation of leaf teeth in the sister groups of Hydrangeaceae is potentially inter- esting. Hydrangeaceae and Cornales share the ten- dency to form inflorescences with showy, sometimes white leafy organs at their periphery. In Cornales Cornus spp., Davidia) these organs are large bracts, differing only slightly from normal foliage leaves. In Hydrangeaceae these organs are the se- pals (genera of Hydrangeeae; Engler, 1891). This tendency has not been coded in our matrix (the “~ Volume 85, Number 1 1998 Nandi et al. Combined Cladistic Analysis of Angiosperms organs involved are not homologous). The synor- ganization of flowers into pseudanthia is a recurring phenomenon in asterids IV and II (sensu Chase et al., 1993; character not used in this analysis). Taxa having iridoid compounds that are not in- cluded in asterids s. str. (sensu Cronquist, 1981) are Hydrangeaceae, Cornales, Icacinaceae, Eucom- miales, Escalloniaceae, Loasaceae, Fouquieriaceae, Symplocaceae, Ericales, Sarraceniaceae, and Ac- tinidiaceae (Hegnauer, 1962-1994). These are all asterids in the rbcL and combined trees (Figs. 3A, 4A), and also, with one exception, in the non-mo- lecular tree (Fig. 2 The presence of a theoid exotesta (152), i.e., an exotesta with lignified and often pitted radial and inner walls (cf. Huber, 1991), links some asterid taxa: Sphenostemonaceae—Aquifoliaceae (Ilex; Cor- ner, 1976), Solanales (Solanaceae, e.g., Atropa, Bro- wallia, Cestrum, Lycium, Mandragora, Nicandra, Nicotiana, Petunia, Solanum p.p. Withania; Corner, 1976), Dipsacales (Caprifoliaceae, e.g., Lonicera; Corner, 1976), Gentianales (Loganiaceae, e.g., Strychnos, Gentianaceae, e.g., Fagraea; Corner, 1976), Pentaphylacaceae (Pentaphylax; Huber, 1991), Marcgraviaceae (Souroubea; Huber, 1991), Symplocaceae (Symplocos; Huber, 1991), Diapen- siaceae (Diapensia; Netolitzky, 1926), Ericales (Empetraceae, e.g., Corema; Huber, 1991), Sarra- ceniaceae (Corner, 1976), Clethraceae (Corner, 1976), Actinidiaceae, slightly differentiated in Sau- rauia (Corner, 1976), and Theaceae (Adinandreae; Corner, 1976). The occurrence of cantleyoside (48), an ester of the iridoid glucoside loganin with the secoiridoid glucoside secologanic acid, is restricted to a few taxa of the asterids II (sensu Chase et al., 1993). This compound is known only from Icacinaceae, Dipsacales, and Campanulales (Hegnauer, 1962— 1994; Sévenet et al., 1971; Jensen et al., 1979; Murai et al., 1985; Harborne & Baxter, 1993). The lignan eucommin A (53) is only known from Eu- commiales and Gentianales, a fact that supports the placement of Eucommiales in the asterid I clade (sensu Chase et al., 1993; Hegnauer, 1962—1994; Deyama et al., 1985; Harborne & Baxter, 1993). Oxalate druses (Metcalfe, 1950) are absent from the clade formed by Balsaminaceae (1), Pentaphy- lacaceae (2; no rbcL data), Marcgraviaceae (3), Pel- licieraceae (4), and Tetrameristaceae (5). All but Pentaphylacaceae have the trait of forming oxalate raphides (1, 3, 4, 5), which is unusual for dicots. A subclade of asterid III (sensu Chase et al., 1993) has a persisting free-central column in loculicidal capsules: Ericales (Ericaceae, Epacridaceae; Drude, 1891b, c; Clethraceae; Drude, 1891a); and Thea- ceae (Cronquist, 1981; 252). Ericales and Sarra- ceniaceae are linked by the presence of protruding diffuse placentae. Scytopetalaceae and Lecythida- ceae share stratified phloem (Metcalfe & Chalk, 1950), cortical vascular bundles (Metcalfe & Chalk, 1950), and a nectary disk in the flowers (Scytopetalaceae, Letouzey, 1961; Lecythidaceae subfam. Planchonioideae, Endress, 19944). Rosidae. Relatively minute embryos (com- pared to seed size) seem more frequent in the first- branching dicots (magnoliids, hamamelids, some of the first-branching asterids) than in the more nested clades such as Caryophyllidae s.l. and Rosidae. Only 4 out of 74 rosid taxa for which the character has been coded exhibit minute embryos (163): Paeoniaceae, Saxifragales s. str. (in our study, Sax- ifragales s. str. include Grossulariaceae, Haloraga- ceae, Penthoraceae, Saxifragaceae; without Vahli- aceae, Greyiaceae, Francoaceae, Parnassiaceae, and Lepuropetalaceae; cf. Takhtajan, 1987), Peri- discaceae, and Tremandraceae. A possible syna- pomorphy of Paeoniaceae and Saxifragales s. str. is the presence of an exotestal palisade with thick- ened outer walls (151) in seeds of Paeonia and Ri- bes (Netolitzky, 1926; Corner, 1976). The ridges formed by radial elongation of the exotestal cells in certain Saxifragaceae (similar also in Crassula- ceae), according to Corner (1976), “suggest the ves- tige of a uniformly palisade-like exotesta" in this family. For an extensive study of the Saxifragaceae s.l. and suggestions on their naming see Soltis and Soltis (1997). Another unexpected grouping in the rbcL tree by Chase et al. (1993) is supported in the combined tree. Vochysiaceae/Myrtales (9796 bootstrap) have methylated ellagic acids (62), intraxylary phloem (171), vestured pits in vessels (Bailey, 1933; van Vliet & Baas, 1984; Carlquist, 1988a; 183), and unilacunar nodes (Cronquist, 1981; Dahlgren & Thorne, 1984; 190). Tropaeolaceae, Akaniaceae, and Bretschneider- aceae, in addition to their glucosinolate production (36), are linked by tricarpelly (233—235). Carica- ceae, Capparales, and Salvadoraceae, three gluco- sinolate-producing taxa, each contain taxa with a fibrous exotegmen (Corner, 1976). Capparales and Salvadoraceae concur in the presence of intra- or interxylary phloem (Carlquist, 1988a; character partially represented in 171). The presence of a single crystal layer (with one oxalate crystal per cell) in the endotesta in Caricaceae (Corner, 1976) and some Capparales (Resedaceae; Corner, 1976) could be a further argument for their affiliation (155). The glucosinolate clade is also present in the 162 Annals of the Missouri Botanical Garden 185 rDNA trees (Soltis et al., 1997b) and atpB trees (unpublished). In Malvales s.l, all families have only simple perforations in the secondary xylem (Metcalfe & Chalk, 1950; 185), and all but two families, Cis- taceae and Bixaceae, have representatives with mu- cilage cells or mucilage cavities (Metcalfe & Chalk, 1950; Cronquist, 1981; 119). АП taxa for which the character is known (i.e., all except Sarcolaenaceae and Sphaerosepalaceae) are characterized by the occurrence of centrifugal or rarely (Thymelaeaceae; Heinig, 1951) lateral polyandry (Hirmer, 1918; Gore, 1935; Corner, 1946; Van Heel, 1966; Sattler, 1973; Woon & Кепр, 1979; Cronquist, 1981; Ronse Decraene, 1989, 1992; Bayer & Hoppe, 1990; Nan- di, 1998b; 224). Another synapomorphic character complex for the extended Malvales can be found in seed anatomy. All families for which information is avallable have representatives with the exotegmen differentiated as a palisade layer (Thymelaeaceae, Sphaerosepalaceae, Malvales s. str., maceae, ochlosper- Dipterocarpaceae, 1998a; 157). An exotegmic palisade occurs only rarely out- Bixaceae, Cistaceae, and Sarcolaenaceae; Corner, 1976; Nandi, side of this group (e.g., Trochodendrales, Huaceae, and Euphorbioideae). Malvales s.l. are also linked by the presence of wedge-shaped phloem rays in Thymelaeaceae, Sphaerosepalaceae, Malvales s. str., Cochlospermaceae, and Bixaceae (unknown for Sarcolaenaceae; 170). Moreover, most representa- tives of Malvales s.l. (except Thymelaeaceae) dis- play a stratified phloem (Metcalfe & Chalk, 1950), a character-state known from just 19 other taxa in this analysis. Sarcolaenaceae (1), Malvales s. str. (2), Cochlo- Cistaceae (4), Dipterocarpaceae (5), and Bixaceae (6) share the presence of stellate hairs (in 2, 4, 5, 6) and peltate scales (in 1, 2, 3, ‚ 5, 6; Metcalfe & Chalk, 1988). This group is ali characterized by palmate leaf venation (in Til- spermaceae (3), iaceae, Sterculiaceae, Bombacaceae, Malvaceae, Cochlospermaceae, Bixaceae, and some Cistus spe- 198) and frequent tricarpelly (233-235). Cochlospermaceae, Bixaceae, Cistaceae, Diptero- cies; carpaceae, and Sarcolaenaceae show the presence of a bixoid chalazal region in the seed (Nandi, 1998a; definition see Appendix 4; 159) as a non- paralleled apomorphy. Moreover, the group is char- acterized by the absence of a nectary disk (231), and by parietal placentation (241), and large, curved embryos [Cochlospermaceae, Bixaceae, Cis- taceae, Dipterocarpaceae (Pakaraimaeoideae, Dip- terocarpoideae) have large, curved embryos; Janch- en, 1925; Pilger, 1925a, b; Maguire & Ashton, 1980; Cronquist, 1981; Nandi, 1998a)]. Vestured pits are found in many representatives of the group [Bixa (Solereder, 1899), Cistus, Dipterocarpaceae i Werker, 1981; Pak- araimaeoideae, Dipterocarpoideae), Sarcolaenaceae (Morton, 1995), not in Cochlospermum; O. Nandi, obs.; 183]. Cistaceae, Dipterocarpaceae, and Bixaceae share the absence of prodelphini- dins (60). Bixaceae and Cistaceae share a starchy endosperm (161) with similar structure of larger starch grains (character not coded; Nandi, 1998a). As in the expanded caryophyllids, it is again a set of specialized character-states that establishes the pattern in the non-molecular trees Monotoideae: Baas ете pers. that then parallels the pattern seen in the rbcL trees for the same taxa. Another unexpected clade, identified in the 1993 rbcL trees, was an expanded Malpighiales clade that included families such as Euphorbiaceae, Pas- sifloraceae, Ochnaceae, and Violaceae. This clade is also present in our rbcL and combined analyses Figs. 3B, 4B). Many of these taxa have a fibrous exotegmen (157). Together with the three taxa in the mustard-oil group, this assemblage accounts for ~ 19 of 24 taxa exhibiting a fibrous exotegmen: Соп- naraceae, Oxalidaceae, and Elaeocarpaceae in Cu- noniales; Celastrales s. str.; and Irvingiaceae, Eu- phorbiales (Phyllanthoideae sensu Corner, 1976), Violaceae, probably Kiggelariaceae, Flacourtiaceae Scyphostegiaceae, Erythroxylaceae, Mal- pighiaceae, Linales s. str, Ochnaceae (Sauvage- sioideae), Medusagynaceae, and Trigoniaceae (Cor- ner, 1976; probable indication of the character for Medusagyne by Dickison, 1990). A fibrous exoteg- men was also recently described from Rhizopho- 1992), sup- porting the alliance of this family with rosids having S. str., raceae (Crossostylis; Setoguchi et al., fibrous exotegmen (Malpighiales) and not with Theaceae, as suggested by the non-molecular trees (we became aware of this publication too late to include the character in the matrix). Conti et al. 1996) produced rbcL trees that showed Rhizopho- raceae in Malpighiales. It is noteworthy that Rhi- а zophoraceae and Erythroxylaceae have tropane al- kaloids (character not coded). This alkaloid class is otherwise known only in dicots from Proteaceae, Convolvulaceae, Solanaceae, Cochlearia (Brassica- ceae), and Elaeocarpaceae. The related hygroline alkaloids are confined to Rhizophoraceae and Er- yrthroxylaceae and are known to occur in only two other families [Solanaceae, Brassicaceae (Cochler- aria); Hegnauer, 1962-1994]. With present knowledge, hostplants of the but- terfly genus Cymothoé (Nymphalidae: Limenitinae) include only taxa from a few families near Viola- ceae (character not used in computation); these Volume 85, Number 1 1998 Nandi e iine da cu Analysis of Angiosperms families are Clusiaceae, Euphorbiaceae, Dichape- talaceae, Violaceae, Kiggelariaceae, and Flacour- tiaceae s. str. (Ackery, 1988). This is one of the rare non-molecular patterns linking Clusiaceae to Euphorbiaceae, and both to Flacourtiaceae. Violaceae, Kiggelariaceae, Flacourtiaceae s. str., and Scyphostegiaceae are linked by the presence of septate fibers (Metcalfe, 1956; Miller, 1975; 176). The clade formed by Salicaceae, Flacourti- aceae s. str, and Kiggelariaceae has thin wood- fiber walls (Appendix 4; Metcalfe & Chalk, 1950; Miller, 1975; 175), lack of calyx-corolla differen- tiation (210), absence of alignment of the carpels with the median tepals or petals (232), and locu- licidal capsules (251). The close alliance of Sali- caceae to Flacourtiaceae s. str. is further suggested by the lepidopteran genus Cupha (Nymphalidae: Argynninae) feeding exclusively on this group (on Hydnocarpus, Kiggelariaceae; Homalium, Xylosma, Scolopia, Flacourtiaceae s. str.; and on Salix, Sali- caceae; Ackery, 1988). Another argynnine genus (Phalanta; Ackery, 1988) feeds mainly on Rinorea, Melicytus, Viola (Violaceae), Dovyalis, Flacourtia, Scolopia, Trimeria, Xylosma (Flacourtiaceae s. str.), Rawsonia (Kiggelariaceae), Populus, Salix (Salica- ceae), and Maytenus (Celastraceae). In addition, both Flacourtiaceae s. str. (Xylosma, Poliothyrsis) and Salicaceae (Populus) have representatives con- taining the phenolglucoside nigracin, not known from any other family (Hegnauer, 1962-1994; Thie- me & Benecke, 1966, 1970; 95). Flacourtiaceae s. str. and Kiggelariaceae, two somewhat preliminary taxa derived from the traditional Flacourtiaceae (e.g., Takhtajan, 1966; Cronquist, 1981), are linked by the presence of finely reticulate pollen ectexine (Keating, 1975; 136), a fibrous exotegmen (in Ca- searia, Flacourtia of Flacourtiaceae s. str.; in Оп- coba, uncertain position, probably Flacourtiaceae s. str.; probably also in Hydnocarpus, Kiggelariaceae; Corner, 1976; 157), a hypostase in the seeds (160), ptate fibers in the overwhelming majority of gen- era та (Miller, 1975; 176), opposite in addition to al- ternate vessel side-wall pitting in the wood of the anatomically most basal representatives (Erythro- spermum, Carpotroche, Mayna, Hydnocarpus, Kig- gelariaceae; Azara, Flacourtiaceae s. str.; Miller, 1975; 184), scalariform vessel perforation plates in some representatives (185), and epidermal leaf crystals (196). Within the Flacourtiaceae s. str./Kig- gelariaceae assemblage, there seems to be a nega- tive correlation between genera bearing cyanogenic glycosides of the gynocardin type (49) and the gen- era displaying salicoid teeth (sensu Hickey & Wolfe, 1975; Appendix 4; 201). The two characters seemingly never occur together in the same genus. In addition to the genera described to have salicoid teeth (/desia, Populus, Salix in Hickey & Wolfe, 1975; Prockia їп Morawetz, 1981), twelve other genera from the tribes Homalieae, Scolopieae, Prockieae, and Flacourtieae (sensu Lemke, 1988) have been found to contain species with salicoid teeth: Dissomeria, Byrsanthus, Calantica, Carriera, Flacourtia, Homalium, Ludia (not well developed), Oncoba (in Oncobeae in the system of Lemke, 1988), Poliothyrsis, Scolopia, Trimeria, and Xylos- ma. (O. Nandi, pers. obs.). A giosperm leaves in the herbaria of Zürich (Z and ZT), Geneva (С), and Vienna (WU) indicated that salicoid leaf dentation is a good systematic marker, and that similar tooth types occur only rarely out- side of Flacourtiaceae s. str. and Salicaceae (O. Nandi, pers. obs.; e.g., Tetracentron). The fact that Oncoba lacks both gynocardin-like compounds and has salicoid teeth in addition to glands on the distal end of the petioles (also found in some of the genera with salicoid teeth) indicates that this genus is not well placed among the tribe Oncobeae (the defini- tion of the tribe is based on floral morphology fol- lowing Warburg, 1894). The tribe Casearieae (sensu Lemke, 1988) lacks both gynocardin-like com- pounds and salicoid teeth. Moreover, the two close- ly related butterfly genera Siderone and Zaretis (Nymphalidae: Charaxinae: Anaeini), are known to feed nearly exclusively оп members of this tribe Casearia, Laetia, Ryania, and Zuelania; Ackery, 1988). Other genera of the subtribe Anaeini feed mainly on Euphorbiaceae. This pattern could in- dicate that Casearieae are not immediately con- nected to other tribes of Flacourtiaceae s.l. Con- versely, the genus Cymothoé, mentioned previously, feeds on Casearia (Casearieae), Rawsonia (Ery- throspermeae), Buchnerodendron, Caloncoba (On- cobeae), Kiggelaria (Pangieae), and Dovyalis Flacourtieae), as well as on Clusiaceae, Euphor- biaceae, and Dichapetalaceae (Ackery, 1988). A re- lationship of Kiggelariaceae to Passiflorales is sug- gested by the fact that at least three butterfly species of Acraea subg. Acraea (Nymphalidae: Acraeinae: sensu Pierre, 1984) feed on Kiggelari- aceae, and Passifloraceae tribes Paropsieae (no mo- broad survey of an- ~ no lecular data available) and Passifloreae (an extrinsic character, not in the matrix). The first mo- lecular insights into Flacourtiaceae s.l. using rbcL sequence data information were provided in Chase et al. (1996); a great deal more study of this and related families will be required to establish proper family circumscriptions. A relationship of Violaceae to the flacourtiaceous line is further indicated by the hostplants of Acraea cerasa, found to be an early-branching representa- 164 Annals of the Missouri Botanical Garden tive of the subgenus Acraea in a morphological cla- distic study by Pierre (1984). This species is known to feed on both Rinorea (Violaceae) and Rawsonia (Kiggelariaceae). Other species of subgenus Acraea feed exclusively on Violaceae and Passiflorales. Acraea subg. Acraea thus seems to show loose co- evolutionary correlations with representatives of Violaceae, Flacourtiaceae s. str, Kiggelariaceae, Passifloraceae, and Turneraceae. Linales s. str., Passiflorales, and Euphorbiales also share the exclusive capacity of producing the cyanogenic diglucosides linustatin and neolinusta- tin (character-states not included in the matrix for Euphorbiales; 29, he two compounds are transport-forms of the widely distributed monoglu- cosylated cyanogenes linamarin and lotaustralin (Hegnauer, 1962-1994; Smith et al., 1980; Selmar, 1993; Frehner et al., 1990; Mkpong et al., 1990). They are formed during seed development (Linum), seed germination (Hevea), or tuber formation (Ma- nihot). Quiinaceae, Ochnaceae, and Medusagynaceae have at least some taxa with contorted petal aesti- vation (Touroulia, Quiinaceae, Engler, 1925; Och- naceae, Gilg, 1925; Medusagynaceae, Engler & Melchior, 1925). Links of Ochnaceae to Medusa- gynaceae can be seen in the common presence of stratified phloem (known for Godoya in Ochnaceae and Medusagynaceae, Metcalfe & Chalk, 1950), cortical vascular bundles in the stem, septicidal capsules, and a persistent free-central column in the fruits (Fay et al., 1997) Connaraceae and Oxalidaceae (here in Cunoni- ales; Figs. 3B, 4B) share the absence of ellagic acid and the presence of rapanone, a benzoquinone (Fieser & Chamberlain, 1948; Hegnauer, 1962- 1994). Rapanone is only known from a few angio- sperm families, including Myrsinaceae, according to Hegnauer (1962-1994). Connaraceae and Oxal- idaceae are further linked on the basis of sieve- tube plastids of the PIc-type (Behnke, 1981; 107), the absence of oxalate druses (Metcalfe & Chalk, 1950; 112), a short exotestal palisade [Connaraceae (Cnestis, Connarus spp., Jollydora, Rourea), Oxal- idaceae (Averrhoa); Corner, 1976; 151], endotestal crystals [Connaraceae (Jollydora), Oxalidaceae (Av- errhoa, Oxalis); Corner, 1976; 155], fibrous exoteg- men [Connaraceae (Cnestis, Jollydora, Rourea), Ox- alidaceae (Averrhoa, Oxalis), Corner, 1976; 157], and exclusively uniseriate wood rays (181). Celastrales s. str. and Plagiopteraceae are linked by the common presence of epidermal crystals in the leaves (Baas et al., 1979; 196), the occurrence of weakly crassinucellar ovules in representatives of both taxa (Celastrus and Cassine (as Elaeoden- dron), Celastrales s. str., Johri et al., 1992; Pla- giopteron, Tang, 1994; 139), and an integumentary tapetum (Johri et al., 1992; Tang, 1994; 138). Cephalotaceae, Eucryphiaceae, Brunelliaceae, and Cunoniaceae all have representatives with fol- licles or ventricidal capsules (Engler, 1930; Bausch, 1938; Cronquist, 1981; 251), and Brunel- liaceae and Cunoniaceae have opposite leaves (191) with frequently craspedodromous venation O. Nandi, pers. obs.; 199). A similar cunonioid clade was found in the 185 rDNA trees (Soltis et al., 1997b The taxa with nitrogen-fixing root symbionts in at least some genera [Fabaceae, Cucurbitales (Da- tiscaceae), Coriariaceae, Faganae (Myricaceae, Be- tulaceae, Casuarinaceae), Rosaceae, Rhamnaceae, Urticales (Ulmaceae), and Elaeagnaceae; 105], with the exception of Fabaceae, are placed in a mono- phyletic clade in the combined tree. In the rbcL trees (Fig. 3B), Fabaceae are also members of this clade, but they are placed outside the clade in the combined tree (Fig. 4B). Fabaceae, Myricaceae, Betulaceae, Casuarinaceae, Ulmaceae, and Elaeag- naceae are known to contain nodule hemoglobin (Landsmann et al., 1986; this character was not used in the pe -molecular matrix). — taxa of the ааа bercularia штеа [Rhamnaceae (Rhamnus), Elaeagnaceae (Elaeag- nus), Urticales (Ulmus), Farr et al., orm a monophyletic clade in the trees derived from the combined data set, a fact that could point to a co- evolutionary relationship of the fungus and these rosids. In both the 18S (Soltis et al., 1997b) and atpB trees (Savolainen et al., 6), this same ni- trogen-fixing clade is present. There are many other specific characteristics upon which some discussion could be made, but at this point in time, this is not appropriate. We have focused in the previous section on features that are of particular interest to us. The most significant out- come of these comparisons is that chemical and micromorphological (often palynological) data should be included as equally important characters as developmental and floral morphological traits in macrosystematic considerations. These characters seem to correspond most closely to the molecular results. Gross morphological traits, particularly phyllotaxy, presence of stipules, and perianth ar- rangement, appear especially unreliable for system- atic interpretations at this level within angiosperms. (C) TAXA FOR WHICH rbcL SEQUENCES ARE NOT AVAILABLE Hydnoraceae (not included in our matrices) have been allied to Aristolochiales in some systems (e.g., Volume 85, Number 1 1998 Nandi et al. йен а Cladistic Analysis of Angiosperms Takhtajan, 1987). Characters of Hydnoraceae tend- ing to be ancestral are monosulcate, di- or trisul- cate as well as trichotomocolpate pollen (x-tomo- colpate pollen known in Chloranthaceae, Cabomba, Saururaceae, and monocots), psilate exine, thick endexine as compared to the ectexine, unitegmic, orthotropous ovules (as in Ceratophyllaceae, but probably also correlated with the high ovule num- ber and parasitism), a well developed perisperm (a character more widespread in basal than in ad- vanced angiosperms), a minute embryo in the seed, non-arborescent growth form, and a perianth not differentiated into calyx and corolla. Rafflesiaceae (not included in our matrices) are known to be heterogeneous both in pollen charac- ters and macromorphology (Takhtajan et al., 1985). It is possible that the different families recognized by Takhtajan (1987), i.e., Rafflesiaceae s. str., Apo- danthaceae, Mitrastemonaceae, and Cytinaceae, belong to distantly related groups. The occurrence of ellagitannins (characteristic for eudicots) and 2-, 3-, or 4-porate pollen grains in Cytinus and the tricolpate pollen grains in Pilostyles suggest the ab- sence of a close relationship to Rafflesia, Rhizan- thes, and Sapria, which have monosulcate or mono- porate pollen (the recent 18S rRNA information on Rafflesiales by Nickrent, 1996, confirms the seg- regation of Cytinus from Rafflesiaceae s. str.). The occurrence of both a lamellate endexine and an atectate ectexine in the pollen of Rafflesiaceae s. str. (Takhtajan et al., 1985) is likely an ancestral character combination for angiosperms In the fossil record, epigynous angiosperm flow- ers from the Early Cretaceous of Portugal have been found (Friis et al., 1994). Partly, these flowers are of unclear systematic affinity; some of them have similarities with Laurales. In extant basal an- giosperms epigynous flowers are comparatively rare, although they are present in several families. The fact that Hydnoraceae and Rafflesiaceae s. str. have epigynous flowers and the seemingly ancestral characters found in these two families call for their integration in the research on first-branching an- giosperms. It would be especially interesting to in- clude them in molecular systematic studies, be- cause morphological and anatomical characters are difficult to assess (because of reductions due to par- asitism). Analysis of 185 rDNA sequences would be the most likely source of useful information to address questions about these parasitic plants, but the high levels of divergence for these plants (Nick- rent, 1996) coupled with the low levels of diver- gence for 185 rDNA found in angiosperms in gen- eral (Soltis et al, 1997b) are likely to make sequence evaluations unreliable. Podostemaceae are another family with unusual biology for angiosperms in general, and are thus difficult to assess. In our non-molecular trees, they generally fall near or in the caryophyllids or less frequently Santalales. Despite rather incomplete data, character-states of systematic importance are the occurrence of silica bodies (111) and secretory cavities in the plant body, tricolpate pollen grains with spinulous exine and colpus membrane (Rutis- hauser, 1997; 130, 135), tenuinucellar ovules (139), suspensor haustoria (165), absence of calyx- corolla differentiation (210), rare occurrence of centrifugal androecium development (in Mourera fluviatilis Aubl., R. Rutishauser, pers. comm.; prolonged stamen connectives (229), рден completely free styles [except for, e.g., “Synstylis” (Polypleurum); 237], micropyle formation by the outer integument (247), and septicidal capsules 251). Some of these characters may be seen as adaptations to the extreme habitat of Podostema- ceae. Ueda et al. (1997), using rbcL sequence data, found that Podostemaceae are sister to Crassula- ceae in the saxifragoid clade. Les and Philbrick (1996) reported extremely high levels of divergence for several Podostemaceae, but also concluded that they are sister to Crassulaceae. Balanophoraceae are also highly reduced due to their parasitic ecology. They tend to align in ex- tended caryophyllids. Triangular T" ons (132) and the occurrence of similar e could indicate a link to Santalales (Zweifel. 1939. The 18S rRNA analysis of Nickrent (1996) contra- dicted a close alliance of Balanophoraceae with Santalales. As with Podostemaceae, our matrix for Balanophoraceae has many gaps. Strasburgeriaceae tend to be placed in basal as- terids in the present non-molecular trees. Onco- thecaceae are another small family placed in the asterids near Aquifoliaceae. Preliminary rbcL anal- yses support this position for Oncotheca (Savolainen & Chase, unpublished). Paracryphiaceae cluster with basal asterids or Eucryphiaceae. Rhizophoraceae either fall near to the Stachyu- raceae group or often are the sister group to Thea- ceae. Published and unpublished rbcL analyses support a placement of Rhizophoraceae in Mal- pighiales (Conti et al., 1996) near Erythroxylaceae (Chase et al., unpublished). The fibrous exotegmen in Rhizophoraceae, described by Setoguchi et al. (1992), would be in good agreement with the mo- lecular results (see Discussion, section b Sarcolaenaceae align with Malvales s.l.; this is also confirmed with recent rbcL analyses (Conti et al, 1996). Cochlospermaceae also clearly align with Malvales s.l. as sister to a branch containing жт 166 Annals of the Missouri Botanical Garden Bixaceae and Cistaceae. Other analyses of rbcL also support this placement (Alverson et al., in press). Bonnetiaceae, Elatinaceae (non-molecular data), and Clusiaceae (also with rbcL ) are kept together by an exotegmen with lobate facets in tangential section (158). All three families have representa- tives with septicidal capsules (251). taxa are linked in the non-molecular trees. The ese three close relationship of the three families was empha- sized by Stevens (1991). Clusiaceae and Bonneti- aceae, in addition, are united by having represen- tatives with arils (perhaps a vestigial aril in Ploiarium; Corner, 1976; 146) and protruding diffuse plac entation (242). Preliminary rbcL studies s.l. (Chase, unpublished), but other Bonneti- aceae may not be related to Ploiarium (A. Weitz- man, pers. comm.). (D) CONCLUSIONS Larger data matrices call for improved compu- tational facilities, both in tree searches and in as- sessing confidence in the resulting clades (e.g., jackknife program, Farris et al., 1997). recognized in these searches that the stronger the phylogenetic signal in a matrix, the easier it is to It has been obtain reasonably short trees. In a sense, once one has found all the strongly supported clades, then the search is complete. Regardless of the manner in which weakly supported branches are arranged, there can be no confidence in the patterns so pro- duced. In experiments with combining large rbcL, atpB, and 18S matrices, it has been noted that tree searches have become faster and production of a reasonably short tree length appears relatively easy (Soltis et al., 1997b; Chase & Savolainen, unpub- lished). We are optimistic that, as we add more data as well as more taxa, searches will in fact become easier rather than more difficult. Hillis's (1996) re- cent simulations and predictions also support the notion that increased sampling, both of genes and taxa, produces more accurate topologies; increase in accuracy by sampling more genes has been ac- cepted for some time, whereas it has been a hotly disputed topic whether increased sampling of taxa also produces more reliable topologies (see for ex- ample Graur et al., 1996). On the molecular-systematic side, improved to- pologies will be obtained by integrating and com- paring more sequence-information from different this should to have more confidence in the relationships obtained and to evaluate whether reticulate evolution through an- genomes; allow из cient hybridization or horizontal gene transfer has macrosystematic effects. e work on the “classical” side is equally chal- lenging. Cladistic analyses using non-molecular data should rely if possible on original observations of living plants, herbarium material, and anatomical slide collections, but also on primary and synoptic literature. Literature searches ideally should also include the older comparative literature (e.g., works by Baillon, Bentham & Hooker, Eichler, Engler & Prantl, Payer, Troll), which contains much useful and recently overlooked information. Biochemical work can be refined, and new tech- niques will doubtless permit more detailed com- parison of the different molecule classes. This promises to be a fruitful field, especially for fami- lies on which not much biochemical work has been done, such as Ceratophyllaceae, Hydnoraceae, Raf- flesiaceae, Amborellaceae, Eupteleaceae, Sabi- aceae, Didymelaceae, Aextoxicaceae, Strasburger- Sphenostemonaceae, Oncothecaceae, Tetrameristaceae, Pellicieraceae, Pentaphylaca- laceae, ceae, Diapensiaceae, Scytopetalaceae, Dialypeta- lanthaceae, Bruniaceae, Plagiopteridaceae, Irvin- giaceae, Sphaerosepalaceae, Diegodendraceae, and Sarcolaenaceae. The study of the form and distribution of solid bodies in cells also reveals additional systematic information (e.g., oxalate crystals, starch grains; preliminary works by Czaja, 1969, 1978). The same holds true for investigations of plant hair structure. Much more information on seed anatomy should also be sampled. Priorities again are small families of restricted distribution, as mentioned above. Seed anatomy has proven to be a good tool for macro- systematics in the present study (see also Corner, 1976; Huber, 1991; Seubert, 1993) A character-rich field that has not yet received much attention from neobotanists is leaf structure (se y & Wolfe, 1975; Klucking, 1986/1987/ 1988/1989/1991/1992/1995). Leaf morphology (leaf dentation, leaf venation patterns) is of great potential usefulness, especially in combination with paleobotany. Rhizome, bulb, and root morphology and anato- my are presently not as well understood as, e.g., floral morphology. Floral ontogeny is a field in which new perspectives have been achieved by the use of SEM (e.g., Endress, 1994a; Tucker & Doug- las, 1994; Erbar & Leins, 1996). Because of prac- tical problems in acquiring different ontogenetic stages, there are still many groups that remain poorly known. Inflorescence types have been stud- ied for many families and are likely to be valuable for phylogenetic analyses. Fruit anatomy has not Volume 85, Number 1 Nandi et al. Combined Cladistic Analysis of Angiosperms received much attention, probably also due to the large size of many angiosperm fruits. Recent works on Oleales (е.р., Rohwer, 1996) and Cornales (Reidt, 1997) show that s of fruit char- acters 1s systematically relevan As with intrinsic characters d angiosperms, ex- trinsic ones from fields such as ecology, paleoecol- ogy, paleobotany, biogeography, and hostplant and mutualistic relationships should also provide useful data. In the la sampled on hostplants of fungi, Lepidoptera, and other groups of organisms that tend to have taxa with restricted preference for particular angio- sperms (perhaps also Orthoptera, Aphididae, and Chrysomelidae). Paleobotany is a promising field for providing insights on early angiosperm radiation and relationships to possible outgroups. It may also st field, more information should be add evidence on the position of controversially po- sitioned clades that cannot be assigned clearly to the asterids, rosids, or caryophyllids as described here and in Chase and Cox (in press). We are optimistic about the prospects for im- proved analyses of all classes of data. This study provides one example of how this approach can succeed, but a great deal more work on methods of coding characters is needed. In which cases can tendencies be coded as uniform characters for fam- ilies in which polymorphisms occur? Should a fam- ily or order, no matter how clearly supported as monophyletic, be used as a terminal? These results appear to demonstrate that this approach can suc- ceed with both molecular and non-molecular data and that the phylogenetic content of characters so coded is not terribly distorted by this type of sum- marization. We suspect that, if the patterns are ro- bust, different codings will provide similar results. What is most needed is not a dogmatic approach to character coding and skepticism of the potential for various coding methods to succeed, but an em- pirical evaluation of real data using consistent methods. Too much emphasis on methodological matters will only serve to impede progress. We maintain that the barriers to creating large matrices and performing analyses on large data sets have less to do with the data collection and analysis than with much skepticism of the process itself. Literature Cited Ackery, P. R. 1988. Hostplants and classification: A ге- view of nymphalid butterflies. Biol. J. Linn. Soc. 33: 95-203. . 1991. Hostplant utilization by 2. pu Aus- tralian butterflies. Biol. Р Airy Shaw, Н. К. 1951. On the ondo eae, a re- . Linn. Soc. bor. 14: 25€ markable new family of flowering plants. Kew Bull. 3: -341. Albert, V. A., S. E. Williams & M. W. Chase. 1992. Car nivorous plants: Phylogeny and structural иез Science 257: 1491—1495. Aldrich. 1. B. Cherney & E. Merlin. 1986. Sequence of the rbcL gene for ni large subunit of ribulose bisphos- wee тете -oxygenase from alfalfa. Nucl. Acids Res. 14: она ra arol, E. Conti & J. Sytsma. 1994. ie ‘umscription ES ў Malvales and its plac py in osids based on rbcL sequence data. Amer. J. Bot. (Suppl. 6): as Urg sos . Baum, M. W. Chase, S. M. Swen- sen, R. McC n K. J. Sytsma. Circumscription of the Malvales and ова to other Rosidae: Evidence from rbcL sequence data. Amer. J. Bot. (in press). Arber, A. 1925. Monocotyledons. A Morphological Study. Cambridge Univ. Press, Cambridge Arora, O. P. & M. Metha. 1981. Chemical investigations of some Rajasthan desert plants. Indian J. Chem., B 20 Baas, P. 1969. Comparative anatomy of Platanus kerrii Gagnep. Bot. J. Linn. Soc. 62: 413—422 1972. Anatomical contribution: to plant taxon- omy I. Тће affinities of ied kins et 1 Blumea 20: 16 975. Vegetative асе and the affinities of m А Phelline and Oncotheca. -407. Blumea 22: rre and Afrostyrax Per- 192. уз egetative anatomy and taxonomy of Ber- be кй sed Streptothamnus (Flacourtiaceae). Blumea x A — Werker. 1981. A т. E "cord of vestured pits in Cistaceae. | АЛМА. Bull., . 2: 41-42. — ———., R. Geesink, W. A. Van . Muller. 1979. The ilfisises 5 Plagiopteron amis Griff. (Plagiop- teraceae). Grana 18: 69—89. Bailey, D. C. 1980. nomalous growth and vegetative anatomy of аа chinensis. Amer. J. Bot. 67 147-161. Bailey, 1. W. 1933. The cambium and its derivative tis- sues. ҮШ. Structure, distribution and diagnostic sig- nificance of TEN pits in dicotyledons. J. Arnold Ar- 2-273. 1957. Additional notes on the vesselless dicot- yledon, Amborella trichopoda Baill. J. Arnold Arbor. 38: 374—380. & B. С. L. Swamy. 1948. Amborella trichopoda Baill. A new morphological type of vesselless dicotyle- don. J. Arnold Arbor. 29: 245-254. Baillon, H 1873. Histoire des Plantes, Vol. 3. Hachette, aris. i "s tta-Kuipers, T. 1976. Comparative wood anatomy of Bonnetiaceae, Theaceae and Guttiferae. Leiden Bot. ser. 3: 76-1 a eu D.. L. R. K. Ibrahim. J. B. Harborne & ( 2% 11. 1988. pec еы flavonoids—An ipaa. p die hemistry 27: 2375-239 Barth, O. P. 1965. rei шы ‘he Beobach- tungen am Sporoderm der Caryocaraceen. Grana Paly- nol. n. s. 6: 7-25. Basak, R. K. € K. Subramanyam. 1966. Pollen grains of some species of Nepenthes. Phytomorphology 16: 334— 338. Batygina, ' ^ Kamelina, A. L. Takhtajan, M. S. Yakov a re С. 2n пина. 1985a. nm Em- 168 Annals of the Missouri Botanical Garden bryology of the Flowering Plants. Butomaceae—Lemna- e. Nauka, Leningrad. [In Russian.] ----- =, 1985Ь. Comparative kabalos of the Flowering Plants. Bru- nelliaceae-Tremandraceae. Nauka, Leningrad. [In Rus- & 985c. Comparativa Embavolagy of the Flowering Plants. Dav- idiaceae— + Nauka, Leningrad. [In Russian.] Bausch, J. A revision of the Eucryphiaceae. Bull. Misc. an la 19 17-349. Bayer, C. € ges 1990. Die Blütenentwicklung von The (Sterculiaceae). Beitr. Biol. obroma cacao L. Pflanzen 65: 301-31: Bedell, H. G. 1981. in Marconi eae Misc. Ser. 160: 62 Behnke, H.-D The bases of angiosperm phyloge- ny: Ultrastructure. Ann. Missouri Bot. Gard. 62: 647— I Ња architecture and foliar sc 1, (Theales). Publ. Bot. Soc. Ame а= ~ 977. Zur 2. pd ни Exine bei drei Centrospermen (Gisekia, Limeum, Hectorella), bei Gy- rostemonaceen und Rhabdodendrac ‘ееп. Pl. Syst. Evol. 21-235 . 1981. Sieve-element characters. Nordic J. Bot. 1 т. сонша to the knowledge of P-type sieve- рыш plastids in dicotyledons Il: aceae. Taxon ie 607—610. Berg, C. C. 1977. Urtic a their cue Hun and sys- tematic position РІ. Syst. Evol., Suppl. 1: 349-374. Beusekom, C van. 197 |. Revision of Места (Sa- biaceae), section Lorenzanea excepted, Ay and fossil, ge rad gis ңе Blumea 19: 355-529. Bhandar 1971. Embryology of "m Magnoliales FS pans on their relationships. J. Arnold Arbor. : 1-39, 285 —304. Blac kmore, 5 ., P. Stafford & V. Persson. 1995. Кал апа a of Ranunculiflorae. Pl. (Suppl.) 9: 71-82. Eucryphi- Blank, F. 1939. Beitrag zur c von Caryocar 2 L. Ber. Schweiz. Bot. Ges. 49: 437—494. Bliss „ТЈ | : R. ч "eon dd h. 1968. .. on the ЕТА га. ~ Mentzeloside, a new din eve pad 1 loydia 3 Boesewinkel 1985. 2. of ovule and seed- pope in ys 2, eae) with notes on some other related genera. Acta Bot. Neerl. 34: 413-4 -----. 1994. Ovule ma seed characters of стран pese and the classification of the Lina les-Gera - niales-Polygalales assembly. Acta Bot. Neerl. 43: 25. & К. Bouman. 1980. Development of the ovule and seed-coat of Dic hapetalum mombuttense Engl. with Ne notes on other species. Acta Bot. erl. 29: 103-115. Воћт, & J. Chan. 1992. 2. апа affinities of „мее eae with discussion of the occurrence of B- ring deoxyflavonoids in аа families. Syst. Bot. 17: 272-281. Boureau, E. 1958. Contribution à l'étude des espèces ac- tuelles de Rhopalocarpaceae. Bull. Mus. Hist. Nat. s 2, 30: 213-221. Brenner, G. J. 1990. An evolutionary model of angio- sperm pollen evolution based on fossil angiosperm E len € е Hauterivian of Israel. Amer. Ј. Bot (Suppl., 6): 82. [Abstract.] ---. Es у. Evidence for the earliest stage of angio- dd ү evolution: А 2. section from srael. 15 in D. W. Taylor & L. J. Hickey (ed- ies жиг Plant Origin, 4. апа Phylogeny. New York. The genera of Cistaceae in the мерен United. States. J. Arnold Arbor. 45: 346— 357 Brüning R. & H. Wa m 1978. Uebersicht über die Celastraceen-Inhaltsstof nie, Chemotaxonomie, Biosynthese, Pharmakologie. Phytochemistry 17: 1821- 358. Burger, W. 977. The P Piperales and the monocots. Alternative шй; for the origin of monocotyledon- ous flowers. Bot. Rev. (Lancaster) 43: 345—393 1955. Comparative morphology a nd rela- V. Wood and nodal Capuron, R. 1974. blyoca rpa Tul., s. 14: 291-292. Une variété nouvelle d'Asteropeia am- Théacée de Madagascar. Adansonia n 2. S. 1964. Pollen morphology and ра of arcolaenaceae (Chlaenaceae). Brittonia 16 1976. Wood anatomy of (жне а ар flabelli- folia (Муос) and the problem of multiper- forate . plates. J. Arnold Arbor. 119-126. — Wood anatomy of Tremandraceae: Phylo- — and ecological implications. Amer. J. Bot. 64: 704—713. 1981. Wood anatomy of Pittosporaceae. Aller- tonia z 2 39]. pem anatomy of Loasaceae ke relation to Niel habit and ecology. Aliso 10: 583—602. — ——. |984b. Wood anatomy of Polemoniace eae. Aliso . Wood anatomy ane relationships of Pen- taphylacaceae: Significance of v Phyto- morphology 34: 84—9( - 844. Wood xii stem anatomy of Bergia suf- Just оза: Relationships of 2. 'eae апа broader sig- nificance of vascular tracheids, vasicentric tracheids, ne fibriform id elements. dan. Missouri Bot. Gard. 71: 232-242. 1988a. essel features. Comparative Wood Anatomy. Springer, . 1988b. Wood anatomy of Seytopetalaceae. Aliso 12: 63-76 1990. Wood Pueri and relationships of Lac- s m тег. 77: 1498-1505. .W a Disc of Sabiaceae (8.1.): Ecolog- ical and siena implications. a 13: 521-549. D. зеКтап. 1985. cal wood anat- omy of the una southern Californian Flora. | A Bull., N. S. 6: 319- 347. ©. J. Wilson. 1995. Wood anatomy of Droso- nien . Ecological апа phylogenetic РА Bull. Torrey Bot. Club 122: 185-189. Carpenter, C. 'kison. 1976. The iie € и af Охооо ға balansae. Bot. € 41-15: 5. nter, J. M. 1988. Choosing paa pup gum mapu c pde Cladistics 4: 29 Chase, M. W. & A. V. Cox. Gene sequences, 2. and d of large data sets. Austral. Syst. Bot. (in press). & 5. M. Swensen. 1995. Relationships of Viola- les sensu Cronquist from the perspective of cladistic Volume 85, Number 1 Nandi et al. Combined Cladistic Analysis of Angiosperms analyses of rbcL Fd see data. Amer. J. Bot. 82 Cuppl | 6): E om t.] D. Lledó, M. B. Crespo & S. M. Sw 1996. us in doubt put it in the Ele вр 'eae": Molecular systematics of Flacourtiaceae. Amer. J. Bot. 83 (Suppl., 6): 146. [Abstract.] —— ———, D. V. Stevenson, Ж Wilkin & P. J. Rudall. 1995. Mie systematics: А co re yw Pp. 685- 730 in P. J. Rudall, P. J. Cribb, F. Cutler & C. J. Humphries (editors), 2. Systematics and Evolution. d Bot a Gardens, Kew. ————, D. Solti is, R. G. Olmstead, D. Morgan, D. H. Hills, Y.-L. Qiu, K. A. Kron, J. H. Rettig, E. Conti, J. D. Palmer, J R. Manhart, К. J. Sytsma, Н. J. Michaels, , K. G. Karol, W. D. Clar Smith, C. R. Furnier, S. H. Strauss, ‚©. -Ү. Xiang, C. M. Plunkett, B 5. Soltis, 5. Swensen, S » Е. Williams. Р. А. A. Albert. 1993. Phylogenetics of seed An pr of nucleotide sequences from the Ann. Missouri Bot. Gard. 80: 528— plants: d gene rbcL. Giorn. Bot. Ital. n.s. 32: 223-314 ¿ E. Francini. 1930. Apomissia in “Ochna ser- ratula” Walp. Nuovo Сш, Bot. Ital. n.s. 37: 1–250. Chopra, R. N. & K. Harjinder. 1965. Embryology of Bixa orellana L. 11 15: 211-214 Conti, E., A. Fischbach & K. J. Sytsma. 1993. Tribal relationships in Onagraceae: Implications on гђе], se- quence data. Ann. Missouri Bot. Gard. 80: 5 — — —, A. Litt & K. J. Sytsma. 1996. ко 'ription of Myrtales and their relationships to other rosids: Ev- idence from rbcL sequence data. Amer. J. Bot. 83: 221- 33. 4.2 А. 1925. Embriologia delle Cistaceae. Nuovo Cook, C. D. К. 1978. The Hippuris syndrome. Рр. 163- 176 in H. E. Street (editor), Essays in Plant Taxonomy. ipi: Press, London . H. 1946. Cantal stamens. J. jer 27: 423-431. 1976. The Seeds of Dicotyledons, Vols. 2. Cambridge Univ. Press, Cambridge. Crane, P. R. 1989. Paleobotanical s E of the early radiation of nonmagnoliid dicotyledons. Pl. Syst. Evol. 162: 16 5-1 91. Arnold l and . Friis & K. R. Pedersen. 1995. and culi dires cin of angiosperms. Nature : 27-33. Тһе angie à edersen, E. M. s & A. N. Drinnan. 1993. Early racc oer to idle 4. pla- tanoid inflorescences associated with Sapindopsis leaves from the 2 group of eastern North fink 'a. Syst. Bot. 18: 328-344. Cronquist, A. 1981. An Integrated System of Classifica- tion of Flowering Plants. Columbia Univ. Press, New Yor с . 1983. AF realignments in the dicotyledons. Nordic J. Bot. : 3. Crossley, N. S. & y: jue 1962. Naturally occurring oxygen heteroc ет Part XI. Veraguensin. J. Chem Soc. 1962: 1459-1462. I J. 2 (Suppl.): 1 вэ. и Fl. Neotrop. Mon- Czaja, А. T. 1969. Mikroskopie der Stürkekórner. Parey, Berlin. . 1978. Stärke und Stürkekórner bei Gefásspflan- . Fischer, Stuttgart Варе, К. М.Т. 1980. А revised ystems of ү» cation of angiosperms. Bot. J. Linn. Soc. 80: 9 1983. General aspects of кр 2. апа тас 2. Nordic J. Bot. 3: 119-149 T. Clifford. 1982. The Tun wea A C ompanaive B Academic Press, London. . S. Rao. 1969. A study of the family Geis- solomatacens: Bot. Not. 122: 207-227. & R. F. Thorne. 1984. The order Myrtales: Cir- cumsc ription, variation, and ment Ann. Mis- 1 Bot. Gard. 71: 633-6 . H. T. Clifford & P. = on The Families of the are ii Structure, ы апа Тах- onomy. Springer, Berlin. avis, G. L. 1 Systematic Embryology of the Angio- spe Viley 8 Sons, New York. Dec ар. К. 1979-1985. Étude anatomique de bois d'Amérique du Sud, Vols. 1-3. Musée Royal de l'Afrique Centrale, Trevuren. Decker. J. M. 1966. Wood anatomy and phylogeny of L ee (Ochnaceae). Phytomorphology 16: 39— s n де R. W. & A. P. Vooren. 1980. Bark anatomy of some байса! лез: and Rhopalocarpaceae and their systematic position. Meded. Landbouwhooge- school 80(6): 3-15. DeVries, P. 1987. The Butterflies of Costa Rica and Their Natural History. Princeton Univ. Press, Princeton. Deyama, T., T. Ikawa & S. Nishibe. 1985. The constitu- ents of Eucommia ulmoides Oliv. II. Isolation and struc- зе new lignan glycosides. Chem. Pharm Bull. 33: 3651–36 Ba Dickison, W. C. 1969. Comparative morphological stud- les in Dilleniaceae. VI. Stamens and young stem. Arnold eg 51: 403-418. 8. Comparative anatomy of Eucryphiaceae. ue | Bot 65: 722-735. A survey of pollen PE а of the Con- naraceae. Polen & Spores 21: 31-7 1981. Contributions to the mo s and anat- omy of кји ишн and a discussion of the taxonomic position of the Strasburgeriaceae. Brittonia 33: 564— 580. 1986. Further observations on the floral anatom and pallen кч of Oncotheca (Oncothecaceae). Brittonia 5. 249— 1990. The onis and relationships of Me- сынға y rede они Pl. Syst. Evol. 171: 27— 55. › Baas. 1977. The morphology and relation- „р. A Paracryphia (Paracryphiaceae). Blumea 2 417 owicke & J. J. Skvarla. 1982. Pollen the ан and Actinidiaceae. mer. J. : ot. 69: 1055 Dile her, D. 9 The e occurrence * fruits with affin- ities to erum 2. in Lower and Mid-Cretaceous sediments. Amer. J. Bot. 76 (Suppl., 6) 162. [Abstract.] & P. R. Crane. 1984. Archaeanthus: An early an- giosperm from the Cenomanian of the Western Interior ta orth America. Ann. Missouri Bot. Gard. 71: 351- "| W. morphology У йй К. & W. Barthlott. 1994. Mikromorphologie der 170 Annals of the Missouri Botanical Garden Epikutic 'ularwachse und die Systematik der Dilleniales, 2. und Theales. Trop. Subtrop. — = — — = =. = a = Ф 7 Domínguez, X. >. Espinoza, C. Rombold, W. Utz € H Acha ч s Neolignans, 2. and other POR UNE from Krameria sonorae. J. Med. Pl. Res. 58: 382-383. Donoghue, M. 1. & J. А. Doyle. 1989. Phylogenetic anal- ysis of angiosperms and the re о of idae. Pp. 17-46 in P. R. Ста: (8. Bla Hamamel- ckmore (edi- tors), әже ition, 2 matics | ar History of the Hamamelidae, Vol. 1. C жазы Pre ress, Oxford. Downie, S. R. . D. Palmer 1994. | 4. DNA phylogeny of the Caryophyllales based on structural and inverted repeat restriction site variation. Syst. Bot. 19: 236-252. Doyle, J. A. 1994. Origin of - ое er А phylogenetic perspective. Pl. ¿vol., Suppl. 8 20. 1996. Seed plant pu and the relation- ships. of Gnetales. Int. J. Pl. 157 (Suppl., 6): 53- 539 & L. J. Hickey. 1975. Pollen and leaves from the Mid-Cretaceous Potomac d and 2. bearing on early angiosperm evolution. Pp. 139— B. Bec (editor), 1 E Калу Evolution кзы Мы к. Columbia Uni ss, New Yor L. Hot on Ж Ј. V. Ward. 1990. Early Creta- ceous енген 51 cate 5. апа Winteraceae. ы Cladistic analysis and implications. Amer. J. Bot. 1558-1568 onoghue & E. A. Zimmer. 1994. Inte- gration a eius 'al and. ribosomal RNA 2. оп the origin of angiosperms. Ann. Missouri Bot. Gard. € 419—450. Drinnan, A. N., P. R. Crane & S. B. Hoot. 1994. Patterns of floral evolution in the early diversific Ru of non- 1. 2 (audio ots). Pl. Evol.. Suppl. 8: 93-122. ———, ——— s & K. Pedersen. 1991. Ее АЧМ Howes Кт tric ee pollen of buxaceous affinity from yu 277 Group (Mid-Cr .. of Eastern North America. Amer. J. Bot. 78: 153-176 Drude, О. 1891а "Clethrac ‘eae. Pp. 1— m an оз Lo ONS K. Prantl ies Die natürlichen O ея, lst ed., Vol. IV/1. Engelmann, Leipzig. 89 . асеае. Рр. 15-65 іп А. Engler & K. Prantl (editors), Die nadie hen Pflanzenfamilien, lst ed., Vol. dis Engelmann, Leipzig. —— ———. 189lc. Epacridaceae. Pp. 1 А. Engler & K. Prantl 1. Die natürlic iur 4... T 1. Engelmann, Lei Durand, R. & ч. H. Zenk. 1974. cleavage pathwa zig. = = The homogenisate ring- y in the biosynthesis of acetate-derived пару of the Droseraceae. Phytochemistry 3: 1483 eu Doa „КС. H. Learn, L. E. Eguiarte € M. T. Clegg. 1993a. S d 12 of rbcL sequences iden- tifies шү calamus as the paa 2, monocotyle- don. Proc. Natl. Acad. J.S : 46 — T. ы | Коён. H. G. Hills, "Smith, B. S. Gaut, E. A. Zimmer E G. H. Learn, Jr. 1993b. Phy- logenetic hypotheses for the monocotyledons construct- ed from ies sequence data. Ann. Missouri Bot. Gard. 80: Ehrendorfer, ч E Morawetz & J. Dawe. 1984. Тће neo- tropical angiosperm families Brunelliaceae and Cary- : First 2. matical data and affinities. Pl. . Evol. 145: 183-192. Eklund. H., E. M. Am P K. R. Pedersen. 1997. Chlor- anthac eous pd struc pis s оез the Late Cretaceous of . Evol. 2 2. is n Тће re 1. live t and sys- Sa position | the 4. eae. Bot. Jahrb. . 101: –43: 19 = Tz T 5 РА т БЕ i oral structure, systematics, and e eny in 1. hodendrales. Ann. Missouri Bot. Gard. 297-324. . 1987. The Chloranthaceae: Reproductive struc- tures 2 phylogenetic position. Bot. Jahrb. Syst. 109: 153-226 1989. Aspects of evolutionary differentiation of the da ‘eae and the Lower Hamamelididae. PI. Syst. Evol. 162: 193-211. 138-140 in K. The Familie 's and Genera of Vascular . 2. Springer, Ber —— Bb. Сеге jin em Pp. 250-252 іп К. кали tio, The Families and Genera of Vascular Plants . 2. Springer, Berlin. 93 a eae. Pp. 296-298 in K. Ku- The Families and Genera of Vascular Le на r. Berlin. 99 Ма. Diversity and Evolutionary ecd of Tropic E a C 'ambridge Univ. Press. Cambridge. ---- ›. Evolutionary aspects of the floral eon ture in 2. РІ. Syst. Evol., Suppl. 8: 183 993a. |. istrobaileyaceae. Pp. Kubitzki а Plants, o bitzki io, Plants, =< 1994c. Floral structure and evolution of primi- tive angiosperms: Recent advances. Pl. Syst. Evol. 192: 19-91. ----- 19944. Shapes, sizes and evolutionary cae in stamens of Magnoliidae. Bot. Jahrb. Syst. 115: 429 460. Igersheim. — & 1997. Gynoecium diversity and systematics "e the 25: OF 168 Laurales. Bot. J. Linn. Soc. 2- & S. Stumpf. 1991. The diversity of stamen structures in ' Rosidae (Rosales, Fabales. Pro- teales, Sapindales). Bot. J. Linn. Soc. 107: 21 7-298, Engler, A. 18 Saxifragaceae. Pp. 41-93 in A. Engler & K. Prantl (editors) Die dable hen пиене 2nd Ж и Vol. 2a. 5. Quiinaceae. Pp. 10€ K. Pa (editors) 2nd i 21 "Lower A LA. Engler Die natiirlic hen 52. 5, Cunoniaceae. & K. P | (editors), 2nd ed., 1 А. Engler Die natie hen aa Ra 1" 18a. Епре 21 Leipzig. . Melchior. 5. Medusagynaceae. Pp. 50- 52 in p E ngler & K. Pranil d Die natürlichen Pflanzenfamilien, 2nd ed., elmann, Leipzig. — & intl (e 4. 1887-1914. Die natürlic he en Pflanzenfamilie n, Ist ed. OE. Leipzig. ———& (editors). 1924—1995. Die natürlichen P ku dX n, 2nd ed. Engel: s 1 Leipzig. Erbar, C. & P. Leins. 1996. Distribution of the character state a sympetaly” and “late sympetaly” within I the “Sympe "alae tetrac yclicae" and presumably allied groups. Bot. Acta 109: 427—440. Erdtman, G. 19 32. Pollen Morphology and Plant Taxon- оту. An Introduction to Palynology. 1. Almqvist & Wiksell, Stockholm. Angiosperms. Volume 85, Number 1 1998 Nandi et al. Combined Cladistic Analysis of Angiosperms A note on the pollen morphology in э оное an 'eae on rt ophyllaceae. Veroff. € ot. Inst. Rübel Züri 7-49. Fairbrothers, D. E. 1966 ). и al correspondence of the genus Lorca with taxa of the Cornaceae, Nyssa- 3: 6: 38 z. F. Bills. G. P. Chamuris & A. Y. Rossman. 1989. Fungi on Plants and Plant Products in the United States. American Phytopathological Society Press, St. aul. | же J. S. 1969. А successive kd bow ers approach character 427 Syst. Zool. 374-385. . V. A. Albert, M. Kallersjó, A T s b&A Kluge. 1997. Каин jackknifing 4. 4... -joining. Cladistics 12: 99-124. Fay, M. F., S. M. Swensen & M. W. Chase. 1997. Taxo- nomic 5. of Medusagyne oppositifolia (Medusa- 11-120. . 1988. Mikromorphologie der Epicuticular-Wachse der Rosales s. l. und deren systematische G “істей Bot. Jahrb. Syst. 109: 407— 428. Felsenstein, J. 1985. Confidence limits on phylogenies: E pps using the bootstrap. Evolution 39: 783- 5 dn E. S., P. A. Gadek & C. J. Quinn. 1995. Sim- aroubaceae, an artificial construct: Evidence from rbcL sequence. bia 2. Bot. 82: 92-103 Fieser, L. г E. Chamberlain. 1948. вА а of embelin, fapanoné 58! related aperi by peroxide alkylation. " Amer. Chem. Soc. 75. Filho, W. W., A. 1. Da Rocha, M. Yoshi da & O. R. Gott- lieb. 1985. E llagic acid derivatives froni 2... (ит 222 Phytochemistry 24: Fitch, W. 1971. Toward defining the course ue evo- lution: Minimal change for a specific tree topology. Syst. Zool. 20: 406-416. Foster, А. S. € H. J. Arnott. 1960. Morphology and di- pega vasculature of the leaf of Kingdonia uniflora. 47; 684-6: 98. T. Horner, Jr. 1980. Calcium oxalate crystals in plants: Bot. Rev. (Lanc diee: 46: 361—428. Frehner, M., M. Scalet & E. E. Conn. 1990. Pattern of the cyanide-potential in developing fruits. Implications for plants accumulating cyanogenic monoglucosides (Phaseolus lunatus) or cyanogenic 4. s in their seeds (Linum usitatissimum, Prunus amygdalus). Pl. Physiol. (Lancaster) 94: 28-34. Friis, E. M. 1984. Preliminary report d Upper Creta- ceous angio m Sweden and their level of а дак, Missouri Bak Gard. 71: 403418. K. Endress. 1990. кале ч a% de of angiosperm 2. Advances Bot. Res 2. — ———, К. К. Pedersen & Р. Crane. Do үе io- sperm floral structures from the Early Cretaceous of Portugal. Pl. c yii eee 8: 31-49. 3 РА = A uin -Е. Кодтап, К. С. Ka nti, R. n Pr . 1992. Affinities of the pene ay endemis l from rbcL се Austral. Syst. Bot. E S. Fernando, C. J. pan Э, B. Hoo ‚ T. Te- rrazas, M. C. Sheahan & M. W. Chase 1996. apio: р Molecular delimitation er им groups. mer. J. Bot. 83: 802—811. Чика F., H. Humbert & H. Leconte (editors). 1907— laceae: : ew evidence 4 2. Flore générale de l'Indo-Chine. 8 Vols. Masson, с E A. 1933. өңү" жа anatomy of the woods of the Myristicaceae. Trop. 35: батта, О. A. 1993. Tvpes of pollen grain sculpture and their significance for systematics of the family Fla- 2.0 Bot. Zhurn. (Moscow & Leningrad) 12: 45- 52. [In Russian.] Geetha, К., I. Umadevi & M. Daniel. 1993. Primulales— A reassessment of the taxonomy and phylogeny of the group. Feddes Керегі. 104: 67-71. Giannasi, D. E., G. Zurawski, G. Learn & M. T. Cleg 1992. Evolutionary relationships of the Caryophyllidae based on comparative rbcL sequences. Syst. Bot 15. Gibbs, R. D. 1974. Chemotaxonomy of Flowering Plants, Vols. 1—4. McGill-Queens, Montrea n E. & F. Hoffmann. 1956. 4th ed. Akademie Verlag, Berlin. Gilg, E 1925. Ochnaceae. Pp. 53-87 in A. En Prantl (editors), Die Bs Pflanzenfamilieti, 2nd . Vol. 21. "are de Leipz Goldblatt, P. & L. J. Dorr 986. Chromosome number in У . 73: 828—829. 1981/1984/1985/ pl! Index to Plant Chromosome Numbers. Monogr. Syst. Bot. Missouri Bot. Gard. 5: 1— 533, 8: 1427, 13: 1-224, 23: 1-264, 30: 1-243, 40: 1-238, 51: 1-267. Gore, U. R. 1935. Morphogenetic studies on the inflores- cence jd ae ws Gaz. 97: 118-138. . Kaplan , К. Kubitzki & J. R. To e rs ioe Chemical — in the gnolialean y Nordic J. Bot. 8 баға Н. & М. Parameswaran. 1966. = sekundiire Xylem der Familie Dipterocarpaceae. Anatomische Un- tersuchungen zur Taxonomie und Phylogenie. Bot. Jahrb. As 85: MEF 2: 18, Die ütherischen E & Gottlieb, . Beitráge zur Anatomie 2” жені әт der UN Bot. Jahrb. Syst. 87: 381. Das sekundäre Xylem und die 2... Stellung der Ancistrocladaceae und 9. Phylogenetic pos- 2. Lagomorpha (rabbit, hares, and al- lies). Dus e 379: 333-335. Grund, U — 1981. Systematic relationships of the 4. revealed by serological characteris- tics of ат proteins. Hy E t. Evol. 137: 1-22 Gustafsson, M. H. G. Bremer. 1995. Morphology and phylogenetic eain. of the Asteraceae, Calyceraceae, Campanulaceae, Goodeniaceae, and re- lated families eee Amer. J. Bot. 82: 250-265. Gutzwiller, M. A. 1961. Die por 7 von Suriana maritima L. Bot. Ja aber. J. M. 1959. The ан an и mor- phology of the flowers and inflorescences of the Protea- I. Some Australian taxa. Phytomorphology 9: 325— 358. 1961. Th and morphology of the flowers ы M кшн of pe Proteaceae. li Some American taxa. Phytomorphology 11: 1—16, 16: 490—527 1966. Th t and morphology of the flowers and inflirestences of the ad Ш. Some African taxa. Phytomorphology 16: 490—527. 172 Annals of the Missouri Botanical Garden Hagemann, W. 1970. Studien zur Entwicklungsgeschichte der Angiospermenblatter. Ein Beitrag zur Klärung ihres |“ Bot. Jahrb. Syst. 90: 297-413 . Gleissberg. 1996. Organogenetic capacity of a The signifi ance " 2. blastozones in angiosperms. Pl. Syst. Evol. 199: 121-152. allier, 19 Der 4. des Pflanzenreichs. 14pp. in L. Vom Nebelfleck zum 3. Verlagsbuchhand- Re прага! a 5 теп. 19. Occurrence of flavonol 5-methyl- ethers іп higher plants and their systematic. signifi- cance. Phytochemistry 8: 5 Н. Baxter. 1993. Phytochemical Dictionary. А Handbook to Bioactive Compounds from Plants. Taylor ‘rancis, London. Hayashi, H., M. Sholichin, T. Sakao, Y. Yamamura & H. Komae. 1980. An approach to chemotaxonomy of the Asarum ш Heterotropa. Biochem. Syst. & Ecol. 8: 109-113 419-423. eel, W. A. van. 1966. м of the androecium in Malvales, Blumea 13: 1¢ atomic к m m netic investigations on the morp оору of the flowers and the fruit of Scy- phostegia borneensis Stapf (Scyphostegiaceae). Blumea 15: 107-125. Ж ————. 1984. Flowers and fruits in Flacourtiaceae. V. The seed anatomy and pollen morphology of Berberi- pus and Sram Blumea 30: 31-37. Hegnauer, R. 1962-1994 2. der Pflanzen. 1-а. | D Bus el. Heimsch, C., Jr. 1942. Comparative anatomy of the sec- ondary xylem in the Gruinales and Terebinthales of Wettstein with reference to taxonomic grouping. Lilloa 8: 83-198. Heinig, K. H. 1951. Studies in s Ben d of the c mur Amer. J 13-13 Hekking, W. H. A. 1988 мш P IN and em Fl. Neotrop. | 46: 1-2 Hennig, 5., Barthlott, 1. Meusel & I. nous 1994. Мова der Epicutic o hse und die Sys- tematik der Magnoliidae, Ranune 2. und Hamame- lididae. Trop. Subtrop. Pflanzenwelt 90: 1—60 Heo, K. & H. Tobe. 1994. Embryology and relationships of Suriana maritima L. (Surianaceae). J. 29-37. Heywood, V. H. (editor). 1978. Flowering Plants of the World. he dr Univ. Press, Oxford. Hickey, L. J. & 1975. The bases of angio- spe дк, ы Negetative morphology Ann. Missouri Bot. Gard. a 538-589. Hideux, M. J. & I. K. M 1976. The stereostruc- ture of the exine e its d significance in 321-378 и . Ferguson & J. The Evolutionary Simificanos of the Press, New Yor Hill, - E. & B. 5. Hill. rue Mineral lons. Pp. 225- 243 in U. Lütt 22. (editors): Епсусіо- pedia ог Plant 1. New Vol. 2: Transport in Plants II, Part B, Tissues n Organs. Springer, Ber- lin. Hillebrand, G. R. & D. E. Fairbrothers. 1966. Phytoser- ological systematic studies of selected g ger nera of the Ru- biales and Umbellales. Amer. J. Bot. 53 . 1970. Serological а on the systematic position of the Caprifoliaceae. 1. Correspon- Wolfe. Saxifragaceae а, егез. 2. Sprir К s ——,R. dence with selected Rubiaceae and Cornaceae. Amer. 15. 1996. Inferring complex phylogenies. Na- ture 383: 130-131. Hirmer, M. 1918. Beitrüge zur 4. дег polyan- drischen Blüten. Flora 110: 140- b: 1972. Pollen iie of Taiwan National Tai- . Botany Department Pr Huber, 2... Leitfaden durch die Ordnungen uod Familicn der Ведве ktsamer. Fischer, Stuttgart 1 129-137 in К. Ku- milies and Genera of Vascular | 993, Aristolochiaceae. Pp. Pu (editor), The Fa Plants, s 2. P Berlin Hufford, L. 992. Rosidae and their relationships to other NH dicotyledons: А phylogenetic anal- ysis using morphologica and Semel data. Ann. Mis- souri Bot. Gard. 79: 218—248 ----. 1996. The mon balas and ev of male терасите structures of Gnetales. Int. Ј. Pl. Sci. 157: et =l le > К. Endress. 1989. The diversity of anther structures and dehiscence patterns among Hamameli- didae. Bot. J. Linn. wá 99: 301-346. Humphrey, R. R. 19 A study of Јата columnaris and Fouquieria splendens. Amer. J. Bot. за. 4–207. Hutchinson, J. 1964/1967. The Genera of Flowering Plants. Dicotyledons. Vols. 1 and 2. Cun Pris, Oxforc А es The Families of и РІапіѕ Аг- ed According to а New System Based on Their Probable Phylogeny 3rd ed. 1. Press, Oxford. Huynh, K.- 69. Etude du pollen des Oxalidaceae. Bot. AER Syst. 89: 272-303 llic, J. 1991. CSIRO Atlas of Hardwoods. Springer, Ber- li in. lonescu, F., S. D. Jolad & J. R. Cole. 1977. Dehydrodi- isoeugenol: A naturally occurring lignan kg Sages chia eine 22. J. Pharm. Sci. 89-1490. Jáger-Zürn, 1. 1966. 2. und 2. gische, sowie embryologische Untersuchungen an Му- rothamnus Welw. Beitr. Biol. "Pflanz zen 42: 241 271. Janchen, % Die Cistaceen 2 Ungarns. Mitt. Naturwiss. Verein Univ. 272 7: 1-12 —313 іп А. WT K. Prani d Die dc hen Pflanzenfamilien, 2nd ed., Vol. 2 y А 5; B Lyse-Pedersen & B. J. Nielsen. 1979. Novel bis-iridoid glucosides from Dipsacus sylvestris. Phytochemistry 18: 273-277. Jensen, ‚ B. Greven. 1984. Serological aspects and - relationships of the Magnoliidae. Taxon : 2- 57 | John, | & K. P. Kolbe. 1980. The systematic position of "The ales ” from the viewpoint of serology. Biochem. t. & Ecol. 8: 241—248. РТА B. M. 1970. Symposium on comparative ЕАО ову of angiosperms. Proc. Indian Natl. Sci. Acad., B 41: 1-385. 1925. Cistaceae. Pp. Jensen, S. R., — & D. Как. 1954. p eee of Tamarix L. ШОО 4: 230— 2. 2 Srivastava. М и Embryology of Aude Vols. m l and il. 5 . P. Bhatnagar, N. N. Bhandari & M. R. 4... (editors). 1967. Seminar оп Volume 85, Number 1 1998 Nandi et al. Combined Cladistic Analysis of Angiosperms Comparative Embryology of Angiosperms. Department of Botany, ч Press, Delhi. Juniper, B. E., R. J. Robins & D. M. Joel. 1989. The Carnivorous Plants. Academic Press, London. Kamelina, O. P., V. A. Poddubnaya-Arnoldi, A. L. Takh- tajan, M. S. 2 « С. Y. Zhukova. 1983. Com- parative Embryology of the Flowering Plants. Phytolac- Thymelaeaceae. au Leningrad. [In : . I. D. Romanov, A. L. Takhtajan, M. S. Yakovlev & G. Y a 1981. Comparative Em- bryology of the Flowering Plants. Winteraceae—Juglan- daceae. Nauka, Leningrad. [In Russian.] Kanis, A. 1968. A revision ide ү Осһпасеае of the Indo- Pacific Area. Bum 6: Kapil, R. N. B 5”. 1991. Embryological evidence in erm classification and phylogeny. Bot. Jab Syst. 113: 309— & R. Maheshwari. 1965. 2. (s m themum 2. aertn. Phytomorphology 1 7-557. Kaur, H. 1969. .. Т е оп uo or- i; ana L. Proc. Natl. Inst. Sci. India 35: 487-506 Keating, R. C. 1972. The comparative morphology of the udi speras ede: Ш. The flower and pollen. Ann. Missouri Se Gard. 59: 282-296. —— Trends of па in pollen of Fla- observations of Cochlo- 9 M. F. Moseley. 1978. Wood anatomy and phylogeny of Paeonia section Moutan. J. Arnold ree 297 Untersuchungen zum B-Asarongehalt handelsüblicher Kalmusdrogen sowie zu den Inhalts- stoffen des asaronfreien Kalmus. Unpublished Ph.D. Thesis, University of Saarbrücken. Keng, H. 1962. Comparative mo 4. ‚к= in Theaceae. Univ. Calif. Publ. Bot. 33: 269- Kessler, P. J. enispermaceae. E 1: 128 in K. қыны (editor). The Families and Genera of Vas- cular Plants, Vol. 2. Springer, Berlin Klucking, E. Р. 1986/1987/1988/1989/1991/1992/1995. Leaf Venation Patterns. 1. Annonaceae/ 2. Lauraceae/ 3. Myrtaceae/ 4. Melastomataceae/ 5. Сотова еае/ 6. Flacourtiaceae/ 7. The Classification of Leaf Venation Patterns. а Stuttgart Kobuski, C 1951. Studies in the Тћеасеае, cm The genus жо; 1. Arnold Arbor. 32: 256—26 Kóhler, E. 1994. 22 evolution of V il in the genus Buxus L. (Buxaceae). Acta Bot. Gallica 141: 223-232. Kolbe, K. P. & J. John. 1979a. Serologische Untersuch- ungen zur Systematik der Violales. Bot. Jahrb. Syst. 01: 3-15. 79b. Serology and syste matics of the puse ind Т. Biochem. Syst. & Ecol. 8 * nean A. J. G. H. 1985. Family status for the Mon- otoideae Gilg and the Pakaraimoideae Ashton, Maguire 35 Salient lines of structural specializa- tion in the wood rays of dicotyledons. Bot. Gaz. 96: 547-557. Kron, K. A. € M. W. Chase. 1993. Systematics of the Ericaceae, Как Epacridaceae and related taxa based upon rbcL ze ce data. Ann. Missouri Bot. Gard. 80: 735-74 Kubitzki, K. еды Cba, Pp. 197-200 in K. Kubitzki (editor), The Families and Genera of Vascular Plants, bus 2. Springer, Berlin Canellaceae. Pp. 200-203 in K. Kubit- zki Modes The dui iid ES Genera of Vascular Plants, Me I^ Springer, Я Degeneriaceae. Pp. 290-291 in K. Ku- ТҮЗ E The Families d Genera of Vascular Plants, Vol. 2. Springer, Berlin. = 5 sh & H.-H. Poppendieck. 1991. Parallelism, its evolutionary origin and systematic sig- nificance. Aliso 1 pe 2” Kuprianova, L. . On the possibility of the devel- opment of бш) pollos from monosulcate. Grana 8: 14. E J.. E. S. Dennis, T. J. V. Higgins, C. A. Ap- y. A. A. Kortt & W. J. Peacock. 1986. Common evo se iin origin of is and non-legume plant НОСА Машге 324: 168. Lavault, runeton. ss Isolement de deux nou- veaux ня alotdes, d pou et O- cec 5'triphy- opeltine. J. Med. 8, Suppl.: Le Quesne, P. W., J. E. b ud R. A e 1980. Antitumor plants. X. Constituents of Nectandra rigida. J. Nat. Prod. (Lloydia) 43: 353—359. Lebreton, P. & M. P. Bouchez. 1967. Recherches chi- miotaxonomiques sur les plantes vasculaires 5. Distri- bution des composés proe chez les Parie- tales. Phytochemistry 6: 160 Ө. Leenhouts, P. W. а notes on the genus Di- chapetalum (Dichapetalaceae) in Asia, Fasc and elanesia. Reinwardtia 4: 75—87. Lemke, D 1988. A synopsis of Flacourtiaceae. Aliso 12: 29-43. Les, D. H. 1988. The origin and affinities of the Cera- art eae. Taxon 37: 326—345. 1993. ee Pp. 246-250 in K. Ku- hala (editor), The Families and Genera of Vascular Plants dur | Sprin erlin ` Philbrick. 1996. The phylogeny of river- weeds одессе, Insights from rbcL sequence data. Amer. J. Bot. El | Abstract. | „р. K. анну: € F. Wimpee. 1991. Molecular sibus history of ancient aquatic angiosperms. Proc. Natl. Acad. Sci. U.S.A. 88: 10119-10123. не К. 1961. Notes sur les Scytopetalacées (Révi- sion 1 des Sc Upon de l'herbier de Paris). Adan- : 106-142. 986. Systematic foliar morphology of Phyl- 2. ыла сенін I. Conspectus. Ann. Mis- souri Bot. Gard. 7 Lin, C. M., Z. Q. n & S. р. King. 1986. Nicotiana chloroplast genome: X. Correlation jm the DNA sequences and the isoeletric 5 pattern of the LS of rubisco. Pl. Molec. Biol. 6: 81—87. Maddison, D. R. 1991. The discovery and importance of multiple islands of most-parsimonious trees. Syst. Zool. 40: 315-328. Maddison, W. P. & D. R. Maddison. 1992. MacClade, Version 3. Analysis of Phylogeny and DNE Evo- lution. 2. Sunderland, Massachuset Maguire, S. Ashton. 1980. е dipter- ocarpacea П. in 29: 225—231. 972. Botany of the Guyana Highland— : пије авина Mem. New York Вог. Gard. 192 Martius, C. F. P. von (editor). 1840-1906. Flora Bras- 174 Annals of the Missouri Botanical Garden iliensis: enumeratio plantarum in Brasilia, Vols. 1-15. Fleischer, Münc Mauritzon, J. 1935. put нщ: der Elaeocarpaceae. Ark. Bot. 26A (10): 1 19: Zur buo 'ologie und systematischen ветно de Reihen Terebinthales und Celastrales. Bot. pa . 1936: 161-211. 1 J Muller & B. Lugardon. 1975. Notes on the 1. and fine structure of the exine of some pol- len types in LN arpaceae. Rev. Palaeobot. Penal 19: 241-280 c Nair, J. E. 1930. The taxonomic and climatic distri- pus of oil and starch in seeds in relation to the phys- ical а chemical properties of boh substances. Amer. J. Bot. е E Mc Alpine J. B., . Riggs & (in part) P. G. Gordon. 968. iu stereoc c of calopiptin. Austral. | Chem. 21: 2095-21( Melchior, H. (editor). 4 А. d Syllabus der Pllnzenfanilien II. Borntrüger, B Mennega, А. M. W. 1982. Stem structure of the New World Menispermac eae. J. Arnold Arbor. 63: 145-171. Metcalfe, C. R. 1952. Medusandra richardsiana de nan. Anatomy of the leaf, stem and wood. Kew Bull. 9: 237— 246. Scyphostegia borneensis Stapf. Anatomy of stem and em in relation to its taxonomic position. Rein- wardtia 4: 99-104. 1962. Notes on the systematic anatomy of Whit- tonia and Peridiscus. Kew Bull. 15: 472-475 987. Anatomy of the Dicotyledons, 2nd ed., Vol. 3. VAM te Illiciales, and Laurales. Clarendon Press, Oxford. & L. Chalk. 1950. Anatomy of the Dicotyledons. Leaves, Stem, and Wood in Relation to Taxonomy with Notes on Economic Uses, Ist ed., Vols. 1 and 2. Clar- endon das Oxfor Chalk. 1988/1989. Anatomy of the Dicot- yledons, 2 ed., Vols. 1 and 2. Oxford Univ. Press, Oxford. Meylan, B. A. & E . Butterfield. 1978. The structure of New Zealand 4 New Zealand Dept. Sci. Industr. Res. Inform. Ser. 222: 1-2: Miller, R. B. 1975. Systematic anatomy of the xylem and comments on the лш of Flacourtiaceae. J. Ar- nold Arbor. 56: 20— Mkpong, О. E., Н. Yan, €. Chism & R. T. Sayre. 1990. Purification, characterization and 1 айайт of 2. marase іп Cassava. РІ. Physiol. (Lancaster). € 181 1— 50. Morawetz, W. 1981. Zur systematischen Stellung der Gat- tung Prockia: Karyologie und Epidermisstruktur im Ver- gleich zu Flacourtia (Flacourtiac cne), Grewia (Tiliaceae) . Evol. 139: 57-76. 1993. 2 rela- tionships among members of Saxifragaceae sensu lato based on rbcL sequence data. Ann. Missouri Bot. Gard. 80: 631—660. 4 . Robertson. 1994. Systematic and subs жоп ations of rbcL T varia- tion in Rosaceae. Amer. J. Bot. 81: 890—% Morton, C. M. 1995. A new genus and species of Dip- terocarpaceae from the Neotropics. I. Stem anatomy. Brittonia 47: 237-247. . Mori, G. T. Pra 1997. ce, К. С. Karol & M. W Phylogenetic nuin of Lecythida- Chase: e: A cladistic 4. using rbcL sequence and mor- phological data. Amer. J. Bot. Ж 530- 540. Murai, Е, M. Tagawa, S Matsuda, T. Kikuchi, $. Џезак & H. | 1985. Abelios 2. А апа В, вес 'oiridoid glucosides from Abelia grandiflora. Phytochemistry 24: 2335. . 1998a. Ovule and seed anatomy of Cistaceae and related Malvanae. Pl. Syst. Evol. —— ———. 1998b. Floral development and systematics of e. PL Syst. Evol. (in press). 1926. Anatomie der Angiospermen-Samen. (in press). т 5 e 3 ~ = or - T Е ~ 0: << 9s Ф = > rlin 1996. Phylogenetic relationships of par- ic Santalales and Rafflesiales inferred. from 185 A sequences. Amer. J. Bot. 83 (Suppl., 6): 212. [Abstract.] Nixon, K. C. € J. 1. Davis. 199]. Polymorphic taxa, miss- ing values, Лаф ви 241. and cladistic analysis. Cladisties 7: 233— .. L. Crepet, D. Stevenson & E. M. Friis. 1994. A re РВИ dion of seed pen phylogeny. Ann. Missouri Bot. Сага, 81: 484—53: Olmstea ; Bremer, K. M. Scott & J. D. Palmer. 199: А parsimony analysis of the Asteridae sensu lato based on rbcL sequences. Ann. Missouri Bot. Gard. 80: 700-72: == Т ісһа К. M. Scott & J. Palmer. 1992. Моло of m Asteridae p 2. of their uS lineages inferred from DNA sequences of rbcL. Ann. Missouri Вог. Gard. 79: 249-265 Dacia; B. & M. Lidén. 1995. er—Evidence EM nuclear ribosomal DNA. Pl. Syst. Evol., a > 89-193. Payer, J.-B. ps d'organogénie comparée de la fleur. ha Paris. Philipson, W. R. 1993. Amborellaceae. Pp. 92-93 in K. Kubitzki (editor), The Familie Plants, V n 2. Springer, Berlin i 901. Il legno e e corteccia delle Cistacee. Nuovo (йл. Bot. Ital. n.s. 8: M. Pierre, J. 1984. Systématique évolutive cladistique et mi- métisme chez les bis pbr a к Acraea. Unpub- lished Ph.D. Thesis, Unive The position бакан s and Genera of Vascular 473-5 Paris. Pp. 313- 315 in A. Engler & K. Prantl (editors), Die natürlicher an Pflanzenfamilien, 2nd ed., Vol. 2 1925b. Cochlospermaceae. Pp. 316-320 іп А. Engler € K. Prantl (editors), Die natürlichen Pflanzen- familien, 2nd “ Vol. 21. Engelmann, Leipzig. Prance, С. E systematic position of Rhabdo- dendro E. A P Bull. Jard. Bot. Belg. 38: 127 5. . 1972. 11: 7 Rhabdodendraceae. Fl. Neotrop. Monogr. К. da Silva. trop. Mono 12: 1-75. Price, . € J. D. Palmer. 1993. Phylogenetic rela- ie of the Geraniaceae and Geraniales from i Ann. Missouri Bot. Gard. 8 Fl. Neo- 1973. Caryocaraceae. Ц e comparisons. 661-671. Proctor, M. C. F. 1955. Some chromosome counts in the European Cistaceae. Watsonia 3: 4. Puff, C Weber. 1976. КЕСЕ АЙШӘ, and karyology of Rhabdodendron, pe a reconsideration of the systema er posu ion of the Rhabdodendraceae. Pl. Syst. Evol. 125: 195-22 Qiu, Y.-L., M. W. Chase, D. H. pura E Parks. 154-159 Со im butona to the mor- 1993. Volume 85, Number 1 1998 andi et al. Combined Cladistic Analysis of Angiosperms Molecular phylogenetics of the Magnoliidae: Cladistic analyses of nucleotide sequences of the E stid gene rbcL. Ann. Missouri Bot. Gard. 80: 587-606 Rao, К. V. & Е. M. Alvarez. 1982. Chemistry of Saururus cernuus. |. Saucernetin, a new neolignan. J. Nat. Prod (Lloydia) 45: 393-397 Rao, T. 199]. € же йон of Foliar Sclereids in An- a: Morphology and Taxonomy. Wiley & Sons, New Delhi. Raynal-Roques, A. 1981. Contribution à l'Étude Bio- morphologique des Angiospermes Aquatic iW Tropi 'ale sai d'Analyse y | Évolution, Vol. 1. Unpub- lished Ph. D. Thesis, University of е Atélier Duplication, Montpellier. Кесога, S. J. 1933. The woods of Rhabdodendron and анодго. =з Woods 33: 6-10. Reidt, 997. Fru e the Studien an den “klassisch” Ronse Decraene & E. gefassten Cornaceae s.l. in E. Smets, Robbrec ht (editors), 13th Si man of Morphology, Anatomy & Systematics, Pro- gram & Abstracts. Sc jio Bot. Belg. 15: 134. Rendle. А. В., С. Baker & S. M. Moore. 1921. Systematic account of the plants collected in New Caledonia " E Islands of Pines by Prof. R. H. кына М. 914. J. Linn. Зос., Bot. 45: 246—417 бы, I. 1957. Мого. e costituzione chimica dei peli nel genere Cistus e loro importanza nella sistema- tica di alcune specie. Ann. Bot. (Rome) 25: 540—566. Rice, K. A., M. J. Donoghue & R. G. Olmstead. 1995. A reanalysis of the large rbcL dataset. Amer. J. Bot. 82 (Suppl.. 6): 156-157. [Abstract. Rodman, J., R. A. Price, K. Kenneth, E. Conti, K. J. Syts- J. D. Palmer. 1993. Nucleotide sequences of the rbcL gene indicate monophyly of mustard oil plants. Ann. Missouri Bot. Gard. 80: 686—699. Rohwer, Ј. С. 1996. A preliminary survey of the fruits and seeds of the Oleaceae. Bot. Jahrb. Syst. 115: 271- 291. Ronse Decraene, L. P. 1989. The floral development 4 Cochlospermum tinctorium апа Bixa orellana with s cial grow on the androecium. Amer. J. Bot. рб. 1344—135‹ ——. on. Characterization and Systematic ина lished Ph.D. Thesis, University of Leuven, Belgium & E. F. Smets. 1992. Complex sine in the — Definition, |“ and systematic val- ue. Nordic J. Bot. 12: 621— “Жыш R. 1997. es m developmental di- versity in Podostemaceae (river-weeds). Aquatic Bot. 0. The Androeceum of the Magnoliophytina: npub- 57: 29-7 & R. Sattler. 1987. Complementary and heuristic value of contrasting models in structural botany. H. Case study on leaf whorls: Equisetum and Ceratophyl- lum. Bot. Jahrb. Syst. 109: 227- Sáenz de Rivas, C. papa o morphology of Spanish staceae. Grana 18: 9 3. 1993. Polen кенінің of the Ambor- Monimi- Hortoni ortonioideae: aceae). сша 32: 154-162. — — — & Р. К. Endress. 1984. Не morphology in the Trimeniaceae. Grana 23: 129— Sandwith, N. Y. 1962. 2. to the flora of tropical America. LXIX. A new genus of Peridiscaceae. Kew Bull. 15: 467-47 Satabié, B. 1974. С йрй de la palynologie à l'étude des Irvingiacées d'Afrique tropicale. Adansonia n.s. 14: 289. Sattler, R. 1973. Organogenesis of Flowers, a Photo- i .T to. d: o < в zA e -E ы = os = ns to the floral n: plan: Illustrated from Cista- . New Phytol. 35: 47-67. 937-1939. pu Morphology. A New Outlook with Special ier to the Interpretation of the Gy- naeceum, Vols. 1 and 2 НА V., J. К. Атпа . Spichiger. 294. Molecular phylogeny of families ind to Ce- 2 ue on eov 5' flanking sequences. Molec. —37. Рһу w & Ae 1, S. B. Ноо! & M. W. Chase. 1996. Ane энш vof ио patterns of plastid atpB gene sequences among eudicots. Amer. J. Bot. 83 (Suppl., 6): 190. [Abstract.] Schaeppi, Н. 1953. Morphologische Untersuchungen an den Karpellen der Calycanthaceae. Phytomorphology 3: 7 112-117. Schmid, R. 1964. Die systematische Stellung der Dion- cophyllaceen. Bot. Jahrb. Syst. 83: 1—56. 3l. Vergleichende Embryologie der An- giospermen. Borntraeger, Berlin. Schweingruber, F. H. 1990. Anatomy of European Woods. H aupt, bern. Selmar, D. 1993. Transport of cyanogenic glucosides: Linustatin uptake by Нетеа cotyledons. Planta 191: 9]- cn M. A G.S. B. K. Chowdhury, J. F. Morton & G. J. Kapadia. 1976. Identification of volatile constit- apac uents of Sassafras albidum root oil. Phytochemistry 15: 1773-1775. Setoguchi, H., H. Tobe & H. Ohba. 1992. Seed coat anat- omy of Crossostylis Rhizophoraceae: Its evolutionary an systematic implications. Bot. Mag. (Tokyo) 105: 625- Seubert, E. 1993. Die Samen der Araceen. Koeltz, Kón- igstein. Sévenet, T., C. Thal & P. Potier. 1971. Isolement et struc- ture du cantleyoside, nouveau glucoside terpénique de Cantleya cornic qug i a ) Howard (Icacinacées). Tet- 3–668. Shiklina, I. A. s The comparative anatomy of the f the genus Oncotheca (order Theales). Bot. . (Moscow & Leningrad) 62: 1273-1275. [In Rus- unn, ГР 19 Nymphaeales. I. Preliminary surve of а Aliso 7: 243-261. Comparative serology of the order Nym- de II: oae c of : ymphaeaceae and Nel- umbonaceae. Aliso 7 Sinnott, E. W. a ES on n phylogeny 4 the angiosperms. 1. The anatom node as an a in ue lst of angiosperms. mnm J. Bot. 303— Smith, 3 c3 D. Weisleder & R. W. Miller. 1980. Linustatin es necios Cyanogenic glucosides of е de that protect animals against selenium tox- 7 70. Comparative serology of the order y on the relationships anic Chem. 45: 507—510. EE H 1899/1908. Systematische Anatomie der Dicotyledonen, Vols. 1 a ¿nke, Stuttga & J. F. Meyer. 1928. P tische ad der Monokotyledonen. Heft III. шаа is hae-Spa- thiflorae. Borntrüger, Berlin 176 Annals of the Missouri Botanical Garden Soltis, D. E. & P. S. Soltis. 1997. Крейг 2. ships in Saxifragaceae sensu lato: А comparis to- pologies b on 185 rDNA and rbcL sequences. mer. J. Bot. ы 504-523. .-Y. Xiang & L. Hufford. 1995a. 2. апа evolution o Hydrangeaceae based on r quence data. Amer. J. Bot. 82: 504—5 E C. Hibsc h doller P. S. Soltis l. Chase & J. 5. Pun 1997a. Molecular no relationships among angiosperms: Ап overview based on rbcL and A sequences. Pp. 157-178 іп К. ee iki & P. H. Raven (editors), 2. and Diversification of Land Plants. Springer, Toky , P. S. Soltis, D. R. Morgan an, S. M. Swensen, B. C. Mullin, J. M. Dowd & P. С. А 1995b. Chloroplast gene sequence data suggest a single origin of the pre- disposition for symbiotic nitrogen fixation in angio- sperms. Proc. Natl. Acad. Sci. U S.A. € ----.Р. S. Soltis, M. Mort, 5. B. Hoot & С. М. Мопоп. Inferring complex phylog- enies: An empirical approach using three large DNA sets for angiosperms. Syst. Biol. (in press). , D. L. Nickrent, L. A. Johnson, W. J. a J. Gillespie, W. J. Kress & K. J. Sytsma. 1997b. Angiosperm phylogeny inferred from 185 ribo- somal DNA sequences. Ann. Missouri Bot. Gard. 84: 14 Souéges, К. 1937. Développement de Pembryon chez oo guttatum. Bull. Soc. Bot. France 84: E bh ). МсКеу ag aa Legume eni Part 5 Botanic Gardens, Kew. Stearn, F. С. 1946. A Study of the Genus Paeonia. The Royal Horticultural Society, London. Stevens, P. F. 1991. On the phylogeny of the Elatinaceae— Bonnetiaceae-Clusiaceae. Amer. J. Bot. 78 (Suppl., 6): 220. [Abstract.] Sutter, D. & P. K. Endress. 1995. Aspects of gynoecium structure and macrosystematics in Euphorbiaceae. Bot. Jahrb. Syst. 116: 517-536. Swensen, 5. M., B. C. Mullin & M. W. Chase. 1994. Phy- logenetic affinities of Datiscaceae based on an analysis of nucleotide sequences from the plastid rbcL gene. Syst. Bot. 19: 157-168. Swofford, E = 1993. 2. Phylogenetic Analysis Us- ing Parsimony, Version 3.1.1. Computer program d tributed by de Illinois Natural History Survey, Chan paign, Illinois Sytsma .& p. A. Baum. 1996. Molecular phyloge- nies and the е of the angiosperms. Pp. 314—340 in D. W. Taylor ickey (editors), Flow- ering Plant Origin, 1. Mud Phylogeny. Chapman & Hall, New York Syvanen, M. 1994. Horizontal gene transfer: Evidence and possible consequences. Annual Rev. Genet. 28: 237-261. 1994. Advances in The Nitrogen Factor. Royal n & P. К. Stevens. 1989. Classical slant аи «ambiguities extend to the molecular level. J. 4 Evol. 28: 536—544. Takahashi, 1985. Wood . anatomical studies of Poly- carpicae. | Magnoliales. Sci. Rep. Osaka Univ. 34: 29— K. & K. Sohma. 1982. Takahashi, Pollen morphology of the Droseraceae and its related taxa. Sci. Rep. Tohoku Imp. Univ., Ser. 4, Biology 38: 81-156. Takhtajan, A. L. 1966. Systema et phylogenia Magno- 22. Nauka, Moscow. [In R 1987. The System of the Magnoliophytes. Nau- ka, Генай, [In Russian.] (editor). 1985/1988/1991. Comparative Anatomy of Seeds, Vols. 1—3. Nauka, Leningrad. [In Russian.] 1997. Diversity and Classific: ‘ation of Flowering Plants. Columbia Univ. Press, New Meyer & V. N. Kosenko. 1985. Тһе роПеп КҮЛЕ кы апа classification іп Rafflesiaceae s.l. Bot e urn. (Moscow & Leningrad) 70: 153-168. [In Riis: n.] Tang. Y. 1994. Embryology of У НИЦИ Ол suaveolens Griffith (Plagiopteridaceae) and its systematic implica- tions. Bot. J. Linn. Soc. 116: 145-157. Taylor, D. W. & Hickey. 1996. Evidence for and pn ps EE an herbae 'eous origin for angiosperms. Pp. 116-140 in D. W. Taylor & I Hickey (edtiors), «nt Pag Origin, Evolution & Phylogeny. Chap- man & Hall, New York. Taylor, F. H. 1972. The secondary xylem of $4 2.” "eae: А comparative study. Bot. Gaz. 133: 230— Thanikaimoni, C. Evolution of 2... Canad. J. Bot. 64: 3130-3133. Theisen, 1. & W. Barthlott. 1994. Mikromorphologie der Epicutikularwachse und die Systematik der Gentiana- les, Rubiales, Dipsacales und Calycerales. Trop. Sub- trop. Pflanzenwelt 89: 1—62. Thieme, H. & R. Benecke. 1966. Isolierung eines neuen Phenolglykosides aus Populus nigra L. Pharmazie 21: 59—00. ussian ж- 970. Ueber die Identitát der Glu- coside Ж und Poliothyrsid mit Nigracin. Phar- mazie 25: 492. Thorne, R. F. 1983. e new realignments in the angiosperms. Nordic J. Bot. 3: 85-117 ----- 1992. Classification and Rory of the flow- ering planta, Bot. Rev. : 225—348. Tobe, H. & C.-I Peng. 1990. The cholos and taxo- nomic е of 2. (Bretschneidera- ceae). Pe Linn. Soc. 19 9-152. . Raven. T 1. and relation- ships of 4. (Akaniaceae). Bot. J 118 201-274. Todzia, C. A. 1993. Chloranthaceae. Pp. 281—287 in K. Kubitzki (editor), The Families and Genera of Vascular Plants, Vol. 2. ia Berlin. Tomlinson, P. B. Anatomy of the Monocotyledons. П Palmae. 2. Ta Press, Londo Tsou, C.-H. The embryology, кийа mor- phology, and systematis of Lecythidaceae. Mem. York Bot. Gard. 1-110. Tucker, S. C. cis, 1994. Ontogenetic evidence and phy logerietic о. among basal taxa of leg- umes. Pp. 11-32 in I. K. Ferguson & S Tucker (editors), Advances in Legume Systematics, Vol. 6. Roy- al Botanic л Kew. yuda, A. Nakano, T. Siuchi, A. Seo, H. us зо Қ” P Hak 1997. Molecular phylogenetic po- sition of Podostemaceae, a marvelous aquatic flowering plant family. J. Pl. Res. 110: 87-92. Uesato, S., M. Miyauchi, Н. es H. Inouye. 1986. Bio- 5 thesis of iridoid glucosides in Galium mollugo, G. spurium var. 2. апа Deutzia crenata. Inter- ~ = - = = . Linn. Soc. Volume 85, Number 1 1998 Nandi et al. 177 Combined Cladistic Analysis of Angiosperms mediacy of deoxyloganic acid, m and iridodial glu- coside. А 25: 2 521. Ukraintseva, V. V. 1993. Pollen ea of the family Cistaceae in 4. to its taxonomy. Grana, Suppl. 2: Veldkamp; J. Е 1984. ANT (Flacourtiaceae) in Australia. Blumea 30: 21— Vestal, P. А. 1937. The os of comparative anat- omy in establishing the relationship of the Hyperica- ceae to the Guttiferae and their allies. Philipp. J. Sci. 64: 199-252 Vijayaraghavan, M. R. € U. Dhar. 1976. Scytopetalum tieghemii—Embryologically unexplored taxon and affin- ities of the family Scytopetalaceae. Phytomorphology 6-22 Vliet, G. J. C. M. van & P. Baas. 1984. Wood anatomy and classification of the Myrtales. Ann. Missouri Bot. —800 986. Phytoserologische Untersuchungen zur Systematik der Euphorbiaceae; Beiträge zur intrafami- liáren Gliederung und zu Beziehungen im extrafamili- üren Bereich. Diss. Bot. 98: 1-124. Walia, K. & R. N. Kapil. 1965. Embryology of Frankenia Linn. with some comments on the 2. position of the dozen Bot. Not. 118: 4 29. Walker, J. W. € G. Walker. 1984. LORI of Lower Cisl edi angiosperm pollen and the origin and early evolution of flowering plants. Ann. Missouri Bot. Gard. 71: 464—521 Warburg, O. 1894. Flacourtiaceae. Pp. 1—56 in A. Engler & K. Prantl = Die natürlichen Pflanzenfamilien, ngelmann, Leipzig. 55; Wood anatomy of the American frankenias е Systematic and evolution- ary 2. ev Bot. 74: 1211-1223. Williams, S. E., Albert & M. W. Chase. 1994. Re- lationships of кыы A cladistic nicis of rbcL sequence and morphological data. Amer. J. Bot. 81: 1027-1037. Williamson, P. S. & E. L. Schneider. 1993. Cabomba- ceae. Pp. 157-161 in K. Kubitzki (editor), The Families and Genera of Vascular Plants, V Wilson, C. L. > ceae. 1. Hibbertia Andr Phytomorphology 15: 248-274. Wojciechowska, B. . Seed morphology and anatomy of some . species. Monogr. Bot. 29: 121- 135 Woon, C. & H. Keng. 1979. еа, E the Dipterocar- paceae. Gard. Bull. Singapore 32: Wu, C.-Y. & K. Kubitzki. 1993. cin caeasteraceae. Pp. 288-289 i in K. Kubitzki (editor), The Families and Gen- era of Vascular Plants, Vol. 2. y Ail Berlin. Mor Xiang . E. Soltis, n & P. S. Soltis. 1993. рањене —€— ot a L. sensu lato and putative relatives inferred from rbcL sequence Nau ‚ Ser. 7, Morfol. Anat. Rast. 5: 250-259. [In Кай, | Zenk, M. Fürbringer & W. Steglich. 1969. Oc- currence and шака of 7-methyljuglone and plum- bagin in the Droseraceae. Phytochemistry 8: 2 200. Zhang, Z.-Y. 1987. A study on the pollen morphology of — Pr and its systematic position. Acta Phyto- tax. Sin. os [In Chinese. | Zurawski, e rrot, W. Bottomley & P. R. Whitfeld. 981. The structure of the gene for the large subunit of ribulose-1,5-bisphosphate carboxylase from spinach chloroplast DNA. Nucl. Acids Res. 14: 3251—3270. Zweifel, 1939. _ Cytologisch-embryologische Untersuchungen an Balanophora abbreviata Blume und Pan Pm vp Wall. Vierteljahrsschr. Naturf. Ges. Zürich 84: 2 Annals of the 178 Botanical Garden issouri M olviviviviviviviviviviviv viviviv v v PE TEDE BE sejeonun i 6e828J8U02 Il у у | у | м | м У y y У |“ v v vyv v v у у зејенфпопд 3 lviviviviviviviv ‚[у|у | у уо [5 7 d E lviviviviviviviv гу | У ЈУ / | У yiviviyv у еевогошецн И Б ERE olviviviv I y viv 7 7 ШЕ оееоезон J у 3 m у 7 252 5 Е уу [уу [у у o v [о [у [у [у v ШЕ; У viv viv 7 ШЕ ns 's зејебезихеб Ж ууу viviv м volv v v vyv v v у SeeDensseJo Ж 7 Il viviviviviv v v v Il vi yi vy v v v 1 ВЕ ЖЕГЕН! өеөоеюоөесј Ж I vy у v у 7 sejeiuouno E. eeeo2eieunig әвөдв! 3 Ж му | у | У [у (vivi viviviviviviviviviviviviviv I| vy v v CTN Ж ERE у / КИГЕН” ВЕ ВЕ ESSE Ж / viviviviviv viviv v v v ууу уу [у [у 7 ЖЕГЕН зејерцешешен 5 гу [у I v yvi vy v v v Гу Гу ' v y v У v ВЕ aeeoeuayuel4 J ууу хм | у | ум viv I v vyv ЖЕГЕН” әвәзеоиеше | s у 9e82€JesoJ( 4 = у 2 у әғәоереродвіочу Ж EST ою Е ИЕЛЕ: vivi у | у Уу м у yvi vyv vv viv THERE aren! 3 у У v әгәори! а J eeeoeiedojejsy у 2| viv ivi viviviviviviviviviviviviv у / | У | У I y 7 зөүепАцдойте: a у (viv м у у | хм yivli vyv v v v viv у v әгәоевриошш!5 a әгөоруриәрордецы әеөорләиип=у E НЕ У ум v у vivivivio violviviv / | v eeeoepisdopueqieg Ж eeeoeuueuio. eeeoejeuÁpi a BE eeeoeiqes a ЖЕГЕН” (viv у у у | у yvy v v v v DE 7 7 оееоехпа a /|viviviviviviviviviv viviviv v v : 7 у оевовеоја _ гуу ууу (ууу [у [у [уу у у ШЕ; зејезрџероц2о | 8 у 2 |: евеоеиодштем у [уу [у [уу уу [уу у [у [уу [уу Гу 7 у өвөовиеүе |91- 3 ИЕШЕ: ‚/|у|у | у у у | у |: ЖЕЛЕ: : I viv У | У р eepynounued | Pi TIERE им | у у у | у | у viviviviviv DE у 5 sejeyou tt 7 7 1 гу y 1 1 У | У ЖК у ] t зејеоошу | Zb V 1 у Є V SLOITEIMSUÁN t ' |м [МУ хм у у | у | уу v у [у [у DE : у о |: sepe] | Ot "ERE yivivivy{[vivjvwiyv THERE ШЕЛ DE eeuenpopisuy | 6 — ЖЕЛЕ] olviviviviviviv ЖЕНЕ! у ШЕ; olv : 7 sep, | L "ERE yivivivijviviviv / |м | у [У У | У | У TIERE seem) | 9 ЖЕ: ГІТЕЛІСІПІТІІГІГІ a E eeeoegejquy | P 7 у 5 22 7 у A | Z / | У seee Аудојеје5 TUAW обје (1808 |'Autp 0104 17912 | Wes lund 50101 |joeu по теор пешү unea [peo |ејеч eoid Mixer [Bur (1010 inyp Due 95 [0295 1095 0295 0295 |0195) x от [ее веј 22 | зе| осе | се | ге | „е|ос|62|92| 22 | 92 | se Ре | 6 2] 22 | „с |ог| 6: | 8: | 2: [s r [se re [ет з | го | | elz] sieur np "“хідеш вер ів|поэ(Тош-поу 1 x1puadd y 179 Combined Cladistic Analysis of Angiosperms Nandi et al. Volume 85, Number 1 1998 ojojv|jv|v|v|v го [о [е | у | у | Ју [у о | о [о Ју | у | у | У | У ЈУ | У | У ЈУ ЖЕЛЕДЕ ЕСЕЈ Е о[у |5 21: E 5 ЗЕЛЕ: гу [у [у 5 У [У \ J У |У [у AE 4 DE 2|2|2|9|2|2|: 5 у у DEJ ӘЖЕ ig у 5 21: у у у У [У Ж У/|У|УІ|У у / [0 [оо |< 515 үә |: f ] у 4 У [У J НЕДЕ] У [У о[у у |у [у [у EME ЖЕНЕ] E у у у |У | У|ТУ|У уо [уу о[у [у 2|02|У|У|У|У|У|У|У у у ' |У [У 3 7 5 oj|vivi|vi|vi|vi|vI|viv ] 1 /|У|Ү a у : ВЕДЕ: " | У | У[М [У Ж 7 5 у a r'Iviviviviviv]v у Алыша. У | У |У У | У MX оу A p 25227) Ж у у ВЕДЕ) eeeoeepg | 66 /jviviviviviv ] £ у у У [У Бәрердәшешен Ж 7 / A | LE У [У АКАК] ] у |у | У | М ЈУ 4 ЖЕ у У | М у MX әвәоеоџеше! | 96. у = ў у евезејезоја у „|“ | У | У /|Ұ|У|4|У|У)|ҮУ у / у _оееоецџедем | PE 5 | әгәовреүоодвюиу | EE у | og | 26 AE. ] riviviolvi|: ] АКАК о[у рм ТЕЛЕ ЗЕ ЕТ „| у [у [Уу У [У зево' od 5 7 z : у У [У өгөоеш poe evecriedaelsy 21У у У viv у DE У | У [У /|viviviviviviviviv BENE] BEBSEASE] sajejAydofieg | 8z- у jv огеоевриошш | 22. E 9eeotJpuepopqeug у | | seeoepisdopueqieg | PZ [| Гаа | eeuwen | Ez | ЕЕ EREFRERERERERES эшзәшра [zz ele ИЛЕ ЕН ЕЛЕ ПОШ pe] seceqes | 15 ЖЕЛЕ; у у |м|јм |У|У y l92]2]2]|2| aesoexng | 02: vjviviviviy у |х |у |м | У |У у ( ‚|у|у|у [v[viviv] әғәзеәюіі | 6b DEAE] 4 yivivivij[v[viviviv E [ate Peje sejespuepoyoos] | Bb c ERERERER — (Ls vjvi[vjviviv ] E У vv |у |« |« |% | eeeoeueeld | 91 Гаа ае EE RET E ] У | У у 21У1У у [у [у [о [уо у eepinounuey | Pt 9195 [3 [y yi viv у | у [ојм |[2|9] зејецомбед | Є1_ КЯКЕКАЕН — Й 2.8 E у I Y IE E EN | | eeeoeonsuAN EN [у [о [уч ESEA! ERER — [2E | / 7 у / | 7 | у | E | зеџецуоојозџу | 6 E | у ЖЕЛЕДЕ; | suopejüooouow | 8 | / | | - зејеајим | 2 /iIviviviviviv viv AJ TEAK оу [у [у |у {ууу у У |у | У | У | У То |v sejer | 9 У | У У | У | У 7 У |v У |v | | eeeoeKejeqonsny | $ | | | eeeoejexquy | P 5 7 | у 214 y |у |У|У|МУ |У | a У | У ЈУ “У | У | У v[Iv|iviviv[v|vivi|viviv| b SERERE |v | | mes H- vv 1 ЖЕ: 3 SERERE eeeoelAudojeJe) zueq оеш |210u |гхпе |2nQ | ејиа члер | ezje Dael |Dida еб oyo | eJ шә jue 1522 |2e1s | дәш Jeje pod esaa уез! "Хур | ould UAS |29ne [spex | 1sne juwis одо ue |упер [jenb ио |ОЛш [usab 2595 Jese x ог|62|82][44| 94] S| Hcl 62]. | 12104 |69]89| 2 9| 99 | 59|] Рој 2 9| 2 9| 19| о9|659| 95] 25 | 95 | ||| 2 2 | 1 | оз| беј ве | Ly] оу | се| ерер | ср [тв z sieur pp ‘panunuoy 71 xipuaddy Missouri Botanical Garden Annals of the 180 (ре ке joe [pee gt as ЖЫ DR | Џ | | | 75157-11415] | 14-1-1144" í £ Б Б ЖЕСЕ E JEJEJE /|vivivic У [У E SALIUN евеовје 05 £ р ЖЕ: Е JEJEJE Ж 5 зегеифпопо ' | У | У о /ivivivioic у 2 [э |: сеџебе j [ / E у [уо [: у E E зевоеошецн [ ЖЕ: у E 7 vivivivivic DEJE 5 ӘеӨОе5ОН E 7 ГТТТЛТЕН у : eweoeejKudeis | E 7 ЖЕЛЕЛЕЗЕЗЕ у E / JEJEJE] ns `8 зејебејухеб Б Е 7 E у 7 AEBSEZSEZ ӘЕӘОР|П58ғ/) ЕСІП І22И 5 у у уу |“ у |: у ' у viviviv viv LBJ E < у у У | у | у | У | У у у /lviviviviviv sejeiuoun?) Є g ТЕЛИ: ©]: TEDE eeeoelud/uon3 E TE г | vivivic /|у|у|у|у /| vi viviv viv ЖЕ: BLEYA E ZEE Б У | У E diviviviviviviviviviviviviviviviv vv aeaee] E у у ЕЕ! yivivi[viv ЖЕДЕ: гу | У ТУ У | У зегерцешешен E TE TON] у У У | у у DEDEDE] BE: eeeoeoueu | / JE у ” у ZEIEN] | BEDE 1 ЖЕ: у ум хм viv vv eesoeyueden | DEDE] BE YE ЖЕ: 1 ‚|у|у | У | У у E у У | У әвөоерејоодѕіоиу 216 1 у / v ЖЕ y | viviv I y 7 ШЕ eeeoejAudoouoig у E ЖЕЛЕ? Е у JEJEJE г уу Туту у ' | У У | У eeeoeuoBAod у уу : у 5 у АҚСАҚ! у У viv У [У әеәоешбеашгіа у 1 V eeeoeiedoJeisy у JEPE ЖЕ; ЖЕ v"iIviviviviviviviv ИКИ / у |v sejejAudo je? у v voit ” 7 "vi у у eeea2erspuoululs у TEJE зееоерџерордецн c 2 c € 1 / еееовјеџипо) | У ojeg Я ZH ЕЈ у У | У eveoepisdopueqieg у 7 | У ' | У | У у у у У | У : у y v ШЕ; 7 E у E 5 у ум viviviv у E 2 E 1 ЖЕ; E yivivivivivivivi[vi[vf[viv ЖЕ: E у |У|У у У [У JEJE 7 Б у |у | ЖЕ [2 |: 216 ЕЕ у Б ЖЕЛКЕКИКЕНЕЙЕ ееооеџејеја у овеовеје па | 2 1: ЖЕ E DE E DE 215 әерцпоипџиен EME: BE ЕЈ ЕЈ E] XE sejeyouBey E у E AK 5 у у 5 viv з |: зегеџоџиу ' | у | У [2 |: У | У У v у [у [< E E seeoeonsu AW ӘЖЕ р 7 : у Б / | viv 9 [rs sejene] E з |3 [= 7 Е у 7 | У ' |“ [У Е веџе!Ц2ој ову 977% E у < у ЖЕДЕ ЋЕ зиореу оооџојј T у DEDE BEJ зејезе А р < у / | У 7 у / | У ЕЗБЕЛІ у у [viv eeeoeAejreqosny р у E DEJ /viviviviviviviviviv viviviv £ E < z у JEJE] у 7 nsa опш (э |ю АШ |? 15А12 1105 | џоје | ехо >06 ОМ рој hens paas оло | oyu [11198 [pue чује ор пазе бии топо | шер эпе | eejo [usoa |2295 [ори jopu! |'uje | Ioue 1020 |20111 JOUR |џеој OZ ife t HEL rizo 1191 Stipe ЦЕ TZ ELE HOt пјбогјво rjo 1|9 0 во H|F O HE о тотто i]OO I| 66 | #6| 2 6 | 96] se| у б| с 6| 2 6 | 1 б|об| бејве | 2 ај за | сеј реј се| св | 18] c гропициођ "ү xipuadd y 181 Nandi et al. Volume 85, Number 1 1998 Combined Cladistic Analysis of Angiosperms / 5 : c sejeonin у әвөоееџоо || f у E : зејеуфпопо |1 E E z < о [< J И ЕЕ ЖЕ ( өевоеошецы || : c : әеөсезон 2 7 2 өсеоеибеөн3 | c 9 | ЖЕЛЕНЕ у ( Е TEJE Е / БИК : a/v ШЕ c ИКЛЕЛЕ; © |: € : ' Е t TE] 7 2 р зејемомпо || "Д A | eeeo2eieunig o[t eeeoenud/uon3 | € р о |: әвәоеил | ( у c WE ad 5 E " с E 5 [3 r = “2 [3 гу ууу Гу Ту [о v I v yv у ЖЕГЕНІ TE ууу v / [у 7 D Е ў / À / / Y ОМ] V у V £ V V у у 1 < ЕВЕ гуу 7 / [у ЖЕ E 7 р 7 У [У c ‚|у viv у ЖЕЛЕ: У У "ЖЕК E 5 |: Е у | У |: у /|У|У|У|У / AK eeeoepisdopueqieg | | у V d 4 4 4 V 1 1 © eeeoeuueuoJAWN & 9t c seeosjeuApiq / у с сЕ: c eeeoeqes | > 7 : ЭТ? ; scu na [4 РИК JEJE о |: овеовецола || Є у TIERE 9 |t _ sepipuepouoo;] || [3 РИК \ 5 ЕД ( E БИЕ; E : ЕД : E 2|9|У ‹ ПІ E Е c : E E c : t E : БИКЛЕ; E [t E z РИ E E [t E 7 © [= 9 9v [o ја € / 9eueglu2ojoisuv E E ШЕ 9/2 € suopeyjooouow | | DEDE] Ð |: зегејешм |. I| у [У РИК : 5 seni | 9 E € 9 |t E 7 eweoeKejeqonsny | 9 әвеоејәюашү | P : у о [у [уу ЖЕЛІ: БИКЛЕ; өтөогөеүйшА | € _ c CE 5 [ғ viv ЖЕЛКАЕ E р 7 EZ q 7 HERE DEJE ом] V 3ued |зопџ |50 Nixes [Bue [чеш uen | e 109 [19141 ие | ou hup [216d И И | 9118] 09169 Hest SHS TSS STS TESTIS TOS TSP TSP ПЕР ПЕРУ ПУТЕВЕ У ПОР ТЕ СИНЕ ЕСТЕ nec iiec ife oet апааа y gieur qup "'panunuo?) "Т xipuadd y Missouri Botanical Garden Annals of the 182 sejesnn 98028J8U07) olo ololo solo зејенафпоп) о| 3 olo o « elo olololol+ solo у <<< <|< ooo olo olo < оо olo зејешоипо тб аі. оо 9eeoeluelQ lolo olololc olo › €! clc ӘБРӘОЕ:ө501( Oo К olo әгәоереүзод5оиу Seen uoi ololo 5 <<< < < <<< < <<<) < S olo Olo olololo 9o!o 9 sejeyouBey 300 « III << << < << ololo Sejeuouuy eveoeonsuAN olo зејепе] 8EUPEIU2OJOISUV olo < oo <|<|<|< olololololo aloalo 9 о ДЕ 2 ololo , ў VOIS | Jea 15 | јеој | jee] Uap |55ол [sseA |55ел | рџе |овел SAA |1015 2) ШОЧ ШЕЛ узап |оел |794 7941} 00м [1925 у 00 әш! Д zr 6 LIS 6 Li? 6 LEG 126 ire LOG 168 88 2 e 99 lise Lit 8 LE e 28 18/08 6; 2 LOL 122 1192 1152 blo 2 ЦЕ HE Z HE 2 102 | ЭЗ т $T 557 9 Lit 9 12911291191 'penunuo) І 5 sieur ир r a Бы *|'^| xipuadd у 183 Combined Cladistic Analysis of Angiosperms Nandi et al. 1998 Volume 85, Number 1 L 2 у / зәп / eeeoeeuo) | PS c зејенфпопо | ES : веџебеа | ZS o [0] olo olo « olo << к-|соісікік- olo » (90 бо) 3 3 olo E o Џ sejeiuoun? ГЕ ү евезецешга вевоиц бола | Ty o x E o olo o pa < olo o (210 < = < m o > < « OÍDO о ololo olo « әгеәоешәхие1- ӘРӘОЕОПЕШР] EENEN] У E р eeeoequedeN | v€- aeaoepejoonsiouy 'eeecepAudoouog | ZE с 2 / ^ eeeoeuobAog | LE Е у у eweoeubequng | 0E вевотедојеј у | 6c C olololo olola 5 elo ole ololo <|<|< | <|<| < = olo [0] « оо о1о! Х olo о b] ole о « [um о о [9] а Ф = Е т” o | Џ су су! се сосу 5 « « E c 2/У 9/0 t V eeeoeJeuunc) bs у eeeoeprsdopuequeg E E OV 2/У 7 E JE eee»euueuoAN | EZ ] € ВЕ eeeoweuApig | ZZ 9/V|9/9 y | 5 еееоекјес 2 Е 5 € eee»exng z E g £ Б ч евезеејоја зејезрџгроц20) | 4 eeeoeuoquinieN 2 c у ESTA] € 5 / c жаға ILI E р өерупоипиен < [e зәүеңоибеүү Sejeuouuy У |“ ју | : у eeaoeonsuAW | € : : : 1 : ЕСТІ) ееџе!20јојвџ у suopejfiooouo | 8 c c ЕСІП) L : sejeionii eeeoeKejreqoisny | " eeeoejeJoquy c O/N|l/V : у вевоевецашаћ | € ИЕЛЕ: [1 [ом ОУ р 27У v ПЕЛЕ; / [У c eeeoeuueouo | Z “ |У [У L viv i fv ИК авеоеуАчдојеле) | t JONS [HONS |) јело јео јео јео |>џ 199U |*AJBA | ШОО хе Әли | YIU | 1090 |1092 [WOS эш [ZBA |210 weds еј eN les эшід | YOu | uu | ey O- |2844 | 0d 'џоје |епјој шөл! JES] uey ІО|ЦО (DES m Oveleczecezecocescepecececcelecocceccscezccecesecpecececeeiecoceetcetedztestesiepiceteectet clot cleococzozsozsocpoceoczoctoz y gieur iip К olo 2 ololo › › « « ola 3lo о E] о ola | | К olco “рәпицио”) 'I xipuadd y ШЕ: р 7 ; / евеоејеџ 05 < E Б зејенфпоп) | 5 Е 3 E Missouri Botanical Garden Annals of the sola ооо Ф “ 5 к : 1/9 9e62BJesoJQ olo Џ | әеөоерејэодѕюи 5 D р 1 ою 7 Е у € d € ееесеш а евезе! ualsy saye|Aydohied SESIEISPUOWWWIS 184 ш евеоезрџерораечн еееоејеџипе) eeeoepisdopuequeg re € евезеџшецјојлуј E eeesejoÁpIg eeeoeiqes eeeoexng olo « ч ololo!c « о < [0] ololo ol olo Џ « ololo fs әеәотопвиАуү 5 |9/9 sejene] 2/У! 1 eeueiu»ojolsuv OW ie 5 1 зејеје им [2 Р 1 зеје ои £ əesvekapeqonsny әеәоеһәғодшү y 5 € E sejeuouuwy | Z Е Т c 0 oo E] olololo у | 9 |9/v 1 5 ov ololelol> = o 1777») c 1005 una, [3885 |1140 [1914 |4nAO |упло | тело [1685 | пир Z S ZIL S |05 TE y јер сј t Z|9 y Zls р Zlv y ZE v ЈЕ y Zip 4 sieur yp 'penuguo?) т xtpueddy 185 << << <= <<< <<< < < әеөоеиц2о әеөоееоойео | | евеоселорелмеб eweoejoeedo;| || Combined Cladistic Analysis of Angiosperms Nandi et al. = sejeojisseq e << ¡A AA < eeeoeBuuow | eweoedieooee[3 | | ӘБӘОЕОЦЕС әвөоеџејәббу | | <|<4|< |< < <|<|<|< <<< < 4|<|< < <<< < <<< < «j««« 9P62PIJnO2P|3 өеөоејодл евезен ша зејед5еје5 'sejepudes || <|<|<|< <|<|<|< <|<|<|< ЕСІ Soje íw || eeeoersÁuooA | t 98628]EUJOJOSSIOC) « « eeeowunAuoeig | ( « eeeoepyexo Е еееоејеџ о) “| 8| "| | < Моште 85, Митбег 1 1998 “"рәпицио”) аена | syeur qup `1 xipusddy Annals of the 186 Botanical Garden issouri M I| у [У "HERE y у | у | у ТУ TEDE ЖЕГЕН! Iviviviviviviviviv 1 у c у у [у [уу [о [у [у [и [у [у ЕЕЕ / [2 [У У хм у у | 7 | У [У Ii olv v ıl viviviv [у | ү BEJE 7 7 ЖЕГЕН ЖЕЛ ' |“ | У |М / у | м ЈУ | У 7 у у |: 7 у 7 7 t c 7 ВЕ (| хм |“ у |“ [0 | ВЕ 7 ' | у | у | у | У | у |м | у | о [о у | у Ју | У [У ERE lv ([viviviviviviv I y 7 ' у | му v | У ИЕЛЕ; ЖЕГЕН eeeoejAxong 3 | 1 ___евеовшоби/ | 7 7 eescejeiedeyog |! у у у 7 әгөови uo |2 22-427 19) 7 ЖЕ: оееоеџце!3 | | c eeeoeneuuog | i Ж [о [о [у [5 7 DE У у |“ у [у [У v |: / |м |м у у [2 | у | У ЖЕ; ИЕЛЕ; / | у v зевое ви) | | хм у | у у +“ 7 ВЕ ' хм у [ју | у | у |: ЖЕЛЕ; ЖЕЛЕ; У [У “ASS saeun | | у у | у v v / ЗЕ ИЕЛЕЛКЛКАКДЕ; TIERE ЕГЕН” TEDE әгесеңібійеуі || еееоеилбезпрер | уу уо |: 7 у хм |м viviviv 7 ШЕ: өтеоешро |. E Ls eeeseJeso Je) у 7 [уу [уу у зееоезорвлеб || 7 7 / vyv y у у v viv viv ' |м |м [У sejeseddeo |1 a 7 2 |: vivivio{lviviviviviv I| v oly у зејекџочапа | : Б /|violviviviv 7 sejesoyisseg | | olviviviv 7 eeeoe2ue) || У [у | м | У евеоебиџој || p 7 7 7 (viviviviviviviviv / | у | м | У у Е у DEJE '" | у | У ТУ 7 п / v ‚ | у | у | У 7 7 ШЕ; У | vivi у 7 E JEJE ВЕЛ у У |м |“ | У | У (| viviviviviviviviv / [у v ШЕ; olviviv / [у ВЕ olviviviviv I Уу |м | у | у | у у |У у |+“ у 7 / | У | : p : у | у ујујојују|јо [у У | у [2 |: ov ‚ |У [У о[у [у [о [у [о [о [о [о 1 t У viviviviviviviviv TIERE ME 7 у у [уу [уу [у [у [у [у TEJE ЕВЕ: 7 7 1 p BEDE niviviv гуу ЗЕ ' ум | у | у | У | У | viviviviviviviviv / |м [У | У У |у /|viviolvioilvic olvivivivivic У vily vy v o yiyi vyv o y viv v v 7 7 7 у У | У | У 7 Y у | у | у [у | У 7 РИК | у | у | к | У | У ИЕЛЕ; TEJE У [У ЖЕ] 7 2 [о |“ [2 | |: 1 c у 7 / | У | у 2 у Д Д 4 4 ӘЕӘОРІРШО808801//) тџед [оеш |210џ |2xne зто | eq aep /әгје Daen | Bide hjjeb 10yo Jes ПЕ ЛИЛ 1525 | 2215 | пеш Е)ӘЛ | JE2I "Ацір оша [11А$ |2опе |spex |16ПЕ uws ооло иез [nep 1enb huid |OAW шәй sas Jese ш ов|62|82| 22 | 94] 94 | У 2] 62] 22] 1 2] 04 ее [ез| 2 9 so [so | у ој со | 29 | 2 о | о9| 65 | 89| 2 9 9 5 | ој у ој со | 22 | 2 го | бујоу | 27 | оу [етет 6 өзеш-інр 'penuguo) 71 xipuaddy 187 Nandi et al. Volume 85, Number 1 1998 Combined Cladistic Analysis of Angiosperms У |У ] 4 011 c / с / 2 ЖЕ: 2 eeeoeunug |601 £ У [У E 2 [У әгеоешерород |801 £ aE 7 eeeoeuefunqsens [201 5 у E A EE ] 2|У|У|У ЖЕЛКИРЕЕ у | У PRE У [У у "да 's зејелјеуј |901 d d $ $ у у 1 у у 4 / / әгөзгіеі) |901 4 ] у vor у ЖЕ у у У | У о [у /|У|У|У|У eeeoexig [£01 1 у у (viv eweoeuesjooes |201 4 E E ] ja | У JEI 101 ЛЕРИ 1 : у ' |м | У | У|У У | У У [У у 7 ] 001 [ ] ] A y 4 7 ; [y „|“ | У |У ВЕДЕ: / [У У [У ЕВЕ овеоеухозциоа | 86_ / 1 [26 у | 7 / [у [у у У [У | 56 7 у o[v у | [v up |86. : / ЕЛЕДЕ ] | овестипо |*6 у у у у 7 [2 | |У|У |У /Iviviviviv eeeoeune|3 | £6 ЖЕ; [2] | ЕНЕН | | ewomeuuog [t6- € 9|v|v у | У |< ЖЕЛЕДЕ! EJ o[v [v | | у [и | eeeoeisnig | 16 5 У | У |У | У | ЖЕ |v ІК? | [viviv “6% sejeun | 06 E E У |v У | у | У [У | 7 | У | | ЕЛЕДЕ әгезецбийеуі | 68. E | ЕН | [efe] — | 68 D У | У v c /[viviv | у | | м | У | У | # | У | У | ЕЈ [у | У|У | У ewewupo | ¿8 E a [fv] vis | | EREREREBER | 4121 «o [9s У [viv | | | у | у |у|м |У [v iIviviviv овесеорелеб | 98 E | Le [9] |41 | ЕЕЕ кенені | 7 6_ : E ] '|ү|о| у] ЖЕЛЕ | [ум | БЕС ШЕИ ME E ( ом о |“ | у] оуу |“ [у [о |“ [у [| У [о | [vio] з |28. 5 У [У [У У | У | viv о |у ју | У [2] JEJEJ! sejejoyisseg | ! 8 [e | | SALPLRE [у | г [ ују | ујУ [У] — os. | | [viv ТЕ 171%) Cowell 62 E ЋЕ 7 | У [У [2 | | У | eeeoedeooee3 | 82- : | ЖЕЛЕ | у | КЕ әвәоеоңес̧ | 22. : р 5 у у ууу у Lv] овесеџејебо | 92. 9|: Є У | У [У ВЕДЕ у eeeoemnooe|j | $2 ZI] AK У |У|У v AI өвөоејол [72 < ӘРӘОРІОШІЛІІ c E З : ] у у 7 sejensejeo |21. E E р Е у у 7 7 7 БІЗ E : { у olv о |» ‹ 2|41|У1|< ВЕ: a E р ] ] у rial: ВЕДЕ у |v У | У J у У [У у ЖЕЛЕ : У [У J y ] У | У |У / | У [У |У a E (vv 2|У|У|Ч|У|У x BE Е 5 |: BEBE] ofviviviy E о |“ |“ [у BEBE Е 5 |: У [У У [У 8 / 9 |: у |у |у |у |у [у [у [уу 4 У | У [У у ум |у |у [у [у y Р 5 | / 5 E ] “|у|у|у у [у |у [ууу 1 ум | У |У|М |У B 7 ] у у |У | У|У|У|У viv |у | у|у|у|У|У|У|У у |У|У | У |У E о |: olo[vi|ivi|v 7 viviviviy у | 1 Д Л | JE | z у G у | viv ЖІК? ЖЕ? vIviviz|viv Y E 3 nsa [опш |2 jio эзАш |:ц08 |/8Л12 [105 | uoje ехо |2016 [191 pos jens paars | 0442 [олуи [11198 ТЕ чује [SOJP | aque иби топо | wep ine | еејо [иод |2286 [ори Гори |шео | roue [10:9 рош | JOUR |џео oz ile t (81 (zt H9 (ns E He Het Hz rir Ц01Ц60Ц80,210,90,50/У0Цс0Ц20Ц10,00/ 66 | 96 | 26 | о в| з еее се| 2 6 | | | об |вејее | „ве | ов| зе |ув|сајгв |: 8] о) Ѕзеш' р 'penumquo?) `1 xtpueddy Botanical Garden issouri Annals of the M 188 у ЖЕЛЕ; у c E у у / / z lolo]: п lviviviviviviviviv ШЕ; У | У 7 5 |» : 2/У 7 "EHE c БИЕ: : olviv о | у | У [у У | У JE э [9 |: 7 z : 2 [о [у iviviviv ШЕ; 7 БЛЛИР] E 7 Б olov уу [уу ( / € 2121У у У | У / | у / E E : E z : з |: 7 у БИЕ 7 о [= : : c 7 TE 7 / : c 5 у € у ВЕ / |" м |м [МУ Б РИС E Б әгәог|хоцАіз | 86 у € / / E әғәзешобы) |26 у / c E Е: eeeoepjedeuog | 96 | У |v | E : | ӘЖЕ өгөовиеүедовАцо | $6 - 9 [= : 9eeoeumnp 7 z о |: c Б c р eeeoeune3 | E6 о [у 7 с € E E eee»eneuuog | Z6 E у у E : eeeoeemo |16- о [= 7 c ‚у v v БИК о |: Б "ASS seeur E у iv viv | әвеоецбијеу | 68 € у 2 о | әвәоеи/беѕпрәүү | € c € 7 E z seewupo | L8 Р ЭШЕ: 7 eeecejeoo ie) | 98 G ] | У У [У | әгәзеюрелее | $8- | € о |= | е eeeoejoeedo, | [€ < | c sep;edde) [£8 - э [е [~ ve[o|v 7 E | c c Ov sejequoydn3 |28 1/9 7 7 2 E E c sejelojjissed о vivit 5 7 5 < < овеовоџе) |08 ^ [3 c eeeoeBuuowN к — z € 7 ole E eeeoed;eooee3 | 84 о tc eeeoexes | LL z о [2 v € E зевовџвјебеу | 94. е с a y [3 c |е 4 әеәоегипоэр|-] Буш E € E ЗЕ E c eeeoeoA | PL [ E Е әвәоебшл | Е2 E о Гу [у БИЕ: 5 |: зејелзвје5 | zl 5 |: о viv 9/V|L/O о |: E seepudes | tL : 2 [о | 3 |: 5 E. : : 9/v БИКЛЕ; JE : Ж 5 БИШ; / 9 [= ENE E E 2 |: c E E c 7 < E 7 у / 5 5 E у BE о [= : 5 2 [5 F у у ЖЕ: olv у E 2/У DEJE у DE olv у ом|19 c БИЕ: ЖЕ; * B : JE о [= 5 A | 09- JEJEJE уу [у vivi oi viv 7 7 о [є p 7 у AUS [6S |у [о [у [о [у [у [у [о [у [у [у [у о [о |е у У ЈЕ 9 [= 9 [9/9 c 7 ‚|у | у әееоерџехо У |у [9 [2 | о | у у violvivioa|viailt 7 5 [= 7 о[у [5 өеөоелейшос E И < 2 2 у / 2 d У i< / у 2 E 2. у 2 2 ӘРӘОЕ|ІРШ0О5058017) ELM y [1OX1q |10x8 | 10x8 »pue |урџе | seu иие] оец; | 10x8 | 10: 10X8 | iwn |2Jes ІМЕ| pue |1220 |1220 |1220 pua (зи 39nu qut ueuj uen EL Г жәр у 1282} | 2ЦЕ| 09 je s 185125195 155 и јео YE S rz S пи 105 re y јет ЦІР ЦӘРЦӘРЦРРЦЕРЦЕРЦІРЦОРЦӘСІ ТЕГЕРЕТЕТ Eije crize dre vog тег recia vez ilse е |е гц] || Seu 'penuguo) сү xipuaddy 189 / 7 евеоеџебпазело |101 7 "де 8 sepepew |901 7 aeeoe [901 yol 2 oo 3 b ә « < <<< < olo с! с 91 9 сс < olo 5) (5) (5) с < olola olo « ole о < <<< < <<<) < о < < olo ole < о о1о!» о Combined Cladistic Analysis of Angiosperms Nandi et al. i 3 1 | Џ ololo olo omol, › 5] | --т----а-і (o0 « olo а-ы = = | < < < Se|EJO|JISSP d € Е ; әгәогоше2 « olo < < o « « —4 9) о eones eeeoenunooel|4 о = < olololo'o 3/010 I ееезејод L ` ` o « | зө ololalol "9 E 5 и ојооооо< ЖЫ ШШ ed 3 > 9 E: oio eeeoeqe Volume 85, Number 1 1998 / өгеоешеәк) өеөоејешојоѕѕәе у | " eeeoeJnÁuoels у | у ӘЕӘОЕРЦЕХО) р V | V ГУ | v ET) J E а V TEDE] 1liivivivivio V V 7 7 ӘЕӘОЕ|ЕШО50850/0 nz mu | pide | 2015 | eaj 5 | jee] EX жәр |SS@A |55ӘЛ |SSG^ | pue |55ƏЛ |150 3101$ | SÄLJ | ШОЦ ша! EM Jequ |ед у роом |1}25 | поо юш! о! ns |quie | qwe Dpue pue 00 евге геев етее пи пи сите s 1[98 (со про 1|£ 8 2818108 rez je 21122 192 t| ZH] e је г 122 1]: 2 1021 e SE nes ST РА sieur tip |__| | _|_|_|_|___|_(___|_|____|_-____ ee | Е У әеәоецщіиецәүү | | | | « | | | | | Buba | REEERE RTA | | | ЕНЕҢ | | | | | | | | | | | | | | | | | | < 2 у : y |: AE Д L 2 2 2 L L 2 2 < Vil 1 L eio 2 < tol < E Оа < ОООО! «ОО Oj«|« гропициој “1 xipuadd y Annals of the 190 Missouri Botanical Garden o о olola евезешгв | 601 ОО o eeeoeuejsopog |801 eeeoeueDunqseuis |201 olo "As `8 зөүелүгүү [901 ole 9/v << <|< <<< < еееоеџевјоојес | 201 евезефеоојеј и) |101 eeeoeeejeu/Au; |001 веао eeeoejÁxoJuu3 F. OLE! 3i olol olo w у A oio « Tal al el sol | евеое пл) ole < <|< << “ASS se[eur] ооо эвөоеџХбеѕпрәуң [88 әғәоезеооҝео ДЕН (Ее ем | | sejerquoydn3 olololo Se|eJo|Jissed eee2e2ue о өеөоебиџоу a e I: o «j«j«««x ees әеәзепеүәбб!ы eeo2enno2e|4 o |»... евеоен бил | olo E EZ агеовјод | vs TE: EX зәүел$е|ә2) ololala зеет sejeuin [69 ololo ӘРӨОРІ5АЦООД elo olo liu ole lenll со со olo olo «ol, olo 9/V|1/V V 1 V эчАБ дәи [AJEA |ju09 ЕСЕЕДСЕСЙІ ЕС 062062 2 ӘЛІ | UIUP |: 2092 Tass flues aoe ИЛ 2сјсгсјггсјі г ЕГІ (5 (5) с с L BHO} шәл 0250 у е E БИЛЕ 9210 "ропициој ОЈ ТОГ ЕЈ] | El sieur ир т xipuaddy 191 Nandi et al. Volume 85, Number 1 1998 Combined Cladistic Analysis of Angiosperms ee weg |011 eeeoeiunug |601 eeeoeuejsopod |801 « ololo eesoeuebinqsens |101 у oo (91 9! <<<) < өвөоејАхоЈцйэ әгәоешоби | чаа olo < ве: eveoeuejeqos uu? өеәовеинпгу еееоеицеј3 olola еееоеи о әвәоввооАе евезејорелјеб евеовјоведо | sejeseddeo sejeiqioydn3 Se|eJoO|jisseq olo olololc Cem Guan [1 зәједзејә sejepuides ololo зеен Е о зеје и | 69 өгәоеүзАцоод |89 —евеоеуАчаови7 | 29 эевовюцботцн | 99 —евезејебдод | $9- eeeoeueung ӘРӘОЕ]2Ш0]055 eeeoeinÁu»eis РД / 7 - EE 2 2 ујео 28014 1 un pees 042|6Р 4 L miqo | 1914 NAO [NA Y ZlZ v zl9 y Zls y < у о [тело 12505 пір | Oe} Zie * ZIE y ziz ај vi ӘЕӨОЕЦЕШ08088012 'penuguo) СТ хгриода у Botanical Garden issouri Annals of the M 192 өеөоезриешә;ј |191 етеоејзџешедеј [091 " 6s! 9eeo2eJeldo! est eeeoeosipueg |281 eeeogjemwedg |961 eeeoeueuye] |991 eeeoeuejsie] | #91 евезеџешеу | ESL у ееесепн |291 у 7 2 2 2 2 2 2 2 2 2 2 eee! 191 еееоејергеуовјеа |061 евеоешеху |691 у у у у 2 V / у 2 2 2 y / 7 2 2 у 2 2 2 2 2 2 2 с 2 2 2 2 2 2 2 2 2 V / y |: 2 2 у 2 / / 2 2 V у у 2 Y |< [e 2 2 / 2 у 2 2 2 2 y 2 у sejedesdig |191 евезешојео53 | OF L < < 2 2 2 2 у 2 у 2 2 y 2 2 зәецеіу |661 2 < 2 әваое:обѕоціа | BEL 2 2 у у 2 V / / / / d d әғәоғәц) |461 eesoe io [9€1 әеәоеілезбоеу |661 2 2 у / 2 2 y V 2 2 2 y у 2 у зејеоиз | vel 2 y 2 у 2 2 V 2 2 2 2 2 2 2 әғөоешәоеиес |ЕЕ| " 2 әеәоерішц2у |261 Я евеоелдајо |161 1 у у 2 / / V / / / у 2 у 2 C ѕәјејпшиа |061 у у у / әрәоРоОР1Ю06 |621 / 2 2 V 2 әееовиваз |821 2 d у 2 2 2 у 2 2 2 2 у 2 2 2 121 2 2 7 у евезерцизеј |921 < eeeoe[et 26 |921 е yz! 2 2 2 2 2 2 2 2 2 2 2 2 y |< ў өе есі d d d d d d d d у 1 у у у sepuo) |221 2 < eeeoeseo] | 121 / / / 2 V / / / 2 2 2 2 је еееоешошејод |021 V 2 2 әеәоеџән 4 |611 2 2 / eesoeumesieg |811 d у әеәоешоео |411 2 2 2 2 2 2 2 2 2 2 2 2 у 2 -eeeoeuouiejsou 911 eeeoe»euioouo |611 9eooeoe| ed | #11 eeeoeiudíoe;eg JELL н eeeogoixoixey |211 / < 2 у у / J V / d Д Д 1 Д / y |< sejejeywes |111 ¡Xu ое пезе kyip ојоц [191 10au (¿nu heroj теш шел pies |ејеч [ead Mixer Dii 1010 пур ише [e 95 [0195 es! 95 oles] ov|ec|sc[z2tc sefte re[ose[ez[ez[zz|se|sz|vz|ez[ez|vrz[oz]er|er|zr]s r|sr| ve r|e v [e io oT 8| 2] y] 28] sep `рэпицио?) `1 xipuadd y 193 Combined Cladistic Analysis of Angiosperms Nandi et al. 1998 Volume 85, Number 1 5 / 4 eeeoespuewes) |191 4 етеоезџешелеј |091 4 6st 851 eeeoeosipueg |251 ӘеәоеЈәЮШӘс |96! eeeoeueuye] |991 eeeoeuejsoe] | St етеоеџешеу |661 TE TEDE] /iviviviv евесепн |291 ] ] у 191 eeeoeJepieu»sieg |061 өеөоеиеуү |67 | [у ] Ж [0 | 2 |У 2|У|У|У|У 4 [у [о [у |у [у [у [у |у у 5 E ev! / / АКАКАКАЕ] viviviviv у 171 : у у Е 5 | 9v! у у iv |У|У E У |У|МУ У | a [9 | / sejeuejos |S*! у у у |“ |: о [у | у | хм | У | У | У | | [ү |: | | | ( sejeuenuydosos | РӮ І [о[у ([viviv гу |У |У У|У | [2 | КИЕ) | [2 | | 7 | r | о е е еер | [v] у 7 [уу /Iviviviviv Iv[v у | хм |у |у |У|У | У | У [v| viv | | | | | у < р ЖЕЛІК [ Е] “ | У [У / | У] E ууу [уугу у [У ууу уу | ] | уу уо [у [уу [о |: / o[w|¡vw|vjo|c / |у | у | У | 4 АКАКАКАКАКЛАЕ; у [о[у У |У уу уу [2 5 у д ЖЕЛЕ” у У |У | У |У у у | У у У | У у у хм |У|У|У [ууу у у 7 [у [у [уу [у [уу 7 ИЕ] 7 Е ВЕ viv у У | У у |у |У | У |У у | У 7 7 q E о[у ууу [уу У [У у | У у [у [уу |у [у [у [у / у |У|У|У|У E /|У PEJ ] у у : ] ] (|jviviviviviviv у | У у ВЕ] У |У |У|У у | У | У |< у [ууу [у [уу уо |у TE ojv|v|v|v у У | У | У AE / [о[у /|У|У|УІУІУ у у ДЕДЕ: у у | У DEJE] у у У | У у у У |У|У|У у у у ] /iviviviviviv DE ] у у у о[у АКАКАКАКАКАКЛЕ; ЛЕЛЕ У У |У|М |У у у у eg [ett әгәр! ое, | ELL | евеоемхојхеу |211 vi{vivivivivivi[y о [у [у [у [у уу (ууу mm [у |м у уо {ооуу [у sejejejes | 111 zueq [оеш [210u |гхпе [zno | eq |члер | ezje Daen | Bide |н поро [es ute] unia 1522 |2815 |деш LJAA | JED Ayip | оша [115$ |зопе [spey [1sne juwis 249 иеэ |упер |jenb иш! | OAW | WED |2505 | Jese ог|62|602|22|94|9. |» 2 | ето [ез [ео 2 9 | о9| го | 7 о [со 29 | ro|oo|es|es|zs|o s|ss|es]es[zs|rs]|os|ev|ov|2v|9v|sv|vv|ev|zv|tv]| 9) sie 'penunuo?) 71 xipuaddy Botanical Garden issouri Annals of the M 194 < Р $ / әееоезриешә;ј |191 7 п 7 7 у eeeoejsueuJje| |091 $ әеәоеоѕіриәа |161 у 2 2 евеоејерцјед |961 Z 7 " eeeoeueuye] |991 < eeeoeuejsie] |961 < с 5 евеоеџешеу |661 i 7 / d вевоепн |261 | t d d у V 2 / / 2 4 у eeeoed;e»ouA0) |181 ke / / / eeeocJepieuosie;gg |061 евеоешеху |691 у у у / әгәр. ¿yt < 5 Е t 2 / S |91 5 2 V sejeuejos |6891 : V у зејеџејпудолос | web 2 E] bl е о о о P › « O о « << aja a| < ¡A a a a a егезе ушао [9€1 әеәоетлезбозеуү |661 < ES / 2 2 d 1 1 < V V 2 2 Y |< 2 / у 2 2 sejeou3 [rel < V V V 2 / ЖЕР eeeoemeenjes |££1 2 < V 2 ЖЕР 2 у 2 2 у еееотршпоу |261 < / / 1 LDL | FEL зејепшџа |061 еәоеоеАс |621 eecoeueq3. |821 <<< << < о < > <|<|< << < < « «<< <<< о < | A a < x IIS < A <<< << <|< << <<< << <<< << <<< a < oo ololo о « « « 5 о ~“ ~“ - о olo « « « << << << << / / £ d d d у у 2 у у у oed | ELL ; әгәзезіхоіхәу 211 Д c < < V / 2 d у |< / / Д - x / / у Д sejejeles |111 11883 опш |2 Но Аш [2yds |5. “=== I SSS Ra TOR TSO TSE чује |Soup пае 4——— 2H e |1020 pony |Joue |џе0ј 021161 81 141191 161 EPL LE L LZE LIL ESOL бо цао го 90 tjs o Ро со го rro roor 66|86| 2 6| 9615 61 Ӯ 6[ erizo | т 6| о6| oe 88] 28 vi 15 *8|t8|[28|!8] z; зеш "ур penunuo) *[ хгризаду 195 Combined Cladistic Analysis of Angiosperms Nandi et al. Volume 85, Number 1 1998 2 / At 2 y |" з d / 1 / у 5 овезејрџеше) | |191 ў ” ” " E / с с / [5 eeeoejsueueJe; [091 е / € 1 / V / у у у у A d у у 5 2 у y |< E 2 2 X < X е / / E E у | У [ у ЖІ у у ЖЕ у © |€ 5 £ TO |991 7 7 € 7 eeeoeueisioe] |#91 t / у 7 2 / 7 | | әеөоеџәшеғу |661 у g 9 [= р | | een [291 r 7 eeeoed)eoouA02) |161 c / с ә [= Г eeeoeJepieuosie.g |0681 c у 7 БИЕ E | eewexy [6v E 7 о [ov CAE 2 зајејпоеашео |891 5 ” E Үр eee»uueAuew |191 с t 2 әеәоеоојіш/с̧ |991 © / ЖЕ X 5 зејеџејоб |691 2 / В = с 9 |Е sejeuejnydosos | РТ Р 7 є с sajeuenuey [ЕРТ є ^ с зејтшшоопа |291 E а € ENE < E o зејеоезфа |191 € 2/2 р эсде морзе Ort E у Е TE] £ Ð |€ 5 С | Е ______зејеџеју |661 / i iv ЭШЕ: | c | eeeoejodsonig | SEL / |: 7 € у " с | | eeeoeeu|] | LEL 7 | | авеоеуило СІЗ 2 |: € | әғәогілгіболеуу | SEF < " Es ЕЛЕНА < eeeoeiueoej)eS | EEL е 2 БИЕ 7 5 Я eweoeipiunoy |261 Z / 2 әвәое:щәјоә |161 2 c / / E E c зејепшџа |061 < / / / £ әвөоеові/ 15 |621 < 7 7 Р) с ” Р] с — ezi у E ZI Е 121 7 / э | 7 5 1: авва 921 < с < < 7 Б eeeoejejedoios |621 ЖЕ 2 ЈЕ D а [#21 7 4 Д 2/У р |E 9/9 р eeeoeeDue)pÁH |601 d d ov 5 д с |. sejeuo) |201 d ov 2 L E £ < eeeoeseo] |121 п с с eeeoeiouejod |021 / / 5 E E | t eeeoeuembnoj |611 / у < € с с £ eeeceuesjeg |811 r ” Г. < < 2 әвәовиово| | 211 / у € " < Ё | nby-eeeoeuoujeisou 911 T В g | eeeoe»euouO |911 " с | eeeoeoejÁudejued |911 v{viviviviv ІК у MEME | 7 у | У eeeoeydhoeed |611 / 7 | әғәоеохохәу |21 АКАКАКАКАЕ АЕ DEDEDE] У |У У V Л ЕЈ 5 1 2 |У у зејејеџеб |111 y юхю [30x8 | 10x8 pue рие | Sew |име) cer joxe | 30x8 | unu |2208 ІМЕ| pue |1220 [nooo |1220 [pue [зи ueu jueu ЕД пош щие ET 2yed | ose, | nel 091165 185 125 1199 199 ies LIE S 1125 LIL S 05 LIG y ele y LIZ y H9 y HS He y ЦЕР LZ y ЕРУ LO y | o ПЕРИ ЕТ E SET SHOTLIZZTLIOT LIST LPS LIETLIZT це а! grew nup 'penunuo) ст xipueddy Annals of the 196 Missouri Botanical Garden ееезејрџеше: | |191 oio « olco Ole eeeoeisueuene; |091 әеөое! eeeoeJejdoibejg | 861 <<< << <<< < < < <<< < <<< < < <<< < eeeoeosipueg |261 «j«j« xx eeeogBJeroyeg |991 eeoeueuje] |661 olo olo eeeoeuejsoe] |9651 әеәовиәше |661 әтәоеп |261 <<< <) < < ole о eio olo eeeoedjeoouA0) |161 « eeeoeJepieuosjegg |091 oai. евеоешеху [6*1 FI ву! әеәоеціие, ¿vt olol I < serenely 6€1 eeeoejodsonidg |861 eeeoeeu| |161 eeeoejuAo [9€1 oio одееоелеЈбојеу |661 _ вејезиз [УЕР ӘРӘОРШӘОЕИЕС | EEL olola olo ololo|a о olo о o o о ool, <<< < x <<< a < о <<< << <<<) < < <) <| << oo <<і<|<|< ololo « ola ++ — RÍAS uos јео 61 gei ТЕ rd 6 Lit 6 LJEG 2611: вові ў 2 2 2 V 2 2 INS | јеој | JES] ер 55ел |55ӘЛ |SSOA | рџе [SSA 15$ӘЛ |31035 ег ШОЧ 2 2 2 шаң рел једу zm ЖОМ [1995 | 09 | әш! 8 L|8 8 |28 198 са Lit e LJE e 12811 e 108 LIG Z 1182 1|2 2 1192 L]S Z 17 21 ee < 4 quie | quie pue pue 91189 |291 == 9 Lit 9 291291191 г"репициог) зејејешео |141 sieur np `1 xipuaddy 197 Combined Cladistic Analysis of Angiosperms Nandi et al. Volume 85, Number 1 1998 € L ЖЕ 5 g V Е с 1 | & y У | У|У у eeesejpuewes] |191 ја c S/O] 1 |< 2 с 9/9 1/9 2 у d ў у әвәое}ѕшәшедәј |091 2 t | V c с £ с V / / / у вееоен $ |661 DP 1 t E р с с р |: V 2 у d әтөәог›ә1до!бе|д |8581 € L C E 5 V eeeoeosipuedg |4691 £ TE Е Е V £ 7 eeeogjenueg |961 € V с Є ја [ у 2 ў eeeoeueuje] |991 € € < : а 7 1 / eeeoeuejsoe] | #61 с 1 е £ / eeeoPueueJy |661 5 £ € у 2 2 әгәоепн |261 £ З 7 ЖЕ: әвәовфвоо0оо |68} T THE 5/V|1/9 < < евеоејерецозјеа |091 £z L g евеоешеху [6*1 7 ч z 7 E j = 7 < sejejnu О | srt J < 6 g eesoeyueluan | Leb ON 1 [3 Б 1/9 2 еееоеоојдшло [9*1 Е € € ом 7 ғ зајеџејос | Stl L = V 4 Б зејеџејпудолос | #® І EM € Е Г 1 E sejeuenuey | tpl € 5 [2 sejeiuuoon3 |Z? L E 4 [t g | E | 7 5 зејеоезфа ПІРІ / | ШЕ: E | с Е әвәовшоЦвР253 | 0ғ1 с g | с " sajenely |6ЕІ po AN um 2 É С | 7 L əeasesodsolid |861 а RR E _ E | 7 с 7 |» ææ] |261 3] - | eeeoejuA) |9€1 ге LAS 21 ! [ әеәоетлебоеи [SEI E T [5 / 2 | | sepou3 [РЕТ [3 Є 2/У р с T әвәовшиәовивс |БЕ| 5 t 5 1 2 2 " eeeoeipiunoy |261 5 7 E |. әғәоецәјо [EEL E | 7 5/2 L V Е зејејпшиа JOEL Б E Т Є евеоеое 5 |621 E 1 E £ Е A с с вевовиваз |801 B 1 TE: fe g g 1 је ro eeeoejodes |221 Є C 1 ШЕ 5 X 1 £ / а eeooepinÁoe] |921 £ £ 1 Ё 5 с eeeoejejedoJÁos |621 с AE TE Ж ШЕ: |I | | eesceisuadeig |%01 КЕЕ < € c о |» oo viv. Го | v омо | | 7 peeseabueiphy |621 € 9/V|1/9 ] oo) V | V | [у | У jus ERE [5 | ____зејешоо |221 с D/V|1/9 < 9/NV|9/V|9/N D/V © I | eeeoeseo] |121 mE с E у | 9/9 mE aesoejoweajoy |021 ЭШЕ: с 5 < 5 | eeeoeuembnoj |611 1 E © L аевоешшезјед |811 n 9 ||] 5 р [2] L E 5 __ әеәоеиоеој |111 1/9 £ 5 © |2/У L c nby-eeeoeuouejsoueuds |911 € € с и 9 L је | гевоеоешозио |911 EN 1 9 |] c 5 L eeeoeoe|Audeiueg [ett c E J J 2 Е у D 5 евезенудАјоејеа [£11 1 [e / © L < eeeoeoixoxey |211 с 9 || : п/у |у О/У 95 о | у | У L9 у sees [tb Abus |udns |1 ез [Јело ез |;uAD | 198U лел | JUD |edxe | әли | ue "ДОС |1u82 |1u89 [wos эшщ |2UBA [21190 eS | eli эшщ | yOu |шпи |e: = У оем | шоо |'UOJ8 | e10} шәл = иел 10/40 [Oes opceccjeeczeziot ет ФЕСЕЕС2Е21162|0646204822):20920522|Р20162212201122020|612|812 aa t L |121 2]! до 26 0280 огјзогјоогј тог јео 2 02] 02] oz gieur' [np 'penuguo) СТ xipuaddy Missouri Botanical Garden Annals of the 198 9/2 L V g 9 eeeoe)ypueueJ| | 191 КЛЕ у ә јо у уу 5] евезејзџешедеј |091 9 Liviv L © әтәогібен 691 у |2 | Vv Е 9 eeeoejeidoi 851 L V eeeoeosipueg |461 “wilitoltwiletletototvin € евеоезетијед [991 ШЖ ЕИ ЕЖЕ ЖЕ ЖЕЛ То [2 у 15 [у |] | eeeoeueuje] |6681 < € eeeoeuejsoe] |961 еі [у [е [оо Те | у] у [у] о] саманнан [ESE " с вевоепн |261 .v[1]9j|v|v[oo[v[| ' [У ] 2 | eeeediezouÑo) | 161 7 Е с | с eeeobJepieuosie)g | 061 КЕЕ E "у [о [2 Ту ји viso] әгесешеуу [6*1 Div! € зегејпи ev! ЕК 1 | F го [1]9 То vyl] eeeoeuiue 191 i | 2111: ! | v Tio! Seen: s [oer жер E | | |9/V| € sejeuejos | Sel с L lo/v € 9 | зејеџејпудоо | PPL RILSRAESRSLSERSSEARALREA ант Eg 2 | ЖЕ -- e] seje'uuoon3 [| 271 | < Б sejeoesdig | Lvl ЕЕЕ: sE T | | BE | | aesoewoyeos3 |091 | 75 7 T Te Te ѕәје!еу |6€1 ES пера о г | 5-1 | eeeoedsong [861 72191: ZEN ШЕ | | o] әвеоеәці |161 } 1. d _ | EE HE с SEXE | [5 egeeogjjuÁ) |961 L 1 әғәоріЛЕІбо/еу) | 661 ERER ppe эшш | о |1/9 евеоешеоејес J EEL E AERE | GE гуу [у [у әвәзеірішроу [eet Е: | | 9 |: wk rlols әгезелдәр) |161 є © зејејпшџа | 061 К | о | ERE НЕШЕ lylo] ____евеовое лао [eet 7У|2|4|9|У|2|12|9121У1|У19)| әкеме |821 _У | |5 21) КЖЕ (9| DEJES 9 121 5 евезерцизеј |921 "“ Г2 13 уо | ә tit. DEAE евезеје! 621 с с [3] t [24 AEREAS ЖЕЗЕМЕЛЕЛЕИЕ: ке, у 1/9 sepu) |221 _Ујојујл|у|9 |]. > | 1 јо ју |У five eeeoeseo] |121 2/2 1 © еееоешошејод |021 ojo[v|i]| 1/9 eeeoeuemnbno3 [611 C L 9 eeeoeumesieg |811 “Ұ|1|2|уУ E гу [ә eeeoeupeo| |211 —AA2A+ == 1 9 [nby-eeeo»euouersou 911 1 £ с [5] еееоезешооио |611 2|91У|91:% Е о [о[у |у [у [ә өеөоеоеј ey |ті 5 / | © евеотцда јоејеа |ELL Lic / € 1 өәтөзто!хо}ухәу |211 “wilifofily 2/V| C vivio seres |111 1000 ип; |эөө |1140 | JOH {пло [NAO | тело | JONS грр ©5218 846% [117] * b Zit v ZIE y Ziz y ZlL y 12 $зешир penunuo? СІ xipuadd y Volume 85, Number 1 Nandi et al. Combined Cladistic Analysis of Angiosperms Appendix 2. Taxon circumscriptions. 1) Ceratophyllaceae 2) Chloranthaceae 3) Nymphaea- ceae 4) Amborellaceae 5) Austrobaileyaceae 6) Illiciales (Schisandraceae, Illiciaceae) 7) Winterales (Canellaceae, Winteraceae) 8 8) invi кеј 9) Aristolochianae (Sau- Aristo iaceae, Lactoridaceae) aceae, “Мос 'eae, Hernandi- Gyr ; Trimeniac OR ded in меча peo 11) Ма меде 12) án nnon (Eupomaticeae, Annonaceae) 13) Magnoliales (Himantan- draceae, Degeneriaceae, Magnoliaceae) 14) Ranunculidae = б E idaceae, Papaveraceae) 15) Eupteleaceae 16) Platanaceae 17) Nelumbonaceae 18) Trochodendrales (Tetracentra- ceae, Trochodendraceae) 19) Proteaceae 20) Buxaceae 21) Sabiaceae 22) 2 23 h аааз у ieu ~ ~ = © E @ = i=} e e о o % кч — ceae, Cactaceae, Didiereaceae) 29) Asteropeiaceae 30) Plumbaginaceae 31) Polygonaceae 32) Dioncophyllaceae 33) Ancistrocladaceae 34) Nepenthaceae 35) Droseraceae 36) Tamaricaceae 37) Frankeniaceae 38) Hamamelidales (Altingiaceae, Cercidiphyllaceae, Daphniphyllaceae, Ha- mamelidaceae) 39) Dilleniaceae 40) Vitaceae 41) Eucry- (Ени eae n н С aceae, Francoaceae, Parnassi- aceae, and Lepuropetalaceae) 48) Staphyleaceae 49) = Ha нв 50). пла. 51) harinas eae 5, Faganae (Nothofagaceae, Fagaceae, Balanopaceae, B 5. то 22. Rhoipteleaceae, Juglandace- e) 53) Cucurbitales (Datiscaceae, Begoniaceae, Cucur- iaces) 54) Coriariaceae 55) Urticales (Ulmaceae, Mora- eae, Cecropiaceae, Cannabaceae, Urticaceae) 56) WD и а ae 57) Connaraceae 58) Oxalidaceae 59) Stachyuraceae 60) Geissolomataceae 61) Geraniaceae 62) Melianthaceae 63) Fabaceae 64) Surianaceae 65) Poly- galaceae 66) Rhizophoraceae 67) Zygophyllaceae (incl. aceae) 68) Trapaceae) 70) Rutales кетесің Simaroubaceae ithout Picramnioideae and Alvaradooideae, Rutaceae Moliacese: Cneoraceae) 71) Sapindales (Aceraceae, Hip- pocastaneaceae, Sapindaceae) 72) Celastrales (Goupi- aceae, Celastraceae, Stackhousiaceae) 73) Irvingiaceae 74) Violaceae 75) Flacourtiaceae (Flacourtieae, Oncoba, Homalieae, Scolopieae (without Banara), Casearieae; (асеае Appendix 3. Characters and character-states. Serology 1 serological reaction with Nelumbo antiserum (1. group) 2 serological reaction with Nelumbo antiserum (2. group) 3 serological reaction with Victoria antiserum 4 serological reaction with Saxifragaceae antiserum 5 serological reaction with Hydnocarpus antiserum 6 serological reaction with Passiflorales antiserum (1. group) 7 serological reaction with Passiflorales antiserum (2. group) without Aphloia, Soyauxia; latter two taxa not included in id dimi 16) Kiggelariaceae {Flacourtiaceae with cyclo- eny lic ogenic compounds rythrospermoideae, ~ ceae, Capparaceae, Brassicaceae) 84) у пи еае 85) Salvadoraceae 86) Caryocaraceae 87) Осћпасеае 88) Me- dusagynac eae 89) Malpighiaceae 90) Linales s. str. (Hu- goniaceae, Linaceae) 91) Clusiaceae (incl. Hypericaceae) 92) Бопе аве 93) Elatinaceae 94) Quiinaceae 95) phaerosepalaceae 100) Thyme- laeaceae 101) Dipterocarpaceae 102) Sarcolaenaceae 103) Bixaceae 104) Cochlospermaceae 105) Cistaceae 106) Malvales s. str. (Tiliaceae, Sterculiaceae, Bombacaceae, — 107) Strasburgeriaceae 108) 2. 09) Bruniaceae 110) Balanophoraceae 111) Santalales 1. Opiliaceae, Santalaceae, hace 'eae, Vis- caceae, Eremolepidaceae) 112) Aextoxicaceae 113) Par- acryphiaceae 114) Pentaphylacaceae 115) Oncothecaceae ud ши (including Sphenostemon) 117) Icaci- e 118) Balsaminaceae 119) Fouquieriaceae 120) Po- pecia 'eae 121) Loasaceae 122) Cornales (Alangiaceae, Nyssaceae, Davidiaceae, Mastixiaceae, Cornaceae) 123) e 124) Diapensiaceae 125) Seytopetalaceae 126) Lecythidaceae 127) Sapotaceae 128) Ebenaceae 129) Styracaceae 130) Primulales (Myrsinaceae, Theophrasta- ceae, Primulaceae) 131) Clethraceae 132) Actinidiaceae 133) Sa daceae, Erica- c tosporaceae 139) Araliales (Araliaceae, Apiaceae) 140) Escalloniaceae 141) таана (Adoxaceae, Sambuca- ceae, Caprifoliaceae, Viburnaceae, Dipsacaceae, еки 57. = Eucommiales (Eu Майса 'eae, Lamiaceae) 145) за ракун (Convolvulaceae, Boraginaceae, Hydrophyllaceae, Solanaceae, Nolanaceae) 146) Symplocaceae 147) Menyanthaceae 148) | lales (Goodeniaceae, Brunoniaceae, Calyceraceae, Stylidiaceae, Asteraceae) 149) M 150) Бенікгішейділіміседе 151) Corynocarpaceae 152) Hu- rameriaceae 154) Lacistemataceae 155) Leit- ) Pellicieraceae 157) Peridiscaceae 158) Plagiopteraceae 159) Scyphostegiaceae 160) Tetrameris- 161) Tremandraceae A: absent; C: present A: absent; C: present A: weak or absent; C: present A: absent; C: present 200 Annals of the Missouri Botanical Garden 8 serological reac tion with E “uphorbiac eae antiserum (1. group) : absent; C: present JO 7. % 4 ~ © зз ж ы ii a 157) = 5% m o = =] z e i ul mtr Ss “= = =" = © 4 е, "E = 5 с O £z = z =. т. Ed M 3 3 N: ys ~ = c = a c x > de с Е- 5 T O - = z O "2 = Ф 4 % = -— = : abs - : weak or absent; C: present : weak or absent; C: p weak or absent; C: present weak or absent; C: present weak or absent; C: present : weak or absent; C: present 14 serological reaction with Primulales antiserum 15 serological reaction with Theaceae antiserum l6 serological reaction with Hydrangeaceae antiserum тйс йн т т Chemical compounds 17 Al accumulation A: absent; C: present 18 amides A: absent; C: present 19 dhurrin A: absent; C: present 20 proteacir A: absent; C: present 21 triglochinin A: absent; C: present 22 taxiphyllin A: absent; C: present 23 proacacipetalin A: absent; C: present 24 heterodendrin A: absent; C: present 25 cardiospermin A: absent; C: present 26 valine- and isoleucine-derived cyanogenic compounds A: absent; C: present 27 linamarin A: absent; C: present 28 lotaustralin A: absent; C: present 29 linustatin A: absent; C: present 30 neolinustatin А: absent; C: present 3l tyrosine-derived cyanogenic compounds A: absent; C: present 32 prunasin A: absent; C: presen 33 sambunigrin А: absent; C: present 3 zierin A: absent; C: pres 35 holocalin А: absent; C: presen 30 glucosinolates A: absent; C: present 37 dihydrosterculic acid A: absent; C: pre 38 acetylenes A: absent; C: present 39 eleostearic acid A: absent; C: present 40 myristicin A: absent; C: present 4l asarone A: absent; C: prese 42 sesquiterpene lactor A: absent; C: present 43 died rane- like and A: absent; C: present myo tol A: absent; C: present 45 pin A: absent; C: present 46 ene hitol A: absent; C: present 47 deutzioside A: absent; C: present 48 cantleyoside A: absent; C: present 49 cyclopentenylic cyanogenic glycosids A: absent; C: present 50 ae d compounds A: absent; C: present 51 ust ке ы A: absent; C: present 52 in A A: absent; C: present 53 euc 'ommin м A: absent; C: present 54 и ШЫ A: absent; C: present 55 sn resinol A: absent; C: prese 56 a des 'ubebin А: absent; C: present 57 galbacin A: absent; C: present 58 licarin А A: absent; C: present 5 veraguensin A: absent; C: present 60 prodelphinidins A: absent; C: present 6l ellagic acid A: absent; C: present 62 methylated ellagic acids A: absent; C: present 63 stachyurins A: absent; C: present ) casuaricitin A: absent; C: present 65 tellimagrandin I A: absent; C: present 66 tellimagrandin П A: absent; C: present 67 pedunculagin A: absent; C: present geraniins A: absent; C: present 69 chlorogenic acid A: absent; C: present 70 рае acid A: absent; C: present 7l epigallocatechin-3-gallate A: absent; C: present 72 flavonoid sulphates A: absent; C: present Volume 85, Number 1 1998 Nandi et al. 201 Combined Cladistic Analysis of Angiosperms 73 — 74 davidige 75 biflavonoids or biflavanoids ins c 87 indole . 88 iridoid c 89 scold eae com 91 an Ds acid & derivatives 92 arjunolic acid & derivatives 93 maranes 94 cucurbitacins 95 nigraci 96 utin 97 naphthoquinones 98 гарапопе 99 plumbagin 100 агоѕег 101 bud 102 phenanthrenes 103 acetophenones 104 actinidine Characters at cellular level 105 nitrogen-fixing nodules 106 chromosome n x= Тогп=богп=8 107 sieve-tube plast 108 ерісшісшаг bi s waxes stratified 109 epicuticular leaf waxes rod or tube shaped 110 epicuticular leaf waxes arranged in rosettes 111 SiO2-bodies in wood or leaf 112 oxalate druses 113 elongate oxalate crystals 116 sphaerocrystals 117 myrosine cells 118 oil cells 119 mucilage cavities or cells 120 resinous cavities or cells 125 nonglandular 2-5 armed hairs 126 glandular scales Embryology 127 anther tapetum 128 microsporogenesis 129 pollen organization 130 type of triaperturate pollen 131 polar pollen oo 132 triangular polle 133 triangular n. ora deepened absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present : present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present God € 433333320741 eo Печ БЕ 74-4 3 4229-8372 2 3:13 5- un Ф > о absent; C: present absent; C: present P-type; C: S-type absent; C: present absent; C: present absent; C: present absent; C: present nt; C: present A: absent; C: raphides; absent; C: present absent; C: present : present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present absent; C: present >БРРРРРЕРРРРРРРОРРРРРРРУ> ы m 3 | D [2] El С» oeboid; C: secretory ^ дељење s simultaneou А: mme te; C: ө сең pm G: triaperturate; T: po- А: not la C3 em or polycolpate; G: tri- or polycolporate; T: trip A: less than 20p; C: 20-301; G: more than 30p. A: absent; C: present A: absent; C: present 202 Annals of the Missouri Botanical Garden 134 angulaperturate pollen absent; C: present 135 sexine texture A: psilate or granulate; C: spinulose; G: reticulate; T: striate 136 type of reticulation A: not reticulate; C: finely reticulate; G: а EE 137 integument number O C: uniteg 138 integumentary tapetum resent 139 nucellus type A: crassinucellar; C: weakly такива Лас С: tenuinu- сеПаг 140 perisperm ог nucellus-derived surrounding tissue А: absent; C: present 141 endosperm development A: nuclear; C: cellular 142 Caryophyllad type of embryogeny А: absent; C: present 143 Piperad type of embryogeny A: absent; C: esent 144 Asterad type of embryogeny A: absent; C: present Seed anatomy 145 end of ovular or seed vascular bundle A: chalaza; C: beyond chalaza, far from micropyle; G: near micropyle 146 aril A: absent; C: present 147 la A: absent; C: present 148 sarcotesta A: absent; C: present 149 ruminat 08 A: absent; C: present 150 exotestal hairs or papillae А: absent; C: present 151 exotestal palisade A: absent; C: present 152 theoid exotesta thickenings А; absent; C: present 153 exotesta tanniferous or with brown contents : absent; C: present 154 mesotesta A: unspecialized; C: sclerenchymatous or thickened walls 155 endotestal crystals A: absent; C: present; G: undefined 156 endotesta A: unspecialized cells; C: elongate cells; G: lignified, not elongate; T: tracheids 157 exotegmen A: М. cells or with lobate facets; C: sclerified or tracheidal; G: fibrous or tangentially elongate; T: as palisade layer 158 exotegmen with lobate facets A: absent; C: present 159 bixoid к in chalazal region A: absent; C: Enn 160 hypost A: absent; C: present 161 dor fae type A: oil or proteins; C: saroh; | ылы uon T undefined 162 Rr ei haus t; C: present 163 embryos A: less than half seed ree C: pw than in A, endo- sperm copious; G: bigger than in A, endosperm scanty; T: no endosperm 164 embryo form A: straight; C: curved 165 suspensor haustoria : absent; C: present Stem morphology and anatomy 166 growth form A: no vine; C: vine or creeping axis 167 dispersed vascular bundles A: absent; C: present 1 nomalous secondary growth A: absent; C: present 169 phloem stratification A: absent; C: present 170 wedge-shaped phloem rays А: absent; C: present 171 internal phloem A: absent; C: present 172 cortical vascular bundles A: absent; C: present 173 — sclerenchymatous idioblasts in cortex or pericycle : absent; C: y 174 wood parenchyma A: scanty or absent; C: diffuse; G: aggregat 175 fiber wall A: thin to moderately thick; C: thick to s thick 176 fiber eee A: absent; C: present 177 trachei A: absent; C: present 178 libri iform fibers bsent; C: present 179 ray-type : heterogenous type I; С: 1... type Па; heter- ogenous type IIb; T: o pu 180 homogenous multiseriate rays A: absent; C: present 181 rays maximally biseriate Á: absent; C: present 182 storied wood structu A: absent; C: diis 183 vestured pits in 5 side walls : absent; C: prese 184 vessel side pitting A: circular, only one or two e C: 2. or tran- sitional; G: opposite; T: alte 185 end wall perforation of pit A: tracheids, no perforation; с: poa йыш G: mixed sca- ariform and simple; T; simple; 186 vessel end wall angle A: highly oblique; C: slightly oblique; G: horizontal Volume 85, Number 1 Nandi et al. 2 1... Cladistic Analysis of Angiosperms 187 vessel shape in transverse section 188 vessel aggregation 189 dendritic pattern of vessels Leaf characters 190 leaf traces 191 leaf arrangement 192 stipules 193 glands on distal petiole 194 leaf organization 195 stoma type 196 epidermal crystals 197 mucilaginous epidermis 198 palmate venation 199 craspedodromous venation 200 leaf teet 201 salicoid teeth 202 chloranthoid teeth 203 kranz structure 204 leaf sclereid 205 vein terminating foliar sclereids 206 foliar tracheoids Floral and fruiting characters 207 elongated floral base 208 cortical and axial vascular bundles in floral base 209 bracts instead of perianth or bract-like perianth 210 К-С differentiation 211 petal or tepal aestivation 212 number of calyx or tepal organs 213 high calyx or tepal number 214 trimery in calyx or tepals 215 sepal union 216 petal number 217 petal union 218 scales on upper side of petal 219 perianth or bract to stamen outline change 220 variation of isomerous patterns 221 trimery in androecium 222 івотегу in androecium and perianth 223 centripetal polyandry 224 centrifugal polyandry 225 polyandry associated 2 ш stamen pairs 226 anther to stamen length-ra 227 inverted anthers 228 expanded stamen 229 connective tip 230 valvate anther dehiscence 231 disk 232 gynoecium position 233 carpel number I 234 carpel number II 235 сагре! number Ш 236 ovary to e length-ratio 237 carpel un 238 degree of synca 239 stipitate free ие or stipitate unicarpellate fruits 240 stigmatic crest А: па. C: slightly angular; С: ov А: more than 85% solitary; C: less 5 85% solitary; С по vessels А: absent; C: present А: one; C: three; С: more than three А: alternate; с opposite; С e whorled absent; C: present A: absent; C: present A: и ан C: idis A: absent; C: pres absent; C: three; G: two or E T: T other M : not сита C: discontinuou imbricate; C: contorted; С: ан T: open ог petals/ tepals absen A: zero; C: three or more than five; G: four or two; T: five A: absent; C: six to ten; G: eleven to twenty; T: more than 'nty dia A: absent; C: present : absent; C: present А: none; C: four or two; С: m T: six to ten absent; C: present absens C: present А: discontinuous; C: continuou A: no isome a percal G: two or n isomerous or doubled 4. T: obhaplostemo опу А: аЬ : ргевеп! : present : present : present abse : present A: more than half; C: less ^e half A: absent; C: present A: absent; C: present A: not prolonged; C: prolonged not as in G or T; G: mem- branaceous; T: expan j or massive : absent; C: pen: ^ pis C: pres A: not antepetalous (or antetepalous); C: ac RE or antetepalous, not oblique; G: oblique : one; C: two; G: three; T: more than three А: three or less; C: four; С: five; T: more than five А: five or чана C: six to ten; С: eleven to twenty; Т: more than А: Бане 1:2; С: 1:2 to 1:3; С: NN A: unicarpellate or totally apocarpous; C: o n ie G: partially styles free or partially fused; T ncs fully А: ne totally syncarpous; C: Шу dex А: absent; C: pre А: да. C m 204 Annals of the Missouri Botanical Garden 241 placentation А: marginal or laminar on apocarpous carpels; C: apical; >: axile, free central or basal; T: parietal 242 diffuse placenta A: absent; C: laminar diffuse; T: protruding diffuse 243 stigma decurrent А: absent; C: pres sent 244 ovary position А: superior; C: infer 245 ovule to carpel number A: less than one; C: one; G: two; T: more Bend two 246 ovule curvature A: orthotropous; C: anatropous or campylotropous 247 micropyle formation A: outer or both integuments; Са inner integument; С: by the only integument; T: no integuments 248 obturator T absent; C: present 249 seed to m number A: less than one; C: o : two; T: more than two 250 fruit ty A: dehiscent fruit; C: nid 251 type of ect fruit A: follicle, pod or ventricidal capsu ale: C: capsule types other than in A and T; G: septicidal ни T: inde- hiscent fruit or schizocarp 252 central column in fruit А: absent; C: present Data errors in non-molecular matrix that could not be corrected un ла 129, А/С instead of А (we became aware of Sampson, 1993, too late to include this poly- morphism); Myristicaceae: character 251, A instead of С; Fabaceae: character 105, C instead of A; Myrtales: character 87, C instead of A Appendix 4. Character definitions. Characters and character-states requiring further explanation are given below. Characters on cellular level 122 fasciculate or stellate hairs : definition following Metcalfe & Chalk, 195€ 123 peltate scales : definition following Metcalfe & Chalk, Qs 124 dendritic hairs definition following Metcalfe & Chalk, 1950 125 nonglandular 2-5 armed hairs : definition following Metcalfe & Chalk, 1950 126 glandular scales : definition following Metcalfe & Chalk, 1950 Seed anatomy 152 theoid exotesta thickenings : exotesta showing thickenings on radial and inner tan- gentia Ü 4. but not on outer tangential walls (cf. Hu- ber, 1 159 bixoid exotegmen in chalazal region : seeds i exotegmen organized as palisade layer, the latter showing a typical inward curving in the chalazal region associated with a hypostase plug differentiated into core and annulus region (see Nandi, 1998) Stem morphology and anatomy 175 fiber wall : thin to moderately thick means sum of wall-thickness is smaller than fiber lumen diameter Leaf characters 199 craspedodromous venation : secondary venation running into leaf teeth 201 salicoid teet : leaf teeth showing a proximally rounded hyaline gland with concave gland body in herbarium specimens; see also as & Hickey, 1975 202 chloranthoid teeth : definition following 2. « |“ ‘key, 1975 205 vein terminating foliar sclereids ы definition following Rao, 206 foliar tracheoids : definition following Rao, Yen Floral and fruiting characters 225 polyandry associated with outer stamen pairs : definition following Ronse Decraene, 1992 228 expanded stamen : thecae of anthers widely == or stamen laminar 230 valvate anther dehiscence депин following Endress, 1994 gynoecium position efined as antetepalous if one Rm is in line with a median tepal or petal 240 stigmatic crest : broad, decurrent stigma, deeply furrowed into two parts 242 diffuse placenta : definition following Endress, 1994a Volume 85, Number 1 Nandi e 2 2. М: Analysis of Angiosperms 243 stigma decurrent : stigmatic surface unilaterally running down more than three times the stigma lobe breadth Appendix 5. Procedures of character-state assignment. The procedures for assigning character-states to matrix fields are indicated in the overview given below. In characters that are not listed, presence was favored over absence (monocots: presence only favored if occurring in Acorus, Arales, Arecales, or Alismatidae; the restriction to the presumed basal monocot clades reduces parallelisms). Serology Characters 1-16: strongest serological reaction > [“>” means favored] Chemical compounds Characters 17-114: Presence of a chemical compound > absence Characters at cellular level 107 sieve-tube plastids ‚ч As equally >, character-state of presumed basal s > in ачаг Magnoliales, Laurales Embryology 127 anther tapetum Both types equally > 128 тісгоѕроговепеѕіѕ Both types equally > 129 pollen organization Character-state of presumed basal members > 130 type of Es ad ри Character-state of presumed basal members > 131 polar pollen diam least polar pollen diameter > 135 sexine texture Character-state of presumed basal members > 136 type of reticulation Character-state of presumed basal members > 137 integument number А> С 138 integumentary tapetum Character-state of presumed basal members > 139 nucellus type А 141 endosperm development Both types equally > Seed anatomy 145 end of ovular or seed vascular bundle G>C>A 154 mesotesta CA 156 endotesta T, G, and C >A 157 ехоіертеп T, G, and C > A 161 endosperm storage type АП types equally > 163 embryo size A > 164 embryo form A Stem morphology and anatomy 166 growth for Both types equally > 168 ап ia secondary growth Character state of presumed basal members > 171 internal phloem Character state of presumed basal members > 174 wood пен A>C>G 175 fiber wall А> С 176 fiber septation AC 178 libriform fibers Absence > presence 179 тау-іуре A>C>G>T 180 homogenous multiseriate rays Character state of presumed basal members > 181 rays maximally biseriate Character state of presumed basal members > 184 vessel side pitting А ап С> GT 185 end wall perforation of pit A>C>G>T 186 vessel end wall angle А>С>6 187 vessel shape in transverse section A>C>G 188 vessel aggregation G>A>C Leaf characters 190 leaf traces Character state of presumed basal members > 191 leaf arrangement Character state of presumed basal members > 194 leaf organization oth types equally — 195 stoma type Character state of presumed basal members > leaf teeth Character state of presumed basal members > Floral and fruiting characters 209 bracts instead of perianth or bract-like perianth Character state of presumed basal members > 206 Annals of the Missouri Botanical Garden K-C differentiation 211 petal or tepal aestivation 212 number of calyx or tepal organs 213 high calyx or tepal number 214 trimery in calyx or us 215 sepal union 216 petal number 217 petal union 219 perianth or bract to stamen ш change 220 variation of isomerous pattern 22] trimery in androecium 222 івотегу in androecium and perianth 223 кош polyandry 224 centrifugal polyandry 226 anther to stamen length-ratio 229 connective tip 231 disk 232 gynoecium position 233 carpel num 234 carpel number П 235 carpel number I 236 ovary to carpel length-ratio 237 carpel union 238 degree of syncarpy 244 ovary position 245 ovule to carpel number 246 ovule curvature 247 тісгоруіе formation 249 seed to carpel number 250 fruit 251 type of dehiscent fruit Appendix 6. Sources. Literature used for taxon delimitations and for finding character-states. Taxon delimitations Albert et al., 2; erson et al., 1994; Baas, 1972, 1975; Carpenter 3 Dickson 1976; Chase & Swensen 1995; Chase et al., 3. 1995; Conti et al., "1993. Corner, 1976; Cronquist, А de Dahlgren, 1980, 198 Dahlgren & Clifford, 1982; Dahlgren & Thorne, 1984; Dahlgren et al., 1985; Downie & Palmer, 1994; Duvall et al, 1993b; Engler & da (eds.), 1887-1914, 1924— 1995; Fernando et al., 1995; 1. et al., 1992: Geetha et al. 1993; . & Bre 995; Hegnauer. 1962-1994; Huber, 1991; Hid 1964/1967, 1973; Kolbe & John, 1979a, b; Kron & Chase, 1993; Lemke, 1988; Melchior (ed.), 1964; Miller, 1975; Morgan & Soltis, 1993; өче et al., 1992, Ke ng & Palmer, 1993; ; Rodman et al., 3; Savolainen et al., , 1995b; a f 1995; Swen- 4; Takhtajan, 1966, 1987; а 1983, 1992; van as 8 Baas, 1984; Xiang et al., Serology Fairbrothers, 1966; Grund & Jensen, 1981; Hillebrand & Fairbrothers, 1966, 1970; Jensen & Greven, 1984; John & Kolbe, 1980; Kolbe & John, 1979a, b; Simon, 1970, 1971; Vogel, 1986 Chemical compounds Arora & Metha, 1981; Barron et al., 1988; Bliss et al., = A à Character state of presumed basal members > Character state of presumed basal members Character state of dome basal members > Character state of ти basal members nce > presenc Character state of presumed basal members > presenc Character state of presumed basal members ер - Б с [e] = Ф ч © E Ф © =, "2 5 c Ф = 5 = Ф E Ф B 3 O a - Ф = wn VVVVV Character state of presumed basal members > GandC >A Character state of presumed basal members > A>C Character state of presumed basal members > Character state of presumed basal members > Character state of presumed basal members > A> C ; A>C>G>T > С Character state of presumed basal members > Character state of presumed basal members > А > С Character state of presumed basal members 2 Character state of presumed basal members > T Character state of presumed basal members > A>CandG>T 1968; Bohm & Chan, 1992; Brüning & Wagner, 1978; Crossley & Djerassi, 1962; Deyama et al., 1985; Durant & Zenk, 1974; Filho et al., 1985; Fieser & Chamberlain, 1948; Gibbs, 1974; Gildemeister & Hoffmann, 1956; Har- iid 1969; Harborne & Baxter, 1993; Hayashi et al., 0; Hegnauer, 1962-1994; Keller, 1982; Lavault & 1. 1980; Le Quesne et al., 1980; Lebreton € Bouchez, 1967; McAlpine et al., 1968; Murai et al., 1985; Rao & Alvarez, 1982; Sethi et al., 1976; Sévenet et al., 1971; Smith et al., 1980; ras zs 1966, 1970; Uesato et al., 1986; Zenk et al., 1 Characters on cellular level Baas, 1972, 1975, 1984; Baas et al., 1979; Behnke, 1975, 1977, 1981, 1985; Carpenter & Dickison, 1976; Cron- quist, 1981, 1983; Dickison, 1978, 1981, 1990; Dickison & Baas, 1977; Ditsch & Barthlott, 1994; Ehrendorfer et al., 1984; Engler & Prantl (eds.), 1887-1914, 1924—1995; Fehrenbach & Barthlott, 1988; Franceschi & Horner, 1980; Goldblatt & Dorr, 1986; Goldblatt & Johnson (eds.), 1981/1984/1985/1988/1990/1991/1994; Gottwald & Par- ameswaran, 1966, 1967, 1968; Hennig et al., 1994; Hu- ber, 1991; Hutchinson, 1973; Keng, 1962; Metcalfe, 1956, 1962, 1987; Metcalfe & Chalk, 1950, 1988/1989; Miller, 1975; Proctor, 1955; Puff & Weber, 1976; Ricci, 1957; Schmid, 1964; Solereder, 1899/1908; Solereder & Meyer, o = & McKey (eds.), 1994; Theisen & Barthlott, 1 4. Baas, 1972; Barth, 1965; Batygina et al., 1985a, b, c; Bhandari, 1971; Boesewinkel, 1985, 1994; Boesewinkel Volume 85, Number 1 1998 Nandi et al. 2 Combined Cladistic Analysis of Angiosperms & Bouman, 1980; Carpenter & Dickison, 1976; Chiarugi, 1925; Chiarugi & Francini, 1930; Chopra & Harjinder, 1965; Corner, 1976; Cronquist, 1981; Davis, 1966; Dick- ison, 1979, 1981, 1986, 1990; Dickison & Baas, 1977; Dickison et al., 1982; Endress, 1993a, b, c; Erdtman, 1952, 1958; Carla, 1993; Gutzwiller, 1961; van Heel, 1967, 1984; Heo & Tobe, 1994; Hideux & Ferguson, 1976; Huang Tseng-Chieng, 1972; Huber, 1991, 1993; Huynh, 1969; соси 1966; Johri, 1970; Johri & Kak, 1954; Јоћп et al., 992; 2. et al., 1983; Kapil & - 1991; Ма 1965; RE. ur, 1969; Keating, 1972, 1975. Köhler. 1994: Kubitzki, 1993a, b, c; Les, ‚ 1993; Maguire & Ash- ton, 1980; Mauritzon, 1935, 1936; Maury et al., 1975; Melchior (ed.), 1964; Netolitzky, 1926; Philipson, 1993; ance, 1968, 1972; Puff & Weber. ‚ 1976; utishauser, 1985/1988/1991; Tang, 1994; Thanikaimoni, 1986; To A 1990; Tobe € Raven, 1995; Todzia, 1993; Tsou, 4; Ukraintseva, 1993; 27 & Dhar, 1976; wd & Kapi a VAR Walker & Walker, 1984; Williamson & Schneider, 1993; Wu Cheng-Yih & Knbitsii. 1993; Yof- fe, 1962; Zhang Zhi-Yu, 198 Seed anatomy Baas, 1972; Blank, 1939; Boesewinkel, 1985, 1994; Boe- sewinke Bouman, 1980; 1976; Cronquist, 1981; Dickison & Baas, 1977; Erden. 1980, 1987; En- gler & Prantl (eds.), 1887-1914, 1924—1995; van Heel, 967; Huber, 1991; Keng, 1962; Mc Nair, 1930; Netolitzky, 1926; Seubert, 1993; СОА (ed.), 1985/ 991; Thanikaimoni, 1986; Tobe & Peng, 1990; Tobe & Raven, 1995; Wojciechowska, 1969 Stem morphology and anatomy Baas, 1969, 1972, 1975, 1984; Baas & Werker, 1981; Baas et al., 1979; Bailey, 1980; Bailey, 1933, 1957; Bailey & Swami, 1948; Baretta-Kuipers, 1976; Berg, 1977; Blank, 1939; Canright, 1955; Carlquist, 1964, 1976, 1977, 1981, 1984a, b, c, d, 1988a, b, 1990, 1993; Carl- quist & Hoekman, 1985; Carpenter & Dickison, 1976; Baas, 1977; Endress, 1993a, b, c; Garratt, 1933; Gottwald Parameswaran, 1966, 1967, 1968; Gutzwiller, 1961; Heimsch, 1942; Hekking, 1988; Huber, 1993; Humphrey, 1935; Шс, 1991; Keefe & Moseley, 1978; Keng, 1962; Kessler, 1993; Kribs, 1935; Kubitzki, 1993a, b, c; Les, 1993; Maguire & Ashton, 1980; Maguire et al., 1972; Mennega, 1982; Metcalfe, 1952, 1956, 1962, 1987; Met- calfe & Chalk, 1950, 1988/1989; Meylan & Butterfield, 1978; Miller, 1975; Philipson, 1993; Piccioli, 1901; Prance, 1972; Prance & da Silva, 1973; Puff & Weber, Takhtajan, 1966; Taylor, 1972; Tomlinson, 1961; Vestal, 1937; van Vliet & Baas, 1984; Whalen, 1987; Williamson & Schneider, 1993 Leaf characters Arber, 1925; Baas, 1969, 1972, 1975, 1984; Baas et al., 1979; Bedell, 1981; Berg, 1977; Blank, 1939; Canright, Crane, 1984; Engler & Prantl (eds.), 1887-1914, 1924— 1995; Heywood (ed.), 1978; Hufford, 1992; Humphrey, 1935; Hutchinson, 1964/1967, 1973; Keng, 1962; Kluck- ing, 1992; Kostermans, 1985; Les, 1993; Levin, 1986; Maguire & Ashton, 1980; Melchior lad), 1964; Metcalfe, 1956, 1962, 1987; Metcalfe & Chalk, 1950, 1988/1989; Prance, 1972; Prance & da Silva, 1973; Puff & Weber, 1976; Rao, 1991; Schmid, 1964; Sinnott, 1914; Takhtajan, 1966; и ари, 1986; Tomlinson, 1961 Floral and fruiting characters Airy Shaw, 1951; Baas, 1972; Baas et al., 1979; Baillon, 1873; Batygina et al., 1985a, b, c; Bausch, 1938; Bayer & Hoppe, 1990; Berg, 1977; van Eoi 1971; Blank, 1939; — En Brizicky, 1964; i ison, 1976; Corn 1981, pid Cuarrecasas 1985; Dahlgren et al., , d; Endre pf, 1991; Engler, 1930; Engler & dia (eds.), 1887- 1. 1924-1995; Friis, 1984; Gagnepain et Аі. (eds.), 1942; Gore, 1935; pen 1961; Haber, 1959, pe e = Pp T m = ürn, 1966; Janchen, 1909; Johri et al., 1992: Kamelina et al., 1981; Kamelina et al., 1983; Kanis, 1968; eating, 1972: Keng, 1962; 1951; Kos- 40-1906; Melchior (ed.), 1 1956; Payer, 1857; Philipson, 1993; Pi Prance, 1972; Prance & da Silva, 1973; Puff & , 1921; Ronse Decraene, 1989, 1992; e& Senes ishauser, 1997; Sand- with, 1962; Sa Шег, 1973; 1937-1939; Schaeppi, 1953; Schmid, Sutter & Endress, 1995; Takhtajan, obe & Peng, ; To 995; Todzia, 1993; lere 1984; Williamson & Schneider, 1993; Wilson, 1965; Woon & Keng, 1979; Wu Cheng- Yih & Kubitzki, "1993 Hostplants of fungi and butterflies (not included in com- tion) Ackery, 1988, 1991; DeVries, 1987; Farr et al., Pierre, 1984 1989; Annals of the 208 Missouri Botanical Garden 1oded simi Y ‘25 25995 ӘРӘОВ161”) “а (9025) umaojipuni8 шпшәутиоЦӘҢ 821111 2661 ‘SUTOS 9 иво 112-92 9 "doaj руцололј әвәовив|в406А1Ц”) 71 09091 спирјрдовАл у) 0У9211 £661 “TP 19 ni) NON %02 2504) әвәовццивло]чг) РІ94916 snoruodpf спујиолоју) €rv10f 1861 “18 19 smeny umouyun авадовтродоиоц“) "ү va2n4aqo тлорилас £661 “Te 19 ni) ПОМ 22016 тд 0е0/7И 1661 “ТЕ 19 5971 NNOD “us 597 әвәов|Ацаоувлә”) "1 шпѕәшәр шт]уХүаотрләг) v68101 2661 “TE 19 мәчүү NON ‘ZI 2504) әвәоғцо|вцдә”) иде змотоујој гпојоцаг 11:38! 2661 “ТЕ 19 95840 ПЭМ ZEL 2504) dal uf РІ94916 (quay) snop snwAuony jaded siu AN 26622 “OW әвәовІвООХІВ”) "ы штп440|9 10204107) 1/9<6И 2661 “ТЕ 19 чешром суд utsuoostA ILIMIY “1 оАрара оар» 696107 2661 “TE 19 рвәѕшүо) HOIN ‘016 uesupf oeooei[ojude?) "] эуоЙләор штитал 2661 “Te 19 nu) NON 21006 то) әвәовП|әшв”) ‘unary (71) DUDLAJUIM njj2uv?) ІЕ6101 2661 “TP 19 мечу HOIN ‘686 чәѕирГ oeooe[nuedure?) “p U2: царот 669211 2661 “Те 19 nÒ NON 29 тд oeooequeoA[e7) xurT (7) xooovad спујириошту) 6661 “ТЕ 19 95845 ПЭМ 202 25949 әвәовхпЯ "XYII suaquim204d DIPUDSÁYID A l6£v11 2661 “Те 19 peaisu[o pup “u's 22144 овоовштлу "uduo1g (71) »souiZnup, vijoz42g ESZSOM 2661 “Те 19 95849 SIM ‘922 UY] y ne] эвәовтәргәицәв1ә1ч quuin ини Duet age 2661 “Te 19 рводешјо) uMouyun ӘРӘОВ2О166ғ16 71 5415әйшр> 02158044 jaded sug OSIA “us uosiaa]y ӘВӘОРХІҢ "1 рирјјало xig Jaded sup Y ‘SGG asm?) aeaoepisdopuoqiag ‘J X4ooH ритүогоо sisdopiuaquag 2596 11 £661 “TP 19 Md) NON “pz MÒ эвэәернәспәц 18) (әипџо4) 12]Daq ртиоцоуү 2661 “ТЕ 19 9524) ПЭМ FII 95995) 9e2oPurures[eg "Фәәрұ sisuadpo SUDAN] 299211 2661 “18 19 NIC) ПЭМ '02006 тб оваовАоедодету IYA İL `2 Suapunos pXapqoaisny jaded sup y “us ]әлХәа1г) әвәәвтәЧолә15ү 0412]$0121u1 тлодолој y 062711 2661 “ТЕ 19 9584) əuou IBIDBIYIO|O]SLIY "] ISUIPDUDI штаргу 991111 2661 “Те 19 95840 SA ‘IZEI nequnjq — Rt "] vsouids DIDAY S69 16M ee661 “Te 19 [eng HO ZEZ 421244 ey Ш быны ee 826101 z661 “ТЕ 19 HOV ПЭМ “611 989) әғәовцойпһү A РРА 1292 11 2661 “TP 19 ni) ПЭМ “sI MÒ әвәәвиоццү [gunq (71) 940141 оитипѕү 1oded sty) ON “£0ZS 12 12 np242^) оваоврвјоодетошу пвәләсу 9 зешоцј "A ‘q s?$uadnaos snpnjoou1si2uy 839311 2661 “TP 19 Md) ON “005 vary] nds | canas зена d ЛЕЯ ородоц ла оуглодшу 8962 11 2661 “TP 19 хәред ММП 90913 uum) з Opuvusa y әвәдеиеуү qqey (330p) 12//101р19 vIuDyy лодеа siy} М 6S6 28047 оваовотхојхо у ва Y zin] шпидоипа uooixo1xay 288101 2661 “Те 19 ueqpy NON 2112 чолу AY аны, Бы ery UO0ISSIDIR Suorneiro 21181911] әәлповуләцәпод Árure y sardadg Хива из“) “(1861 “1smbuor) о] Surp1oooe Ápsoui) вәгтшр} Aq А||еоподецаје рәдивие әле 9894 | "1291 103 ѕә[ішеѕ exe] jo ојдеј, “хе 11244 Jo әүдер 7) xipuadd y 209 Combined Cladistic Analysis of Angiosperms Nandi et al. Volume 85, Number 1 1998 1661 “ТЕ 19 чоор “алу ‘sumag "ешојце“) jo N оваовџотбпо4 88ојјом siunuum]joo Disainbnoy Joded зид A '£6/9 әпәмә]]ог) 9e29Ptuaxuei J "1 руиәүпләаупа ттигуирла Jaded sug ПЭМ "ws yooping y әвәовципоәв|4 JoX4ooH suaus 51845070104 OSEETT €661 “TE 19 95845 £g uowuryseyy п әвәәвЗр J 1519) (UA) 1Хәфшор sn?ofoyioy 9861 “Te 1 qoup[y uMouxun әвәов4вД “1 201205 одолрагуј STOCTI 2661 “е 19 ni) NON '92006 MÒ зваовајеј па “әәп7 X pjoqers vipupdjod ргјејат Jaded sum V69I£9 9 'долј ррцәло4 овзовтфоцап boef оштдашгјит рудолој 816101 2661 “Те 19 ноу 0SZ0-98 "фу 2urq&us әвәәвтцЧАләп sona vpron) оту оп 2161071 2661 “Te 19 ноу ПЭМ %2016 "10 — m" AUO PU винил 28117 2661 “81106 че®лор ЄЄЄТ`бС ОН FIO *N — P "8 Ашом sisuaquimboo 0110110954 €8I£TI 2661 “ТЕ 19 95940 21618; YÍMQUIPA DAN A М8. ы oy 1oded sug MSNA “us uum) авзовф 18909814 snibn9124 snd402020]3 лодеа sy} Y 958 28047) овоовидезе| 4 nijofioips грудодтн €19Z11 £661] “9SPY) x чогу ПЭМ %00с чогу әвәәвиәч "1 оиотл8па soa£dsoiq 116107 7661 “Те 19 sure ONT ПРО sumas — | | 1987 070019 01950] 6661 “Te 19 95945 AD ‘9q чоририохо( аваовадвоолој а (1 uojusy (SIJNBMY |) оотирјкаг DILOYS Jaded siy} у ‘£99 25995 әвәов||Ацаоооцот( меце Алу ([orz[e(] Y uostqoinH) un ad штујАцдоХцаај £06101 2661 “Te 19 ноту ПЭМ “FEE 95990) әвәрә] "[ Doipul шәп] Jaded si OW “и ^s трирг] аваовјашАрцј привәт 12128 зајашАрци Jaded siu у 0:5 по авоовјејодвцоц] 18иң иойтоолорш штуподоцоц 219211 2661 ‘SLYN Y чогу NON 81068 ПН әрәәрізиәйігі (1 "| »2ruoddp, pisuadmg 006101 2661 “18 19 мәдру NON “us чогу IBID) 1 220jfnua2D4 DÁN) c68101 2661 “18 19 моду SVD ‘SEIZ 42112N aeaortuoun’) ews штајшшта штүрзәйорогә”) 8£61731 6661 “ТЕ 19 95945 2uou овзовифтопо) 71 odad опдтото 6/1117 £661 ‘SUNOS = uezio]y 96 Luy MULS oyoury 9922P]BUIOSOSSOI") "nw шпотилой|ро vuosossoan) 281111 £661 “5106 Y UPBIOA 10829 9g uoiZunung әвәәрүп55в17) џелојј (uosjeA 'S) ортоѕта рАгјрт лодеа sty) доцопол ou овоов еооџАло 8104 74) Y 35104 МГ 070891490] sndan2ouKao7) 022111 €661 “ТЕ 19 SuerX у-19-ртр “GV ррошу әрәовшо”) пмәдивд 14297]Da. зпилођ 268101 2661 “Te 19 ueqpy ПЭМ ‘Ske 9522 а а, т эцойиАш 20105 862900 #661 “TE 19 ueZ10]y НЯ “709 149 әвәәвлеииог) "епу 4 $?d402042u02 snapuuor) 8b6101 2661 “Te 19 мочту HOIN “us иочгриу “Y M әвәәруәлушог) "Тори sippnbsm() £29C77 1661 “Te 19 Ае] ПЭМ ‘TRE #5992 "— He [YPIS пуорјрипа тзт] 609cTI 2661 "95840 Y uory ПЭМ ‘#881 чогу эвәәвицә) "1 vijofiupo 04127) ЧО1$5Ә2ЭР биоцеџо əmən золпов/лоцопод K[rure 4 satoadg xueguar) 'penunuo) у xipuaddy Annals of the 210 Missouri Botanical Garden 2661 “Те 19 тфу NON 62016 MÒ ссог И 1661 “18 19 sa] NNOD “us 527 әвәәвәрцашА к uojty Dopo рәрцішхду 926101 6661 “Te 19 ueq[y ПЭМ ‘SHI 25942 әвәовцциәдәқ ooue[g 27010 sayruadon] 2661 “ІР 19 тфу NON 82016 MÒ GEOLLW 1661 “Те 19 5971 NNOD “u's 597 ас ша мәд CPIM) 2921) 0quinjay 1oded зид 4 “us 100H әвәәвишещцолАр ‘ds snuumiioakpy jaded в М “608 25945) IBIDPUISIA A 9[[I249T saproruisz&ur DSID HA] 6092 11 £661 “Te 19 ni) ПОМ 12016 MÒ оваовопаџлу MEDIE 01314210] тн 6SELTT 2661 ' Bs 19 ueuipoy SIM 10506 $11 IBIJPÍULIO W “ше D42fiojo DÍULO y €£6101 2661 “ТЕ 19 19q1Y 9uou IBIIBIO NA 71 2419 sno 900t T1 €661 i 19 peajsuo auou әғәовццивАиәр( "1 опојћај saumip&uapy 1aded suu Y 211 2804) оеаовцивцоу ']Azs&zg suaon] numsaag 0498977 L661 “Te P 424 Y 029 25995 9PooeuÁSesnpoa]y Jayeg 210finsoddo ou&2psnpopy 1oded siy} ПЭМ ‘TEE 25945) 9€32EIAEIZOIE]A "u2ue[d 9 eueug 40/1224 DIADLFIID y 98I£TlTI €66[ “ТЕ 19 95945 OSI “U's Japuay 9e32PA[E]I "репу Y пиозилдол штаХ«вог) 268 101 2661 “Te 19 моју HOIN (08918 9 741 pp aesoeniidpe y чипу (71) 0170/1520 рштиоѕіКя сс9с171 2661 “ТЕ 19 nif) ПОМ Ра тд овооепоизру '2207 X р|04916 ропајоа« у vyousoy 189677 2661 “Te 19 Ав ПЭМ “ГЇ 9504) SVIRU] 71 әииәгәа штит] 2661 “TE 19 uoo] ТОРЕР-0961 мә DIM әвәәвріціќәәт le[qny sisuaupimmz рпапото? £661 “ТЕ 19 95942 ПЭМ '22016 MÒ аваовџошта/] шағц”) риррмој ni2up2] 1oded sii M 289 19 12 uojiguruuad әрәәруешә]1втәв”] Áqsny шп,029ә1920 DUI] 2661 “ТЕ 19 95845 HOIN 1-1-<0-99 uosduig авдовџошелч мој, D]D]022UD] омләшоіу ООУРТЛ €66[ “ТЕ 19 ueuipoy SIM “us грдцоус-ју оваовшлодоом ооп7 psouids 7р11/12420% c661 “Te 19 оривилој y ‘892 uosduig ӘРӘОРІЗШАЦ uuog хә “ALO DUDÁD DU viguia4j ZS9Z 11 €661 “Te 19 nu) NON '28 m) әвәәвтә[[ "Ju9A хә 'хцогд штаојалра шт Jaded siy} DVM “Ор uanag ира овоовшовој АЦО пиирш DULIDI] 181111 2661 “SUJOS 9 ULeZIow SA ‘0812 UDI LO y оваовазивлран мој, оујацаолоош рәдирірАң 1oded siy} QVM 2210 обима әвәовПН PIM за хә злог 1100022 оту 226101 б661 “TP 19 ueqpy NON “6016 MÒ энзоерцәшЕшЕн AO Cure temer Jaded siy} Y £99 aspyy әвәовләпипс) шең `$ ^A хә үү ттио11шюу D42uun-) 026101 2661 “Te 19 мочту ам “u's әд о али, "qui umaojfipup4d штирләг) 96211 2661 “Те 19 peaisu[o 9uou әвәовивциәс) ШОҢ D42204d DUDA) 1404ип *uaure[oAeg uMouxun 9€92E]EUIO[OSSI2^) "ssnf ^y (T) штити vuiojossia«) Jaded sup М 6921 2504) эвә2еие[ә88гу вәнемці, олирјАаг DIUIPDYNAL, 101559008 Те | аолпов/лацопод Аше вәтдәйс xueguar) 'penunuo) +) xipuaddy 211 Nandi et al. Volume 85, Number 1 1998 Combined Cladistic Analysis of Angiosperms ес9с11 6661 9584) 9 чогу ПЭМ 2002 чолу әвәвәәв1А15 "ше DUDIIaWD хюл41$ Jaded si ПЭМ ‘ГІ 2804) әвәәвә[Ацар1с Jakeg то ofin әрус лодеа зид Y ‘008 2804) IBIDRAINÁYOIBIS ‘ооп Y "qaiG xovapid SNANÁYIDIS taded siu Y “Оов 25949 aeaoe[edoso1oeudg ‘ds sndapoopodoty 9861 “P 19 ur] uMouxun зезовивјос "1 шпордој DUDUOMN 1aded sup jJ “us 100g IBIDPISPUOUINIS 1opreuuog “Y *) (Яш) sisuauty2 DISPUOWUNS 206101 2661 “TE 19 ueqiy auou әвәәвивүпцдоләс "1 vaindind sonda 1oded sup Y 025 “1D 12 nv242^) оваовјејодој Кос ‘J Joyeg 27010 pimn2upqn() лодва siy} Hg “us sang ^f әвәәвтдә1воцаАәс jdeig sisuaausog 7182180ца << £S610T 2661 ‘SOS 29 uez1ojy SA “ESZE 5106 Y 5105 OS оон ют]о/ыдәуи1 3p4fixvs 256101 2661 “18 19 мәү NON “PET 25995) оваовгшаовллес “1 пару DIUIIDLIDS C€6101 2661 “Те 19 мету ПЭМ %21 2504) avaovjodeg uakoy (71) 070002 оаоурируу Jaded si ПЭМ ‘SIT 25995 aesoepurdes "uxeT DID]NITUDA DLIIMAL]IOY ‘[qndun *uaure[oAeg Y EU'S 1n02p424 IBIOBIOPBATPS "1 221s42d DIOPDA]DS Jaded si Y ‘OFS 2804) әвәәвәтүес̧ “Т 070]п21ә1 XD 299311 £661 “Те 19 asey) NON ‘2016 ™O нні ас "ds 01905 £661 “Те 19 95245) NON 211 2504) ide Jeu (T) 2mwyofn вплоиод v38901 7661 “TE 19 ueZ1o]y SA ‘ОТРС 51105 X sujos лен APO T poon vsoy 681611 2661 “ТЕ 19 95945 ПЭМ “001 #5942 id E AA женшоту 1oded siu M 2911 91290 аваовлризрордеци 19qNH UNIUOZDUD uo4puapopquiy 689627 2661 “TP 19 Ае] x ‘9881 чо18шииәд әвәовипп(у зала (pe) 97/4ydoprard vunne) 061111 E661 “51106 9 UB3IOA пемен “94 dol] “EN 3B3IB3]014 ug y шә ттаодшт] ZOLLLIN 2661 “ТЕ 19 івеширіс) umouyun оваовиовАјоД дадоватод 9 рплалоцј, штаојто x шпәуу сФ6101 2661 “Te 19 мәү ПЭМ ‘SST 25945 двзовјевАјод "1 91010142 D pÁ od jaded siy} У (046 ?SP47 аваовшошајод "виола с (ysing) 070991990 рт) I0LLLIN 2661 “ТЕ 19 rSeuuer) uMouxun авзовшввашпја "qunu[ sisuadpo ободитја еРбі0і 2661 “Te 19 мәчүү NON “00064 710 љиљани id E Rin AA jaded sup Y CEET 25095) 9voogpuo1dorZe[q ju) suajoaapns иолојаот8 ја 202111 €66[ “51106 9 uezio]y VSM “us Әләдәзәту әвәәовлойзво1{ [Sg 7) хә “HOR шпоиодог unaodsojnq jaded sm Y ‘98G 10 12 иојдишига 9ваовлотоц 9g тию, x uoqoue[d D4o)doznj4 n42121]]2d 0%6101 2661 “Te 19 ueq[y NON '000g чогу IBIDBIOHISSE 1 1 suo mauniponb a0jfissoq іөтеті E661 “Te 19 9584) NON ‘SIIZ чогу ILIMU d T ођојтиг ютиоәр,] 926101 2661 “TE 19 19q|y UNI “us дома әвәоврЦех() '"boef пигјпр six) SOCTTI 2661 “SUJOS ў ue210]y ТІП ‘6682 1u2421N ARO "Puy y `[ 14092428 Difdooqos 2661 “Te 19 9seu7) NON 626 25DY/) әвәовицо() "за огоуттш риуод Џ01559008 биоцеџо əmr] додпов/лоцопод Á[rure y вотоз 46 Хива из) 'penunuo) 7) xipuaddy Annals of the 212 Missouri Botanical Garden 'S9uo рлеривја əy} әле SWÁUOIJL unrrequau +Хивләлтип ‘A цеойод “dolj, ївпәрлес) ошо [e4oy “Ody :ечоцеи “ПЕК tuapaes 9 tuapaes јкошејод “Dg *um]310q.e “(іу :рэѕп SUO}BIAIIQGY “pelt 16.) SEM UOXP?] siy} 10) әәиәпһәѕ 1244 ue QqotuM ut uomneor[qnq $ <06101 0е2с0П 096101 ТОП 90/7Г1 РР6ТОЛ 2092 11 899311 089201 т29СГІ 2661 “ТЕ 19 95845 2661 TE 19 H9q[V 2661 TE 19 peaisui[() 2661 “TP 19 моју 2661 “Те 19 peaisui[( 2661 чәшірд Y зома 2661 “ТЕ 19 9285 2661 "18 19 95849) 2661 '95940 9 поту 1oded зид 2661 “TE 19 UOLOW 2661 “ТЕ 19 95240 jaded siy} с661 “e 19 оривилә y 2661 “PSPYD Y uoy HOIN "ws uoszopuy ^y M ПЭМ 91006 т0 HOIN 099£[ uosiapuy "M M NON '922 #5942 3uou ПЭМ ‘ETT 25045 HOIN ‘9921 uoszopuy "M M имоиўип NON ‘008 чогу әРәов||Ацао8А2 әвәәвләзи! үү әвәоғәцЦ | әрәовәр|әшАЦ p овоовјаџошведај, әрәовдиәовдәТ овдовоџешеј аваовивипс оваовоојашАс “Т "umjoups штэрттх) 75104 79 Y 725104 'M Т мојшт síun] ‘ds noppn() UM EET СПА PUN "eser ом “Т ош шпүоәрао4| "Ssaquie7) pamu DIUOFIL] "Teg ртој јао тада DIAYIAID] "1 рлиоаоГ ођјешрђ "user, DuD1D222Q омрутђу ‘ds nisuaumaja АЦО 51849115 иолтидорјај asung D4pupjuad XUDUD], М 91unapuu DUDUNG ‘by ототоира sn2ojduikg UOISS3228 quegqua^) диоцеџио олпјелој 1 долпоз/улоцопод Apure y saradg 'penuguo) 7) xipuaddy STATISTICAL SUMMARY OF SOME OF THE ACTIVITIES IN THE MISSOURI BOTANICAL GARDEN HERBARIUM, 1997 Vascular Bryophyte Total Acquisition of Specimens Staff Collections 20,611 6,523 27,134 Purchase 18,330 3,065 21,395 Exchange 29,942 3,992 33,934 Gifts 11,830 1,169 12,999 Total acquisitions 80,713 14,749 95,462 Mountings Newly mounted 49,706 12,095 61,801 Mounted when received 18,234* 0 18,234 Total specimens filed 67,940 12,095 80,035 Repairs Specimens repaired 27,241 n/a 27,241 Specimens stamped 1,881 n/a 1,881 Total repairs 29,122 n/a 29,122 Specimens sent On exchange 49.071 312 49,383 As gifts 15,560 1,075 16,635 Total 64,631 1,387 66,018 Loans sent Total transactions 410 27 437 Total specimens 32,846 4,586 37,432 To U.S. institutions Transactions 256 18 274 Specimens 22,374 2,856 25,230 To foreign institutions ransactions 154 9 163 Specimens 10,472 1,730 12,202 To student investigators Transactions 59 4 63 pecimens 10,160 509 10,669 То — investigators Transactio 326 23 349 uer. 22,553 4,077 26,630 Loans received Transactions 319 20 339 Specimens 33,697 2,204. 35,901 * The 18,234 “mounted when received” vascular plants are specimens of Chinese plants purchased directly from From U.S.A. From abroad Total Visitors 342 107 449 On 31 December 1997 the total number of mounted, accessioned specimens in the herbarium was 4,777,217 (4,482,859 vascular plants and 294,358 bryophytes). The Garden's herbarium is closely associated with its database management system, TROPICOS. For example, many of the numbers in the preceding chart are taken from TROPICOS, since it is used as a herbarium management tool. Herbarium labels for newly collected specimens are generated through TROPICOS, and the information is retained there for further use. The charts below summarize some of the statistics from TROPICOS both for the calendar year 1997 and as year-end totals. Note that the specimen records in TROPICOS are primarily based on MO specimens, meaning that about seventeen percent of the bryophytes and twenty-six percent of the vascular plants in the herbarium are now com- puterized, with an overall total of about twenty-six percent. Distributional records are taken both from herbarium specimens and from literature records, and these are distinguished in TROPICOS. Similarly, information concerning types is taken both from the literature (protologues) and from specimens. TROPICOS is essentially complete for the names of mosses, except forms, and contains a few thousand records for hepatics, for which no comprehensive effort has yet been undertaken. The 1997 additions to the names for bryophytes, 504, reflects pretty accurately the number of nova published for that group. TROPICOS records—1997 additions Bryophytes Vascular Plants Total Specimens 1,685 92,778 100,463 Names 504 24,545 25,049 Synonyms 944 15,725 16,669 Distributions 1,869 22,545 24,414 Types 123 16,475 16,598 Bibliography 1,348 2,399 3,747 TROPICOS records—Year-End 1997 Totals Bryophytes Vascular Plants Total Specimens 50,338 1,179,217 1,229,555 Names 93,121 697,741 790,862 Synonyms 57,614 335,957 393,571 Distributions 36,330 692,513 128,843 Types 6,671 208,981 215,652 Bibliography 18,284 53,178 71,462 Specimens in herbarium 294,358 4,482,859 4,777,217 Percent computerized 17 26 26 —Marshall R. Crosby Volume 85, Number 1, pp. 1-214 of the ANNALS OF ge 2. JRI BOTANICAL GARDEN was published on June 23, 214 Flora of the Venezuelan Guayana Located in the southeastern half of Venezuela, the Venezuelan Guayana is the core area of what has been called “Тће Lost World." The area is dominated by massive table mountains known as tepuis and includes many endemic species and genera, with much of the area still in pristine condition. There are nearly 10,000 species in the flora area, and over half will be illustrated by line drawings. Volumes 3 and 4 of the Flora of the Venezuelan Guayana are now available from Missouri Botanical Garden Press: Berry, P. E., B. K. Holst, and K. Yatskievych, editors. Flora of the Venezuelan Guayana. Volume 3, Araliaceae-Cactaceae. 1997. ISBN 0-915279-46-0. 774 pp. 1113 species treated. 628 line drawings. $67.95. Volume 4, Caesalpiniaceae-Ericaceae. 1998. ISBN 0-915279-52-5. 799 pp. 1329 species treat- ed. 621 line drawings. $67.95 Also still available: Volume 1, Introduction (includes Vegetation Map and Topographical Map). 1995. ISBN 0- E -313- 3. 320 pp. of text, plus 44 pp. of color plates, 10 b/w photos, 51 line drawings. 95. Volume 2, Pteridophytes, Spermatophytes (Acanthaceae-Araceae). 1995. ISBN 0-88192-326- 9. 706 pp. 1285 species treated. 618 line drawings. 9». TE Map and орков Map, 2-map set: rolled and shipped in tube $17. 00: or professionally folded $15.00 | Price Мо. Total ~ 3 (8: 2-map set) 152.95 ved checks payable to Volum $6798" LL. ен souri Botanical сед Ргеѕѕ | Volume à MIE 1344 4 Shaw Blvd. Volume 4 TON. 222. St. Louis, МО 63110-2991 2-map set, rolled I Lu V LE 2-тар set, folded — 4500) э: 22222 Or cider ше Visa. or шесі iE Add for shipping and handling: : d : Check one: Visa е: Манасын... Within the U.S. "Ou Free us NN. — ы ы мейт . for the first book ta. $4.00 ы =» Signature "nuu A ^- (oxides poi, - RENE О С О MN Е UR тты anada and Mexico ; ; ne d for the first book $6.00 2: 4 fax, 314-571-9591. i E а 22 for each additional bosk — 3130. а c-mail, mbgpress@mebotorg o International - T 5 ¡añ web site, meL e d ed for the first book О - LU 55-ы Ship to: — аа мы ы. for each additional book QUE: c E RAS нЕ - —— Air rates available on нефес, y e ос RNC PME Total enclosed um с oe ‘State кш ы и dnd E от Lip CONTENTS New Tools for Investigating Biodiversity, the 43rd Annual Systematics Symposium of the Missouri Botanical Garden Introduction to the Symposium e P. Mick Richardson An Ecological Perspective on Biodiversity Investigations: Examples from Aus- tralian Eucalypi Forests ы. а-ы Mike P. Austin Deciphering Landscape Mosaics of Neotropical Trees: GIS and Systematic Sam- pling Provide New Views of Tropical Rainforest Diversity == а Қо E T ыа ADR AEG ОЗИ CLR E Үр. а M eee Deborah A. Clark Large-area Mapping of Biodiversity ... J. Michael Scott & Michael D. Jennings What Satellite Imagery and Large-scale Field Studies Can Tell about Biodiver- sity Patterns in Amazonian Forests. A к ae M Hanna Tuomisto Selected Proceedings from the 1997 Midwestern Rare Plant Conference A ЖН SELLAM epit et ie Sn arie qn Iw Ыза ш a алы С Kayri Havens Factors Affecting coupes Success in a Rare Grass, Calamagrostis porteri subsp. insperata — _ Kayri Havens 4: Douglas L. Holland Microhabitat Relations of tis Rare нее Bent Grass, Calamagrostis porteri subsp. insperata (Poaceae), with Implications for Its Conservation __------ | Bittner & David J. Gibson берене ia in Running Buffalo Clover (Trifolium stoloniferum: F abaceae) Using Random Amplified Polymorphic DNA Markers (RAPDs) ------ Ме Daniel J. Crawford, Elizabeth J. Esselman, Jennifer L. Windus & Carol S. Рап | Pauline Discos of the батасы Endemic Astragalus bibullatus (Faba- -ceae) Using Isozymes = Carol J. Baskauf & Sharon Snapp ! Genetic Variability in the Federal Threatened Mead's Milkweed, Asclepias mea- | dii сезе Чен) as Determined by Allozyme Electrophoresis 1 ке со A Diane L. кеш, Jenny L. McBride, Marlin L. Bowles & _ E Daniel L. Nickrent 6: t ÓN and Nosti Ecology of the F ederal Threatened Mead's Milk- | weed, ласере meadii Macs. Ри Wn eng ce е er gi Фе Pn "А Bowles, 1 L. McBride & R. F Betz % 7 The Pollination ‘Ecology of Five eue 2 Penstemon (Scrophulariaceae) i in the _ | | Tallgrass Prairie Richard R. Clinebell II & Peter Bernhardt | к aS Combined Cladistic Analysis of Angiosperms Using rbcL and Non-Molecular Data : Owi І. Nandi, Mark W. Chase & Peter K. Endress | SEN аа of Some of the Activities i in the Missouri Botánical Garden Her- 2575.43 ; ; barium, 1997 . | d | аудар R. Crosby 19 AT Cove ion. pues by теби Myers s Systematic тираны: reins Botanical Garden. 1 ir A EY de ғы” A924 а 9 se кә 4 _ Annals of the _ Missouri Botanical Garden 1998 W% Number 2 Volume 85, Number 2 Spring 1998 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating out- side the Garden will also be accepted. All manuscripts are reviewed by qualified, independent reviewers. Authors should write the Managing Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Committee Michael H. Grayum Editor, Missouri Botanical Garden Amy Scheuler McPherson Managing Editor, Missouri Botanical Garden Diana Gunter Editorial Assistant, Missouri Botanical Garden — Vicki Couture Secretary Ihsan A. Al-Shehbaz Missouri Botanical Garden = Gerrit Davidse Missouri Botanical Garden Roy E. Gereau Missouri Botanical Garden Peter Goldblatt Missouri Botanical Garden Gordon McPherson Missouri Botanical Garden P. Mick Richardson Missouri Botanical Garden Henk van der Werff Missouri Botanical Garden For subscription information contact ÁNNALS OF THE MISSOURI GARDEN, % Allen Marketing & Management, P.O. Box 1897, Lawrence, KS 66044-8897. Subscription price is $120 per volume U.S., $130 Canada & Mexico, $155 all other countries. Four issues per volume. The journal Novon is included in the ир price of the ANNALS. amepher@admin.mobot.org (editorial queries) http://www.mobot.org © Missouri Botanical Garden 1998 The ANNALS OF THE MISSOURI BOTANICAL GARDEN (ISSN 0026-6493) is published quar- terly by the Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, MO 63110. Pe- riodicals postage paid at St. Louis, MO and ad- ditional mailing offices. PosrMASTER: Send ad- dress changes to ANNALS OF THE MISSOURI BOTANICAL GARDEN, % Allen Marketing & Management, Р.О. Box 1897, Lawrence, KS 66044-8897. The mission of the Missouri Botanical Garden is to "tUe and share knowledge about plants and their environment, in order to preserve and enrich li Ө This paper meets the requirements of ANSI/NISO 239.48-1992 (Permanence of Paper). Volume 85 Number 2 1998 Annals of the Missouri Botanical Garden NZ POLLINATION OF PETALOID GEOPHYTES BY MONKEY BEETLES (SCARABAEIDAE: RUTELINAE: HOPLIINI) IN SOUTHERN AFRICA! Peter Goldblatt?, Peter Bernhardt?, and John C. Manning’ ABSTRACT Field observations, floral dissections, and pollen load analyses of insects indicate that pollination by hopliine beetles (Scarabaeidae: Rutelinae: Hopliini) has орыш UE in many genera of herbaceous perennials in southern Africa. Beetle-pollinated flowers are identified by a suite of characters including a salver- to shallow bowl-shaped perianth p pigmentation emphasizing bright colors ied x orange, cream). Stereotyped "beetle marks" of = pale or dark color frequently present at the bases of tepals or petals. These flowers are typically odorless and rarely nectar. Beetles, iic gi consume anthers and pollen, which are m a contrasting color from the perianth. Taxa that are pollinated by hopliine beetles include species in genera of the Hyacinthaceae (Daubenya, Ornithogalum), pei (Aristea, Homeria, Ixia, Moraea, Ro- mulea, Sparaxis, Tritonia), and Hypoxidaceae (Spiloxene) in the monocots and Aste ytes is closer to that described in the herbaceous perennials ‘ters associated with beetle pollination in these herbaceous of the eastern Mediterranean -— and the woody flora of eastern Australia than it is to the classic series of features associated with magnoliid angiosperm The consumption of floral rewards (e.g., pollen, nectar, starchy food bodies, epidermal tissue) by Coleoptera has been well documented, and the me- chanics of consumption and digestion of pollen, in particular, are extremely variable in beetles. Most beetles studied have either a pollen-cracking “то- lar" on their mouth parts or swallow pollen grains whole in the presence of hydrating nectar. In a few cases beetles may consume hard trichomes with pollen and use these plant cells as a pollen crack- ing grit (see review in Bernhardt, 1996). Knowledge of the role of beetles as pollinators of angiosperms has, however, changed radically in the last 15 years. In the classical view of beetle pollination, ! Support for this study by grants 4816-92 and 5408-95 from g^ Vesp ce Geographic pen is dying acknowl- edged. We thank Holger Dombrow, Worms, Germany, and M. J. identifications of hopliine beetles ker, University of Cap ‚ for their help with ? B. А. Krukoff Curator of African Botany, айе Botanical Garden, Р.О. £s 299. St. Louis, Missouri 63166, U.S S.A. a текелі of Biolo uis Univer St. Lo * Compton Herbarium, ы Botanical xen P. Bag. uls, 2. 63103, cp ин 7735, South Africa. ANN. MISSOURI BOT. GARD. 85: 215-230. 1998. 216 Annals of the Missouri Botanical Garden reviewed by Faegri and van der Pijl (1979), beetles were associated primarily with the pollination of basal angiosperms, especially magnoliids, Araceae, and Cyclanthaceae (Armstrong, 1979; Bernhardt & Thien, 1987). Beetle pollination is traditionally as- sociated with chamber- or urn-like flowers or inflo- rescences, absence of bright coloration, strong, un- pleasant odors, and anthers that often extrude their pollen upon dehiscence. “Beetle flowers" shelter their pollinators, e.g., Stapfia (Gottsberger, 1977), but are not usually associated with true cretion. The major pollinators of such flowers are comparatively small nitidulid, curculionid, and staphylinid beetles. Large-bodied dynastine scarab beetles have been associated with the pollination of Victoria (Prance & Arias, 1975), Cyclanthus (Beach, 1982), and a number of species of Araceae (Gottsberger & Amaral, 1984) and Annonaceae (Gottsberger, 1989a, 1989b) This view of beetle pollination has expanded rad- nectar se- ically with ongoing research in temperate-tropical Australia and in the eastern Mediterranean. Work in Australia (Hawkeswood, 1987) showed that large brightly colored buprestids, cerambycids, and scar- abs consumed the nectar in bowl-shaped flowers of the Myrtaceae and Burseraceae. Unlike the mag- noliids and palms, these plants have flowers with anthers elevated on long stiff filaments, and the beetles often reach the nectar by pushing the fila- ments aside or crawling between them (Hawkes- wood, 1987; photograph by Hawkeswood in Bern- hardt, 1993). To the human eye, these flowers are usually white or light pastel shades, and strong fruit-like odors suggesting fermentation are not de- tectable. In Israel, fieldwork and experimentation (Dafni et al., 1990) have shown that flowers with bowl-shaped, red to orange perianths, blackened tepal bases and/or pollen, and no discernible scent are pollinated almost exclusively by vernal scarabs in the genus Amphicoma. These insects are far hair- ier than the majority of beetles associated with the classic syndrome of cantharophily. Plants with flow- ers showing this suite of characters comprise a guild of herbaceous perennials dominated by Ran- unculaceae and some petaloid monocots (Dafni et al., : Early work by Scott Elliot (1891) appears to con- tain the first reference to the importance of hopliine beetles in the pollination of the South African flora. Peringuey (1902) also remarked on the frequency of beetle pollination in southern African plants, noted floral foraging in many genera of native bee- tles, and suggested that their membranous mouth parts implied a diet emphasizing nectar. Peringuey noted that such beetles departed from flowers cov- ered with pollen, and that “оп a bright day in the spring (August-October) no flower is without a ten- ant." He maintained that few insects were better adapted for flower pollination than such genera of hairy beetles as Anisonyx, Lepithrix, and Peritri- chia. Curiously, Vogel (1954) did not cite Peringuey in his mammoth review of pollination systems in southern Africa. Vogel did note that some scarab genera were pollen- and flower-eaters but made few overt references to beetle pollination, and the sub- ject remained virtually dormant for the next 40 years. In their review of insect pollination systems in the Cape Flora (the winter-rainfall climate zone of southern Africa), Whitehead et al. (1987) derived most of their references to scarab pollination from Vogel (1954), although they did note that cetoniids, nitidulids, and staphylinids visited the flowers of some shrubby Proteaceae. Recent evidence, however, now strongly suggests that scarab beetles in the subtribe Hopliini (tribe Rutelinae) comprise an important pollinator guild in southern Africa and that a suite of floral char- acters is associated with “monkey-beetle” pollina- tion. Among the few works available to date on the pollination of southern African plants by the Hop- liini is that of Picker and Midgley (1996), who list- ed some 25 species of plants as putatively monkey- beetle pollinated. These included both monocots and dicots representing some 10 families. More im- portantly, Picker and Midgley recognized three sys- tems of monkey-beetle pollination, based on differ- ences in beetle hairiness, flower color preferences, and whether foraging was restricted to pollen. Gold- blatt and Manning (1996) described the foraging behavior of hopliine beetles in the genera Anisonyx and Peritrichia (as Lepithrix), concluding that they were most likely to be the dominant (or sole) pol- linators of two species of Drosera (Droseraceae), and one species each of Aristea and Moraea (lri- daceae). These authors also suggested that other species of monkey beetles were likely to be the pollinators of many more species of Iridaceae in genera such as Aristea, Homeria, Moraea, Romulea, Sparaxis, and Tritonia. Studies by Steiner (1998 and pers. comm.) also show the importance of mon- key-beetle pollination in the so-called peacock mo- raeas, M. villosa and its close allies, as well as in Sparaxis and genera of Asteraceae including Arc- totis. Obviously, additional fieldwork on beetle polli- nation in southern Africa is required. The problem is that while we have a number of observations of monkey beetles visiting flowers, there remains a paucity of data showing that these beetles transport pollen of their host flowers and actually contact Volume 85, Number 2 1998 Goldblatt et al. Beetle Pollination stigmatic surfaces. As Hawkeswood (1989) has shown, scarab beetles may pollinate the flowers of some species while destroying those of other co- blooming species. For example, while Diphucepha- la affinis (Scarabaeidae: Melolonthinae) regularly visits flowers of Hibbertia (Dilleniaceae) in western Australia, these scarabs fail to transport Hibbertia pollen or contact the stigmas. Here, we present our own observations on pollen foraging by beetles on native southern African geophytes and compare beetle pollination in southern Africa to that else- where in the world. MATERIALS AND METHODS Fieldwork was conducted during August to Oc- tober 1995, and during the same months in 1996 and 1997 at several sites (Table 1) in the south- western Cape (Cape Floristic Province) and the western Karoo, South Africa, areas of Mediterra- nean climate with wet winters and d Observations of insect foraging involved 4—20 hours per plant species, and included recording of floral attractants (pigment patterns, scent), the be- havior of insects on the flower, and the taxonomic identity of floral foragers. Insects were not collected unless they were observed to contact the sexual or- gans of flowers while foraging or mating. Insects were captured and killed with ethyl acetate fumes for subsequent identification and analysis of pollen loads. To prevent contamination of one insect with pollen carried by another in the same killing jar, individuals were isolated by wrapping in tissue pa- summers. r. Removal of pollen from insect bodies involved either gently scraping pollen off the body with a dissecting needle or gently washing the insect bod- ies in drops of 95% ethanol. The residue from nee- dle probes or washes was collected on glass slides and mounted in 1-2 drops of Calberla's fluid (Og- den et al., 1974). The pollen of a particular plant species was scored as present on the body of an insect if more than 10 individual grains (or polyads) were counted on the slide (Tables 3, 4). Pollen grains were identified by comparison with a refer- ence set of pollen-grain preparations made from plants flowering at our study sites. Field determinations of nectar (if present) were made by withdrawing nectar from the base of the floral tube with 2 pl capillary tubes after separating the ovary from the perianth. Nectar samples were dried on filter paper and sent to B.-E. van Wyk, Rand Afrikaans University, Johannesburg, for HPLC analysis. The percentage of sugars dissolved in fresh nectar was recorded on a Bellingham & Stanley hand-held refractometer (0-50%) using nectar extracted from flowers in the manner de- scribed above. When volumes were too small to measure or to determine sugar concentration, pres- ence of nectar was established by brushing nectar- iferous areas of flowers against the tongue. Identifications of beetles were made by M. Pick- er, University of Cape Town, and H. Dombrow, orms, Germany. Flies were determined by J. C. Manning and bees by Robert Brooks, Snow Ento- mological Museum, University of Kansas. Voucher specimens were made of plant species visited by beetles when necessary; these specimens are de- posited at MO and NBG (Table 2). Insect vouchers are deposited at the Snow Entomologial Museum and/or the South African Museum. RESULTS FLORAL PHENOLOGY AND HABIT Flowers visited most often by hopliine beetles (monkey beetles) are largely restricted to the win- ter-rainfall region of southern Africa, namely the southern and western coast of the subcontinent and the near interior. Flowering there is concentrated in the late winter and late spring, August to early November (Table 2). The majority of flowers ob- served to be visited by beetles belong to herbaceous perennials, especially geophytic petaloid monocots, and subshrubs (Aizoaceae subfam. Mesembryan- themoideae, some Asteraceae). hese species typically form fairly dense popu- lations locally, with over 10 individuals per square meter not uncommon. In some species, e.g., Glad- iolus meliusculus and Ixia polystachya, plants tend to be much more scattered, typically of the order of 1-2 m apart. FLORAL PRESENTATION AND REWARDS The majority of flowers visited by monkey beetles have salver- to shallow bowl-shaped, actinomorphic perianths or involucral inflorescences (Asteraceae). Species of Iridaceae subfamily Ixioideae studied (Table 2) have a short, cylindric or more or less funnel-shaped perianth tube, 1.5-10 (rarely to 20) mm long. In species of Ixia sect. Ixia (I. curta, I. dubia, 1. maculata, 1. cf. polystachya), Romulea, Sparaxis, and Tritonia the tube is filiform below and blocked by the style, and sometimes the mouth of the tube is closed off by the fused or coherent fil- aments. These nectarless tubes appear to be inac- cessible to the mouth parts of the foraging insects described below. Floral colors are extremely variable (Table 2), but 218 Annals of the Missouri Botanical Garden Table 1. Plant species pollinated or visited by hopliine beetles and study sites. Dates of observation are included in column 3. Parentheses in column 1 indicate species apparently visited casually by hopliine beetles and in column ect species other than 21 Apis mellifera (Apidae), ). halictid bees (Halictidae); Musco etles. Their orders and fa hopliine be s follows: Apoidea: An idea: Philoliche c (Tabanidae); Musca, Orthellia (Muscidaeidae); milies Scathophaga (Sarcophagidae); Anthomyia (Anthomyidae). drena (Andrenidae), Plant species Hopliine beetles (other insects) Study site (date of observation) Homeria ochroleuca Babiana rubrocyanea, Gladiolus mel- iusculus, Ixia maculata, Romulea ехита, К. Квин Spiloxene ca- pensis Ixia framesii, Arctotis acaulis, Orni- thogalum thyrsiflora Ixia maculata, Ornithogalum thyrsi- flora Ixia maculata, Ornithogalum thyrsi- flora Ixia dubia, Moraea bellendenii Ixia curta, Ursinia sp., Gazania kreb- siana, Monsonia speciosa Tritonia crocata, Ornithogalum du- ium Tritonia deusta Tritonia squalida, Agathosma sp. Homeria elegans, Aristea teretifolia, Hesperantha falcata Aristea lugens, Moraea villosa, Aga- thosma sp., (Geissorhiza aspera) Aristea жыл еи Moraea cf. urida, Drosera spp. Sparaxis eleg Arctotis acaulis, Ursinia cakilefolia | (Homeria bifi- a Sparaxis elegans, Ursinia cakilefolia Hesperantha 5. Romulea mon- adelpha, Sparaxis pillansii, Bulbi- nella икан Arctotis acaulis, Ber- Romulea monadelpha, Arctotis acau- Romulea sabulosa Romulea sabulosa Homeria vallisbelli, Romulea mon- tana, Spiloxene capensis, (Oxalis btusa) Daubenya aurea, Romulea subfistu- losa Anisonyx ursus, (Apis mellifera Scathophaga stercoraria, Orthellia sp., Anthomyia, Calliphoridae, Syrphidae) Lepisia rupicola, Pachycnema crassi- nes, (Andrena sp Lepithrix ornatella, (Philoliche atri- cornis) Pachycnema crassipes, Lepithrix lon- gitarsis, L. ded Des pes, Hetero pti us, Sc en ysa militaris, Pachycnema сга ochelis Lepithrix ornatella 2. arthriticus, Pachycnema crassipes, Lepithrix ornatella, Het- erochelis unguicularis Pachycnema crassipes, Lepisia rupi- cola Pachycnema tibialis Peritrichia hybrida Peritrichia sp. 1 Peritrichia pseudoplebeia, (Apis melli- fera, Orthellia sp., Scathophaga stercoraria) hein 2-5 A. ursus, Lepithrix ornatelta Peritrichia pseudoplebeia, Anisonyx ursus, (Musca sp.) Lepisia sp. 1, (Philoliche atricornis) Anisochelus inornatus, (Philoliche atricornis Philoliche atricornis) ~ Lepisia sp. 1, Lepisia sp. 1 Lepithrix stigma Lepithrix stigma Anisochelis inornatus Lepisia sp. 2, (Halictid bees) pm Sir Lowry's Pass Village (Aug. 1995) Waylands Reserve, Darling (Sep. 1995, 1996) Camphill road, Malmesbury (Sep. 1995 Ysterfontein, Clanwilliam (Sep. 1995 Sandberg, Leipoldtville (Sep. 1995) Darling, renosterveld (Sep. 1996) Versveld Reserve, Darling (Sep. 1996 ) Riversdale commonage (Sep. 1995) Swellendam (Oct. 1997) e 205 road, 2. (Осі. 97) Fairfield Estate, Bredasdorp (Aug. 1995, Sep. 1996) б. commonage (Sep. 1995, 996) ты Lowry's Pass (Aug.—Sep. 1995) Bokkeveld Plateau, Glenlyon renos- terveld (Sep. 1995, Oct. 1996) Nieuwoudtville church yard (Sep. 1997) Bokkeveld Escarpment, m dolerite (Sep. 1995, Near Calvinia (Oct. 1996) Bokkeveld Escarpment, Oorlogskloof road (Sep. 1996) Bokkeveld Escarpment, Grasberg road (Sep. Bokkeveld Escarpment, Keyzerfon- tein road (Sep. 1996) ~ Roggeveld Escarpment (Sep. 1995) Volume 85, Number 2 1998 Goldblatt et al. Beetle Pollination 219 Table 1. Continued. Hopliine beetles Study site Plant species (other insects) (date of observation) Moraea insolens, Ixia flexuosa, Aris- Anisonyx lepidotus tea biflora Aristea biflora, Drosera pauciflora, Anisonyx lepidotus Spiloxene capens Thereianthus racemosus Khoina bilateralis Ixia cf. polystachya, Ornithogalum ubium, Prismatocarpus pedunculatus Sparaxis grandiflora, Asteraceae spp. Peritrichia subsquamosa, (Pachycne- ma saga—o Peritrichia rufotibiali is, Anisochelus Drayton, Caledon (Oct. 1996) Near Drayton, Caledon (Sep. 1997) Zuurvlakte, Grootwinterhoek (Nov. 1995) Brandvlei hills, Worcester (Nov. nly Prismatocarpus) Citrusdal—Clanwilliam (Sep. 1997) inornatus, (Philoliche atricornis, Hal- ictid bees intense yellow, bright orange to red, or purple shades predominate at most sites. Contrasting pigmentation may be seen at two different levels. The majority of beetle-visited flowers have dark, or sometimes pale, marks on the tepals or petals (Figs. 1—6), sometimes superimposed on a calloused epidermis (Table petaloid geophytes, these markings may take the form of a central blotch encompassing the bases of all the tepals and sometimes the filaments (e.g., Aristea can- tharophila, Ixia maculata), or one or both tepal whorls may have quite discrete marks composed of ovate ar- eas of contrasting pigmentation, sometimes with hazy edges (e.g., Aristea teretifolia) or sometimes with a pal- er or darker central line that resembles the line be- tween the elytra when at rest (Figs. 1, 4, 6). We pro- visionally call these markings *beetle marks" both for the frequent resemblance to the shape of a beetle and for the presumed function of attracting beetles to flow- ers. The color of the markings may be black (Aristea lugens), light to dark brown (A. teretifolia, Ixia curta, I. maculata), or greenish or even yellow on a darker background, and then most often with median dark lines. The markings on the tepals of dark red-flowered Romulea eximia and R. obscura are light green and closely resemble the beetle Lepisia rupicola often seen on their flowers (Goldblatt & Manning, 1996 with col- or photograph). The floral markings on Aristea biflora and Тиота crocata subsp. hyalina consist of trans- parent oval areas at the lower edges of the tepals, which appear dark when viewed from above. The presence of beetle marks on flowers of Babiana rub- rocyanea is questionable: the deep blue flowers have a uniform, large, bright red center rather than a dis- crete dark, beetle-like mark. The presence of beetle marks in the flowers of Ixia dubia varies from popu- lation to population. Flowers observed near Ronde- berg have typical dark markings at the tepal bases, whereas these marks are absent in plants from near Darling. The second level of contrasting pigmentation consists of anthers or pollen of unusual color. The anthers and pollen may be bright orange (Table 2) and thus prominent against dark-colored perianths or filaments, and sometimes the anthers may be black, then presumably forming part of the beetle marks (/. monadelpha, 1. cf. polystachya). The an- thers are sometimes unusually large, particularly so in Homeria elegans (8-10 mm) and some species of Aristea (4.5-7 mm) and Jxia (6-10 mm), com- pared with anthers in other species of these genera. Floral fragrances were not noted in the majority of species. Flowers of Homeria elegans have a sweet odor reminiscent of shredded coconut, whereas those of H. ochroleuca have a mild, slightly acrid, musk-like odor reminiscent of flowers of Rhus spp. (Anacardiaceae). Gladiolus meliusculus has a strong, sweet, honeyed fragrance like that of Viola odorata (Goldblatt & Manning, 1998) The majority of species studied have no discern- ible nectar glands, and floral nectar does not appear to be secreted. Trace amounts of nectar are present as a wet sheen toward the base of the floral tubes of Ixia јтатези and species of Romulea, Sparaxis, and Tritonia. Gladiolus meliusculus secretes nectar at the base of the floral tube [0.8-1.2 pl, 29.2% (SD:1.3) sucrose equivalents, sucrose dominant, n — 5], while Homeria ochroleuca secretes nectar on the lower sur- faces of the tepals (0.2 wl, concentration not mea- surable, equal quantities of fructose and glucose and no sucrose). Nectar and/or fragrance were evident only in those species that were visited by a combi- nation of beetles and other insects. BEETLE DIVERSITY AND PHYSICAL PARAMETERS Coleoptera captured totaled 26 species in nine genera (Figs. 1—6, 7A-D), all of which belonged to tribe Rutelinae, subtribe Hopliini (Scarabaeidae). 220 Annals of the Missouri Botanical Garden Table 2. Floral characteristics and voucher data for species pollinated by puros beetles, including shape, perianth color and marking, presence of nectar, anther color, and flowering time. | b ymorphic in different populations, tr — , ce, + = pol Abbrevi ations: b = to measure volumetrically. Plants collected by P. Goldblatt without voucher are indicated voucher numbers are those of P. Goldblatt. bowl, f = funnel-shaped, trace amount too little by the abbreviation n/v; Flower Anther/ eetle ollen Flowering | Voucher Shape Color marks Nectar color time number Hyacinthaceae Daubenya aurea Lindl. b red = - yellow Sep. n/v Ornithogalum dubium Houtt. 5 white + tr white Sep.-Nov. n/v thyrsiflora Jacq. 5 white + їг white Sep.-Nov. n/v Hypoxidaceae Spiloxene capensis (L.) Garside s cream + — yellow Aug.-Sep. п/у serrata (Thunb.) Garside 5 yellow = - yellow Aug.-Oct. n/v Iridaceae: Iridoideae and Nivenioideae Aristea biflora Weim. s mauve + = orange Aug.-Sep. 8898 cantharophila Goldblatt & J anning s стеат/ Пас + = orange Aug. 10284 teretifolia Goldblatt & anning s lilac + = orange Aug.-Sep. 10250 lugens Ker Gawl. s white/blue + - orange Sep.-Oct. 10311 Homeria bifida L. Bolus 5 pink - tr yellow Sep.-Oct. 969 elegans (Jacq.) Sweet b yellow + = yellow Зер. 10255 ochroleuca Salisb. b yellow tr yellow Aug.-Oct. 10248 vallisbelli Goldblatt b yellow/pink = tr yellow Sep.-Oct. 4032 Moraea bellendenii (Sweet) N. E. Br b yellow + - yellow бер.-Осі. n/v insolens Goldblatt s orange t — orange Sep. 4880 a urida Ker Gaw b white + Е red pes -Зер 10281 боза (Ker Gawl.) Ker Самі. g purple + = orange Sep. 6215 Iridaceae: Ixioideae Babiana d (Jacq.) Ker Gawl. b blue/red +? {г brown Sep. n/v Hesperan falcata ч. f.) Ker Gawl. s yellow - tr yellow Sep. n/v vaginata (Sweet) Goldblatt b yellow + ? yellow Sep 40: Gladiolus meliusculus (G. Lewis) Goldblatt & J. C. Manning g pink t + yellow Sep. 10386A Ixia curta Andrews s orange + = yellow Sep.—Oct. 10358 dubia Vent 5 огапре ал = yellow Sep.—Oct. 10338 SS = Bolus 8 orange + tr yellow : г 10333 maculat 8 orange + - yellow р--Осі. 10349 cf. ric M L. 8 сгеат T = blackish A —Nov. 10568 Romulea eximia de Vos b red + - yellow Sep. 10361 monadelpha (Sweet) Bak. b red + = yellow Sep. 4036 montana Schltr. ex Bég. b yellow = yellow Aug.-Sep. n/v Volume 85, Number 2 Goldblatt et al. 221 1998 Beetle Pollination Table 2. Continued. Flower Anther/ Beetle pollen Flowering Voucher Shape Color marks Nectar color time number obscura Klatt b red t x yellow Sep 10317 sabulosa Schltr. ex Bég. b red + = yellow Aug.-Sep. пу subfistulosa de Vos b red t ? yellow Sep. 10305 Sparaxis elegans (Sweet) Goldblatt s salmon * tr brown Sep. 4286 grandiflora (D. Delaroche) Ker Gawl. b yellow = tr yellow Aug.—Sep. 2438 pillansii L. Bolus s rec 4 tr yellow ct. 327 Thereianthus racemosus (Klatt) G. Lewis s blue - = blue Nov 10454 Tritonia crocata subsp. hyalina .f.) de Vos b orange + tr yellow Sep.-Oct n/v deusta (Aiton) Ker Gawl. b orange + = yellow ct. 10782 squalida (Aiton) Ker Gawl. b pink - tr white Oct. 9790 Campanulaceae Wahlenbergia capensis (L i A. DC. 5 blue + ? blue Sep.-Oct. n/v Prismatocarpus pedunculatus кай 8 сгеат = = cream Oct.—Nov. 10569 Droseraceae Drosera cistiflora L. s cream/pink + = orange Aug.-Oct. 10282 pauciflora DC. s cream/pink + = orange Aug.-Oct. 10283 These beetles ranged in length from 6 to 14 mm. Body hairiness varied among genera and species, with Anisonyx having the densest and longest hairs (e.g., Figs. 1, 2, 4). The shortest beetles were Het- erochelus arthriticus (collected on Ixia dubia) and Lepthrix stigma (collected on Romulea sabulosa); the longest were Anisonyx ursus, collected on Dros- era cistiflora. А total of one to five beetle species were captured on 40 species of herbs in four fam- ilies (Table 3). Ixia maculata was the only species recorded with as many as five beetle species on its flowers. Less than half (40%) of the plant species, however, were consistently visited by just one spe- cies of beetle (Table 3). BEETLE FORAGING BEHAVIOR Monkey beetles are common on warm days in late winter and spring when ambient temperatures are above 18°С. Individual beetles were observe in flight as early as 9.30 hr and as late as 16.00 hr, but peak activity on flowers was between 11.00 and 15.00 hr. Monkey beetles fly slowly and over relatively short distances. Beetle populations ap- peared to be most dense on inflorescences of As- teraceae and the larger flowers of Aizoaceae sub- fam. Mesembryanthemoideae. In contrast, beetles captured on the flowers of species listed in Table 2 rarely occurred in groups of more than two or three per flower. In these flowers, beetles were most often seen either foraging on pollen directly on the an- thers or pushing their heads into the flower center, leaving the posterior portion of their abdomens prominently displayed. Since the anthers are usu- ally positioned close to the center of the flower and above the beetle marks on the perianth, foraging beetles were usually observed positioned on the beetle marks while they fed. en more than one beetle of the same species was present on a flower, they often displayed intra- specific agonistic behavior, and one or more of the beetles might be driven off as a result. The beetles also used the flowers as sites to assemble and cop- ulate. Compared to other pollinators, beetle visits to flowers lasted a long time, at least several min- utes, or more when mating or evidently at rest. Bee- tles were often observed moving both to another Annals of the Missouri Botanical Garden *s 1-6. Hopliine beetles foraging on flowers. —1. Anisonyx longipes on Aristea 2. --2. Anisonyx ursus on гоа cistiflora. — 3. Anisochelus inornatus on Homeria vallisbelli. —A. Anisor тух ursus on Moraea cf. lurida. — 5. Pachycnema tibialis on Tritonia crocata subsp. hyalina. —6. Lepisia sp. on Hesperantha шш Arrows indicate stigmas of flowe Volume 85, Number 2 1998 Goldblatt et al. Beetle Pollination 223 flower of the same species and to flowers of differ- ent species. Beetle contact with stigmas occurred in one of two ways depending on the length and position of the style. In Aristea spp. and Drosera cistiflora and D. pauciflora (Figs. 1, 2) the style is twisted to lie parallel to, and above, the perianth surface. The stigmatic areas are thus removed from the center of the flower. In this case, beetles brushed against the stigma or crawled over it when they moved across the flower. In the second, more common, case the style is short and the stigma barely pro- trudes beyond the short floral tube or cup. The beetle contacted the stigma ventrally while crawling over it or dorsally when climbing into the floral cup, while either foraging or engaging in agonistic or copulatory behavior. As the color of the pollen is often so distinctive and contrasts so sharply with that of the beetles and the stigmas, pollen could easily be seen clinging to the hairs of the beetles and on the stigmas after the beetles departed. The style branches of Moraea species are broad and arching, concealing the anthers on their abaxial surfaces (Fig. 4). Moraea pollen was deposited on the abaxial stigmatic lobe only when a beetle dust- ed with pollen crawled under a style branch to lie in the center of the flower. The prominent “nectar guides” and dark tepal claws in some species of Moraea may in fact be beetle marks encouraging these insects to move into the center of the flower directly under the gynandrium to contact both pol- len and stigmas. As female beetles continued to feed while mating, both males and females some- times became dusted with pollen and brushed against stigmas. POLLEN LOAD ANALYSES А total of 294 monkey beetles were collected on 40 species of flowering herbs (Table 3) representing 14 genera. More than 90% (270) of the beetles car- ried the pollen of the host flower on which they were collected. However, of these only 28% carried their host plant's pollen exclusively (Table 3). The majority of beetles carried a minimum of two and a maximum of five recognizable pollen taxa on their bodies. The only beetle to carry five pollen taxa was an individual of Pachycnema crassipes, 10 mm long, collected on Gladiolus meliusculus, which had the pollen of G. meliusculus, Romulea eximia, Drosera cistiflora, Spiloxene capensis, and an unidentified member of the Asteraceae clinging to its body sur- face. Pollen washes showed that 28 beetles each car- ried pollen of more than one species of Iridaceae. Of these, four specimens of Anisonyx longipes, col- lected on Aristea lugens, each carried pollen of three species of Iridaceae: А. lugens, Geissorhiza ?aspera Goldblatt, and Moraea villosa. OTHER VISITORS In Sparaxis elegans, S. grandiflora, and S. pil- lansii, beetle species (Table 3) appeared to share flowers with the tabanid fly, Philoliche atricornis. In contrast to the flower-visiting Philoliche gulosa and P. rostrata (Goldblatt et al., 1995; Manning & Gold- blatt, 1997), which have mouth parts 20-30 mm long, P. atricornis has a proboscis only 3-5 mm long. This fly appeared to forage on the flowers of Sparaxis species for nectar exclusively, and carried ample quantites of pollen of the host flower, which in 5. elegans and S. pillansii is a distinctive red- brown color, easily visible to the naked eye as the flies foraged or flew from flower to flower (Table 4). Ixia framesii and Ornithogalum thyrsiflora are visited by both the beetle Lepithrix ornatella and the tabanid, Philoliche atricornis. It also forages for nectar and carries the pollen of both host flowers Tables 3, 4). The beetle Peritrichia pseudoplebia may share Homeria elegans with the muscid fly Orthellia sp. and the native honey bee, Apis mellifera, all of which may contact the stigmas of H. elegans and transport its pollen (Table 4). Homeria ochroleuca receives the most diverse assembly of floral forag- ers. The beetle Anisonyx ursus may share the flow- ers with Apis mellifera and as many as six dipteran taxa. However, the particularly large anthers, prom- inent beetle marks, and depauperate nectar of flow- ers of H. elegans suggest that beetle pollination is more important in that species than in H. ochroleu- ca, with its wider range of visitor species and more ample nectar production. Gladiolus meliusculus, Romulea subfistulosa, and aubenya aurea are visited by a combination of hopliine beetles and solitary bees in the families Andrenidae and Halictidae (Tables 3, 4), and Ar- istea biflora Weim. by hopliine beetles and occa- sionally by Apis mellifera. All three bees, Andrena sp. (Andrenidae), Patellapis sp. (Halictidae), and Apis mellifera, appear to be polylectic foragers, but they do contact the stigmas of their respective flow- ers. “~ LOCAL FLORAL GUILDS At some study sites, there was a tendency for floral pigmentation patterns to converge. This was striking at Sir Lowry's Pass, where Aristea canthar- ophila, Drosera cistiflora, D. pauciflora, and Moraea 224 Annals of the Missouri Botanical Garden Volume 85, Number 2 1998 Goldblatt et al. Beetle Pollination 225 sp. aff. lurida all had cream or lilac flowers with dark centers and orange pollen. At Malmesbury commonage, Aristea lugens and Moraea villosa flow- ers were blue to mauve with very dark markings on the outer tepals. Near Caledon, A. biflora, Drosera pauciflora, and Spiloxene capensis all had whitish to pale mauve, salver-shaped flowers with dark markings near the center. Along the Bokkeveld Es- carpment, yellow-flowered species dominated the beetle-pollinated guild that includes Homeria val- lisbelli, Romulea montana, and Spiloxene serrata, as well as other small-flowered dicots including Ur- sinia sp. (Asteraceae) and Oxalis obtusa Jacq. At other sites obvious color convergence is not evi- dent, and color patterns are broadly mixed. For ex- ample, at sites on the Bokkeveld Plateau, Romulea monadelpha and R. sabulosa have dark red and black flowers, those of Hesperantha vaginata are deep yellow and chocolate, and those of Sparaxis elegans and S. pillansii are pink to salmon with dark red or purple and yellow markings. DISCUSSION Pollination by hopliine monkey beetles obviously conforms to a pattern distinct from classical can- tharophily in the magnoliid angiosperms. In partic- ular, flowers and inflorescences in the pollination systems described above do not have urn-like, hap- lomorphic perianths or overlapping bracts. Polli- nation by monkey beetles in southern Africa more closely parallels beetle pollination by the large scarabs, buprestids, and cerambycids in Australia and the eastern Mediterranean. Perianths are usu- ally open and shallow, anthers do not extrude or shed pollen, and strong odors are uncommon. In fact, similarities between the red-flower guild of the eastern Mediterranean and the monkey-beetle flow- ers of southern Africa are particularly marked. Bright orange to red colors, salver-shaped flowers, and absence of floral odor are well distributed in the beetle flowers of southern Africa and the dark, beetle-like marks of the southern African species may be comparable to the blackened stamens or blackened tepal bases in some of the Mediterra- nean flowers. Few of these flowers, however, appear to secrete nectar as do the Mediterranean species of Anemone, Ranunculus, and Tulipa. А primary difference between beetle pollination in the Mediterranean and in southern Africa is the taxonomic diversity of the Coleoptera involved. In the Mediterranean, pollination of the red-flower guild involves only six species of the genus Am- phicoma (Dafni et al., 1990). The southern African guild of beetle pollinators is far broader, with at least nine genera of floral foragers representing a total of over 20 species. Our results suggest that plant species visited by Hopliini may be specialized for beetle pollination to varying degrees. Thus, where plants offer nectar in shallow floral bowls, generalist entomophily oc- curs and beetles are members of a wider pollinator spectrum that includes native Diptera, Hymenop- tera, and sometimes Lepidoptera. This would ap- pear to be the most likely scenario in Homeria ele- gans, Н. ochroleuca, Gladiolus meliusculus, Ixia framesii, Sparaxis elegans, S. grandiflora, and S. pillansii. Pollination by a range of different organ- isms is known in many flowers; for example, some plant species in the Western Hemisphere and in Australia are pollinated by a combination of birds and bees (Armstrong, 1979). In southern Australia the flowers of a number of woody genera appear to be pollinated by a combination of syrphid flies and small colletid bees (Bernhardt, 1989). Pollination strategies combining beetles and other insects are perhaps less well known, but may be much more common than previously anticipated. For example, Schneider and Buchanan (1980) found that the magnoliid flowers of Nelumbo lutea are pollinated by a combination of bees, flower flies, and can- tharid beetles. It would appear that monkey beetles are a predictable part of generalist entomophily in the flora of southern Africa, much as syrphid flies and small colletid bees are a dominant part of gen- eralist entomophily in southern Australia (Bern- hardt, 1989). In other instances, however, monkey beetles appear to be the sole pollinators and flowers are highly specialized for beetle pollination. The high incidence of pollination by monkey beetles among the Iridaceae of southern Africa has not been widely appreciated. The literature dealing with pollination ecology of the Iridaceae has em- phasized the prominent role of bees, moths, birds (Knuth, 1909; Vogel, 1954), and nectarivorous flies with moderate to long mouth parts (Goldblatt & Bernhardt, 1990; Goldblatt et al., 1995; Manning & Goldblatt, 1996, 1997). However, work by Picker and Midgley (1996), Steiner (1998), and our own €— Figure 7. C. Pachycnema crassipes. Dorsal and/or lateral views of hopliine beetles. —D. Lepisia sp. (Nieuwoudtville). Scale bar = 5 mm —A. Peritrichia subsquamosa. —B. Lepisia dl — . (Drawn by Y. Wilson-Ramsey 226 Annals of the 1 Botanical Garden Table 3. Pollen load analysis of collected beetles. Table 3. Continued. Number of beetles carry- ing pollen load(s) Host Host flr + Other No flr other decas of beetles carry- ng pollen load(s) Host Host flr + Other No sp. pol- other sp. pol- Plant and beetle taxon only sp. only len Plant and beetle taxon only sp. only len IRIDACEAE Pachycnema crassipes 0 2 о 0 Aristea Scelophysa ornatella 0 l 0 0 iflora monadelpha Anisonyx lepidotus l 2 0 0 о, fulvipes 0 1 0 0 cantharophila possa Gora xus 0 3 0 0 Peritrichia subsquamosus 0 3 0 0 Peritrichia pseudoplebeia 0 9 0 1 огаеа bellendenii ugens Anisonyx longipes 1 9 0 0 Неше аи A. ursus ] 9 0 0 . latus : " ш ! Lepithrix ornatella 2 1 0 0 insolens e Anisonyx lepidotus 1 1 0 0 teretifolia Petririchia pseudoplebeia 1 2 0 0 ш окна | etririchia pseudoplebeia rue 3 4 2 1 Babiana рш Ма pseudoplebia 0 6 0 0 rubrocyanea villosa achycnema crassipes 0 1 0 3 Anisonyx longipes 1 6 0 0 Gladiolus A. ursus 0 3 1 0 meliusculus Romulea Lepisia. rupicola | 2 4 0 1 eximia Pachycnema crassipes 0 6 0 0 Lepisia rupicola 0 4 0 0 hirsutus Pachycnema crassipes 0 4 0 0 Anisonyx ursus 0 1 0 l monadelpha Hesperantha Lepisia sp. 1 6 8 0 3 falcata sabulosa Peritrichia pseudoplebeia 1 0 0 0 Anisochelus inornatus 3 3 0 0 vaginata Lepithrix stigma 0 5 1 0 Lepisia sp. 1 7 Б 0 0 subfistulosa Homeria Lepisia sp. 2 0 7 0 0 elegans Sparaxis Peritrichia pseudoplebeia 6 0 0 1 elegans chroleu Lepisia si 1 5 о 0 Anisonyx ursus 2 0 0 0 Anisochelus inornatus 0 5 1 0 vallisbelli grandiflora Anisochelus inornatus 3 5 0 l Peritrichia rufotibialis 0 2 0 0 Ixia Anisochelus inornatus 2 4 0 0 curta Peritrichia sp. 1 0 2 0 0 Lepisia rupicola 0 3 0 0 pillansii Lepithrix fulvipes 0 5 0 0 Lepisia sp. 1 1 7 1 1 Расһуспета crassipes 0 1 0 0 Thereianthus ubia racemosus Heterochelus arthriticus 3 2 0 0 Khoina bilateralis 2 1 0 1 Lepithrix ornatella 0 1 0 0 Pachycnema crassipes 0 2 0 0 Tritonia framesii eusta Lepithrix ornatella 0 5 0 1 Peritrichia hybrida 3 0 0 0 maculata hyalina Heterochelis sexlineatus 0 2 0 0 Pachycnema tibialis 0 6 0 0 Lepithrix longitarsis 2 2 0 0 squalida L. ornatella 0 5 о о Peritrichia sp. 2 7 2 0 0 Volume 85, Number 2 1998 Goldblatt et al. Beetle Pollination 227 Table 3. Continued. Number of beetles carry- ing pollen load(s) Host Host flr + Other No fir other sp. pol- Plant and beetle taxon only sp. nly len HYACINTHACEAE Ornithogalum ubia Peritrichia subsquamosus 1 2 0 0 thyrsoides Lepithrix = 1 0 0 0 L. longita 0 2 0 0 P кену 0 1 0 0 Daubenya aurea Lepisia sp. 2 0 3 0 0 DROSERACEAE Drosera cistiflora Anisonyx ursus 3 1 0 0 pauciflora Anisonyx lepidotus 1 2 0 0 CAMPANULACEAE Prismatocarpus pedunculatus achycnema saga 0 3 0 0 Peritrichia subsquamosus 3 1 1 0 Wahlenbergia capensis Lepisia sp. 3 0 1 0 Total 75 195 8 16 research indicates that beetle pollination must now be accepted as being widespread in the southern African flora. This is especially marked in Irida- ceae, which have undergone their greatest adaptive radiation and speciation in western southern Africa, where flower-visiting Hopliini show their greatest diversity. Modification of the irid flower for pollination pri- marily by monkey beetles has occurred in several genera with diverse floral morphology. In most, the shift seems to be relatively minor, based more on morphological reduction than enlargement. This applies particularly to genera in which an actino- morphic, bowl-shaped flower is ancestral, including Hesperantha, Homeria, Ixia, Moraea, and Romulea. In a few genera with primitively zygomorphic flow- ers, change in symmetry has been necessary; for example, in Sparaxis and Tritonia the adaptive shift has been more pronounced. The Iridaceae polli- nated by monkey beetles are more likely to have Table 4. Pollen load analysis of insects collected on the same species as beetles. Taxonomic affiliations are as follows: Diptera: Philoliche (Tabanidae); Orthellia (Mus- on Scathophaga (Sarcophagidae). Hymenoptera-A po dea: Andrena (Andrenidae); Apis (Apidae); Palins (Hali ictidae). Number of insects carry- ing pollen load(s) Host Host flr + Other No other sp. pol- Plant and insect taxon only only len Gladiolus meliusculus Andrena sp. 0 2 0 0 Homeria legans Apis mellifera 2 0 0 0 Orthellia sp. 0 2 0 1l Scathophaga stercoraria 0 0 0 3 ochroleuca nthomyia 0 0 0 1 Apis mellifera 3 1 0 0 Calliphoridae 0 0 1 0 Orthellia sp. 5 0 0 0 usca sp. 0 2 0 1 Scathophaga stercoraria 0 2 0 3 Syrphidae 0 1 0 0 Ixia framesü Philoliche atricornis 3 3 0 0 Moraea aff. lurida ?Musca sp. 1 0 0 0 Romulea subfistulosa Patellapis sp. 0 2 о 0 Sparaxis elegans Philoliche atricornis 0 © 0 0 grandiflora Philoliche atricornis 0 3 0 0 Patellap 0 3 0 0 pillan Philoliche atricornis 0 2 0 0 Total 14 29 1 9 prominent, dark nectar guides and produce less nectar than the African lridaceae pollinated by long-tongued bees, flies, or other insects (Goldblatt et al., 1995; Manning & Goldblatt, 1996, 1997). In Iridaceae subfam. Ixioideae, which is characterized y the presence of a perianth tube, the tube is also reduced in some way in monkey beetle pollinated species, either in length or diameter, resulting in a 228 Annals of the Missouri Botanical Garden role change from nectar reservoir to pseudopedicel. Flowers pollinated by long-tongued flies in south- егп Africa also typically lack a discernible scent, Gladiolus, Lapeirousia, Nivenia (Goldblatt, 1993; Goldblat & Manning, 1998; Manning & Goldblatt, 1995, 1996, 1997). The main features that distinguish species of Iridaceae as having beetle-pollinated flowers appear to be the distinc- tive beetle-like marks often combined with partic- ularly bright flower colors, which have evolved convergently in many other families; a reduction in the amount of nectar produced; and floral actino- morphy. Salver- to shallow bowl-shaped perianths are also a frequent aspect of this syndrome. Adaptive radiation in response to monkey-beetle pollination is evident in some lineages within sev- eral genera of the Iridaceae, most conspicuously in Ixia sect. Ixia. Nearly all members of that section have spreading tepals, contrasting central marks, a filiform perianth tube, and lack nectar. The tube is blocked by the style and the mouth is closed off by the central filaments that are either coherent or united. Some 20 species are currently included in section /xia, out of a total of 50 species in the genus (Lewis, 1962; de Vos, 1988). Most other species of the genus have campanulate or cylindric perianth tubes that contain nectar in the lower part, which is accessible to nectar-foraging insects (Lewis, 1962; Manning & Goldblatt, 1997, and unpub- lished data), but at least 1. framesii (sect. Morphix- ia) is also visited by monkey beetles. In Sparaxis and Tritonia, floral zygomorphy is most likely an- cestral (based on outgroup comparison, Goldblatt & Manning, unpublished), but zygomorphic flowers or at least zygomorphic perianths characterize species pollinated by monkey beetles or a combination of these beetles and Philoliche atricornis. In these species, the perianth tube is also filiform and blocked by the style and appears to function only as a stalk for the flower. The actinomorphic, beetle- pollinated flowers of these species appear to be de- rived in both genera, an unexpected phenomenon. Pollination in Moraea is, as far as recorded, pre- dominantly by bees (Goldblatt et al., 1989), but pollination by monkey beetles has been document- ed by Steiner (1998) within subgenus Vieusseuxia, notably M. villosa. Several allied species, loosely called peacock moraeas (for their prominent dark tepal markings often with a central pale eye), also have flowers that do not produce nectar, and in ad- dition often have a sterile flap of tissue at the base of the large outer tepal, the limb of which is broad and outspread. This lineage includes some eight species, of which at least M. gigandra L. Bolus, M. neopavonia R. Foster, M. tulbaghensis L. Bolus, and M. villosa have flowers adapted for monkey-beetle pollination. Our observations on M. villosa mirror Steiner's conclusions. Other species of this appar- ently monophyletic group include M. amissa Gold- blatt, M. calcicola Goldblatt, and M. loubseri Gold- blatt, also likely, on the basis of their floral pigmentation, to be pollinated by beetles. Our own observations show that monkey-beetle pollination in Moraea is not confined to this group of species. At least M. bellendenii, M. insolens, and the new taxon here allied to M. lurida also appear to be adapted for monkey-beetle pollination, and accord- ing to Scott Elliot (1891), so does M. tricuspidata (L. £.) G. J. Lewis. Moraea lurida itself has flowers with livid red tepals, sometimes marked with yel- low, a fetid odor, and which produce nectar on the tepal claws. The flowers in our study population were whitish with small yellow nectar guides, dark style branches, and produced neither noticeable odor nor nectar. In other respects, the plants appear similar to M. lurida. In Aristea, four of the seven species of section Pseudaristea currently recognized have flowers adapted in different ways for monkey-beetle polli- nation. The ancestral condition in the genus is pol- lination by pollen-collecting female bees (Goldblatt & Manning, 1997), and the species of all other sec- tions have dark blue tepals, small yellow anthers, and yellow pollen, including as well A. pauciflora Wolley-Dod of section Pseudaristea. Four species of section Pseudaristea have whitish, pale blue, or lilac tepals with contrasting markings and elongate anthers with orange pollen, and beetle pollination has now been recorded for all of them (Table 3). Even at sites where beetles were not observed for- aging on Aristea flowers, pollen washes have shown ample quantities of distinctive Aristea pollen, in- dicating visits to species. For example, the beetle Anisonyx lepidotus, collected on Moraea insolens, showed the presence of pollen of coblooming А. biflora, which grew nearby. e situation in Romulea also suggests that ra- diation and speciation based on monkey-beetle pol- lination are fundamental to the genus. Most of the approximately 80 species of Romulea in the south- ern African winter-rainfall zone have bowl-shaped flowers and a perianth tube with a filiform base, and many have beetle-like marks (de Vos, 1972). Pollinators of these species are either monkey bee- tles exclusively, or a combination of beetles and pollen-collecting bees (Apidae, Halictidae), or in some instances (e.g., R. flava, the flowers of which lack markings) possibly only bees (Goldblatt et al., unpublished data). The floristic diversity of the Cape Floristic Re- Volume 85, Number 2 1998 Goldblatt et al. Beetle Pollination 229 gion is greater than that of such Mediterranean regions as the California Floristic Province, Central Chile, and southwestern Australia (Goldblatt, . One reason for this diversity may be that beetle pollinators have acted, and may continue to act, as unusually powerful mechanisms of natural selection as plant populations become isolated due to dispersal and/or vicariance. hy do monkey beetles, in particular, appear to play such an important role in the radiation of the flora, since they lack the long, specialized mouth parts and rapid flying speeds of large, long-tongued flies and bees (Goldblatt & Bernhardt, 1990; Gold- blatt et al., 1995)? The answer may be that monkey beetles, for all their apparent limitations, are op- portunistic foragers that contact flower stigmas, just like nemestrinid flies and anthophorid bees. Our collections suggest that the majority of beetle-pol- linated geophytes may depend on only one or two beetle species to effect pollination. However, no beetle species appears dependent on the flowers of any single geophyte species as a food source or mating site. This is reflected further by the fact that the overwhelming majority of beetles carry mixed loads of pollen. Consequently, while monkey bee- tles probably find levels of floral diversity adequate, we suggest that the geophytic flora finds the density of beetle pollinators less so. This presumably re- sults in competition between geophytic species for the limited pollinator resource, e.g., fruit set in species of the Cape Flora is known to be асаа (Johnson & Bond, 1997). Speci- ation in the geophytic members of the Cape Flora may thus be driven, in part, by this competition. Floral morphology in the monkey-beetle polli- nated species of the Cape Flora seems conservative, while scent and nectar production are negligible. These floral trends become comprehensible in the light of beetle morphology and behavior. Monkey beetles lack both manipulative forelegs and elon- gated glossae; they do not appear to respond to flo- ral odors, but require a flat surface to mate. Some flower scarabs may have color vision equal to, or much broader than, for example, that of bumble- bees (Dafni et al., 1990). Consequently, the con- vergent evolution of the guild of monkey-beetle- pollinated flowers in southern Africa emphasizes flattened, radial symmetry combined with complex patterns of pigmentation and perianth colors often contrasting with colors of the anthers and/or pollen. he pollination of flowers by monkey beetles in southern Africa appears to have shaped the flora in two ways. First, it is another factor that may help explain the unusually brilliant and broad range of floral colors and contrasting patterns in the Cape Suppl. : 211-226. -----. 198 Flora in general. Second, competition for monkey beetles as pollinators has very likely encouraged both adaptive radiation and convergent floral evo- lution within several plant families, in particular the Iridaceae. Literature Cited ib uie, 1979. ~ pollination mechanisms in the stralian flora—A review. New Zealand J. Bot. 17: 461-5 )8. Beach, J. H. 1982. Beetle pollination of Cyclanthus bi- paritus (Cyclanthaceae). Amer. J. Bot. 72: 346-356. Bernhardt, P. 1989. The floral ecology of Australian Aca- cia. Pp. 127-135 in C. H. Stirton & J. L. Zarucchi (editors), Advances in Legume Biology. Monogr. Syst. Bot. ees Bot. Gard. 29. 993. Natural Affairs: A Botanist Looks at At 22. Between Plants and People. Villard b. New York. — 1996. Anther adaptation in animal pollination. Pp. 192-220 in W. G. D'Arcy & R. C. Keating (editors), The Anther: Form, Function, and Phylogeny. Cambridge Univ. Press, Cambridge, England. . & L. Thien. 1987. Self-isolation and insect pol- lason d in the primitive angiosperms: New evaluations of older hypotheses. Pl. Syst. Evol. 156: ag 176. Bernhardt, A. Schm ida, Y. Ivr baum, C. O'Toole & L. Losito. 1990. Red = е еа fl inati d Israel J. Bot. 39: 81-92 „ van der a> The Principles of Pollination pete Ed. 3. Pergamon Press, New York. M ce P. . The Woody Iridaceae. Timber Press, ortland, ени 1997. Floristic diversity in the Cape Flora i Y Africa. Biodiversity € Conservation 6: 359-37 & P. Bernhardt. 1990. Pollination biology of Niv. enia (Iridaceae) and the peu of heterostylous self- compatibility. Israel Ј. Bot. 3 -111 & Manning. 1996. Aristeas and beetle pol- lination. Veld & je 82: 17-19 & New species of Aristea (Irida- ceae) from South His and notes on the taxonomy and pollination biology of section Pseudaristea. Novon 7: 144. iterranean region. Faegri, г —————. 1998. Gladiolus in Southern Africa. ian. 2. Cape Town лаша: rnhardt & * la Мана 1989, Notes on the 4. mechanisms of Moraea inclinata and M. brevistyla (Iridaceae). P P. ña Evol. 163: 201-209 „J.C anning & P. Bernhardt. 1995. Pollination biology af Lapeir irousia subgenus Lapeirousia (Iridaceae) in southern Africa: Floral divergence and adaptation for long- с fly Шан, Ann. Missouri Bot. Gard. 82: 517-53 Gottsberger, A 1977. Some aspects of beetle pollination in the evolution of flowering plants. Pl. Syst. Evol., 9a. Beetle pollination and flowering xe == spp. (Annonaceae). Pl. Syst. Evol. 167: . 1989b. Comments on flower жігін = beetle Annona and Ro a (Anno- Жарака). 1984. Pollination strategies in Bra- Annals of the Missouri Botanical Garden zilian Philodendron. Ber. Deutsch. Bot. Ges. 97: 391— 410. Hawkeswood, T. J. 1987. Pollination of Leptospermum fla- vescens SM. (Myrtac а. by beetles (Coleoptera) in the Blue Mountains, New South Wales, Australia. Giorn. Ital. Entomol. 3: 261-269 989. Notes on Diphucephala affinis (Coleop- tera, Scarabaeidae) associated with flowers of Hibbertia and Acacia in Western Australia. Pl. Syst. Evol. 168: 1-5 Johnson, S. J. & W. Bond. 1997. Evidence for widespread pollen limitation of fruiting success in Cape wildflowers. Oecologia 109: 530— Knuth, P. 1909. Handbook of Flower Pollination. Clar- endon Press, Oxford. Lewis, G. J. 1962. South 2. Iridaceae. The genus Ixia. J. S. African Bot. 28: 45-195. ‚&Р. сав. 1995. Cupid comes in The not-so-humble fly and a pollination guild in фк боа. Veld & Flora 81: 50—52. —— & 1996. The Prosoeca peringueyi (Dip- tera: Nemestrinidae) pollination syndrome in southern Africa: ued flies wes ~ tubular flowers. Ann. Missouri s Gard. 83: 6 A id longi- rostris (Diptera: Nemestrinidae) pollination guild: Long- tubed flowers and a specialized long-tongued fly-polli- nation system in southern Africa. Pl. Syst. Evol. 206: _69. Orden, E. C., G. S. Raynor, J. V. Hayers & D. M. Lewis. 1974. Manual of Sampling Airborne Pollen. Hafner ss, London. - L. 1902. Descriptive catalogue of the Cole- of South Africa (Lucanidae and Scarabaeidae). Trans 5. 2. Phil. Soc. 12: 1-920. i & J. J. Midgley. 1996. Pollination by mon- ae. eee (Coleoptera: Scarabaeidae: е Flower and colour preferences. African Entomol. 4 14 Prance, С. & J. R. Arias. 1975. А study of the floral biology of Victoria amazonica (Poepp.) Sowerby (Nym- phaeaceae). Acta Amazonica 5: 109-139. Scott Elliot, G. 1891. Notes on the fertilisation of South African and Madagascan flowering plants. Ann. Bot. 5: 5. Schneider, E. L. 8 J. D. Buchanan. 1980. Morphological ne of the Nymphaeaceae. XI. The floral biology of e dp tap Amer. J. Bot. 67: 182-193. Seine er, K. 1998. Beetle pollination of peacock mo- sts Africa. Pl. Syst. Evol. 209: 47-65 Vogel, 5 1954. Blü пенные 'he Р als Elemente der Sippenglioderung ud. 1: 1-338 Vos, M. P. de. 19 ios genus quu. in South Africa. J. S. African Bot. Suppl. 9. ree new species of Ixia L. (Iridaceae) from the Cape Province. S. African J. Bot. 54: 5 602. Whitehead, V. В., J. Н. . & A. С. Rebelo. 1987. Insect pollination in the Cape Flora. Pp. 52-82 in A. G. Rebelo (editor), A Preliminary Synthesis of Pollina- tion Biology in the Cape Flora. CSIR, Pretoria. POLLINATION ECOLOGY AND MAINTENANCE OF SPECIES INTEGRITY IN CO- OCCURRING DISA RACEMOSA L.f. AND DISA VENOSA SW. (ORCHIDACEAE) IN SOUTH AFRICA! S. D. Johnson”, К. E. Steiner’, V. B. Whitehead’, and L. Vogelpoel’ ABSTRACT The orchid Disa racemosa was found to be pollinated by xyloc opine ae anthophorine bees (Xylocopa e te жми c anism ensures that p always attached to the middle pair of legs on the pollinator. Flo D. ra sa are nonrewarding, bu alos attract bees searching for new food sources. Levels of pollination and fiting success were low, varying from 4 to 48% mong the eight study populations. Disa racemosa is often sympatric and co-flowering with its very similar, though sister species Disa venosa. The oniy of the sepals. Although the two species appear одагай ter r that is consistently different between the taxa is the width arrier, rather than ethological or mechanical barriers, is responsible for the maintenance of species integrity in mixed мини. «а Coexistence of closely related species is possible only if there are effective barriers to hybridization (Levin n, 1978). The orchid family is renowned for the apparent ease with which thousands of artificial hybrids have been created by hobbyists (Dressler, 1981). Yet, closely related orchids often occur in sympatry without hybridizing, and an intriguing question is how these species manage to coexist when genetic barriers to hybridization appear to be weakly developed in the family. It is generally thought that specialized pollinator relationships and elaborate floral mechanisms in orchids prevent, or at least minimize, export of pol- len to stigmas of other sympatric species (van der Pijl & Dodson, 1966; Dressler, 1981). Several stud- ies have reported divergent pollination systems in sympatric orchid species that seldom form natural hybrids (Stoutamire, 1974; Smith & Snow, 1976; Chase, 1986; Manning & Linder, 1992; Steiner et al., 1994; Bower, 1996). Isolation of many of these species seems to be based on ethological or me- chanical barriers only, since they can be crossed easily by artificial means. However, there is some evidence that sympatric orchid species that share pollinators may occasionally possess sterility bar- riers, as in the case of the sympatric Cryptostylis species studied by Stoutamire (1975). In this study we focus on a pair of closely related species—Disa racemosa L.f. and Disa venosa Sw.— which often occur sympatrically in the Cape moun- tains of puo Africa. Most authorities have recog- zed th o species as being distinct, although Schlechter (1591) reduced D. venosa to a variety of . racemosa, a treatment that has not been adopted by any subsequent authors (Linder, ) Disa racemosa is one of the more common or- chids in the Cape floral region, growing in marshes and seepage areas. Flowering in D. racemosa is strongly stimulated by fire. Hence, populations flower only at intervals of 5—30 years, which cor- responds to the frequency of fires in the Cape fyn- bos vegetation. Flowering occurs during November and December. The inflorescence of D. racemosa bears from 1 to 15 pink-magenta flowers, which are ! We acknowledge funding from the University of Cape Town Research Council and the Smuts Botanical Fellowship. p Eardley, National Collection of Insects, Pretoria, is thanked for assistance with the identification of Amegilla specie 2 A of Botany, University of Cape Town, Rondebosch 7700, South Africa. Present address: Department of Botany, e is of Natal, Private Bag X01, Scottsville 3209, Pietermaritzburg, South Africa * Compton Herbarium, National gen Institute, Kirstenbosch, Private Bag X7, (Даладан 7735, South Africa. 4 South уты Museum, Саре 5 7 Sunnybrae Rd, Rondebosch, 7700. South Africa. n 8000, South Africa. ANN. MISSOURI Bor. GARD. 85: 231-241. 1998. 232 Annals of the Missouri Botanical Garden 24° Figure 1. Тһе distribution of Disa racemosa (circles) localities of the stu Чу sites are indicated by capital letters i Point Nature Rese : s Bay, D = ip, F = Betty’ Ysterkli and D. renosa (triangles) in the Cape floristic 2. Тһе іп open circles. А - Silvermine Nature 2. 3- - Franschhoek, F — Bains Kloof, G artberg Pas Cape — Gydo Pass. Populations of D. racemosa occur at all of the study sites, and D. venosa is sympatric with D. racemosa at sites F and ( about 40—60 mm in diameter. The flowers do not produce nectar or any other floral reward. Most Disa species have a spur formed from the dorsal sepal. However, the spur in D. racemosa (and also D. venosa) is virtually obsolete, consisting of a mere shallow depression at the back of the dorsal sepal. Disa venosa is remarkably similar to D. race- mosa; the only character that consistently separates the two species is the much narrower dorsal and lateral sepals of D. venosa (see below). The tw species have overlapping distribution ranges and are sometimes found flowering side by side in the same marshes after fire. However, D. venosa is much less common than D. racemosa and is not as often seen or collected. Our interest in Disa racemosa and D. venosa was generated by the observation that the two species are very similar morphologically, yet appear never to hybridize, despite sharing the same habitat and flowering time. As in many other South African or- chids, no previous investigation of pollination bi- ology in the species had been undertaken. This study had the following aims: (1) to confirm whether or not the taxa are readily diagnosable as separate species in the field; (2) to characterize the pollination biology of each species; (3) to determine the levels of pollination and fruiting success in nat- ural populations; (4) to establish the mechanisms that allow coexistence between the species. MATERIALS AND METHODS FLORAL CHARACTERISTICS Floral parts were measured to the nearest 0.5 mm in populations of Disa racemosa at Fran- schhoek and D. venosa at the Bains Kloof site dur- ing 1993 (see Fig. 1 for localities of study sites). Further measurements were also made of dried specimens in local herbaria (BOL and NBC). The flowers of Disa racemosa and D. venosa ap- pear to have a very similar color in the field. To obtain objective measures of floral coloration in the visible spectrum (400—700 nm), we measured the reflectance of sepals of each species with an ACS Volume 85, Number 2 Johnson et al. Pollination Ecology in Disa 233 550m spectrophotometer. Bee vision is known to extend to wavelengths shorter than 400 nm, but no spectrophotometer capable of measuring ultraviolet reflectance was available. Instead, we took photo- graphs on Tri-X film with and without a Corning 7- 60 "black" filter, which only transmits ultraviolet light. The gray scale described in Kevan et al. (1973) was used to standardize exposure of prints. The photographic method has the advantage of be- ing able to reveal floral patterns in the ultraviolet wavelengths, unlike the spectrophotometer, which takes average measurements of reflectance. POLLINATOR OBSERVATIONS Observations of pollinator visits to the flowers of Disa racemosa were made between 1990 and 1995 at eight sites in the Cape Floral Region (Fig. 1). Populations at these sites varied from ca. 100 plants at Betty's Bay to several thousand plants at Franschhoek. Small populations (ca. 20 plants) of D. venosa were found to co-occur with D. racemosa at the Bains Kloof and Swartberg sites (Fig. 1). At all of the study sites, the vegetation had been burnt during the previous season, thus triggering a flow- ering display of the orchids. Pollinators (defined as insects that remove and deposit pollinaria of the orchids) were captured ei- ther directly on the orchids or else while foraging on nearby food plants. A herbarium voucher from the Bains Kloof population (Steiner 2274) 18 depos- ited in NBG, while the other populations are rep- resented by existing collections in NBG and BOL. Insect vouchers are deposited in the South African Museum, Cape Town. POLLINATION SUCCESS Pollination success was measured in most of the populations of Disa racemosa by determining the frequency of pollinarium removal and pollen de- position on the stigmas of randomly selected sam- ples of flowers between 1990 and 1995. We also recorded fruiting success where possible. CROSSING EXPERIMENTS Since our initial observations indicated that Disa racemosa and D. venosa may have the same polli- nators, we made reciprocal crosses between the two species to determine if they are capable of hybrid- ization. Crosses were performed between D. race- mosa at the Franschhoek site and D. venosa at the Bains Kloof site. Although the two species occur sympatrically at the Bains Kloof site, no plants of D. racemosa flowered there in 1994, making it nec- essary to use plants from another site. At each site, we covered inflorescences with pol- linator-exclusion bags while flowers were still in bud. Following anthesis, flowers were randomly as- signed to one of the following treatments: (1) un- manipulated to test for autogamy, (2) hand-polli- nated with pollinaria from conspecific plants, and (3) hand-pollinated with pollinaria from the sister species. To make the latter crosses, we transported freshly cut inflorescences between the sites, a dis- tance of ca. 30 km, and withdrew pollinaria from the anthers immediately before the hand-pollina- tions. This method ensured that all pollinaria used in the experiment were in optimal condition. АП the crosses were made on 29 November 1994, and the fruits were harvested on 31 December 1994 before dehiscence had taken place. To test seed viability, we used standard tissue culture procedures that have been found to work well for germinating seeds of Disa species. Before opening each fruit, we sterilized the outside with 1096 sodium hypochlorite to minimize the chances of fungal and bacterial infection. The fruits were then opened in a laminar flow cabinet and the seeds from each fruit placed in separate 50-ml tissue cul- ture flasks containing a sterile agar-based nutrient medium. The medium consisted of %4-strength MS solution (George & Sherrington, 1984) fortified with 20g banana per litre and 2g peptone per litre. The flasks were placed in a dark cabinet for three months, followed by a 12 hr light/12 hr dark cycle at 20-25°С. Germination and development of pro- tocorms were noted in some flasks 3—5 months after the commencement of the experiment. Seeds that had not germinated after 12 months were consid- ered to be unviable. RESULTS MORPHOLOGY Disa racemosa and D. venosa can be distin- guished by the narrower sepals of D. venosa. Anal- ysis of existing herbarium specimens of the two taxa showed a bimodal distribution of the dorsal sepal width/length ratio (Fig. 2). More detailed measure- ments of floral dimensions in populations of D. racemosa and D. venosa confirmed that the absolute width of the sepals differs markedly between the species (Table 1). The flowers of D. venosa at Bains Kloof were slightly smaller than those of D. race- mosa at Franschhoek, resulting in statistically sig- nificant differences in the means of several char- acters (Table 1). However, sepal width was the only character that showed no overlap in the range of measurements from individuals of the two species (Table 1). 234 Annals of the Missouri Botanical Garden Number of specimens 05 06 07 08 09 1 11 12 Disa venosa Disa racemosa Dorsal Sepal width/length gure 2. Frequency distribution of dorsal sepal width/length in a sample of herbarium specimens of Disa racemosa к, D. venosa. Dorsal sepal width/length is the main diagnostic character separating the two species. SPECTRAL REFLECTANCE The reflectance spectra of flowers of Disa race- тоза and D. venosa are remarkably similar, i porting the impressions gained in the field. The genta-pink color results from strong a of blue and red wavelengths (Fig. 3). The magenta- pink coloration is due to anthocyanin pigments in D. racemosa and D. venosa (Vogelpoel et al., 1985 and is a common color among bee-pollinated flow- ers in the Cape and elsewhere (e.g., Thien & Marks, 1972; Nilsson, 1983). Photographs of the flowers with ultraviolet light showed that the petals, labellum, and rostellum in — both species are UV-absorptive, forming a contrast with the relatively UV-reflective sepals (Fig. 4C, D). POLLINATOR OBSERVATIONS Our observations indicated that Disa racemosa and D. venosa are both pollinated by medium-sized anthophorid bees and large xylocopine bees. Car- penter bees (Xylocopa rufitarsus Lepeletier and X. caffra L.) were observed to visit D. racemosa at five of the sites (Silvermine, Theewaterskloof, Betty’s Bay, Bains Kloof, and Swartberg Pass). Smaller an- thophorid bees (Amegilla niveata and Amegilla spi- lostoma) visited D. racemosa at the Swartberg and Table 1. Measurements of floral characters in Disa racemosa (Franschhoek population) and D. venosa (Bains Kloof). All units are millimeters. NS. = not significant. Disa racemosa Disa venosa Character (n = 10 plants) x + S.D. (range) (n = 8 plants) x + S.D. (range) Dorsal sepal length Dorsal sepal width Lateral sepal length Lateral sepal width Lip length Petal length Distance between viscidia Distance from rostellum to top of stigma Length of pollinaria 20.4 + 1.9 (17-24) 19.7 + 1.8 (15-21) 22.1 + 2.2 (18-26) 14.2 + 2.3 (10-18) 11.8 + 1.4 (10-14) 8.8 + 0.8 (7-10) 10.0 + 0.0 t P 19.4 + 1.4 (18-22) 1.35 NS. 9.3 + 0.7 (9-10) 17.39 ж 20.0 + 1.0 (19-21) 2.70 * 9.0 + 0.0 (8-10) 6.65 ыыы 9.5 + 0.7 (8-10) 4.54 т.ж 10.3 = 0.7 (9-11) 3.69 жж 4.3 + 0.4 (4-5) 0.73 NS. 8.1 = 0.6 (7-9) 2.24 NS. 10.0 + 0.0 0.00 NS. * P « 0.05, ** P < 0.01, *** P — 0.001. Volume 85, Number 2 1998 Johnson et al. Pollination Ecology in Disa 235 100 T Reflectance (%) © Disa racemosa j %- Disa venosa 700 Wavelength (nm) Figure 3. Bains Kloof sites. All of these bees except X. caffra (which may be too large to act as a pollinator) car- ried pollinaria attached to their middle legs. The rarity of D. venosa made it difficult to observe pol- lination in this species. A single carpenter bee (probably X. rufitarsus) was seen to visit flowers of D. venosa and D. racemosa in succession at the Swartberg Pass site. The behavior of the bee was identical on flowers of D. venosa and D. racemosa. Since the orchids in this study have no floral rewards, the bees obviously need to rely on other plants in the community for pollen and nectar re- quirements. The observation that most floral visitors to D. racemosa are female bees (Table 2) suggests that the orchid primarily exploits pollen-seeking in- sects. Pollen-rewarding flowers that occurred at the same sites as D. racemosa included pink-flowered Chironia jasminoides L. (Gentianaceae) and Dros- era regia Stephens (Droseraceae). Flowers of C. jasminoides are buzz-pollinated by female Xylocopa bees (S. Johnson; unpublished). At Swartberg, how- ever, the primary food source for the smaller Ame- gilla bees was Moraea ramosissima (L.f.) Druce (Ir- idaceae), a nectar-producing species with yellow flowers. FUNCTIONAL MORPHOLOGY Pollinaria of Disa racemosa were consistently at- tached to the middle pair of legs on the bees, im- plying a precise interaction between the morphol- ogy of the bees and the flower. Unlike other Disa species, where the dorsal sepal forms a galeate Reflectance spectra for Disa racemosa and D. venosa. chamber, and a spur in some species, the dorsal sepal of D. racemosa is almost flattened and ap- parently serves no function other than visual at- traction. The flowers of D. racemosa and D. venosa differ from those of most other Disa species in hav- ing a floral chamber formed by the petals, rather than the dorsal sepal. When alighting on a flower of Disa racemosa, bees grasp the petals and insert their heads forcibly into the floral chamber (Fig. 5A, B). The bees are presumably attracted to this part of the flower by the contrast between the strongly UV-absorptive petals and the relatively UV-reflective sepals (Fig. 4). In addition, the inner surface of the petals has an alternating pattern of dark and light stripes that may function as "nectar guides." While settled on the flower, the bees clasp the petals with their front legs, while the middle legs rest across the rostellum and the back legs are placed on the lateral sepals. Pollinaria become attached to the basal segment of the middle legs by means of a large sticky visci- dium. It was interesting that pollinaria were always attached to the middle legs, regardless of the great variation in bee size. There appears to be space for just one pollinarium on the first segment of each middle leg, as none of the captured bees carried more than two pollinaria (one per leg), despite be- ing observed to visit several flowers in a sequence. After withdrawal from the anther, the pollinaria are positioned so that the tip is correctly angled to strike the stigma, which is tucked underneath the projecting rostellum. The sectile pollinaria remain 236 Annals of the Missouri Botanical Garden Visible Figure 4. The gray scale is use venosa. Scale bar — E10. attached to the bees and gradually become worn as massulae are torn away from the tip after each con- tact with a stigma (Fig. 5D, E). POLLINATION SUCCESS Bees were relatively uncommon at most of the sites, except Swartberg Pass where seven bees were 277 of reflectance of flowers of Disa racemosa and D. venosa іп e visible and ultraviolet light. Ónsure a comparable range of contrast in each photograph. / C Dis ‚ B Disa racemosa, С, D Disa caught in two days. The paucity of pollinator visits was reflected in the low levels of pollination and 4% and ru varied between 48% (Table 3). The median level of pollination suc- iting success, which cess in seven populations was only 13.6%, while the median level of fruit set in a smaller sample of four populations was 30.3% (Table 3). Since orchid Volume 85, Number 2 1998 Johnson et al. 237 Pollination Ecology in Disa Table 2. Bee species that visited flowers of Disa racemosa at the study sites. Observation Study site time (hrs.) Floral visitors to D. racemosa Food plants for the bees (P ollen source, Sex and pollinarium load N — nectar source) Swartberg ca. 15 Xylocopa rufitarsus "icd Amegilla niveata (Frie Amegilla spilostoma Case Bains Kloof ca. 15 X. rufitarsus Xylocopa caffra L. . rufitarsus ^ [е5] — — ~ sl Silvermine Franschoek ca. 10 X. rufitarsus Betty's Bay ca. 5 X. caffra Gydo Pass ca. 5 A. spilostoma 3(1) E. 9(2), 9(2) Moraea ramosissima (N) 2(0), ?(2), %(1). %(1), 9(2) M. ramosissima (N) 9(2), d (0) Drosera regia (P) Chironia jasminoides (P) 9 (0) 2 (2), 9(0) — 9 (0) жа 9 (0) M. ramosissima (N) 9(2) z flowers are long-lived (ca. 7-14 days in D. race- mosa), “snapshot” measures of pollination success may lead to an underestimate of the final levels of fruit set. HYBRIDIZATION EXPERIMENTS Flowers that were bagged and left unmanipulated did not form fruits, indicating that both species are incapable of autogamy. Both intra- and interspecific crosses resulted in the formation of well-developed fruits with seeds. While seeds resulting from intra- specific crosses germinated and formed vigorous seedlings after five months, seeds resulting from crosses between D. racemosa and D. venosa showed no signs of germination after twelve months (Table DISCUSSION POLLINATION BY DECEPTION The observations reported in this study show that Disa racemosa is pollinated by xylocopine and an- thophorine bees that visit the flowers even though they do not contain a floral reward. The large pink floral display seems to be sufficient to attract bees that enter the general vicinity of the population. These bees are probably sampling potential new food sources and after probing a few empty flowers they usually fly off again. There is no compelling evidence that D. racemosa is a mimic of other re- warding species, although the flowers do bear a general resemblance to pink buzz-pollinated flow- ers, such as Chironia jasminoides, which was sym- patric with the orchid at two of the sites. The pol- lination system of D. racemosa and D. venosa can best be characterized as generalized food-source deception (Ackerman, 1981, 1983; Boyden, 1982; Dafni, 1984; Nilsson, 1992) The Disa racemosa—D. venosa pair have inter- esting similarities to many of the bumblebee-pol- linated northern hemisphere orchids. The most striking similarity is the possession of large pink- magenta flowers (Thien & Marks, 1972; Nilsson, 1980, 1983; Fritz, 1990). A characteristic that ap- pears to be shared by all deceptive orchid species is a very low level of pollination success (Nilsson, 1980; Boyden, 1982; Nilsson, 1983; Ackerman, 1986; Gill, 1989; Fritz, 1990). This may be a con- sequence of insects learning to avoid the unre- warding flowers (Nilsson, 1992). The low levels of fruit set in D. racemosa are clearly due to pollen- limitation, as supplemental hand-pollinations in two populations led to significant increases in fruit set at a whole plant level (Johnson & Bond, 1997). This was most pronounced in the Franschhoek pop- ulation, where hand-pollination led to an increase in fruit set from 4% of the flowers in control plants to 63% of the flowers in hand-pollinated plants (Johnson & Bond, 1997) MECHANISMS OF COEXISTENCE IN DISA A plethora of artificial hybrids has been made be- tween the species of Disa sect. Disa, to which D. racemosa and D. venosa belong (Vogelpoel, 1992 and references therein; Linder, 1990). Disa racemosa has been successfully crossed with several other Disa species, including Disa uniflora Berg, D. cardinalis Linder, D. atricapilla (Harv. ex Lindl.) H. Bolus and D. bivalvata (L.f.) Durieu & Schinz (Vogelpoel et al., 1985; Wodrich, 1995). Disa venosa has been suc- cessfully crossed with D. cardinalis and D. tripeta- loides (L.f) N. E. Br. (Vogelpoel, 1992). Geographical, seasonal, and ethological barriers probably explain why genetically compatible Disa species seldom hybridize in nature. Disa racemosa 238 Annals of the Missouri Botanical Garden Figure 5.— arpenter bee (Xylocopa rufitarsus) settled on a flower of Disa racemosa. The front legs are used to grasp the a iie middle legs are placed over the rostellum, and the back legs are resting on the lateral sepals. The bee's head is inserted into the chamber formed by the petals. —B. Carpenter bee settled on a к of D. venosa. — C. Floral шыш кы of D. racemosa. 2. DS - dorsal sepal, PT — petal, P — pollinarium, V — viscidium, R = rostellum, 5 = stigma, L = lip, LS = lateral sepal. Sc ale = 10 mm. —D. Amegilla niveata (Anthophorinae) with two pollinaria of D. racemosa attac We to its middle legs. 5 Scale = 5 mm. —E. Xylocopa rufitarsus (Xylocopinae) with a single attached pollinarium of D. racemosa. Scale = 5 mm. and D. uniflora, for example, are highly interfertile, Һу anthophorine and xylocopine bees (this study). but have different habitats and pollinators; D. un- А possible natural hybrid between D. racemosa and iflora is pollinated exclusively by butterflies (John- D. atricapilla (a wasp-pollinated species, see be- son & Bond, 1994), while D. racemosa is pollinated low) was discovered recently (Wodrich, 1995). Volume 85, Number 2 Johns al. 239 Pollination е іп Disa Table 3. Pollination and fruiting success in populations of Disa racemosa and D. venosa. Median values are given in bold type Flowers with Flowers with Number of pollinaria pollen on the Flowers that owering stigma set fruit Study site Date individuals % (n) % (п) % (п) Disa racemosa Franschhoek Dec. 1994 ca. 2000 4.9 (41) 7.3 (41) 4.1 (48) Bains Kloof Jan. 1991 ca. 250 — 33.9 (115) 45.3 (589) ape Point Dec. 1991 ca. 100 22.2 (81) 4.9 (81) — Silvermine Nov. 1992 ca. 100 33.8 (59) 10.1 (59) 38.6 (210) Swartberg Dec. 1992 ca. 400 63.6 (44) 47.7 (44) — Y sterklip Jan. 1992 ca. 50 — 21.9 (160) Betty's Bay Dec. 1993 ca. 40 4.8 (42) 9.5 (42) — Gydo Pass Jan. 1995 ca. 150 23.8 (88) 17.0 (88) — 23.0 13.6 30.3 Disa venosa Bains Kloof Nov. 1993 ca. 30 49.1 (57) 19.2 (57) — Steiner et al. (1994) showed that the rarity of natural hybrids between Disa bivalvata and its sym- patric sister species D. atricapilla can be attributed to ethological factors. Although these sexually de- ceptive orchids are interfertile, they are pollinated by different wasp species, thus preventing the for- mation of hybrids. The formation of occasional hy- brids was attributed to beetles that visit the two species indiscriminately. By contrast, no ethological barriers appear to ex- sa. These species have very similar flowers with closely matched reflectance spectra. They appear to share pollinators and, importantly, their column and pol- linarium morphology is identical, thus ruling out the possibility of mechanical barriers to hybridiza- tion. In the absence of ethological or mechanical barriers between these species, the only plausible explanation for the lack of natural hybrids is a ste- rility barrier. This hypothesis was supported by the crossing experiments. Crosses between D. racemosa and D. venosa resulted in seeds that failed to ger- minate, while seeds resulting from intraspecific ist between Disa racemosa an . veno crosses in the same populations germinated readily to form healthy seedlings (Table 4 There is little other evidence for sterility barriers among closely related orchids. Stoutamire (1975) found that several sympatric Cryptostylis species in Australia share the same wasp pollinators without forming natural hybrids. Crossing experiments sug- gested that a sterility barrier may prevent hybrid formation. Genetic barriers are known to occur among less closely related orchids. Nilsson (1980), for example, showed that natural hybridization be- tween bumblebee-pollinated Dactylorhiza sambu- cina (L.) Soó and co-flowering Orchis species is pre- vented by a sterility barrier. Dressler (1981) pointed out that it is difficult to estimate the extent of sterility barriers in the Orchidaceae, since un- successful attempts to hybridize species are seldom reporte The basis for the apparent sterility barrier be- tween Disa racemosa and D. venosa is not known. Differences in cytology can be ruled out as the two species share a diploid chromosome number of 2n — 38 (Pienaar et al., 1989). It is curious that while Table 4. Results of reciprocal crosses to determine the compatibility of Disa racemosa and D. venosa. Floral measurements taken in the parent populations are given in Table 1 Number of Number of Number of Number of Number of asks with seedlings per Pollen recipient Pollen donor crosses swollen fruits fruits flasked seedlings flask D. racemosa D. venosa 8 8 5 0 0 D. venosa D. racemosa 6 6 3 0 0 D. racemosa D. racemosa 3 3 3 3 > 100 D. venosa D. venosa 7 7 3 2 >100 Annals of the Missouri Botanical Garden less closely related Disa species hybridize easily, these two sister species should be inter-sterile. cause of the rarity of D. venosa, we were not able to replicate the crossing experiments on a large scale, but it would be useful to attempt further crosses to determine if the sterility barrier between the two species is absolute or not. Since crosses between D. racemosa and D. venosa resulted in ap- parently normal fruits and seeds with embryos, we assume that the isolating barrier is postzygotic. DIVERGENCE OF DISA RACEMOSA AND D. VENOSA It is difficult to determine if the evolutionary di- vergence between Disa racemosa and D. venosa has an ecological basis. The only consistent external difference between the species is the width of the dorsal sepals. There seems to be no difference in the habitat, pollination biology, or flowering time of the two species. This situation is quite unlike that in the rest of the genus Disa where speciation has been clearly associated with shifts between polli- nators (Johnson et al., . We can only guess at the factors that promoted speciation in the D. ra- cemosa—D. venosa pair. The width of the sepals in D. racemosa and D. venosa does not have any ob- vious adaptive significance for bee-pollination. Pre- sumably this character diverged through non-adap- tive processes, such as genetic drift in small isolated populations, while the overall divergence in the genomes of the two daughter species was profound enough to cause a sterility barrier. We doubt that a sterility barrier between D. racemosa and D. venosa could have arisen through natural selection, as it is difficult to imagine why a hybrid between species that share near-identical floral morphology and habitat would suffer reduced fit- ness in terms of pollinator attraction or seedling establishm ent. imately, it is difficult to determine with cer- tainty blo a sterility barrier between congeners arose through natural selection (reinforcement), as a single mutation that preceded divergence in sym- patric populations, or as a pleiotropic consequence of character divergence in allopatric populations (Grant, 1994). The present-day distribution of closely related species offers few clues about the mode of speciation. For example, although D. ra- cemosa and D. venosa are often found sympatrically, this does not exclude the possibility that they di- verged in allopatry and later expanded their ranges to become sympatric at some sites. The findings of this paper contradict much of the current dogma about isolating mechanisms in or- chids. The Disa racemosa—D. venosa pair is one of the few known cases where sterility barriers, rather than divergent pollination systems or floral mech- anisms, are responsible for species integrity of sym- patric orchid species. Literature Cited Ackerman, J. D. 1981. De 2. of Calypso bul- bosa var. occidentalis eri ae): А food deception system. Madroño 28: 101—11( Я 1983. e bee pollination of the orchid Coc Моана lipscombiae: A food source mimic. Amer. J. Bot. 70: 830—834. 1986. Mechanisms and evolution of food-decep- tive pollination systems in orchids. Lindleyana 1: 108— ГІ: Howes CoG; reproductive isolation in sexually decepti Chiloglottis (Orchidaceae: С aladeniinae EN Bot. 44: 15-33. 1996. Demonstration of Jue CARI най e species o Austral. J. Воудеп, T. C. 1982. The pollination biology of Calypso bulbosa var. americana (Orchidaceae): Initial deception of bumble bee visitors. Oecologia 55: 178-184. :hase, M. У. 1986. Pollination ecology of two sympatric, — ната flowering species of Leochilus іп Costa . Lindleyana 1: 141-147. Dun. A. 19 Mimicry and и in pollination. Annual Ж Ecol. Syst. 15: 259-278. Dressler, R. 1981. The Orc hidas Natural History nd Classific dor Harvard Univ. Press, Cambridge, Mas заспизе 5. ritz, A. L. 1990. Deceit pollination of Orchis d (Orchidaceae) on the island of Gotland in the Ba ui i syste m, 2. : J. Bot. 9: 577-587. seorge, E. F. & P. а Tes Plant Propa- 1 ч P C e Egegetics, Wiltshire. Gill, ;. 1989. Fruiting failure, pem efficiency id 5. іп orchids. Pp. 456—481 іп D. Ой A. Endler (editors), Speciation s из Consequences. Sinauer, Sunderland, Massachus yrant, 294. Modes and origins el mechanic P and ethological isolation in angiosperms. Proc. Natl. Acad. Sei. U.S.A. 90: 7729-7733. Johnson, S. D. & W. J. Bond. 1994. Red flowers and butterfly pollination i in the fynbos of South Africa. Pp. 137-148 in M. Arianoutsou & R. H. Groves (editors), e rie Interaction in т Ecosystems. Kluwer Academic Press, Dor & == 1997. iden псе e for widespread pollen limitation of fruiting success in Cape wildflowers. Oeco- logia 109: 530—534 a > т , H. P. Linder & К. E. Steiner. 1998. Phyloge ny and bos radiation. of ied systems in Disa (Orchidaceae). Amer. J. Bot. 85: 402411. vevan, P. G., N. D. Grainger, G. A. Mulligan € A. R. Robertum. 1973. e for measuring reflec- tance and color in the insect se human visual spectra Ecology 54: 924—926. The origin of isolating mechanisms in flow gud plant Evol. Biol.11: 185-317. Linder, H. P. 1981. Taxonomic ла on the Disinae ІП. A revision of Па Berg excluding sect. Micranthae. Con Bolus Herb. 9: 1-370. 19 4. in Disa (Diseae-Orchidoideae). L айву yana 5: 226-23( Manning, J. C. & H. p. deus 1992. Pollinators and Volume 85, Number 2 1998 Johnson et al. 241 Pollination Ecology in Disa evolution in Disperis, or why are there so many species. S. African J. Sci. 88: 38-49. Nilsson, L. A. 1980. The pollination ecology of Dacty- orhiza sambucina (Orchidaceae). Bot. Not. 133: 367— 385. -----. 1983. Аптека of Orchis mascula (Orchi- daceae) е Bot. 3: 157-179. 2. Orc Sm palla biology. Trends Ecol. Evol. | 255-25 Pienaar, R. V., G. d Littlejohn, M. S. Hill, C. du Plessis & S. Cywes. 1989. Chromosome numbers of various Disa hybrids and their ае and complex hy- brids. S. African J. Bot. : 99. Pijl, L. van der & C. H. dh 1966. Orchid Flowers: Their Pollination per Evolution. Univ. Miami Press, Coral Gables Schlechter, R. 1901. Monographie der Diseae. Bot. Jahrb. 313. . E. Snow. 1976. Pollination ecology of Phxanihera ( baña ciliaris and P. (Orchidaceae). Bot. Gaz. 137: 133-140. blephariglottis Steiner, K. E., V. B. Whitehead i S. D. Johnson. 1994. Floral and pollinator divergence i tive South African orchids. AW r. Stoutamire, W. P. 1974. fringed orchids Platanthera psycodes and P. grandiflora. Brittonia 26: 42-58. 1975. dem i prp E раи orchids. ree Orchid Soc. Bull. 44: 226— Thien, L. B. & B. G. Marks. 1972. Чы floral biology of Arethusa bulbosa, Calopogon tuberosus & Pogonia ophioglossiodes (Orchidaceae). Canad. J. Bot. 50: 2319 А ics Disa hybridization in ps western status and future к Part 5. The 2. Mosa: ^ al qp isa v d Disa caules- cens, and future trends. Arica Orchid J. 23: 82- 87 ). W. Van Der Merwe & B. p Anderson. 1985. A gum brn of Disa racemosa—A rare се with a great future. = Orchid Soc. Bull. 54: 47—50. 7. h, K new natural келн from the western 1995. e? S. Џина pus J. 26: TRIBAL PHYLOGENY OF THE ASTERACEAE BASED ON TWO NON-CODING CHLOROPLAST SEQUENCES, THE trnL INTRON AND trnL/ trnF INTERGENIC SPACER! Randall J. Bayer?? and Julian R. Starr’ ABSTRACT Asteraceae are the largest family of dicotyledonous plants and have long been known for their taxonomic complexity. The ubiquitous parallelisms in morphology within the fam ieae, and of the Asteroideae, the Inuleae—Plu lianthoid clade (Helenieae, Heliantheae s. str. of the Asteroideae, and the paraphyly of the Tem s.l. by our analysis. Our study illustrates the utility of the ies on and trn among tribes of the Asteraceae. Using approximately 8 resolution to phylogenies based o rphological and molecular data sets. Cichorioideae are D be The primary clades of the aa are the Mutisieae-Cardueae, aea Astereae—Anthemideae, г. The Inuleae-Pluchee Пу have made phylogenetic reconstruction and tribal circum- e Asteroideae are a mo a dde? group, but the iabeae- Vernon- Seni Graphics and the he- ade is sister to the remainder alieae, Inuleae s. str., x Plucheeae) i is firmly supported Fi intergenic эрас CI for p. we were able to produce a phylogeny of comparable on well-known coding m such as rbcL and ndhF. For phylogenetic inference at the family level the trnL intron and trnL/F spacer provide 2. levels of resolution to longer coding sequences (e.g. rbcL, ndhF), while having the advantage of = muc o amplify and sequence due to their short lengths and nd deletions ок found in this region are easily aligned and аге phylogenetically informative, thus adding du) to the information content per base pair sequenced. Asteraceae are the largest family of dicotyledon- ous plants (ca. 23,000 spp.) and have long been recognized for their taxonomic complexity. Ubiq- uitous parallelisms in morphology within the family have made it difficult to find conservative (non- d characters that can be used reliably phylogenetic reconstruction (Carlquist, 1976). Casaini (1826) was the first to divide the Asteraceae into tribes (19 tribes), and the first to suggest their natural relationships. Significant early contribu- tions were also made by Bentham (1873), who re- duced the number of tribes to 13, and Cronquist (1955), who placed Heliantheae at the base of his 12 recircumscribed tribes. Hoffmann (1894) rec- ognized two distinct lineages within the Asteraceae: the Liguliflorae, in which he placed the single tribe Lactuceae; and the Tubuliflorae (= Asteroideae of modern authors), in which he placed all the re- maining tribes. Subsequent authors have continued to recognize two lineages within the family, but their circumscriptions have differed dramatically. Among these major revisions, Carlquist (1976) was perhaps the first to recognize an expanded Cicho- rioideae (— Liguliflorae) by placing 6 tribes within each of his subfamilies Cichorioideae and Astero- ideae. Beginning in the late 1980s, the discovery and subsequent analysis of a phylogenetically in- formative inversion in the cpDNA of Asteraceae ! We thank Bruce Bohm for supplying leaf material of Chuquiraga and Doniophyton, Mike 2. rd becaria and nd Shaw Geoff Burrows for Stuartina, used in this stud We also thank Brett Purdy, Travis Minish, a cis for help in the field during the collection of other leaf material and Reneé LeClerc and dips Ramprasad n 21. in the lab. The manuscript was improved through reviews by Lily Ainouche, Káre Bre on. The horticultural skills of Ann Rolls and Steven Williams are 154 Leigh Johns ве Ста is Bruce Ford, and eatly appreciated. This research was supported by NSERC grant A3797 from the Natural Sciences and “ме ын ыз рен. Council of Canada to R. J. B. ви of Alberta, Department of Biological Sciences, Edmonton, Alberta, ТӨС 2E9, Canada. штеп! Address: CSIRO—Plant Industry, Centre for Plant нашы Research, Australian National Herbarium, BH Box 1600, Canberra, 2601, Australia. Email: r.bayer@pi * University of Manitoba, Е of Botany, Winnipeg. Manitoba. R3T 2N2, Canada. il Email: starr@cc.umanitoba.c ANN. Missouni Bor. GARD. 85: 242-256. 1998. Volume 85, Number 2 1998 Bayer & Starr Asteraceae Phylogeny 243 (Jansen & Palmer, 1987), in addition to the mor- phological work by Bremer (1987) and others, dem- onstrated that the former Barnadesiinae (in Mutis- ieae) was monophyletic. This work also indicated that this subtribe was the basal group in the As- teraceae and worthy of being recognized as the ar- chaic subfamily, the Barnadesioideae. As a result of these and other morphological and molecular studies (Bremer, 1987; Michaels et al., 1993; Gus- tafsson & Bremer, 1995; Kim & Jansen, 1995), it is becoming clear that the Asteraceae arose in South America (Bremer, 1992) and are probably sister to the South American endemic family Ca- lyceraceae. Phylogenetic relationships within the Asteraceae have long been an area of debate, beginning with Cassini (1826) and continuing to the present day. Although much has been accomplished over the past 15 years to resolve phylogenetic relationships among the tribes, the taxonomic limits and relationships of many tribes are still unclear. In particular, the ques- tion of the monophyly of the Cichorioideae and the “old” Inuleae are important relationships that have not been resolved. In addition, the tribal circum- scriptions of tribes such as the Helenieae and Eu- patorieae are still very much in doubt. Since the advent of molecular systematics, pro- tein-encoding gene sequences have been very use- ful for resolving higher-order questions (e.g., Chase et al., 1993). However, in groups such as the As- teraceae that have undergone a rapid radiation (Carlquist, 1976), coding regions may not always provide sufficient information to resolve relation- ships. In this study we explored the utility of using two relatively short, non-coding chloroplast DNA sequences, the trnL intron and trnL/trnF intergenic spacer, to resolve phylogenetic relationships among tribes of the Asteraceae. The availability of rbcL- (Kim et al., 1992) and ndhF- (Kim & Jansen, 1995) derived phylogenetic trees allows for a direct com- parison of the phylogenetic utility of the trnL intron and trnL/trnF intergenic spacer relative to these widely used sequences. MATERIALS AND METHODS OUTGROUP SELECTION Outgroup taxa were selected on the basis of the ndhF analysis of Kim and Jansen (1995), the re- striction fragment length polymorphism (RFLP) studies by Jansen and Palmer (1987, 1988), the rbcL analysis by Kim et al. (1992), and the mor- phological works of Bremer (1987, 1994). Although attempts were made to use groups from outside the Asteraceae to polarize trees, alignments were am- biguous and could not be used for phylogenetic re- construction. Two members of the Barnadesioideae (i.e., Chuquiraga and Doniophyton) were thus cho- sen as a functional outgroup (Watrous & Wheeler, 1981). The basal position of this subfamily is con- firmed in all the above-mentioned studies, and its use as an outgroup for the remainder of the Aster- aceae Is not without precedent (Jansen et al., 1990, 1991; Keeley & Jansen, 1991). INGROUP SAMPLING Tribal circumscriptions and nomenclature are based on the treatment of the Asteraceae by Bremer (1994). One or two members from each of the rec- ognized tribes were sequenced. For the 26 taxa used in this study, all sequences were generated by us (Table 1) from fresh leaf material, except for representatives of Artemisia, Chuquiraga, Donio- phyton, Liabum, and Osteospermum, which were ob- tained from dried material. Material was collected in the field for some genera, whereas other samples were obtained from commercial sources; herbarium vouchers are cited in Table 1 DNA ISOLATION, AMPLIFICATION, AND SEQUENCING Total DNA was isolated as outlined in Bayer et al. (1996). The trnL/F region was amplified via the poly- merase chain reaction (PCR) using Tag DNA роју- merase on a GeneE? thermal cycler (Techne Incor- rated, Princeton, NJ). The PCR reaction mixture consisted of 5 pl of 20X reaction buffer, 6 pl of 25 mM magnesium chloride solution, 16 wl of a 1.25 mM dNTP solution in equimolar ratio, 25 pmol of each primer, 10—50 ng of template DNA, and 1.0 unit of polymerase іп a total volume of 100 wl. The PCR samples were heated to 94?C for three minutes prior to the addition of DNA polymerase to denature un- wanted proteases and nucleases. The double-stranded PCR products were produced via 30 cycles of dena- turation (94°С for 1.0 min), primer annealing (48°С for 1 min), and extension (72°С for 2 min). A 7-min final extension cycle at 72°С followed the 30th cycle to ensure the completion of all novel stands. The trnL intron and trnL/trnF spacer, hereafter referred to as trnL5'/F (Fig. 1), was amplified as a single piece using primers “с” and “f” of Taberlet et al. (1991). Primers “а” and “b” (Fig. 1) were used to estimate the approximate size of the trnT/ L intergenic spacer in the Asteraceae, but these were not sequenced. Double-stranded PCR prod- ucts were cleaned by differential filtration using Millipore Ultra-free®- MC tubes (30,000 NMWL fil- ters) prior to sequen The ы PCR products were then 244 Annals of the Missouri Botanical Garden Table 1. Collections of Asteraceae used in the trnL/trnF sequencing study. Presented are species, origin (loc ation of voucher), and accession numbers. All voucher numbers beginning with two letters (signifying a s state, province, or from cultivation (GH)) followed ad 5 а are collections of Bayer or Bayer et al. GenBank accession numbers for the sequences (intron, spacer) are giv Species Accession numbers and (voucher location) Source GenBank (intron, spacer мене 1) Ageratum houstonianum Mill. 2) Antennaria luzuloides Torr. . Gray 3) Artemisia tridentata Nutt. ) Aster novae-angliae ) Calendula officinalis L. ) Chuquiraga aurea Skottsb. ) Cirsium subniveum Rydb. 8) Crepis tectorum 1 ) I )) ) ) 3) G Doniophyton anomalum (D. d. Echinops exaltatus Schrad. Gaillardia aristata Pursh gens R. Br ~ Ја ы 8 = = 5 ~ % x > & ч 5 ~ > = - S a E E 5 5 z E Е ч ess. 18) Matricaria matricarioides ess.) Port 19) Osteospermum clandestinum ess.) Norl. 20) Petasites frigidus (L.) Fr. 21) Senecio vulgaris L. 22) Stokesia laevis Greene т. Жозе cylindriceps (1. Black) Dunlop Рн Skiori muelleri Sond. 25) Tagetes pa 26) с ехзсара (Richard- son) Porter GH-95011 (CANB) OR-91002 (ALTA) CO-90072 (ALTA) AB-95003 (CANB) GH-95009 (CANB) Stuessy et al. 12911 (OS) WY-90044A (CANB) AB-95002 (CANB) Stuessy et al. 12857 (OS) AB-95005 (CANB) GH-95006 (CANB) GH-95012 (CANB) GH-95004 (САХВ GH-95007 (CANB GH-95013 (CANB AB-95007 (CANB —— — МУ Dillon & Sánchez 6253 (F) AB-95005 (CANB) WA-94070 (CANB) Starr 96001 (WIN) AB-95006 (CANB) GH-95014 (CANB) WA-94049 (ALTA) Burrows s.n. (CANB) Bayer s.n. (CANB) СО-93037 (CANB) Commercially grown plants U.S.A.: Oregon Colorado Commercially grown plants Commercially grown plants Argentina U.S.A.: Wyoming Canada: Alberta Argentina Commercially grown plants Commercially grown plants Commercially grown plants Commercially grown plants Commercially grown plants Commercially grown plants Commercially grown plants Peru: Prov. Huancabamba Canada: Alberta Australia: Western Australia Canada: Manitoba Alberta Commercially grown plants Canada: Australia: Western Australia Australia: New South Wales Commercially grown plants U.S.A.: Colorado U82012, U82013 U82014, U82015 FB2DID, U82017 082026, U82027 202028, 082029 ‚ 082031 32, U82033 U82034, U82035 U82036, U82037 U82038, U82039 U82040, U82041 U82042, U82043 U82044, U82045 U82046, U82047 U82048, U82049 U82050, U82051 U82052, U82053 U82054, U82055 U82056, U82057 U82058, U82059 U82060, 082061 082062, 082063 used as templates in cycle sequencing reactions, which employed three primers (Taberlet et al., 1991) to sequence the two regions ... terminal primers “с (Amersham). and “f” and an internal primer “d” (Fig. 1). Sequencing primers were 5” end-labeled in a preliminary dae oe T4 polynucleotide kinase and [y? The double-stranded d were then cycle-sequenced using the dideoxy chain ‚ including the termination method (Sanger et al., 1977) with use of Promega's fmol®*1 Sequencing System (Pro- mega Corporation, Madison, Wisconsin). Ап an- nealing temperature of 57?C was used for primer "f," while temperatures ranging from 60 to 62°С were employed for primers “c” and “4.” The су- TAL V (Higgins et al., cle-sequencing protocol followed the manufac- turer's instructions. Termination products were separated in 6.096 polyacrylamide gels (0.4 mm thickness; 1X TBE buffer); the gels were fixed in 10% acetic acid for 20 minutes, washed in dis- tilled water, and allowed to air-dry. They were then used to expose Kodak BIOMAX®-MR film for 8-48 hr depending on the intensity of the radio- active signal from the gel. SEQUENCE ANALYSIS AND PHYLOGENETIC RECONSTRUCTION Sequences were aligned initially using CLUS- 1992), then adjusted man- Volume 85, Number 2 98 Bayer & Starr Asteraceae Phylogeny 5'>GGGGATAGAGGGACTTGAAC<3 | 5'>CGAAATCGGTAGACGCTACG<3' | 5'>ATTTGAACTGGTGACACGAG<3' U Intron | "и (UA L| spacer тү А) (САА) Figure 1. exon (UGU), the trnL/T intergenic space and the trnF exon (САА), copy regions, and the in primers ( ually (Swofford & Olsen, 1990) to minimize gap number using SeqApp vers. 1.8A (Gilbert, 1992). Several divergence weights [20%, 4096, 60% (the default), and 80%] were explored during sequence alignment (Delay Divergence Option of Clustal V), including several combinations of the gap-opening penalty (GOP) and gap-extension penalty vae op- tions of CLUSAL V (Higgins et al., 1992). GOPs of 10 (the default) and 100 were explored in t; per- mutations with GEPs of 5 (the default) and 10. The different permutations resulted in very similar alignments, and one was chosen as a starting point to continue with manual adjustment of the align- ment. The alignment of the sequences necessitated inference of many insertions and deletions (Table 2). 424 - 453 Бр э |е 255 - 345 bp Торассо (Nicotiana tabacum) cpDNA 156 000b.p. Structure of chloroplast DNA in Nicotiana tabacum L. r, the trn intron, the ігпі, relative to the commonly sequenced genes rbcL, r verted repeats (two bold semicircular regions). Re c, d, and f) used in PCR and sequencing are indicated, along with their base sequences (Solanac eae). Presented are positions of the trn T 3' and 5' exons (UAA), the trnL/F intergenic spacer, lative positions of the Taberlet et al. (1991) Small portions of the trnL and trnF genes were also sequenced along with the intron and spacer sequences. No variation was observed among the taxa for any of these gene regions with the excep- tion of a single point mutation (CT) at the 3" position of the 5' segment of trnF. Artemisia, Aster, ctuca, Matricaria, Petasites, Senecio, and Town- sendia have “Т” at this position, whereas all other taxa have a *C." This character was included in all analyses. The proportion of nucleotide differences between taxa was calculated using the “Show Distance Ma- trix" option of PAUP. A total of 101 phylogeneti- cally informative base pairs and 32 indels from the trnL5'/F region was available for use in the analysis 246 Annals of the Missouri Botanical Garden able 2. Insertions and deletions in the chloroplast trnL intron and the trnL/F intergenic spacer in the Asteraceae. Presented are type and size of the indel, start point of the indel based on the first bp of the intron sequence (trnL intron is 1-533; trnL/F spacer is 534-913), and the species in which the mutation occurs (numbers of the species refer to those given in Table 1). Also given are the repeat sequences for those insertions that are repeats of adjacent sequences, as well as the locus of the start point from which the repeat is derived. * — potentially phylogenetically informative indels. Indel Туре o Size in Direct repeat Fragment Repeated Mutated number mutation (bp) sequence from base from base species 1* del 1 144 2. 24. 26 2* del l 205 4, 26 3* del 1 264. 5, 19 4* del l 273 4, 6. 7 5% del 1 293 1. 11, 14, 25 6 del 1 316 ж del 1 388 8 8* del 1 402 1, 11, 14, 15, 23, 25 9 del 1 591 1 10* del 1 755 2, 24 11* del 1 844 1. 11, 14, 25 12 del l 887 15 13 del 2 303 9 14 del 2 309 9 15* del 2 606 8, 17, 22 16 del 2 670 2 17* del 2 672 3, 18, 20, 21. 24, 26 18 del 2 679 20 19 del 2 763 20 20 del 3 602 17 21 del 4 189 5 22* del 4 283 1, 11, 14, 25 23 del 4 295 19 24 del 4 332 26 25 del 4 592 6 26* del 4 621 4. 26 21 del 4 660 20 28 del 5 432 21 29 del 5 598 9 30 del 5 743 22 3l del 5 15 13 32 del 6 666 33* del 7 635 9, 16 34. del 8 115 17 35 del 8 598 19 36* del 8 682 15, 17, 22 del 8 806 17. : 38 де] 9 420 13 39 де] 9 617 19 40% del 9 654. l, 11, 14 41* del 9 747 20, 21 42 del 9 198 5 43* del 10 115 5, 19 44 del 10 602 ) 45 del 10 690 9 46% del 10 754 5, 23 del 10 835 5. 19 * del 11 123 6. 7 49* del 11 761 5, 6, 7. 19 Volume 85, Number 2 1998 Bayer & Starr Asteraceae Phylogeny Table 2. Continued. Indel Type of Size in Direct repeat Fragment Repeated Mutated number mutation (bp) sequence rom base from base species 51* del 18 295 3, 18 52 del 19 652 16 53 del 82 730 25 54 ins 1 141 13 55 ins 1 143 3 56* ins 1 613 3, 5, 9, 18, 19 57 ins 1 626 58* ins 1 683 1, 18, 19, 20, 21 59 ins 1 691 60* ins 1 841 1, 11, 14, 21, 25 61 ins 2 666 62 ins 4 704 4 63* ins 6 288 2, 12, 24 64 ins 7 233 11 65 гереа! 4 AAAA 149 145 18 66 repeat 4 (8) AATC[AATC] 337 345 18 (3) 67 repeat 5 AATAC 278 284 4 68» гереа! 5 ТТСАА 327 322 2, 3, 18 69 гереа! 6 СА 389 396 12 70* repeat? 6 C(A/G)TT 606 612 3, 5, 8, 9, 17, 18, 19, 22 (C/T)(A/T) 71 repeat 6 AACTTA 782 776 9 72 repeat 7 GATCAAA 360 380 1 (3% repeat? 7 TA - 619 612 All (except 5, 9) (T/A)C(T/G/A) 74* repeat? ¢ GT(GCA)A(CT)- 673 682 All (except 5, 6, 7, 8, 10, A(CT) 75 repeat 20 GATCAAATCA- 360 380 TTCACTCCAT of the 26 taxa. Invariant sites and autapomorphic base changes were removed from the analysis using the “Ignore Uninformative Characters” option. trnL5'/F sequences for all taxa are available from GenBank (see Table 1 for accession numbers) or can be obtained from the authors upon request. In- sertion/deletion events (indels) were scored as bi- nary characters (Table 2), following the recommen- dations of Wojciechowski et al. (1993), with gaps treated as missing. Primary sequence lengths an G/C contents were determined in Amplify 1.2 (En- gels, 1993). These values were manually recalcu- ated for those sequences with ambiguous nucleotide characters (e.g., N, Y, R), which are un- acceptable to the program. Sequence data were analyzed using PAUP ver- sion 3.1.1 (Swofford, 1993) ylogenetic recon- struction was performed on unweighted characters by heuristic searches with “simple,” “closest,” and “furthest” addition of taxa. Heuristic searches em- ploying a random-addition sequence of 1000 rep- licates were also conducted to search for other is- lands of most parsimonious trees (Maddison, 1991). Three separate data sets were analyzed. The first excluded all potentially phylogenetically inform tive indels, and the second included all indels. The third data set included only those potentially phy- logenetically informative indels greater than 2 bp in length. This follows the recommendations of van Ham et al. (1994) and Lloyd and Calder (1991), who suggested that most of the homoplasy in in- sertion/deletion events is accounted for by smaller indels. Strict and 50% majority rule consensus trees (Margush € McMorris, 1981) were construct- ed for the set of equally most-parsimonious clado- ams. The distribution of phylogenetically infor- mative characters (point mutations and indels) on tree topologies was examined using MacClade ver- sion 3.0 (Maddison € Maddison, 1992). Bootstrap (Felsenstein, 1985) and decay (Bre- mer, 1988; D , 1992) analyses were used to estimate the robustness of clades. Bootstrap onoghue et al. 248 Annals of the Missouri Botanical Garden Table 3. Sequence characteristics of the trnL sequenced in this study. intron, trnL/F spacer, and combined trnL-trnL/F non-coding region trnL intron Combined (trnL intron + trnL/F trnL/F spacer Length range (bp) 424—453 (255)308—345 (685)733-793 Length mean (bp) 437.50 329.54 761.65 Aligned length (bp) 505 369 874. content range (%) 33.6–36.2 33.8-38.1 33.4-36.3 G + C content mean 34.9 35.5 35.1 Sequence divergence (96) 1.1–6.4 1.2-11 1.0-7.7 Number of variable sites 96 (19.0%) 123 (33.3%) 219 Number of potentially informative sites 43 (8.5%) 58 (15.7%) 101 Number of constant sites 409 (81.0%) 246 (66.7%) 655 Number of autapomorphic sites 53 (10.596) 65 (12.9%) 118 Number of indels 3] 44. 75 Indel size range (bp) 1-29 1-20 (82) 1-92 Ratio of indels to potentially informative sites 1:1.39 1:1.32 1:1.57 analyses employed 100 replicates of heuristic : M E addition. а searching. Decay nalyses were pe ed using a converse con- straint (ENFORCE CONVERSE command) method aum et al., 1994). The amount of phylogenetic information in the parsimony analysis was assessed by use of the consistency index (C.I.; Kluge & Far- ris, 1969) and the retention index (R.I.; 1989). Farris, RESULTS Length variation for the entire trnL intron ranged from a low of 424 bp in Matricaria to a high of 453 bp in Gazania (Table 3). The proportion of nucle- otide differences ranged from 1.1 to 6.496 between all species of Asteraceae, and from 2.7 to 6.496 between species of the Barnadesioideae and the rest of the Asteraceae (Table 3). The trnL intron had an average G/C content of 34.9% (33.6 to 36.290) (Table 3). The complete trnL/F intergenic spacer (corre- sponding to рш 49876-50231 in the Nicoti- in length from 255 bp in Tagetes to 345 bp in Aster (Table 3). The great range in length is somewhat misleading, because Tagetes has a unique 82 bp deletion; the next shortest se- quence was that of Osteospermum (308 bp) (Table 3). The proportion of nucleotide differences in the spacer is greater than that found in the trnL intron and ranges from 1.2 to 11.7% between all species of Asteraceae, and from 2.2 to 10.0% between the Barnadesioideae and the ingroup (Table 3). Like the intron, the r has an average G/C content of 35.5% (33. 8i А 38. 1%) (Table 3). Within Asteraceae, the proportion of nucleotide differences in the combined spacer and intron se- quences ranged from 1.0 to 7.796 (Table 3). Total average A/T content was 64.996, whereas G/C con- tent was 35.146 on average (Table 3). A total of 101 sites (11.396 of the sequence length) provided po- all other sites .2%) were either invariant or autapomorphic (Table 3 Seventy-five indels (Tables 2, 3), ranging in length from 1 to 82 bp, were needed to align se- quences. Deletions relative to the outgroup taxa ac- counted for 7196 (53/75) of the indels, unique se- quence insertions 14.596 (11/75), and insertions that are repeats of adjacent sequence also account- ed for 14.5% of the indels (Table 2). Thirty-two of the indels (Table 2) are phylogenetically informa- tive and support relationships based on nucleotide substitutions alone (Figs. 2—4). Many more of the 1 and 2 bp (hereafter referred to as “small”) indels (6496) were homoplasious (Table 2, Fig. 3), when compared with those 3 bp and greater (3596; here- after called “large”indels; Table 2, Fig. 4). tential phylogenetic information; 87 — PHYLOGENETIC RECONSTRUCTIONS All three analyses (Figs. 2—4) show similar phylogenetic relationships within Asteraceae. In the 50% majority-rule trees (Figs. 2—4), branches not appearing in the strict consensus are indi- cated by dotted lines. The phylogenetic analysis of the sequence data excluding all indels yielded 180 equally parsimonious trees of 234 steps (C.I. = 0.61; КЛ. = 0.63; Fig. 2). The data set in- cluding all indels produced 244 trees, 293 steps in length (C.I. 0.61; R.I. 0.64; Fig. 3), Volume 85, Number 2 Bayer & Starr 249 Asteraceae Phylogeny Ageratum Eupatori Tree Statistics ы ) —— ғынды phylogenetically informative Gaillardia ) Helenieae char = 101 ria of most parsimonious trees = 234 А ) Number of most parsimonious trees - 180 Heliantus ) Heliantheae A M таттан Tagetes ) Helenieae ei cido ty и Antennaria Gnaphalieae Stuartina Petasites 7 Senecioneae Senecio Asteroideae Artemisia Anthemideae Matricaria Aster Astereae Townsendia 2 JF Calendula ND Calenduleae Osteospermum 3 / Inula ) Inuleae N Streptoglossa ) Plucheeae 8 : Liabum ) Liabeae Stokesia ) Vernonieae % Сагата ) Arctoteae Crepis 7 Lactuceae Cichorioideae Lactuca 10 AL Gerbera ) Mutisieae Cirsium Cardueae Echinops Chuquiraga Barnadesieae Barnadesioideae Doniophyton Figure 2. The 50% majority rule of sequence data of the trnL intron and the nsensus tree of 180 equa rnl/F intergenic spacer u ally parsimonious trees resulting from phylogenetic analysis sing all in ormative base pairs, ut excluding all e the branches, and Bee с уй given as percentages below each branch. Taxon labels аге from left to right: genera, Со and subfamili whereas the data set including only large indels yielded 258 trees, 267 steps long. The latter trees have the highest consistency and retention indi- ces of all three analyses (C.I. — 0.62; R.I. — 0.65; Fig. 4). Island searches (Maddison, 1991) on the data sets did not reveal any islands of shorter length trees. TOPOLOGY OF MAJOR CLADES АП trees (Figs. 2—4) indicate that Asteroideae are monophyletic and place a clade or clades 2. рап of Cichorioideae, including members of tri Liabeae, Vernonieae, Arctoteae, and Lactuceae, as sister(s) to the Asteroideae clade. Decay index values 250 Annals of the Missouri Botanical Garden 1Е— Ageratum ) Eupatorieae Tree Statistics Gaillardia ) Helenieae n урну informative haracters = 134 Раға of most parsimonious trees = 293 Helianthus ) Heliantheae Number — ap rsimonious trees = 244 Cuneo A XA Tagetes ) Helenieae Retentian ind 2 Length of 50% majority rule tree = 298 6 (5) 2- Antennaria ІҢ Gnaphali 3 | naphalieae F * Stuartina 1 y 12 x. Petasites 1 | 7 Senecioneae 1007. 12 Senecio А E Asteroideae 20 (15) Artemisia д ік Anthemideae = Matricaria Wo "m Aster Astereae 7 Townsendia = Calendul 100 Р ана alendula 19 NV Calenduleae Ik Osteospermum 6 JU Inula ШІ” AL 1(0) Streptoglossa ) Plucheeae [2] • 14 33% T U^ Crepis dix 6 Lactuceae bM Lactuca 1 (0) 7 ы f Liabum ) Liabeae 4(2) 6 32% Nor Stokesia ) Vernonieae 13 a» 5 | Cichorioideae МА Gazania ) Arctoteae 1 Lo Gerbera ) Mutisieae 5 Gs Cirsium \ 1 Cardueae Echinops Chuquiraga p | Barnadesieae Barnadesioideae Doniophyton Figure 3. of sequence data of the small indels. Branches that and bootstrap values (D.I.) of 0-2, synapomorphies (SYN) of 3—4, and boot- strap values (B.V.) of 39% to 49%, provide only weak support for this relationship. A clade containing mem- bers of the Mutisieae and Cardueae is seen at the base of these trees (Figs. 2—4). In most cases, tribes represented by more than one genus (i.e., the An- The 5096 majority rule consensus tree of 244 equally parsimonious trees resulting from мөкү d trnL intron and the trnL/F intergenic spacer using all informative base d both large and did not appear in the strict consensus tree are indicated by dashed lines. “The tree 1. the number of г к арн d (inc ds indels) above e branches, decay index values (in paren ent ach branch. d (1 and 2 are shown with bp length enclosed in | |, disons аге ||, bo type are the homoplasious indels. Taxon labels are from left to euin genera, tribes, and Өрт те ntheses) also above the branches, 2 bp) phylogenetically informative insertions ce lype indels are those with СЛ. of 1.00, and italic themideae, Astereae, Calenduleae, Cardueae, Gna- phalieae, Helenieae, Lactuceae, Senecioneae) are monophyletic. Exceptions to this are Helenieae, which is paraphyletic in all trees (Figs. 2—4), and Cardueae, which proved to be unnatural in the anal- ysis that excluded indels (Fig. 2). Volume 85, Number 2 Bayer & Starr 251 1998 Asteraceae Phylogeny Ageratum ) Eupatorieae Potentially phylogenetically informative Gaillardia ) Helenieae characters = 123 Lengths of most parsimonious trees = 267 Helianthus ) Heliantheae Number of most parsimonious trees = 258 | 2 тент Меле Tagetes ) Helenieae Lengthaf 30% aderit ы ыды Antennaria Gnaphalieae Stuartina Petasites Senecioneae Senecio Asteroideae Artemisia Anthemideae Matricaria Aster Astereae Townsendia Calendula Calenduleae Osteospermum Inula ) Inuleae Streptoglossa ) Plucheeae Crepis Lactuceae Lactuca 12 (26) Gazania ) Arctoteae 11,11 8 100% Шабит ) Liabeae ZO aer ( Cichorioideae 3% NA | ) Stokesia ) Vernonieae 1 ілін Lo Gerbera ) Mutisieae o Cirsium 4 Cardueae Echinops Chuquiraga | Р Barnadesieae Barnadesioideae Doniophyton Figur re 4. The 50% majority rule consensus tree of 258 equally parsimonious trees resulting from phylogenetic analysis of sequence data of the trnL intron and the trnL/F intergenic spacer using all informative base pairs, but excluding all small (1 and 2 bp) indels. Branches that did not appear in the strict consensus tree are indicated by dashed lines. The tree gives the number of apomorphies (inc Miri indels) above the branches, decay index values (in ошире. also above the digg and bootstrap value as percentages below a branch. Large phyloge- cally informative insertions are shown with bp le ee enc losed i in | |, deletions are | |, boldface type indels are Кы with С.І. of 1.00, and italic type are the homoplasious indels. Taxon labels are from left to right: genera, tribes, and subfamilies. TOPOLOGY OF MINOR CLADES trees, but are most strongly supported (SYN = 6- Clades containing members of the tribes Eupa- 9; B.V. = 86-99%) in the analyses that included torieae, Helenieae, and Heliantheae (the helian- indels (Figs. 3 and 4). There is low support for two thoid clade) are common to all most parsimonious additional clades within the Asteroideae, one con- Annals of the Missouri Botanical Garden taining members of thie Gnaphalieae and Seneci- oneae (SYN = 1; B.V. = 57-58%) and another containing de. of the Anthemideae and Astereae (SYN = 5-7; D.I. = 0-3; B.V. = 69- 7296). These clades are found in all the most par- simonious trees from the data sets containing no indels and large indels only (Figs. 2, 4). In two of the analyses (Figs. 2, 3), both of the genera in the Calenduleae are part of the main Asteroideae clade, whereas in the analysis containing only large indels (Fig. 4), they are part of a weakly supported (D.I. = 0; SYN = 1; B.V. = 32%) group that is sister to the Anthemideae-Astereae clade. In both groups of trees derived from data sets containing no indels and large indels only (Figs. 2, 4), the Inuleae-Plucheeae clade is sister to the rest of the Asteroideae, whereas in the third analysis contain- ing all indels this clade is part of a basal polytomy of a less resolved Asteroideae (Fig. 3). Cichorioideae are a paraphyletic group in all analyses (Figs. 2, 4). The Cardueae-Mutisieae clade mentioned above received weak support in all our trees (SYN = 1; D.I. = 0-1; B.V. = 35- 43%). In all the analyses (Figs. 2—4), a clade or clades representing the tribes Liabeae, Vernonieae, Arctoteae, and Lactuceae are patristically closer to the Asteroideae clade than are Cardueae and Mu- tisieae. All of the trees (Figs. 2—4) PO Vernonieae and Liabeae as sister taxa (SYN = 2- 1. = 0- 2; B.V. = 31-8296). One of the trees m indels excluded; Fig. 2) provides weak support for a re- lationship in which the Arctoteae is the sister group to the Vernonieae—Liabeae clade (SYN = 0-1; B.V. = 16-42%). The Lactuceae clade has weakly sup- ported relationships in the three trees, as sister to the Arctoteae-Liabeae-Vernonieae clade (Fig. 2; У. = B.V. = 11%), as sister to the Asteroideae (Fig. 3; SYN = L; B.V. = 19%), and as part of a polytomy (Fig. 4). DISCUSSION This study represents one of the few to use the trnL intron and/or trnL/F intergenic spacer for phy- logenetic reconstruction. The initial study of Ta- berlet et al. (1991) introduced PCR primers for these regions and showed that they could be am- plified across a broad taxonomic range from algae to bryophytes, vascular cryptogams, gymnosperms, and angiosperms. This was followed by a phyloge- netic reconstruction of some Crassulaceae genera using the trnL/F spacer by van Ham et al. (1994), who demonstrated the utility of the sequence to re- construct phylogeny at the family level. Gielly and Taberlet (1996) and Gielly et al. (1996) used the trnL intron to produce a phylogeny for Gentiana Gentianaceae), comparing it to phylogenies for the same group based on sequences of the internal transcribed spacers (ITS) of nuclear ribosomal DNA. They concluded that ITS was more informa- tive than the chloroplast sequence for resolving phylogenies at this level, and that the trnL intron sequences would probably be more useful at the intergeneric level (Gielly et al., 1996). Most re- cently Bóhle et al. (1996) employed both of the regions used in this study, along with the trnL/T intergenic spacer and ITS sequences, to reconstruct the phylogeny of Echium (Boraginaceae) in the is- land groups off the northwest coast of Africa and the adjacent mainland. They obtained good reso- lution of the major clades (especially island versus mainland clades) within the genus, and showed the utility of combining ITS and chloroplast spacers in phylogenetic reconstruction at the generic level. —= In resolving relationships in the Asteraceae, we found that the combined use of base substitutions and large indels produced trees that were better supported and less homoplasious than trees pro- duced using only base substitutions or base sub- stitutions and all indels. Our results agree with those of other studies (van Ham et al., 1994; Lloyd & Calder, 1991) in showing that smaller indels tend to be more homoplasious. The averages of G/C vs. A/T content we found for the trnL intron and trnL/F spacer are nearly 35.196 (Table 3) and 64.9%, respectively]; this compares favorably to the relatively narrow range in G/C content (36— 39%) reported in angiosperm cpDNA (Palmer, 991 identical [combined average = T The topologies of our trees (Figs. 2—4) largely agree with those from other studies (Bremer, 1987; Jansen et al., 1990; Jansen et al., 1991; Karis et al., 1992; Kim et al., 1992; Karis, 1993; Kim & Jansen, 1995) of tribal relationships in the Aster- aceae. Our Asteroideae, consisting of the Anthem- ideae, Astereae, Calenduleae, Eupatorieae, Gna- phalieae, Helenieae, Heliantheae, Inuleae, Plucheeae, and Senecioneae (Figs. 2—4), is the same monophyletic group found by Bremer (1987) based on morphology, and by Jansen et al. (1991), Kim et al. (1992), and Kim and Jansen (1995) based on molecular studies. We have also found that the Cichorioideae is paraphyletic, as reported in most other studies (Bremer, 1987; Karis et al., 1992; Kim & Jansen, 1995). The exceptions to a рагарћујенс Cichorioideae are seen in the rbcL (Kim et al., 1992) and RFLP studies (Jansen et al., 1990; Jansen et al., 1991). The rbcL study, how- ever, lacked representation from critical taxa like Volume 85, Number 2 1998 Bayer & Starr 253 Asteraceae Phylogeny the Inuleae s. str., Plucheeae, and Gnaphalieae, taxa that cause notable topological differences with- in the Cichorioideae when excluded in our analysis (results not shown). The rbcL study (Kim et al., 1992) portrayed relationships within the Cichorioi- deae largely incongruent with those suggested by ours and the above-mentioned studies. А recent reanalysis (Mishler et al., 1996) of the RFLP stud- ies has found a paraphyletic Cichorioideae and has called into question the original methods of anal- ysis (Jansen et al., 1990; Jansen et al., 1991) of those data. The Mutisieae and Cardueae form a monophy- letic group (Figs. 2—4) that is sister to a clade con- sisting of the remainder of the Cichorioideae and Asteroideae. Similar basal positions for the Mutis- ieae and Cardueae are found in morphological (Bre- mer, 1987; Karis et al., 1992) and most molecular- based (Jansen et al., 1990; Jansen et al., 1991; Kim & Jansen, 1995) phylogenetic reconstructions. There is some weak evidence (Fig. 2) that the Car- dueae may be paraphyletic, as suggested by Dit- trich (1977). He split the Cardueae into three sep- arate tribes, of which two, the Echinopeae and Cardueae s. str., were represented in our study (by Echinops and Cirsium, respectively). As in many other studies (Bremer, 1987; Jansen et al., 1991; Karis et al., 1992; Kim et al., 1992), the relationships of the Lactuceae, Arctoteae, Lia- beae, and Vernonieae (LALV), which form the re- mainder of the Cichorioideae, were largely unre- solved in our investigation. We have only weak evidence for a monophyletic LALV group (Fig. 2), and Kim and Jansen (1995) also found only modest support (SYN = 3; deletion = 1) for the monophyly of this group. Most studies (Bremer, 1987; Jansen et al., 1990; Jansen et al., 1991; Kim & Jansen, 1995), including ours (Figs. 2—4), show Liabeae and Vernonieae as sisters, except for the rbcL study by Kim et al. (1992). Although Liabeae were once placed in Senecioneae (Robinson, 1983; Bremer, 1987), it is now clear that they are quite distinct from that tribe and are indeed most closely related to Vernonieae. It has been suggested that Vernon- ieae and Liabeae should be united (Jansen & Stuessy, 1980), although it appears that there are morphological synapomorphies that warrant their recognition as distinct lineages (Bremer, 1987). We now turn our attention to the Asteroideae clade. The recent work of Anderberg (1989, 1991a, 1991b, 1991c) and Karis (1993) has shown that the Inuleae sensu Merxmüller et al. (1977) should be considered as three separate tribal lineages: the In- uleae s. str., Gnaphalieae, and Plucheeae. Although Anderberg presented strong cases for separation of the tribes, some studies (Kim et al., 1992; Jansen et al., 1991; Bremer et al., 1992) have chosen not to address the “Inuleae problem.” Our current long-term research into the molecular phylogenet- ics of the Gnaphalieae necessitates that we first re- solve the sister-group relationships of the Gnaphalieae. We have thus included members of all three of Anderberg's tribes, and our results cor- roborate the morphological (Anderberg, 1989, 1991a, 1991b, 1991c; Karis, 1993) and single-mo- lecular analysis (Kim & Jansen, 1995) in indicating that the “old” Inuleae are not а monophyletic lin- eage. In all of our analyses, the Inuleae s. str. and the Plucheeae are sister taxa, and these in turn are sister to the remainder of the Asteroideae in two analyses (Figs. 2, 4). Kim and Jansen (1995), using паћЕ, showed a strong sister relationship between the Plucheeae and Inuleae, but the base of their Asteroideae was not resolved finely enough to show the sister relationships of that clade. Our topolog- ical relationships on the other hand were nearly identical to those of Karis (1993). Only the RFLP- based study of Keeley and Jansen (1991), which included members of all three tribes, showed the "old" Inuleae to be monophyletic. Therefore, based on the available evidence, the segregation of the Gnaphalieae from the “old” Inuleae seems war- ranted, although the circumscription of the Plu- cheeae is still unresolved. The sister relationships of the Gnaphalieae remain controversial. In our analysis (Figs. 2-4), the Gnaphalieae are sister to the Senecioneae. Karis (1993) showed them as sis- ter to a clade containing the Astereae and Anthem- ideae, Jansen et al. (1991) as sister to the Inuleae (represented by /nula), Keeley and Jansen (1991) as sister to a clade consisting of the Inuleae and Plucheeae, and Kim and Jansen (1995) in an un- resolved clade containing the Calenduleae, Aster- eae, and Anthemideae. The sister relationships of the Gnaphalieae remain unresolved due to the dis- cordance among studies. The sister relationships of the Astereae seem less controversial (Zhang & Bremer, 1993). We have shown them to be a well-supported sister group to the Anthemideae (Figs. 2-4), as also seen in the morphological study of Karis (1993) and the mo- lecular studies of Jansen et al. (1991), Keeley and Jansen (1991), Kim et al. (1992), and Kim and Jan- sen (1995). Only Bremer (1987) portrayed them in a different relationship, as sister to the Eupatorieae. The relationships of the Calenduleae are contro- versial, and in only one of our analyses (Fig. 4) are their affinities to other tribes resolved, i.e., as sister to the Astereae-Anthemideae clade. Most morpho- logical analyses do not show this relationship (Bre- 254 Annals of the Missouri Botanical Garden mer, 1987; Karis, 1993), while other molecular analyses (Kim et al., 1992; Kim & Jansen, 1995) indings. Interestingly, RFLP's in cpDNA (Jansen et al., 1991; Keeley & Jansen, 1991) show the Calenduleae as sister to the Sene- cioneae, which has been the traditionally recog- nized relationship since the time of Bentham (1873). The helianthoid clade, including the Eupato- rieae, Helenieae (Tageteae, pro parte of some au- thors), and Heliantheae, is a strong monophyletic group in all our analyses (Figs. 2—4). Problems arise when trying to resolve relationships and cir- cumscribe tribes within the helianthoid clade be- cause it appears to contain a badly understood se- ries of phylogenetically basal branches forming successive sister groups to the rest. The combined evidence suggests that some of the tribes in the helianthoid clade are paraphyletic and need to be re-examined. Tagetes was treated as part of the Helenieae by Bremer (1994), as a member of subtribe Pectidinae (in Heliantheae) by Robinson (1981), and as the type genus of the tribe Tageteae by many authors from Cassini (1826) to Karis (1993). Our results have part of the Helenieae (Tagetes) as sister to a group consisting of the Eupatorieae, Helenieae (Gaillardia), and the Heliantheae, a disposition common to other molecular studies (Jansen et al., 1990; Jansen et al., 1991; Keeley & Jansen, 1991; Kim et al., 1992). Phylogenetic analyses using mor- phology (Bremer, 1987; Karis, 1993) and ndhF (Kim & Jansen, 1995) did not provide enough res- olution to reveal relationships among most of the genera in the helianthoid clade. The remainder of the helianthoid clade forms an unresolved polytomy containing the Heliantheae, the Eupatorieae, and the Helenieae (sensu Bremer, 1994). The Helenieae are represented by Gaillar- dia, which some authors (Robinson, 1981; Karis, 1993) have included in the Heliantheae (as the type genus of subtribe Gaillardiinae). Our analysis does indicate that the Heliantheae in the sense of Bremer (1994), Robinson (1981), and Karis (1993), are closely related to the Eupatorieae. This is a relationship that is also reflected by a number of additional molecular analyses including those of Keeley and Jansen (1991), Jansen et al. (1991), Kim et al. (1992), and Kim and Jansen (1995). Bre- mers (1987) morphological analysis showed that Astereae and Eupatorieae were sister taxa, whereas Karis (1993) portrayed a close relationship between helianthoid elements and the Eupatorieae. In conclusion, our phylogenetic analysis of the tribes of the Asteraceae produced trees largely con- gruent with other hypotheses based on both mor- phological and molecular data sets. Asteroideae are a monophyletic group, but Cichorioideae are para- phyletic. The primary clades of Cichorioideae are Mutisieae-Cardueae, Liabeae- Vernonieae; those of Asteroideae are Inuleae-Plucheeae, Astereae—An- themideae, Senecioneae-Gnaphalieae, and the he- lianthoid clade (Helenieae, Heliantheae s. str., and Eupatorieae). The Inuleae-Plucheeae clade is sis- ter to the remainder of the Asteroideae. The para- phyly of the “old” Inuleae (sensu Merxmiiller et al., 1977) has been confirmed by our analysis. Calen- duleae are sister to the Astereae-Anthemideae clade in some trees. А clade consisting of Lactu- ceae, Arctoteae, Verononieae, and Liabeae was also present in some most-parsimonious trees. Our study illustrates the utility of the trnL intron and trnL/F intergenic spacer for resolving the re- lationships among tribes in the largest dicot family, Asteraceae. Using approximately 874 bp (Table 3), we were able to produce a phylogeny that shows a similar level of resolution to that produced by Kim and Jansen (1995) using 2200-2300 bp of ndhF. Comparison of the divergence values in the 17 taxa shared by our study and that of Kim and Jansen (1995) revealed that the combined trnL intron and trnL/F spacer evolves at a rate that is 1 to 1.28 times faster than ndhF. Further resolution could also be expected if additional taxa and the ca. 620— 700 bp of trnL/T intergenic spacer were added to our analyses. Another chloroplast sequence, rbcL, which is 1428 to 1458 bp long in the Asteraceae and is often used in phylogeny reconstruction at the family level and above, did not provide as much resolution of the tribal relations in Asteraceae (Kim et al., 1992) as did ndhF (Kim & Jansen, 1995). RFLPs of chloroplast DNA, although providing fair- ly good resolution of relationships in the Astera- ceae, resulted in several equally parsimonious trees that had moderately large amounts of homoplasy (СЛ. = 0.54) (Jansen et al., 1990; Jansen et al., 1991). Additionally, that study was labor-intensive, requiring eleven restriction enzymes to produce 328 phylogenetically informative sites, and the methods of cladistic analysis of the RFLP data (Jansen et al., 1990; Jansen et al., 1991) have re- cently been criticized by Mishler et al. (1996). The sequences used in the present study have three ad- vantages over the other commonly used gene regions: (1) they are easy to amplify across a wide taxonomic range because the universal primers de- signed by Taberlet et al. (1991) are placed in highly conserved tRNA genes; (2) the primers used to am- plify the region can also be used to sequence it entirely using manual methods; and (3) the numer- Volume 85, Number 2 Bayer & Starr 255 Asteraceae Phylogeny ous large indels provide additional phylogenetic in- formation. For phylogenetic reconstruction at the family level the trnL intron, trnL/F intergenic spac- er, and the trnL/T intergenic spacers may represent an ideal sequence, providing levels of resolution similar to those of longer gene sequences (rbcL an ndhF), but requiring much less labor to generate data. Literature Cited Anderberg, A. A. 1989. Phylogeny and reclassification of x tribe Inuleae (Asteraceae). Canad. J. Bot. 67: 2277— 2296. -----. 1991а. Taxonomy and phylogeny of ie tribe Cnaphalieae Ar ciens Opera Bot. 104: Taxonomy and phylogeny "T L е бокае teen Pl. Syst. Evol. 176: 75-123. Taxonomy and phylogeny 2 the tribe 25. (Asteraceae) Pl. Syst. - 176: 145-177. Baum, D. A., K. a & P. C. Hoch. я А phy- атна analysis " “Epilobium пуан based on uclear ribosomal DNA sequences. Syst. Вог. 19: 363— 388. Bayer, R. Ј., L. Hufford & D. E. Soltis. 1996. Phyloge- netic relationships in Sarraceniaceae based on rbcL and Notes on ‘he di history, and geographical distribution of the Compositae. J. Linn. Soc. Bot. 13: 33: Bóhle, U.-R., H. | Hilger & W. F. Martin. 1996. Island colonization and evolution of the insular woody habit in Echium L. (Boraginaceae). Proc. Natl. Acad. Sci. U.S.A. Tribal interrelationships of the Aster- 2 Cladistics 3: 210-253. . The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 5—803. 42: 79 Ancestral areas—A cladistic reinterpre- tation of the center of origin concept. Syst. Biol. 41: 436—445. 4. Asteraceae: m and Classification. nor Press, Portland, O . Jansen, R. K., P. 0. Қыз M. Kallersjé, S. : Keeley K. Kim, H. J. Michaels, J. D. Palmer & R. * Wallace. 1992. A review of the phylogeny and 5. fication of the Asteraceae. Nordic J. Bot. 12: 141-148. Carlquist, S. 1976. Tribal 2. and phylog- eny of the Asteraceae. Aliso 8: 465—492 26. > ade Phywlesiques, Vols. I and 8, i D. Mi shler, M. R. is ll, R. x Price. ut, R. K. Jansen, K.-J. satin e. F. Y. Xiang, G. M. Plunkett, P. S. Soltis, S. p jameda ашы of seed plants: An ма of кош se- quences 5 the plastid gene rbcL. Ann. Missouri Bot. Gard. 80: 528—580. Cronquist, A. 1955. Phylogeny and БЕЛЕТ of the Com- positae. Amer. Midl. Naturalist 53: 4 l. Dittrich, M. 1977. Cynareae—System бије review. Pp. 999-1015 in V. Н. Heywood, J. B. Harbourne & B. L. Turner (editors), The асар к шамы. of the Com- positae. Academic Press, Donoghue, M. J., R. G. Olmstead Ld: F. Smith & J. D. Palmer. 1992. Phylogenetic relationships of ye n based on rbcL sequences. Ann. Missouri Bot. Gard. 7 993. Amplify LE Software for designing, nvolving the ailable free on the The retention index вай the rescaled sistency 4. p o 5: 417-419. кол. J. 1985. Confidence limits on phylogenies: approach using Mr bootstrap. Evolution 39: 783— Gielly, 1. & P. Taberlet. 1996. A phylogeny of the Eu- ropean gentians inferred from chloroplast trnL (UAA) intron жі лд. 'es. Bot. J. Linn. Soc. 120: 57-75 ‚ Y.-M. s ‚ P. Küpfer & P. Taberlet. 1996. Phylo- genetic us coding regions in the genus Gentiana Ls Corea ae (UAA) intron versus nuclear ribo- somal internal transcribed spacer sequences. Molec. соне, Evol. 5: 460—466. Gilbert, 1992. Seqapp version 1.8a, a multiple se- quence ее for Macintosh computers. Published elec- tronically on the Internet (ftp.bio.indiana.edu). Gustafsson, M. H. G. & K. Bremer. 1995. Morphology and pu interrelationships of the Asteraceae, Calyceraceae, Campanulaceae, Goodeniaceae, and re- lated families (Asterales). Amer. J. Bot. 82: pb Ham, R. С. H. J. van, H. Hart, T. 1. Mes & J. M. Sand- brink. 1994. Molecular сок of nonc du: regions of the chloroplast genome in the Crassulaceae and re- о V: Improved | они for multiple AS align- . Computer Applic. Biosci. 8: p -191. Hoffmann. O. 1894. Compositae. /n: Engler & K. Prantl, Die natürlichen наде 4(5): 87-391. Wilhelm ME cu Leipz Jansen, R. K. & J. D. Pa Pos 5087. A chloroplast DNA inversion mais ancient evolutionary split in the sunflower family (Asteraceae). Proc. Natl. Acad. Sci. U.S.A. 84: 1-5 & 1988. Phylogenetic implications of 2. DNA restriction site variation in the Mutis- ae (Asteraceae). Amer. J. Bot. 75: 753-766 & . Stuessy. 1980. omosome counts of Composita from Latin America. Amer. J. Bot. 67: 585- ‚Н. J. Michaels & J. D. Palmer. 1991. Phylogeny nd character evolution in the Asteraceae based on chloroplast DNA restriction site mapping. Syst. Bot. 16: 98-115. E. Holsinger, H. J. Michaels & J. D. Palmer. 1990. Сабр analysis of chloroplast E restric- tion site data at higher taxonomic levels—An example from the Asteraceae. Evolution 44: 2089-2105 Karis, P. O. 1993. Morphological phylogenetics of the Asteraceae—Asteroideae, with notes on character evo- lution. Pl. Syst. Evol. 186: 69-93 256 Annals of the Missouri Botanical Garden — ———, M. Kiillersjó & К. Bremer. 1992. Phylogenetic d al the Cichorioideae (Asteraceae), with empha- sis on the Mutisieae. Ann. Missouri Bot. Gard. 79: 416— 427. Keeley, S. С. & R. K. Jansen. 1991. Evidence from chlo- roplast DNA for the recognition of a new tribe, the Tar- chonantheae, and the tribal placement of Pluchea (As- teraceae). Syst. | 16: 173-181. Kim, K.-J. & R. K. Jansen. 1995. ndhF sequence evo- lution and the major clades in the sunflower family. Proc. Natl. Acad. Pu U.S.A. 92: 10379-1( ar ; , R. S. Wa H. J. Michaels & J. Palmer. 1992. Pleni jai ia i jd. se- e 5, in the Asteraceae . Ann. Missouri Bot. Cai d. 428-44; Kluge, "4 с. & J. S. Faris. 1969. а and the 2. of Anurans. Syst. Zool. 18: 1— жк D. С. & V. L. Calder. 1991. Multi residue gaps, ass of molecular characters with exceptional reli- ability for 2. analysis. J. Evol. Biol. 4: 9-21 Maddis . R. 1991. The discovery and importance of muliple islands of most parsimonious trees. Syst. Zool. 40: : 2. x Maddison. 1992. MacClade. Analysis of Phylogeny por Character кчы Version 3.0. Sinauer, Sunderland, Massachuse Margush, ' R. . Pi 981. Bull. Math. Biol. 43: 23€ P. Leins & e [n essler. 1977. Inuleae— сее review. Pp. 577-601 in V. Н. Heywood, J. B. Harbourne & B. L. Turner (editors), The Biology and Chemistry ol the Compositae. Academic Press, London. Michaels, H. J., К. M. Scott, R. G. Olmstead, T. Szaro, R. K. Jansen & J. D. Palmer. 1993. Interfamilial re 5. ships of the Asteraceae—Insights from rbcL sequence variation, Ann. Missouri i, 1. 80: [m 751. Mishler, B. D., V. A. Albert, /. Chas K. Bremer. 1996. Charac ^ state 4. бог DNA Ge eu n-trees. Merxmiiller, H., restriction site data: Asymimetty, ancestors, and the As- pri 12: 11-19 “ом Ј. )l. ШІП рөөнүн өн Structure and 54. Рр. 5—53 in L. ogorad & Vasil (edi- tors p C ell Culture ond Somatic Cell Genetics in Plants, vol. 7A: The Molecular Biology of Plastids. Academic sats )rlando. 20 H. 1981. А revision of the tribal and subtribal limits of pio Heliantheae (Asteraceae). Smithsonian Сот. а : 1-102. 198: p generic review of the tribe Liabeae (As- ter: ea a Contr. Bot. 54: 1—6‘ Sanger, F., S. Nicklen & A. R. Co 1977. DNA se- quence ps iil chain-terminating inhibitors. Proc. Natl. Acad. Sci. U.S.A. 74: 5463-5467. m D. L. 1993. PAUP: Phylogenetic Analysis Us- ing Parsimony, Version 3.1.1. Illinois Natural History Survey, С hamp aign. & G.J. Olsen. Pp. 411-501 in D. M. lec ш Systematics. 1990. Phyloge ny reconstruction. Hillis & C. Moritz (editors), Mo- Sinauer, Snel Massachu- зеп Таһе det P., Қ G. Pautou & J. Bouvet. 1991. Uni- versal primers for amplification of three non- а ге m of c Шы DNA. Pl. Molec. Biol. 105— 1109 Watrous, L. E. Wheeler. 1981. The DE comparison 2. Svst. Zool. 3 364—300. Wojciechowski, M. F., M. J. Sanderson, B. С. Baldwin & . Donoghue. 1993. Monophyly : the | As- алай (Fabaceae): Evidence from th« internal 11– ar ribo- transcribe i spacer sequences. 722. 1993. A cladistic analysis of the tribe Astereae EE ) with notes on Fee evo- lution and subtribal classification. Pl. Syst. Evol. 259-283. А SYNOPSIS OF THE GENUS ECHINOPEPON (CUCURBITACEAE: SICYEAE), INCLUDING THREE NEW TAXA! Alex K. Monro? and Peter J. Stafford? ABSTRACT Following a palynological and general Rd шы, фик survey, the genus LAL: > (Cucurbitaceae) has been divided into three species groups on the basis of stam and illustrated: 5 tultitlanapaensis А. К. A Monro & 5 (DC.) A. K. Monro & Staff., and E. glutinosus (Cogn o synonymy of E. pubescens (Benth.) and pollen morphology. Monro & Staff., E. belizensis A. к. Мопго & Staff., and Е. micropaniculatus aff.; three new combinations are inem; E. arachoides (Dieterle) e new species of Echinopepon are described A. K. Monro & Staff., E. gemellus .) A. K. Monro & Staff.; and E. floribundus (Cogn.) Rose is reduced Hose. A 2. review is presented, and the 18 species of the genus Echinopepon are listed, together with the specimens examine Echinopepon (Cucurbitaceae) is a genus of 18 New World taxa whose center of diversity is the Pacific coast of Mexico at middle to high elevations (above 1000 m) in the Sierra Madre Occidental and the Sierra Madre del Sur. While preparing accounts of Echinopepon for Flora Mesoamericana, it became clear that this genus was in great need of nomen- clatural and systematic revision. Previous palyno- logical work on the genus (Stafford & Sutton, 1994) and systematic treatments of the family (Jeffrey, 1964; Rangaswami Ayyangar, 1976) indicated that a synoptic revision of the genus based on macro- morphological and palynological observations would be appropriate. Echinopepon is one of four genera to have been separated from the genus Sicyos L. as originally proposed in Hortus Cliffortianus in 1737 (Stocking, 1955). In 1840, Torrey and Gray formed a separate genus of the New World taxa in the Linnaean Si- cyos, which they named Echinocystis. Over the next decade Echinocystis was divided into three sec- tions, Echinocystis (“Euechinocystis”), Echinopepon, and Marah (Cogniaux, 1877; Cogniaux, 1881), which were later recognized at generic rank by Watson (1887, in which he referred to Marah Kel- logg under the synonym: Megarrhiza Torr.). Despite further papers on the nomenclature and taxonomic status of Echinopepon (Watson, 1889; Rose, 1897; Stocking, 1955), the genus has not been comprehensively reviewed since 1881, when Cogniaux treated it as a section of Echinocystis. GENERAL МОВРНОГОСУ The principal characters used to classify subfam- ilies, tribes, and subtribes in the family Cucurbi- taceae relate to the pistil, stamens, tendrils, and pollen (Cogniaux, 1877; Jeffrey, 1990). Within the tribe Sicyeae, anther arrangement, the disposition of the ovules within the ovary, fruit and seed mor- phology, and the branching of tendrils have been used to define the genera (Jeffrey, 1990). Naudin (1866) defined Echinopepon as monoecious, with 6-merous flowers, having three fused stamens (one unilocular and two bilocular), a unilocular ovary bearing 8—10 ovules, a 2-chambered, coriaceous, cylindrical fruit possessing a dehiscent, apical operculum, and seeds that are ovoid-compressed and corrugate. Within the genus Echinopepon itself, inflores- cence disposition, flower size, the disposition of the anther thecae, the relative length of the corolla lobes to the hypanthium, and fruit size, shape, and spine-length have all been used to distinguish taxa Naudin, 1866; Cogniaux, 1877; Rose, 1897; Wat- son, 1889). P нек” ! We thank Helen Greenop for the botanical drawings, Lourdes Rico (K) for help with Mexican localities, Carol Furness (K) for pollen samples of Echinopepon jaliscanus, Philipe Morat for a warm reception at P, Charlie Jarvis (BM) for help with nomenclature, Norman Robson (BM) for help with the Latin diagnoses, Bob Press (BM), Sandra Knapp (BM), and Mary Gibby (BM) for help with the manuscript, and the following herbaria for the loan of reference material: B, BR, Е, GH, GOET, MA, MO, NY, P. and US. ? Department of Botany, The Natural History Museum, Cromwell Road, London SW7 5BD, United Kingdom. ANN. Missouni Bor. Савр. 85: 257-272. 1998. 258 Annals of the Missouri Botanical Garden ж re 1. —a. Anther Edd of the E. panic ns species group. Scale PoLLEN MonPHOLOGY Pollen characters in the family Cucurbitaceae show a high degree of diversity and have long been perceived as indicators of relationships at all tax- onomic levels. In an overview of pollen morphology дер 95: f the Echinopepon racemosus spre ies 8 group. —b. Anther of the E. torquatus species group.— ar — ] mm. in the Cucurbitaceae, Jeffrey (1964) found that his earlier conventional classifications of the family Jeffrey, 1961, 1963) corresponded to a high degree with the natural order of taxa as indicated by pollen morphology, “much more so than any previous clas- sification scheme." Rangaswami Ayyangar (1976) ~ Figure 2. (scale Баг = 25 шт)—В. Е. E (sca = 25 џт).—р. E. racemosus. Sectio E. Е. coulteri. Polar view (scale bar. = 20 --Е E. с Section showing exine structure (scale eos — 5 pum). sho e exine — 'ation (scale ba m). шегі. Pus view (scale bar = 25 jum). — Polypantoc JF to irregularly sync сын pollen type (SE M micrographs).—A. Echinopepon racemosus 5 ит).—С. E. lup canus. Note branching colpi (scale bar = 8 шп). 6-8-zonocolpate pollen. type.— —G. E. coulteri. Volume 85, Number 2 Monro & Stafford 1998 Synopsis of Echinopepon Annals of the Missouri Botanical Garden ШЕ, mne m "T, T > L ж H Volume 85, Number 2 1998 Monro & Stafford Synopsis of Echinopepon also found that dividing the family into seven tribes on a palynological basis agreed with his own pre- vious classification (Rangaswami Ayyangar, 1967) based on karyological characters (although it was not highly congruent with Jeffrey's classification). Except in the case of Echinopepon, pollen mor- phological divisions within the subtribe Cyclanth- erinae otherwise correspond almost perfectly with present generic boundaries (Stafford & Sutton, 1994). In a general account of pollen morphology in the Cyclantherinae, Stafford and Sutton (1994) grouped the pollen of a selection of species repre- senting each of the genera into seven main types, within which taxa can be further identified to ge- neric and, in some cases, to specific level. Pollen of the genus Echinopepon is particularly variable. Stafford and Sutton (1994) indicated that three or more conspicuous pollen types could be identified within Echinopepon, which conform to the groupings of Jeffrey (1980). In the light of these distinct infrageneric group- ings, an analysis of pollen characters in Echinope- pon and related genera has been undertaken which extends preliminary work conducted by Stafford and Sutton (1994). MATERIALS AND METHODS АП the taxa in the genus Echinopepon and rep- resentative taxa from the related genera Echinocys- tis, Marah, Brandegea, and Vaseyanthus were stud- ied. For Echinopepon the extant type material was examined, although it was not always possible to include these in the palynological review. In addi- tion, all of the specimens at K, BM, and P were seen, as were selected specimens from BR, F, GH, GOET, MA, MO, NY, and US. Specimens were ex- amined by eye and at a magnification of X64 to X400 under a Leitz Wild M3C microscope. The following characters were found to be of par- ticular value: indument of the leaf, petiole, stem, and fruit; staminate and pistillate inflorescence type; staminate and pistillate perianth morphology; fruit spine length; seed ornamentation; and pollen morphology. Pollen samples from each of the specimens stud- ied were prepared by acetolysis (Erdtman, 1969). Due to the thin nature of the exine, acetolysis time was restricted in some cases to two or three minutes to minimize pollen collapse. A small portion of the acetolyzed residue was mounted on aluminum stubs for scanning electron microscopy (SEM), and the remaining material was used to prepare microscope slides for light microscopy. Observations were made using a Zeiss Axioplan light microscope and a Hi- tachi S800 field emission scanning electron micro- scope, using secondary electron detection and an accelerating voltage of 8 kV. Material for observa- tion in the SEM was first sputter coated with gold palladium for one and a half minutes at 20 mA. The following parameters were measured: polar axis (P), equatorial axis (E), grain symmetry, number and character of colpi, exine thickness at center of mesocolpium, and ornamentation characteristics. Measurements were made from light microscope preparations in glycerine jelly, and are based on an examination of ten pollen grains from each speci- men. RESULTS Anther morphology falls into three groups: obo- void anthers with the thecae appearing 2—3-sigmoid (Fig. Та); subglobose anthers with the thecae ap- pearing “horse-shoe”-shaped (Fig. 1b); and ovoid anthers with the thecae appearing *banana"-shaped Fig. 1c). The pollen of the genus Echinopepon is remark- ably variable and may be grouped into four distinct pollen types: Pollen type 1 (Fig. 2A-D). Grains polypantocol- pate to irregularly syncolporate; colpi short, some- times branched and fused, dividing the exine into angular plates; endoaperture a weak, circular thin- ning of the nexine, distinguishable in the scanning electron microscope but indistinct in light micros- “~ 1 coarse granules, зехше 1 of sparsely distributed columellae, sexine 2 a rela- tively thin perforated tectum; ornamentation punc- titegillate. Size range: longest axis 85-164 pum. Species: Echinopepon racemosus, E. pringlei, E. tultitlanapaensis, E. jaliscanus. Pollen type 2 (Fig. 3A-D). Grains 10-16-zono- colpate, radially asymmetrical with sunken colpi; colpus long, broad and conspicuous, usually with <-- Figure 3. um).—B. Е. arachoides. um).—D. Е. torquatus. 4. of c рне pollen type.—E. E. micropaniculatus. Polar view bar — 20 лем (scale bar = 20 Polyzonoc E pollen type (SEM micrographs).—A. sini q torquatus. Polar view (scale bar = 25 um).—C. E. olpus showing exine stratification (scale Б. = 5 pm). (scale bar = 20 pm).—G. E. pubescens. Section showing exine ме зри не (sc 54 bar = 12 pm atus. Equatorial view (scale bar — 25 6—9-zonocolpate to zonocol- m).—F. E. belizensis. Equatorial view (scale cirrhopedunc 262 Annals of th Missouri Botanical Garden elaborate, distinctly granular margins and a margo ig. 3A-C); exine of three layers, thickest at center of mesocolpium with corresponding thinning of nexine and sexine toward colpi: nexine thin, without covering of granules, sexine 1 of densely spaced columellae, sexine 2 a perforated tectum; ornamentation punctitegillate. Size range: P60—148 шт, E60-135 рт. Species: Echinopepon torquatus, E. milleflorus, E. gemellus, E. cirrhopedunculatus, E. arachoides, E. minimus. This pollen type also embraces Apatzingania Dieterle (herein ѕупопу- mized with Echinopepon). Pollen type 3 (Fig. 2E-G). Grains 6-8-zonocol- pate, radially symmetrical, colpi not or only slightly sunken; colpus long and narrow without any mar- ginal differentiation of the exine; exine thin and of three layers: nexine very thin, with covering of fine granules, sexine 1 of broadly spaced columellae, sexine 2 a relatively thin, perforated tectum; or- namentation punctitegillate to weakly verrucate. Size range: P85— 5 pm, E95-110 jum. Species: Echinopepon coulter. Pollen type 4 (Fig. 3E-G). Grains 6-9-zonocol- pate to -zonocolporate; colpus long, broad, and con- spicuous, with granular surface, sunken; endoap- (see erture when present a poorly defined circular pore, arranged alternately above and below the equator and rarely on the equatorial plane itself, sometimes characterized by a raised annulus; exine of three layers: nexine thin, without covering of granules, sexine 1 of densely spaced columellae, sexine 2 a perforated tectum; ornamentation punctitegillate. Size range: Р74-104 рт, E70—94 рт. Species: Echinopepon paniculatus, E. glutinosus, E. wrightii, E. pubescens, E. longispinus, E. belizensis, E. mi- cropaniculatus. DISCUSSION On the basis of the above characters we suggest that the genus Echinopepon be divided into three clearly defined species groups, as outlined below. We feel that one of these, the racemosus group, may well represent a distinct genus. In view of the fact that this tribe (Sicyae) and subtribe (Cyclantheri- nae) are among the least well known in the Cucur- bitaceae (Jeffrey, 1964), the related genera in this tribe require further investigation before such a tax- onomic decision can be made. The monospecific genus Apatzingania Dieterle shows many similarities to Echinopepon in both pal- 1974). Apatzingania is distinguished from Echinopepon by the presence of a unilocular, single-seeded, inde- hiscent fruit. We interpret these characters as ad- ynology and macromorphology (Dieterle, aptational consequences of geocarpy. The pollen most closely resembles that of E. cirrhopeduncula- tus, a taxon whose fruit, although not geocarpous, is also borne on an exceptionally long peduncle and whose seeds are very similar in their ornamenta- tion. Although the habit of geocarpy is extremely rare in the Cucurbitaceae (the only other species is the South African Cucumis humifructus Stent), geo- carpy by itself may not be sufficient reason for dis- tinguishing Apatzingania as a separate genus. In consideration of this and the morphological simi- larities outlined above, we include Apatzingania ar- achoidea Dieterle in the genus Echinopepon. CONSPECTUS OF THE GENUS ECHINOPEPON І. E. racemosus species group Flowers campanulate; anther obovoid; thecae ap- pearing tubular prior to dehiscence after which rib- bon-like, folded into 2- and 3-sigmoid curves (Fig. la); pollen polypantocolpate to irregularly syncol- porate (Fig. 2A-D). l. Echinopepon racemosus (Steud.) C. Jeffrey 2. E. pringlei Rose 3. E. tultitlanapaensis A. K. Monro & Staff. 4. E. jaliscanus Rose II. E. torquatus species group Flowers campanulate to infundibuliform; anther subglobose; thecae “horse-shoe”-shaped, appearing tubular prior to dehiscence after which ribbon-like Fig. 1b); pollen 10-16-zonocolpate and radially asymmetrical, the colpus margins distinctly differ- entiated and granular (Fig. 3A-D); or 7-9 sym- metrically zonocolpate, the colpus margins undif- ferentiated (E. coulteri, Fig. 3E— “~ Echinopepon torquatus (DC.) В milleflorus Naudin gemellus (DC.) A. K. Monro & Staff. coulteri (А. Gray) Rose cirrhopedunculatus Rose arachoides (Dieterle) A. K. Monro & Staff. minimus (Kellogg) S. Watson си ЕСЕЈ 7 8 9. 0 1 ке ке III. E. paniculatus species group Flowers campanulate to infundibuliform; anther subglobose; thecae “banana”-shaped, appearing tu- bular prior to dehiscence after which ribbon-like (Fig. 1c); pollen 6—8-zonocolporate and radially asymmetrical with indistinct endoapertures, the colpus margins undifferentiated and not granular (Fig. 3E-G) Volume 85, Number 2 Monro & Stafford 263 1998 Synopsis of Echinopepon 12. Echinopepon paniculatus (Cogn.) Dieterle TAXONOMIC SYNOPSIS 13. E. glutinosus (Cogn.) A. K. Monro & Staff. 14. E. wrightii (A. Gray) S. Watson 15. Echinopepon pubescens (Benth.) Rose This synopsis presents a key to all of the species of Echinopepon discussed above. Full descriptions are only given for previously undescribed taxa. 16. E. longispinus (Cogn.) Rose However, full nomenclature, including synonymy 17. E. belizensis А. К. Monro & Staff. and literature citations, is given for each taxon to- 18. E. micropaniculatus A. K. Monro & Staff. gether with a list of all specimens examined. KEY TO THE SPECIES OF ECHINOPEPON АП inflorescence and floral characters refer to those of staminate flowers only. la. Peduncle of mature fruit thread-like, 60-80 mm in length at maturity. e Fruit сеооа, lacking an apical operculum ....... ¿tt E. arachoides Fruit not geocarpous, with an apical operculum __ E. cirrhopedunculatus lb. Pci of mature fruit not viret like, more robust, 3-45 mm in length at maturity. 3a. ncle of mature fruit glabrous. 2 гуч weakly lobed to entire; inflorescences with fewer than E — TIC E DNE E. milleflorus Leaves profoundly lobed; inflorescences with more than 25 "ii 3b. Кы of mature fruit pubescent, occasionally only sparsely a. Hypanthium glabrous; anther thecae never folded into 2- to )3- sigmoid curves. a. Corolla lobes 6-10 mm in length 6b. С. lobes 2-4.5 mm in length. · Petiole leder although occasionally with small, white plate-like structures; anther the- с rse-shoe”-shapec 8a. Unbranched portion е tendrils 4-10 mm long; pedicel 3-5 mm long _____- E. gemellus nbranched portion of tendrils 10-65 mm long; pedicel 10-20 mm long |... E. minimus 7b. Petiole densely pubescent; anther thecae “banana”-shaped. Оа. Pedicel 3-5 mm long; anthers fused to form an ovoid or obovoid head UOS RE: E. micropaniculatus b. Pedicel 8-10 mm long: anthers fused to form a subglobose head ________- . wrightii . Hypanthium pubescent; anther thecae “horse-shoe”-shaped, "Ђапапа"-вћаред, or folded into 2- to 3- sigmoid curves 10a. Corolla ko than 5 mm in length; anther thecae "banana "-shaped. 11а. Pedicel 4—6 mm long; fused filaments ho p than 0.2 mm long 00 -- E. glutinosus llb. Pedicel 8-10 mm long; fused filaments ca. 1 mm lor E. wrightii 10Ь. C CORE 5 mm or more in length; anther thecae folded into 2-3. sigmoid curves or, “horse-shoe”- aped. 12a. Inflorescence bran 13a. Inflorescences Bet 12-20 flowers; fruit more than 50 mm in length 13b. Inflorescences with 90-120 flowers; fruit less than 50 mm in length 12b. nescence unbranched. . Thecae not folded into 2- to 3-sigmoid curves eae E. paniculatus л т ЖЕН E. belizensis Жолы E. longispinus 15a. Corolla 5-8 mm long; anther thecae * Bom se-shoe”-shaped ___________Е. coulteri 15b. Corolla 12-18 mm long: anther m ‘banana’ ЈЕ пао“ БЕ ад E. pubescens 14b. Тћес ae — into 2- to 3-sigmoid сї 16a. Corolla divided to less than half it its onu peduncle of mature fruit less than 8i mm in leng 17a. Inflorescences 80-340 mm long; corolla (12)16-22 mm diam.; anthers fused to € an obovoid head 3-3.5 mm long |... E. jaliscanus 17b. ошен, 'ences 45-95 mm long: corolla 7-10 mm diam.; anthers fused to ne or obovoid head 2 mm lon E. tultitlanapaensis 16b. Corolla dvdd to more than half its length; E le of mature fruit greater han 10 mm in length. 18a. Calyx lobes ca. 1 mm lor E. pringlei 18b. Calyx lobes 2-4.5(-9) mm ы КЕСЕНЕ БІЗ E. racemosus 1. Echinopepon racemosus (Steud.) C. Jeffrey, 9: 56. 1955. TYPE: t. 94 in Vell., Fl. flumin. Kew Bull. 33: 357. 1979. Momordica muricata 10. 1831 (holotype). Vell., Fl. flumin. 10: t. 94. 1831, non Willd. 5 пи ок 2. Naudin, Ann. Sci. Nat., Bot. sér. 5, (1805). Momordica racemosa Steud., Nomencl. · 19 ТҮРЕ: material сай Кай и e. bot. Ed. 2, 2: 155. 1841. Echinocystis race- pe .. Mexico by Bourgeau іп 1865-1866 тоза (Steud.) Mart. Crov., Notul. Syst. (Paris) Db P: isotype, GH n.v.). 264 Annals of the Missouri Botanical Garden -— 110 |. 90 80 70 _ 60 50 40 Io 20 a 30 - d К y. № ЛИ ДИ FR A қанады смо о EL \ Көзе =p o а| | |] | b у | <> = “САТ У, 20— . ^ же т е % el | ч Е А . | 10 | 7 0 Е E / e aT a " 10 SO ¡Y = Ô қ кене канын (шн CM > iio е 4 200 | a J14 -— A c =н. о 200 400 600 $09 100049 Br Figure 4. Distribution of Echinopepon racemosus. dii ries Cogn., Diagn. Cucurb. Nouv. 2: 90. (F), Williams 40316 (F). 4. et al. 40361 (MO, NY, 1877. TYPE: Venezuela. Ernst 940 (syntype, BM); F). Baja Verapaz: Hawkes 1939 (K); Jutiapa, Steyermark Venezuela. С Colonia Tovar, Fendler 503 (syntype, К); Colombia. Triana 3017 (syntype. P). Echinocystis lanata 4. Diagn. Cucurb. Nouv. 2: 92. Tu TY М Galeotti s.n. (syntype, BR); :0. Liebmann 49 буйура; 1). Symb. fl. argent. 135. кіна d. araneosa Griseb., 1890. TYPE: Argentina, Lorentz & Hieronymus 551 (holotype, BR; isotypes, GOET, K). Distribution. From northern Mexico (Baja Cal- ifornia Sur, Chihuahua) to northern Argentina (Salta Province) at elevations of 300-3500 m (Fig. 4). Additional 2. examined. MEXICO. Chiapas: Breedlove 13896? (F*. K), Breedlove е ue 13157 (K, ). UN 2166 (к). Lira 957 thihuahua: Gentry 2645 (K). Jaliseo: Lott 1315 K pe 573 (К), Lott 609 (K), Bullock 1267 (K). México: Hinton 14694 (K). Hinton 2524 (K), Hinton 4963 (K), Hinton ied (K). Oaxaca: Carlson 4150 (F). Veracruz: Nee 235 nu. Bourgeau 1478 (P), Bourgeau 3266 (K). GUATE MA Alta Verapaz: гоп Tuerckheim 1099 (K), Wilson 1. = * An asterisk (*) denotes source of pollen for this study. 30378 (F `), Heyde & Lux a d HONDURAS. El Par- aiso: Hawkes et al. 2055 (F, К), oe 2. (F. MO, NY), Molina 8688 (F). Rodríguez 1893 (F). Francisco : Chorley 374 (BM*), babar ae (MO), Le- MO), Molina 24563 (NY*, MO), Molina 730 . NICARAGUA. Esteli: Moreno pines (К), 18406 (К), Могепо 22352 (К),: (К, NY). Jinotega: Molina 27288 (F), Molina 22947 (F. Y), Stevens 22580 (BM, К), Williams et al. 24750 (NY), Williams & Molina 42753 (F*). COSTA RICA. Locality unknown: Echeverría 877 (Е), Lankester 130 (К), León 259 (F). Cartago: James 79 (F), Cooper 5775 (К), Echeverría 203 ud Torres 7 (F). VENEZUELA. Locality unknown: André 2809 (K), Fen- dler 5. (К). BRAZIL. Locality unknown: Pohl 1996 (К), Burchell 6122 (K). Distrito Federal: Sid 1537 (K). Mato Grosso: Hatschbach 34110 (K). ais: Glaziou 19380 (К). Rio de Janeiro: Gla- ziou 12739 (K). BOLIVIA. Locality unknown: Buchtien s.n. (К), Lectrae 533 (К). Cochabamba: Eyerdam 25318 (K). La Paz: fei 2750 (K*), Solomon 8921 (K). Santa Cruz: Beck 6487 (K). —— н 10073 (К). PERU. Locality unknown: Cajamarca: Sagástegui 12980 (К), Sagástegui 11419 (K). Gua Stork 10500 (K). Tumbes: Weberbauer 7726 (K). San Martin: Young 241 (K). AR- Volume 85, Number 2 1998 Monro & Stafford Synopsis of Echinopepon 265 Figure 5. cirrhopedunculatus | belizensis (A). Distribution of Echinopepon pringlei (8). E. (A), E. micropaniculatus (Bl). and E mo GENTINA. Locality unknown: Lorentz & Hieronymus 552 (K), Lorentz & Hieronymus 551 (K). Salta: Krapovickas 28245 (K). Pedersen 12846 (K), Pedersen 12842 (K). 2. Echinopepon pringlei Rose, Contr. U.S. Natl. Herb. 5: 117. 1897. TYPE: Mexico. Morelos: Pringle 6183 (holotype, GH; isotypes, BM*, P). Distribution. Southern Mexico (Veracruz to Oa- xaca) at elevations of ca. 1500-2500 m (Fig. 5). dditional specimens examined. MEXICO. Oaxaca: Galeotti 1880 (K), Barnes & Lord 474 (K), Pringle 4958 (BM*). Veracruz: Galeotti 1899 (K). „ Echinopepon tultitlanapaensis А. К. Мопго & Staff., sp. nov. TYPE: Mexico. Puebla: San Luis Tultitlanapa, Purpus 3548 (holotype, BM*; isotypes, F, GH, МО, NY, US). Figure b. 6a, E. pringlei Rose affinis, sed corolla profundiore incisa, antheris maioribus, staminorum longiori columna, fructu parvo, bene differt. Leaves ca. 5.0—6.5 X 4—7 cm, lobate, charta- ceous, upper and lower surface strigose to hispid, the trichomes with broad multicellular bases; lobes 3, 5, or 7, the base subsagittate, the margins re- motely denticulate, the apex acuminate; petiole ca. 1 mm, densely villous; tendrils bifid, ca. 5—6 cm long before branching. Staminate flowers ca. 9-12, borne in a raceme ca. 4.5-9.5 cm long bearing flowers for % of its length; pedicel ca. 4—5 X 0.3 mm, pilose to villous; hypanthium ca. 3-4 X 2-3 mm, campanulate, villous; calyx lobes ca. 2 mm long, spiciform to filiform; corolla ca. 7-10 mm long, broadly campanulate, fused for Y of its ength, inner surface glabrous, outer surface sparsely pilose to villous, lobes ovate, the apices acute; filaments ca. 2 mm in length, fused; anthers ca. 2 mm in length, fused to form an obovoid head 1.2 mm diam.; unilocular and bilocular thecae 3- sigmoid, glabrous; pollen pantocolporate to syncol- porate, colpi 18-24, branched. Pistillate flowers solitary; pedicel ca. 5 X 0.5 mm, villous; hypan- thium ca. 2 X 3 mm, broadly campanulate, sparsely puberulous, constricted at base, constriction ca. 4— 5 X 0.5 mm, puberulous; calyx lobes ca. 2-3 mm long; corolla ca. 10 X 7 mm, infundibuliform, fused for Y of its length, inner surface glabrous, outer surface sparsely puberulous, lobes ovate, the apices acute; ovary ca. 6–7 mm, ovoid, pilose to villous, spines ca. 3—5 mm; style ca. 2-2.5 mm, glabrous; stigma ca. 1 X 1.5 mm. Fruiting peduncle ca. 4—6 mm, puberulous; fruit ca. 22-37 X 10-12 mm; spines ca. 10-17 mm, pilose; seeds 5 per locule, ca. 5 X 3.5 X 2 mm, obovoid. Distribution. tion at an elevation of ca. 2 Known only from the type collec- 000 m (Fig. 7). This species most closely resembles Echinopepon pringlei in that it has a broadly campanulate hy- panthium, 2- and 3-sigmoid anther thecae, and pantocolpate pollen. It differs, however, with re- spect to the perianth and stamens. The perianth of E. tultitlanapaensis 1s larger and the corolla lobes are fused for only % of their length (compared to 2% in E. pringlei), causing them to spread to a much greater extent. In addition, the anthers of E. tultit- lanapaensis are significantly larger, glabrous, and attached to longer filaments. The name is derived from the locality of the type collection, San Luis Tultiltanapa. 4. Echinopepon jaliscanus Rose, Contr. U.S. Natl. Herb. 5: 117. 1897. TYPE: Mexico. Jal- isco: Pringle 4563 (holotype, US n.v.; isotype, K). Distribution. Pacific side of central Mexico (México to Guerrero) at elevations of ca. 17 (Fig. 8 Additional specimens examined. MEXICO. México: Hinton 1968 (K*), Hinton 5004 (K), Hinton 5211 (K), Hinton 8257 (K). Guerrero: Hinton 10978 (K*), Hinton 11621 (K — 5. Echinopeon torquatus (DC.) Rose, Contr. U.S. Natl. Herb. 5: 118. 1897. Elaterium tor- quatum DC., Prodr. 3: 310. 1828. Echinocystis torquata (DC.) Cogn., Diagn. Cucurb. Nouv. 2: 88. 1877. TYPE: tab. 38, fig. C in Мос. & Sessé, Icones Fl. Mexic. ined. (holotype, G* n.v.). ҒА reproduction of the plates in /cones Fl. Mexic. ined. can be found in A. DC., Calques Fl. Mexique. 1874 (BM!). 266 Annals of the Missouri Botanical Garden 254” FLORA OF MEXICO © yu met 1908 BRITISH MUSEUM (NATURAL HISTORY Sample taken for pollen collection id Руз 15), lay Figure 6. Photograph of the holotype of Echinopepon tultitlanapaensis (Purpus 3548, BM).—a. Habit.—b. Staminate flower. All scale bars in mm. Volume 85, Number 2 Monro & Stafford 267 1998 Synopsis of Echinopepon 110 100 90 110 100 90 2 ~ = p- | 77а 3 4 UCET pem көк Ze hui man DNE 224 а как Ji - 7 pem ms. A e T ' m О „ 20 , Ku e d ы » ~ Figure 7. Distribution of Echinopepon milleflorus (8). E. minimus (A), E. wrightii (ІШ), and Е. tultitlanapaensis (Л). инсо и ag Ann. Sci. Nat., Bot. ‚ 6: 18. . TYPE: Воигреаи s.n. ms = ызы d in e teas seed sent from Mexico by Bourgeau in 1865-1866) (holotype, P). Distribution. The Pacific side of Mexico at el- evations of 2000—2600 m (Fig. 9). Additional specimens examined. MEXICO. unknown: Sessé et al. кез (М 290 (P). Baja California 5 ; pas: Ortiz 1207 (F). Breedlove 6769 (F*), Breedlove 12431 (F*). Distrito Feder : Bourgeau 1060 (K, P). México: Hinton 7648 (К). у Locality A n.v., photograph K), Hahn " 6. Echinopepon milleflorus Naudin, Ann. Sci. Nat., Bot. sér. 5, 6: 18. 1866. Echinocystis mil- йога (Naudin) Cogn., Diagn. Cucurb. Nouv. : 88 . TYPE: Mexico. Bourgeau s.n. nd aiiud in Paris from seed sen from Mexico by Bourgeau in 1865-1866) (ho- lotype, P). Distribution. Pacific side of central and зошћ- ern Mexico at elevations of 1500-2640 m (Fig. 7). Additional specimens examined. MEXICO. Chiapas: рақаты 963 (6), 4. 1169 (К). Distrito Federal: K. P). Pringle 6457 (BM*, K, P), Roe . Rose & Painter 7121 (BM). Méx- ico: Hinton 8561 ©, Boon 7366 (К). 7. Echinopepon gemellus (DC.) A. K. Monro & Staff., comb. nov. Basionym: Elaterium gemel- lum DC., Prodr. 3: 310. 1828. Echinocystis ge- mella (DC.) Cogn., Diagn. Cucurb. Nouv. 2: 88. 1877. TYPE: tab. 38, fig. B in Мос. & Sessé, [cones Fl. Mexic. ined. (holotype, G n.v.). yr, Linnaea 30: 56. 1859-1860. xico. Heller 393 (lec totype, here designat- ed, P); d s.n. (syntype, not traced). Sicyos eremoc чле Ре ТҮРЕ e 8. Distribution of Есһіпорероп paniculatus gur ө). E coulteri (Bl). and Е. jaliscanus (A). Distribution. Pacific side of central Mexico at elevations of ca. 2700 m (Fig. 9). Additional specimens examined. MEXICO. Distrito Federal: Schmitz 87 (BM). Morelos: Schmitz 997 (BM), Berlandier 1115* (BM). This species was placed in the poe dd sect. Echinopepon by Cogniaux (1881). Rose did not in- corporate this into the genus Есһіпорероп when this was resurrected by him (Rose, 1897) since he was unable to obtain herbarium material of this taxon. 8. Echinopepon coulteri (A. Gray) Rose, Contr. U.S. Natl. Herb. 5: 116. 1897. Elaterium coul- teri. А. Gray, Pl. wright. 2: 61. .T Mexico. Zacatecas: Coulter 51 (вере. CH. isotype, Ба confusus Rose, Contr. 0.5. Natl. Herb. 5: 115. 1897. TYPE: U.S.A. New Mexico: Thurber 1122 (holotype, GH). Echinopepon nefsoni Rose, Contr. U.S. Natl. Herb. 5: 117. zt TYPE: Mexico. Oaxaca: Nelson 1878 (holo- type, US). Echinopepon p Rose, Contr. U.S. Natl. Herb. 5: 118. 1897. : Mexico. Oaxaca: Сопгаш 139 (holotype, Tu Figure 9. E. pubescens (А), and E. gemellus Distribution of Echinopepon torquatus (0), an. 268 Annals of the Missouri Botanical Garden Distribution. Southwestern United States (New 12. Echinopepon paniculatus (Cogn.) Dieterle, Mexico) to southern Mexico (Veracruz) at elevations of ca. 2000-2350 m (Fig. 8). Additional specimens examined. U.S.A. New Mexico: Metcalfe 1348 (BM*). MEXICO. Distrito Federal: Bour- geau 789 (К, P), Hahn 162 (P*). Guanajuato: Haa, Schmidt 598 (K). Morelos: 1. 1388 (P). Greene 17011 (K). Veracruz: Nee & Soulé 33055 (K*). Z tecas: Wright s.n. (K), Rose 2699 (K). Balls & Gourlay 530 (BM), collected in Puebla, Mexico, at 2600 m elevation, differs from all other specimens examined in having significantly larger flowers and relatively longer stamens. x Echinopepon cirrhopedunculatus Rose, Contr. U.S. Natl. Herb. 5: 115. 1897. TYPE: Mexico. Sonora: Palmer 634 (holotype, US; isotype, K). Distribution. Northern to central Mexico at el- evations of ca. 500-1000 m (Fig. 5). Specimens examined. MEXICO. Chihuahua: Gentry 2355 (K). Guerrero: Hinton et al. 10496 (K, P). Jalisco: Pringle 4562 (BM*). México: Hinton 1364 (K*), Hinton 7 (K). Hinton 6438 (K), Hinton 8483 (K). 10. Echinopepon arachoides (Dieterle) A. K. Monro & Staff., comb. nov. Basionym: Apatzin- gania arachoidea Dieterle, Brittonia 26: 131. 1974. TYPE: Mexico. Michoacán: Dieterle 4379 (holotype, MICH n.v.; isotype, K Distribution. Pacific Coast of Mexico (Michoa- cán to Guerrero) at an elevation of ca. 300 m Additional specimen examined. MEXICO. Guerrero: Hinton 6424 (K*). 11. Echinopepon minimus (Kellogg) S. Watson, Proc. Amer. Acad. Arts 24: 52. 1889. Marah minima. Kellogg, Proc. Calif. Acad. Sci. 2: 18. . Elaterium minimum (Kellogg) S. Wat- son, a 589 mer. Acad. Arts 12: 252. 1877. TYPE: Mexico. Baja California Sur: Cedros Is- land, Streets s.n. (lectotype, designated by Stocking (1955), US n.v.; n.v., NO photograph). isolectotypes, GH Distribution. Limited to Baja California Sur at elevations of ca. 200-1000 m (Fig. 7). Additional specimens examined. MEXICO. Baja Cal- ifornia Sur: Brandegee s.n. (K photo, US), Palmer 719 (K). о 299 (К), Gentry 4125 (К), Palmer 65 (К), s 14436 (K*), Aug. 1859—Jan. 1860, Xantus de Ve- Сн) Fieldiana, Bot. 24(11): 342. 1976. 2. paniculata Сор, Diagn. Cucurb. Nouv. . TYPE: Mexico. Guerrero: Galit. s.n. (holotype, BR). Distribution. Central Mexico (Guerrero) to southern Guatemala (Chiquimula) at elevations of 50—1360 m (Fig. 8). Additional specimens examined. MEXICO. Campe- e: Andres & Nee 157 (K). Chiapas: Lira et al. 930 (BM), Lira et al. 942 (BM), Soto 13269 (BM). Guerrero: Mexía 8703 (К), Hinton 6623 (К), Hinton 8509 (К). BE- LIZE. Bartlett 12882 (F), ee 2190 (F), Liesner & Dwyer 1626 (K, MO). GUA ALA. Locality unknown: Bernoulli & Cario 2838 a Chiquimula: Standley 73737 (F). Huehuetenango: Molina 21325 (F), Molina & Molina 30171 (F, MO), Williams 41113 (F*), Williams 41364 (F*). Jalapa: Standley 77384 (F), Standley 76495 (F*). Petén: Ortiz 547 (MO*). т. 13. Echinopepon glutinosus (Cogn.) A. К. Mon- ro & Staff., comb. nov. Basionym: Echinocystis glutinosa из Diagn. Cucurb. Nouv. 2: 1877. TYPE: Bourgeau s.n. 1866 (mater cultivated in Panas in 1867 from seed sent from Mexico by Bourgeau in 1866) (holotype, P*). Distribution unknown. Mexico; known only from the type collection. This species was placed in Echinocystis sect. Echinopepon by Cogniaux (1881). corporate this into the genus Echinopepon when this was resurrected by him (Rose, 1897), since he had been unable to obtain herbarium material of this taxon. ose did not in- 14. Echinopepon wrightii (A. Gray) S. Watson, Bull. Torrey Bot. Club 14: 158. 1887. Elater- ium wrightii А. Gray, Pl. wright. 2: 61. 1853. TYPE: Mexico. Sin. loc., Wright 1090 (holo- type, US; isotypes, GH-257*, -259* & -260) Distribution. Central Mexico at elevations of ca. 1000-2000 m (Fig. 7). MEXICO. Zacatecas: Additional specimen examined. Emory 397 (K). The GH isotype contains two different types of pollen: sheet 260 has zonocolpate, spinulose pollen of a type found in the Sicyinae, whereas sheets 2 and 259 have 7—9-zonocolpate, non-spinulose pol- len similar to that of the holotype. It is not, how- ever, possible to say with any certainty whether this is the result of a mixed collection or whether this species has dimorphic pollen, since sheet 260 does not have any additional, sufficiently mature flowers from which to draw a conclusion. Volume 85, Number 2 1998 Monro & Stafford Synopsis of Echinopepon 269 15. Echinopepon pubescens (Benth.) Rose, Contr. U.S. Natl. Herb. 5: 118. 1897. Elater- ium pubescens Benth., Pl. hartw. 6. 1839. Echinocystis pubescens (Benth.) Cogn., Diagn. Cucurb. Nouv. 2: 88. 1877. TYPE: Mexico. Aguascalientes: Hartweg 15 (holotype, K; iso- type, BM*) т floribunda Cogn., Diagn. Cucurb. Nouv. 2: 89. 1877. Echinopepon floribundus (Cogn.) Rose, CE U.S. Natl. Herb. 1: 116. 1897. TYPE: Mexico. Oaxaca: Liebmann 53 (syntype, B missing and prob- ably destroyed); Liebmann 28 (syntype, В missing and probably destroyed); Galeotti 1890 7 totype, here designated, P; isolectotypes, G, K, Distribution. Central to southern Mexico (Aguascalientes to Oaxaca) at elevations of ca. 1500-1800 m (Fig. 9). Additional specimens ve iet MEXICO. Locality unknown: Hartweg s.n. . Michoacán: Pringle 4346 (BM*, K). Oaxaca: eee 1890 (K, P), Pringle 4957 (BM*, K), Purpus 4204 (BM*, K), Rose 11313 (K). 16. Echinopepon longispinus (Cogn.) Rose, Contr. U.S. Natl. Herb. 5: 117. 1897. Echin- ocystis longispina Cogn., Diagn. Cucurb. Nouv. 2: 91. 1877. TYPE: Mexico. Veracruz: Schiede 1080 (lectotype, here designated, GH; isolec- totype, B missing and probably destroyed). Distribution. Central Pacific side of Mexico at elevations of ca. Additional specimens waited MEXICO. Mo- relos: Pringle 11301 (K), Ravias 541 (P), Hahn S. 17. Echinopepon belizensis A. K. Monro & Staff., sp. nov. TYPE: Belize. Cayo: El Cayo, Bartlett 11989 сре NY*; isotypes, GH, US). Figure 10a, b. E. pubescenti (Benth.) Rose affinis, sed hypanthio latis- simo brevissimoque, corollae lobis brevibus latisque, cor- ollae glandis pedicellatis, antheris subsessilibus, bene dif- fert. Leaves ca. 5.0-7.5 X 4.5—7.0 cm, lobate, mem- branous, upper surface puberulous and minutely pustulate, lower surface puberulous, lobes 3, 5, or 7, the base cordate, the margins remotely dentic- ulate, the apex acuminate; petiole ca. 30—50 X 1 mm, densely puberulous; tendrils bifid, ca. 4.5 cm long before branching. Staminate flowers ca. 12— 20, borne in a panicle ca. 60 mm long, bearing flowers for % of its length; pedicels ca. 11 х 0.2 mm, puberulous; hypanthium ca. 1 X 2 mm, pa- telliform, glabrous; calyx lobes dentate, ca. 0.5 mm long, glabrous; corolla ca. 5-6 X 10-12 mm, ра- telliform, fused for % of its length, inner surface stalked glandular, outer surface farinaceous, lobes ovate, the apices acuminate; filaments ca. 0.1 mm in length, fused; anthers ca. 1-1.5 mm in length, fused to form an ovoid to obovoid head 0.5-0.8 mm diam., subsessile; unilocular and bilocular thecae semi-sigmoid or “J”-shaped, microechinate; pollen ca. 7-zonocolporate, radially asymmetrical (there may be a raised annulus-like structure encircling the pore, as present in the related genus Rytidos- tylis, clearly seen in SEM but indistinct in LM). Pistillate flowers 1-2, solitary or borne in a fascicle; pedicel ca. 4—15 X 0.5 mm, glabrous, sparsely pu- berulous; hypanthium ca. 2.5 X 4 mm, campanu- late, sparsely puberulous, constricted at base, constriction ca. 3.5—4.5 X 0.5 mm, densely puber- ulous; calyx lobes ca. 0.5 mm long; corolla ca. 7 Х 9 mm, narrowly patelliform, fused for % of its length, inner surface stalked glandular, outer sur- face glabrous, lobes ovate, the apices acuminate; ovary ca. mm, ovoid, puberulous, spines ca. 5-3 mm; style ca. 0.5 mm, glabrous; stigma ca. 0.8 X 1 mm. Fruit not seen. Distribution. Known only from the type collec- tion made at 200 m (Fig. 5). This species most closely resembles Echinopepon pubescens in having ovoid anthers and zonocolpor- ate pollen. It differs, however, іп hypanthium shape, corolla shape, corolla indument, and anther disposition. The staminate and pistillate perianths in E. pubescens are composed of a relatively long, narrow campanulate hypanthium from which spread out the deeply divided, long, narrow and recurving corolla lobes; in Ё. belizensis, however, the hypanthium is very broad, and plate-like, the corolla lobes less divided, shorter, broader and not recurved. The glands covering the inner surface of the corolla lobes are stalked in E. belizensis and sessile to subsessile in E. pubescens. The anthers of E. belizensis are subsessile, while those of Е. pubescens are supported by a filament 1—1.5 mm ong. The two species also differ greatly in their altitudinal ranges, E. pubescens not having been collected below 1500 m while Е. belizensis was col- lected at ca. 200 m. 18. Echinopeon micropaniculatus A. K. Monro Staff., sp. nov. TYPE: Costa Rica. Guana- caste: 2 km E of Hacienda Palo Verde, Com- elco property, Keeler 192 (holotype, MO*). Fig- ure lla, b E. paniculato (Cogn.) Dieterle affinis, sed floribus mi- noribus, antheris ovoideis, bene differt. Leaves ca. 6-10 х 5-9.5 cm, lobate, subchar- Annals of the Missouri Botanical Garden <= (=> PLANTS ОР BRITISH HONDURAS CAYO DISTRICT F THE СавивотЕ INSTITUTION oF pi AND THE UNIVERSITY OF M ее \ ” Hay 9б: k AN у dors . АЛ BOTANICAL > от RO IZA . H. BARTLETT we //7Р) | Figure 10. Photograph of the M specimen of Echinopepon belizensis (Bartlett 11989, NY).—a. Habit.—b. Staminate flower. All scale bars in mn Volume 85, Number 2 Monro & Stafford 271 1998 Synopsis of Echinopepon Echinocystis sp. Richard P. Wunderlin University of South Florida 1977 Figure 11. Photograph of the holotype specimen of Echinopepon micropaniculatus (Keeler 192, MO).—a. Habit.— b. Staminate flower. All scale bars in mm Annals of the Missouri Botanical Garden taceous to chartaceous, upper and lower surface villous, lobes 3, 5, or 7, the base sagittate to cor- date, the margins remotely denticulate, the apex acuminate; petiole ca. 25—40 X 1.5-2 mm, densely villous; tendrils trifid, ca. 2.5-9.0 cm long before branching. Staminate flowers ca. 50—60, borne in a panicle ca. 100 mm long bearing flowers for У— % of its length; pedicels ca. 3—5 X mm, pilose; hypanthium ca. m, broad campanulate to patelliform, glabrous; calyx lobes deltate, ca. 0.3 mm long; corolla ca. 2 X 3 mm, patelliform to broadly campanulate, fused for % of its length, in- ner surface subsessile, outer surface glabrous, lobes elliptic to ovate, the apices acute to obtuse; fila- ments ca. mm in length, fused; anthers ca. 1 mm in length, fused to form an ovoid to obovoid head 0.4 mm diam.; unilocular and bilocular thecae semi-sigmoid or “J”-shaped, glabrous; pollen ca. 6- zonocolporate, radially asymmetrical. Pistillate flowers not seen. Fruiting peduncle ca. 5 mm, pi- lose; fruit ca. 30 X 14 mm; spines ca. 15-16 mm, sparsely pilose; seeds 2 per locule, ca. 6 X 3.5 X mm, rectangular, armored. Distribution. This species is known only from the type collection (Fig. 5). Echinopeon micropaniculatus most closely re- sembles E. paniculatus in the morphology of the perianth and pollen. It differs significantly, how- ever, in flower size and anther structure. The flow- ers of E. micropaniculatus are ca. 3 mm in diam- eter, while those of E. paniculatus are 7-14 mm in diameter. Literature Cited Cogniaux, A. 1877. Diagnoses de Cucurbitacées nouvel- les et observations sur les espéces critiques. Deuxiéme fascicule. F. Hayez, Brussels. . Cucurbitaceae. Pp. 798-820 in A. " A., Monographiae Phanerogamarum 3. С Masson, Ag Dieterle, J. V. . A new geocarpic genus from Mex- ico: Apatzingania Ge 'urbitaceae). Brittonia 26: 129— idinan, G. 1969. Handbook of Palynology. Munks- ажар o nhagen. Jeffrey, C. 1961. Notes on Cucurbitaceae including a 4 2. new classification of the family. Kew Bull. 337-371. . 1963. Notes on Cucurbitaceae including a n. posed new classification of the family. Kew Bull. 483. 1964. А e оп E ишан in Cucur- Бабага, Kew Bull. 1980. А + ot e Cue 'urbitaceae. Bot. J. Linn. 890, 81: 233-247. )90. Appendix: An Outline . of the Cue urbitaceae. Pp. 449-463 іп D. M. Bates, R. W. Robinson & C. Jeffrey (editors), Biology and Utilization of the Сис С Comstock Publishing, Cornell Univ. Press, Ithac Naudin, С. 1866 6. wo 'urbitacées nouvelles cultivées au Muséum em Naturelle еп 1863 1865. Ann. Sci. Nat., Bot., sér. 5, 6: 17-19, "а dines ar, K. 196 = of Cucurbi- 3ull. Natl. Inst. Sci. India 3 )-396. 976. А 'al highlights of Cucurbitaceae. ids 'es Pollen-Spor : 54-59. M J. N. 1897. Studies . Mexican and Central Amer- an plants. Contr. U.S. Natl. Herb. 5: 109-144 Stafford. P. J. & A эл 1994. Pollen morphology of the Cyclantherinae C. Jeffr. (Tribe Sicyeae Schrad., Cuc urbitaceae) s its e significance. Acta Bot. Gallica 141: 71-182. Stocking, K. M. 1955. Some considerations of the genera Echinocystis and тш in ш = Su and northern Mexico. Madrofio 13: 84— 1840. A flora 4 a America 1: Ey S ÑS i ог КЫ! н, 1. М. Ча С. 1831 [71827 `]. Icones. A. Se аге бе. Paris Watson, S. The genera А Megarrhiza and dre odii Bull. Torrey Bot. Club 14: 155-158. 1889. ио to American botany. Proc. . Acad. Arts 51-52. Florae Fluminensis Amer. THE GENERA CESTRUM AND Carmen Benttez de Rojas? and SESSEA (SOLANACEAE: Иал О CESTREAE) IN VENEZUELA! ABSTRACT Solanaceae, tribe Cestreae, is represented in Venezuela by the two genera Cestrum (31 species) and Sessea (1 species). This treatment distinguishes the genera and their species by dichotomous keys. All species are provided with descrip- pe illustrations, distribution maps, and notes on their appearances, phenology, and geographical ranges. A list of ecimens seen that were made in Venezuela is also provided Tribe Cestreae of the Solanaceae is represented superior ovaries.The tribe includes three closely re- in Venezuela by two genera, Cestrum L. and Sessea lated genera, Cestrum (150 species), Sessea (25 spe- Ruiz & Pav. Plants of the tribe are shrubs, trees, cies), and Vestia Willd. (1 species), and two more and vines with entire leaves, and the flowers have distant monospecific genera, Metternichia Mikan tubular corollas that are long-exserted from small (eastern coastal Brazil, and Tsoala Bosser & calyces. The flowers of Cestrum and Sessea are so D'Arcy (Madagascar). Cestrum ranges from north- similar that the genera usually cannot be distin- егп Mexico and southern Florida to southern Chile. guished without fruit. Cestrum has a juicy berry and Sessea is restricted to tropical South America ex- more or less prismatic seeds, while Sessea has a cept for a dubious record from Hispaniola, and its dehiscent capsule with winged seeds. diversity is centered in Andean regions, especially This paper, based on field, herbarium, and Ecuador. Vestia is restricted to south—central Chile. greenhouse studies, revises the Venezuelan species It has flowers similar to those of Cestrum and Sessea in the two genera. À summary of the work with statistics on ranges, ecology, and other relationships is being published elsewhere (Benítez & D'Arcy, in press). Photographs of representative species (Figs. 1, 2) and line drawings and distribution maps for all species (Figs. 3—62) are provided. Data for spec- imens cited in this paper and that support state- ments about distribution outside of Venezuela have : | kid been entered into TROPICOS, the Missouri Botan- 1936, who recognized 257 species. Within Cestrum, ical Garden computer database of scientific infor- а Section Habrothamnus has been recognized, mation, where they may be accessed via the Inter- Which encompasses a small suite of showy species net at http://mobot.org/; h/pick.html ranging from Mexico to Nicaragua, but the remain- der of the genus (sect. Cestrum) 18 still undivided. Sessea was revised by Bitter (1922), who divided the genus into series. It was revised again by Fran- Tribe Cestreae embraces plants with near-acti- сеу (1934), who ignored the series of Bitter and nomorphic flowers, small, persistent calyces, long, recognized 23 species. D'Arcy (1979) suggested narrowly tubular corollas, small, longitudinally de- that Sessea may not have nearly as many as hiscent anthers held near the corolla mouth, and species. but much larger and a capsular fruit with more or less prismatic seeds. The tribe is a member of sub- family Cestroideae, which in recent phylogenetic schemes (Olmstead & Palmer, 1992; Olmstead et al., in press) is basal to the other subfamily, Solan- oideae. Cestrum was revised for Venezuela by Pit- tier in 1932 and in its entirety by Francey in 1935- SYSTEMATICS ! This study was conducted with binational support, and we thankfully acknowledge funding from the Consejo Na- cional de Investigaciones Científicas y Tecnólogicas (CONICIT PI-56) of Venezuela and the National Science Foundation (INT-9116039) of the U.S.A. Many herbaria made their facilities available or sent specimens on loan for our study, and we gratefully acknowledge dies assistance. Thanks are offered to Bruno Manara, who graciously prepared our illustra- tions, Richard C. Keating (Missouri Botanical Garden) for photographic assistance, and C. E. Freeman, University of Texas, El Paso, who provided sugar analyses of Cestrum nectars. Francisco Rojas lent invaluable support to many parts of the project, including an especially good eye for locating Cestrum alot from moving vehicles. ? Facultad Agronomía, Universidad Central de Venezuela, Maracay, 27 Correos 4579, Aragua, Venezuela. * Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166, U.S.A ANN. MISSOURI Bor. GARD. 85: 273-351. 1998. Annals of the Missouri Botanical Garden Figure 1. Selected Cestrum species.—A. Cestrum bigibbosum. Fruits on an are hing branch (after Benítez 508. DESCRIPTIVE NOTES Most Venezuelan species of Cestreae are shrubs or small trees, but some (Cestrum humboldtii, C. lindenii, C. microcalyx, C. racemosum) become large trees, and three (C. scandens, S. strigilatum, C. reflexum) are scrambling or twining vines. Most species coppice readily and root at the nodes when the stems are cut off, thus altering the growth form from a tree with a single trunk to a many-stemmed shrub. The leaves of Solanaceae are estipulate, but those of Cestreae are often paired, а normal “та- jor” leaf accompanied by a much smaller, often ses- sile “minor” leaf, a condition sometimes suggestive of stipules (Fig. 57, Cestrum tomentosum). An in- terpretation of the minor leaves of Cestreae as ho- mologous with those in subfamily Solanoideae (see Eichler, 1875; Danert, 1958) seems reasonable, but morphological study to verify such homology is de- sirable. Minor leaves are more commonly present on seedlings and turions and are caducous in many species. They are a conspicuous and useful taxo- nomic character in some species (C. mariquitense, C. petiolare, C. tomentosum, C. humboldtii). Leaves are often malodorous, even when not bruised. The inflorescences of Cestreae are mostly sev- Cestrum petiolare. T а on branch (after D'Arcy 18235).—B. eral- to many-flowered clusters among the foliage of ascending branches, but in a few species such as C. megalophyllum, they are more or less cauli- florous in the axils of shoots on the main stem or trunk. Bracts are commonly present, often in a se- ries grading from normal leaves to small structures subtending branches of the inflorescence or indi- vidual flowers. In addition, a small, linear, and of- ten caducous bract termed a bracteole subtends each pedicel. Pedicels, which usually appear to be basal contractions of the calyx, sometimes extend several millimeters down to the subtending brac- teole, but in many species they are obsolete, so that the calyx is sessile in the axil of the bracteole, and sometimes on the peduncle or inflorescence branch: in many cases, where there is no peduncle or stalk below the bracteole, flowers are termed sessile, even when a short pedicel is present above the bracteole. [In this sense, the term sessile means lacking a stalk below the bracteole.] The flowers of Cestrum and Sessea all have tu- The flowers of most Venezuelan Cestreae species open bular corollas exserted from small calyces. in the evening or at night and are closed during the day. When open, the corolla lobes spread wide- ly or are reflexed (Fig. 2A, B), and a sweet fra- Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 275 Figure 2. Selected Cestrum species. trum buxifolium. Flowers, midday view, greenhouse plant (D grance is emitted. In these species, thin, mostly pubescent, marginal tissue between the corolla lobes folds out of sight during the day (Fig. 2C). The calyces of Venezuelan Cestreae are usually apically pubescent with the margins ciliate and the tips “tufted.” They may have other pubescence as well. Corollas are mostly inconspicuous greenish, yellowish, or whitish, and nocturnal or crepuscular in opening, but in some species they are showy with bright red, orange, or other colors and open during the day. In Venezuela, only C. lindenii and С. diurnum have showy corollas, the first of these bright yellow and purple, and the second bright white. Red, blue, purple, and pink flowers, found in species of Cestrum in other areas, are absent from the suite of species occurring in Venezuela, although some species have violet or purplish pains on otherwise whitish or yellowish corol- The corollas of Cestreae are inserted about half- way up the height of the ovary, but the distance between the point of corolla insertion and the base of the calyx, either the cupular portion or the nar- rowed, pedicel-like portion, varies between species. —A. Cestrum latifolium. Night view of inflorescence (Benítez 5079).—B. Ces- Arcey 18236).—C. Cestrum alternifolium. Flowers, daytime . greenhouse plant (D'Arcy 18206). Note expansion at top of corolla tube.—D. Cestrum tomentosum. Fruits, green- ibus: plant (D'Arcy 17838). In this treatment, corolla length refers to the dis- tance between the point of corolla insertion and the tip of the corolla lobes, a measurement that re- quires dissection of the flower. Flower length refers to the distance from the bracteole (base of the ped- icel) to the tip of the corolla lobes, a measurement that can be made without dissection. The stamens are of similar but usually slightly unequal lengths, and anthers are clustered just in- side the corolla mouth. Pubescence, size, and shape of the filaments are of taxonomic utility. Fil- aments are adnate to the corolla tube for different distances from the corolla insertion and are often evident as raised areas along the corolla interior. The 0.5-1.5-mm-long tumid region where the fila- ment is inserted in the corolla tube is termed the insertion. In some species, the insertion is not ad- nate to the corolla tube for its entire length, but is free in its distal part. In some species the insertion is associated with umbos or teeth, and when the filament departs abruptly from the corolla wall, it is termed geniculate. Above the insertion, the fil- ament is glabrous, slender, and of roughly uniform 276 Annals of the Missouri Botanical Garden diameter except at the apex, where it narrows into a neck that supports the versatile anther. In Venezuelan members of Cestreae, the stigma is usually situated within the corolla mouth and above the anthers, but in Cestrum diurnum and Ses- sea corymbiflora it is exserted from the corolla. In Sessea, the stigmas of the species examined are po- sitioned laterally or obliquely on the style (Fig. 61H), while in Cestrum and Vestia the stigmatic surface is apical. If this oblique stigma is constant in other species of Sessea, it may serve in addition to characters of the fruit to separate Sessea from the other two genera. Most species of Cestrum have fewer than 16 ovules and many have 8 or fewer, but C. petiolare is unusual in the genus in having as many as 32 ovules. Seed number varies with species and in individual fruits, and the size and shape of seeds appear to depend on seed number, 1- and 2-seeded fruits bearing larger seeds. Seed shape is variable, seemingly depending on the po- sition of neighboring seeds in the berry. In some species, such as C. diurnum, the placental area re- mains juicy long after the mesocarp is dry, provid- ing a moisture reward to dispersers over an extend- ed period. Fruits of most native Venezuelan species have dark and juicy or fleshy mesocarp, but diurnum and a number of species in Central Amer- ica have white fruits with spongy mesocarp. POLLINATION AND DISPERSAL The factors selecting for morphological diversity in Cestrum are largely unknown. In Mexico, red- an yellow-flowered and red-fruited species appear to be associated with hummingbird pollination (D'Arcy, in press), and this syndrome may also be represented in the bright yellow-flowered C. petiolare in Venezuela. Flowers of most other species open at dawn or dusk or at night, and are sweet-scented. Many species bloom sporadically, producing few or many blossoms over a short period and then, after weeks or months, producing another flush of blooms. In some species, such as C. latifolium and C. alternifolium, plants are “mass-bloomers,” producing a display of hundreds of flowers lasting only one or two days. In greenhouse plants we found that flowers of C. alternifolium, C. latifolium, C. mariquitense, C. megalophyllum, C. ra- cemosum, and C. strigilatum had undetectable amounts of nectar. Nectars produced by other species vary greatly in sugar concentration and composition. Nectar content was reported by Percival (1965) and Bernardello et al. (1994), and we studied eight dif- ferent greenhouse-grown species. Nectars in our eight species had concentrations ranging from 11% to 40%. Nectars in most species were sucrose-dominant, but some had more fructose than sucrose. In nectar of the yellow-flowered C. corymbosum Schltdl., which was characterized as hummingbird pollinated, er et al. (1984) reported a content of amino acids. Ber- nardello et al. (1994) recorded phenols, lipids, and amino acids in nectar from Cestrum cf. bracteatum Link & Otto. Thus, while some data have been re- ported about Cestreae nectar, the picture is still poorly sketched. Overland (1960) studied the temporal opening and closing of flowers of Cestrum nocturnum and found that these movements are independent of light but are affected by temperature. She reported that scent emanates from the tips of the corolla lobes. Actual pollinator observations have been pub- lished for only one species, Cestrum alternifolium. n Costa Rica, Haber and Frankie (1989) and White et al. (1994) examined pollination of C. al- ternifolium by hawkmoths (Sphingidae), which are night-flying insects commonly attracted to white flowers with strong fragrance. Similarities of many other species of Cestrum suggest that they, too, are sphingid pollinated. Hummingbirds and sphingids are characterized as being drawn to plants with su- crose-dominant nectars (Procter et al., 1996) such as those in Cestrum, which might facilitate shifts between or sharing by these pollinators in flowers such as those of C. petiolare and C. aurantiacum, where the corolla mouth is large enough to permit — access by hummingbirds. Fruits of Cestrum are mostly juicy berries pre- sumably dispersed by birds, but bat dispersal can- not be ruled out. Bat dispersal of fruits and seeds reported for several species with strongly foetid fo- liage, e.g., Solanum sect. Brevantherum (Fleming, 1988) and Cleome (Ruiz Zapata, 1993), may be en- hanced by the odor of the leaves. Thus, species of Cestrum with foetid foliage and more or less leath- ery fruits (C. glabrescens?) may be suspected of at- tracting bats to disperse their fruits. USES AND TOXICITY А few species of Cestrum, none native to Vene- zuela, are cultivated for ornament. Cestrum noc- turnum is grown for the evening fragrance of its flowers, and C. aurantiacum for its showy orange- yellow flowers. Cestrum racemosum is sometimes cultivated as a city park tree. The fruit pulp of Ces- trum. buxifolium is used as writing ink. The Solanaceae are well known for having a wide array of alkaloids (Romeike, 1978), and species of Cestrum have been implicated in more than one tox- icity system. One system mimics vitamin D attributes Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 277 and influences calcium uptake (Prema & Raghura- mulu, 1994), leading to a condition referred to as cal- cinosis (Wasserman, 1978). Thus, C. diurnum, occa- sionally cultivated in Venezuela, has been implicated in fatal poisoning of horses in Florida (Krook et al., 975). This toxicity is also found in Solanum glau- cophyllum Desv. (S. malacoxylon Sendtn.) in Brazil, producing an illness called *enteque seco" (Wasser- man, 1974). Another system involving liver poisoning has killed cattle in other countries (Atkinson & James, 1979; McLennan & Kelly, 1984) and must be suspected in Cestrum species growing in Venezuela. Most reports of hepatic toxins involve Cestrum parqui L'Her., an Argentine species now naturalized in South Africa and Australia, but herbarium labels and at least one published report (Atkinson & James, 1979) indicate similar properties in red-flowered species in Central America. These hepatotoxic species have rel- atively large showy flowers like those of C. lindenii or . petiolare їп Venezuela, suggesting the possibility that these too may have toxic properties. Cestrum ac- uminatissimum is used in Venezuela for stunning fish. Much of what is known about the toxicity and bi- ology of Cestrum may apply also to Sessea. One rel- evant investigation was carried out by Andrade CYTOLOGY Cytological reports (cited below under species descriptions) for a number of species of Cestrum = 8 is general in the genus. Numbers other than n = 8 or 2n indicate that a chromosome number of n = 16 have been reported for some cultivated species. A chromosome number of n = 8 has also been reported for Vestia, but no chromosome data are known for Sessea. The numbers n = 8, 2n = 16 in Cestreae contrast with the prevailing number in subfamily Solanoideae of n = 12, 2n = 24, sug- gesting a significant taxonomic Berg and Greilhuber (1992, 1993a, 1993b) noted that chromosomes of Cestrum are unusual in flow- ering plants in having cold-sensitive regions. The functional or adaptive roles of these regions are not understood, but variation was found in the five spe- cies they studied. MATERIALS AND METHODS Fieldwork comprised three extended trips with students and other helpers covering most parts of Venezuela, and many shorter, less directed trips to various parts of the country. Specimens were stud- ied from herbaria in Venezuela: CAR, CORO, GUYN, IRBR, MER, MERC, MERF, MY, MYF, PORT, TFAV, UCOB, VEN, and VZU; and in other countries: BM, BR, CM, CORD, F, G, GH, HBG, K, LINN, MA, MO, NY, P, US, W, and WIS. Many species were grown in the St. Louis greenhouse over several years for day and night examination of flowering behavior and as a source of plant parts to examine in the laboratory. Specimens were exam- ined under stereoscopic microscopes in Maracay, St. Louis, and other places. Nectar sugar ratios were obtained by C. E. Freeman (pers. comm.) us- ing the protocol outlined in Freeman et al. (1984). TAXONOMIC TREATMENT Cestreae (“Cestrineae”) G. Don, Gen. Hist. 4(2): 400. 1838. TYPE: Cestrum L. Unarmed shrubs, trees, or vines; pubescence of simple, branched, or stellate, sometimes glandular hairs. Leaves simple, entire, minor leaves some- times present. Inflorescences variously branched panicles or racemes, often with bracts, the flowers mostly subtended by a small bracteole. Flowers mostly 5-merous, calyx cupular or tubular, lobed less than halfway; corolla tubular or funnelform with 5 short, usually spreading lobes; stamens in- serted in the corolla tube, anthers opening length- wise; ovary superior, often basally differentiated into a nectariferous disk and sometimes short-stip- itate; locules 2, stigma positioned near the anthers, ovules 4-20(-37). Fruit a juicy berry or an apically dehiscent capsule. Seeds of various shape, some- times winged; embryo straight. In Venezuela 2 gen- era: Cestrum and Sessea KEY TO GENERA (CESTRUM AND SESSEA OF TRIBE CESTREAE IN VENEZUELA la. Fruit a berry; seeds unwinged ________ Cestrum lb. Fruit a capsule; seeds winged Ss Sessea Cestrum L., Sp. Pl. 191. 1753; Gen. Pl., ed. 5. 88. 1754. TYPE: Cestrum nocturnum L. Parqui Adans., Fam. Pl. 2: 219. 1763. Parquis Raf., e ri ig 56. 1838. D Alkekengi Feuillée (— Ces- етте m.). e “Schltdl. Чү ар 8: 251. 1833, not Меуета ees, in Wall. (1832), Acanthaceae. MR amnus Endl., Gen. Pl. 667. 1839. TYPE: Meyenia fascicu- scd bod dl. (= Сангин 4. (Schltdl.) Loera Ral, Оља bi od 56. 1838. TYPE: Lomeria pur- . (7 Cestrum longiflorum Ruiz & Pav.). Waden Ral. Sylva Tellur. 56. 1838. TYPE: Wadea latifolia Raf. (= Cestrum latifolium Lam Unarmed shrubs, trees, or rarely vines; pubes- cence of simple, branched, or stellate hairs, some- times glandular. Leaves simple, entire, pinnately nerved, mostly glabrate above; mostly short-petio- 278 Annals of the Missouri Botanical Garden late; minor leaves present or not. Inflorescences ax- mostly situated together at the corolla mouth; ovary Шагу and pseudoterminal, few- or many-flowered ^ mostly shorter than the calyx, 2-locular, ovules (1-) racemes, spikes or cymes, often large and appear- 4-16(-32), style slender, mostly puberulent or pa- ing paniculate; bracts often present. Flowers diur- pillose near the apex, style capitate or variously nal or nocturnal, mostly fragrant, mostly 5-merous, lobed, small, sometimes exserted beyond the an- pedicellate, bracteolate; calyx small, cupular or tu- thers or corolla mouth. Fruit blackish, purplish, bular, mostly shallowly lobed; corolla narrowly tu- — red, or white, an ovoid, ellipsoidal or obovoid, juicy bular, much exceeding the calyx, lobed, the lobes ог fleshy berry; fruiting calyx sometimes accrescent narrow, shorter than the tube, spreading or reflexed — and splitting; seeds varying in number, variable in when open; stamens inserted in the corolla tube at shape, even in the same berry, embryo straight. similar levels, the insertion levels varying greatly References to the last revision of the entire genus in different species, the adnate portion mostly ev- by Francey (1935-1936) are found in brackets at ident from the corolla base, the insertion variously the end of descriptions in the following accounts of pubescent, tumid, or denticulate, anthers small, the Venezuelan species. Кеу TO SPECIES OF CESTRUM AND SESSEA IN VENEZUELA la. l — ). Mature leaves pubescent beneath. 2a. Leaf undersides sparingly pubescent or glabrate with minute simple hairs; calyx mostly glabrous out- side. За. Leaves subcoriaceous and rigid; corolla tube stout, > 1 mm wide from near the uds flowers apparently open during the day 0000000000000 0 Cestrum > 3b. Leaves membranous, flexuous; corolla slender . € ] mm wide to more than halfway n flow c losed during most of the day 4a. Flowers subumbellate on leafy (bracteate) shoots, few per inflorescence, > 20 mm long; calyx = 2.5 mm long; corolla > 25 mm long; fruit mostly = 10 mm long .... Cestrum alterni- olium 4b. Flowers in short, leafless racemes or spikes, = 20 mm long; calyx < 2.5 mm 20р, corolla < 22 mm long; fruit mostly = 10 mm long 5 Cestrum latifolium 2b. А ping copiously pubescent n simple or branched hairs; calyx pubesc ent ea Flowers > 25 mm long; corolla pubescent outside; calyx > б mm long; minor leaves wan pas: 6a. Ca lyx teeth cuspidate/apiculate, > 3 mm long; corolla 2.5 mm across at the pos filam unexpanded and straight at insertion area _______ rum strigilotum 6b. кс teeth blunt, < 2 mm long; corolla 3— 3.5 mm across at the mouth; filaments tumi and geniculate at insertion areà | 0000000 estrum olivaceum 5b. Flowers = 25 mm long; corolla glabrous outside; calyx < 6 mm long; minor leaves present. Leaves drying dark with floccose to ам ы ak sh pubescence; leaves n stly with > 12 veins on each side; calyx lobes < 1.5 mm lon rum ооа - ib. Leav es drying green or brown with evenly distributed vellowish or brownish pesen leaves сае ith < 12 veins on each side; calyx lobes > 1.5 mm long Mature leaves 4. beneath. Ва. Leaves with = 10 = of lateral nerves. Оа. Calyx > 7 mm long; minor leaves conspic uous 9b. Calyx < 7 mm long; minor leaves wanting eaves with the veins drying salient beneath, half or more as wide as e аш of mature leaves mostly > 12 mm long; inflorescences held erect or nodding; tre la. Map 'ences terminal; peduncles sturdy; calyx > 4 mm long; corolla mouth >2m wide; fruit a dehiscent capsule; plants found above 2100 m lb. 1 ‘ences mostly axillary along leafy stems; peduncles slender; calyx < 3. long, corolla mouth < 2 mm wide; pu a Juicy етту; plants found below 2100 m. 12a. Inflorescences spicate: filament insertion geniculate and tumid; leaf bases mostly Се RES estrum. tomentosum UM o Cestrum petiolare — ~ ы & — uw. % > € % 5 ~ о Bags 38 з 9-3 S S с — - UNBATE CUu —————— Á——— strum cuneifolium 12b. Dno ‘ences racemose: filaments straight and unthickened; leaf bases mostly rounded. 13a. dud 13-18 mm long; fruit globose: petioles often drying dark, especially at the Cestrum racemosum 13b. С re 28.31 mm long; fruit ellipsoidal: petiole bases mostly drying light- colored. 14а. Leaves with veins ascending at < 70°; stamens pubescent just below the insertion: corolla mouth 2.5 mm across |. Cestrum acuminatissimum 14b. Leaves with veins strongly ascending а! > 70°; stamens glabrous near C the insertion; corolla mouth 3.5 mm across _____ estrum schulzianum Volume 85, Number 2 Benítez & D'Arcy 1998 279 Cestrum and Sessea in Venezuela 10Ь. амы with veins drying — plane beneath; leaves broad or narrow, diminishing in size d the би те often much narrower than half their ne РА ng; inflorescences arching and dangling; slender, willowy 15a. ris кі с 2. along the eu a sions stalk); Peur 1- flowe ves w with > 15 lateral veins on each side ____________- estrum olus 15b. Flower sessile (bracteole pere beneath pus peduncle 2-3-flowered; leaves 181 rrow or broad with < 15 lateral veins on each side __________ estrum bigibbosum 8b. ee with = 10 pairs of ps nerv НЕМ calyx = 35 m — > 20 mm long. 8a. Minor leaves persistent, often conspicuous; leaves < 4 cm wide Tx эш leaves wanting on mature growth; leaves wide or narrow. . Flowers subtended by bracts or bracteoles > 6 mm long (> 3 mm wide 20a. Bracts not folded, not enclosing flowers; ee ce di stalked сіреді оп а terminal, elongate central axis; climbing shrub ________- Cestrum reflexum 20b. A bract or bracteole longitudinally folded As half enclosing the basal part of the flower; inflorescences stalked clusters, terminal or axillary without a central axis; erect shrubs. 21а. Enveloping bract half enclosing the corolla tube and the fruit, per- sistent; leaf bases rounded or truncate _______ estrum jaramillanum 21b. Enveloping bract hardly extending along the corolla tube, caducous; C mostly < 1 m long; is portion of filaments < 5 mm. . Cestrum mariquitense f bases cuneate estrum BORNE 19b. Flowers lacking conspicuous bracts (mostly with caducous bracteoles < 6 m 22a. Leaves < 2 cm wide, much longer than wide; inflorescences few-flowere ascicles estrum neblinense 22b. Leaves mostly > 2 cm wide, if narrower, then < 3 times longer than wide; inflorescences few- or о racemes. 23a. Climbing vines; flowers > 28 mm long |... Cestrum scandens 23b. Erect or sprawling Кы ч ог edi? ^ ps « 27 mm long. 24a. Leaves mostly « 4.5 cm wi mbranous; Sn ortion of filaments > 2.5 mm, t the: inscio toothed; fruit whites corolla lobes < 3 mm long; widespread and cultivated species __ Cestrum nocturnu 24b. Leaves mostly > 5 cm wide, often coriaceous; free portion of filaments < 2 mm, the insertion smooth (untoothed); fruit pur- ple or green?; corolla lobes > 3 mm long; local endemic spe- cies 254: Filaments 1-2 mm free, glabrate; calyx lobes < 1 mm long; corolla mostly > 23 m long; p endemic Ces- m glabrescens 25b. Filaments 4—5 mm free, pubescent; dic 1 obes 1-2 mm ng; corolla mostly < 23 m long; Cordillera de la endemic —. Cestrum potolifolium 17b. dp « 20 mm lon lon 26b. “э mostly closed during the day, mouth < 1.5 mm across, the lobes spreading ог reflexed at night, straight when expanded, pointed, greenish white or purple mm long. ng. Ecce not closing, mouth 3—3.5 mm across, the lobes recurved, rounded, bright white, Ce strum diurnum 27a. Peduncles > 1 ст long; leaves membranous, mostly drying plane, the mi nor venation unobtrusive, d ing similar colored to other venation beneath; petioles rying as the stems (sometimes darker inp at p base) е estrum uA V шш 27b. Peduncles < 1 cm long; leaves coriaceo en drying wrinkled, the эе E ain veins conspicuous and salient 1 minor venation Sii d ish le neath; petioles drying (б. Cestrum eon us 16Ь. 2. calyx = 3.5 mm long; free portion of filaments mostly > 5 mm. rolla « 25 mm long, the lobes < 5 mm lon 29a. Leaves < 3 cm wide; corolla lobes = 3 mm Я 30a. ei mostly < 1 ст wide; petioles = 3 mm long; calyx tube ‚айаш {тее n of filaments 5-9 mm C icd жанаш 30b. ied mostly > 1 ст wide; petioles > 3 mm long; calyx tube ае portion of filaments 4-5 mm 29b. "b mostly > 3 ст wide; corolla lobes mostly > 3 mm long. . Calyx > 5 mm long; inflorescences terminal panicles. 32a. Corolla > 22 mm long, the mouth > 3.5 mm across; filaments glabrous, fre estrum Дени 280 Annals of the Missouri Botanical Garden "s m genic ‘ulate-tumid, lac нын teeth; leaves mostly < 6 т < 3 cm wide 'estrum ruizteranianum 32b. Cin Ша < 22 mm long, the mouth « 3 mm ac ross; filaments pubescent, e insertion T bent or 51. ер ulate; leaves mostly > 6 mm long and > З cm wide _____________. Cestrum lindenii 3lb. Calyx < 5 mm long; inflorescences mostly arranged along the stems, subum- bellate, racemose, or spicate. a orescences at or near the end of elongate, slender, wandlike, often ue clining branches that bear diminishing leaves or terminal, few-fl pedunculate or subsessile clusters; filaments о Cestrum bigibbosum 33b. Inflorescences mostly distributed along branches with leaves not conspicuously diminishing upward and terminal on stiff branc hes filaments glabrous 34a. Leaves 5 large, often > 4 cm wide; free portion P. fila- ments 1-2 mm; flowers < 23 mm long Cestrum glabrescens 34b. 2. coriaceous, mostly < 3 ст wide; free portion of filaments 5- ; flowers > 22 mm lon Hia a ascending, mo PE widest at or below the middle; free pu n Шел glabrous, 6–8 mm; calyx often drying darker ear the apex, the nerves not darker, the teeth ca. 0.5 mm long A NET Cestrum imbricatum 35b. Leaves spreading, mostly widest at or · above the middle; free portion of filaments pilose, ca. 5.5 mm; calyx drying even ily fro base to apex, the teeth ca. 1 mm p Cestrum cuneifolium 28b. Corolla > 25 mm long, uv lobes = 4 mm long. 36a. Vines or scramblers; free portion of filaments 0,5—1 mm, glabrou 37a. Flowers lac lane pedicels (the subtending bracteole nl on the axis); corolla lobes < б mm long; plants drying dark brown or gray; twigs and podus 'ences puberulent; petioles tomentose __________ strum reflexum 37b. Flowers on short, distinct pedic ‘els (below the subtending brac ‘teole); 2. lobes > 6 mm long; plants drying 2 or yellowish; twigs and inflorescences ostly -. petioles glabrate |... estrum MT 36b. Erect ва filaments free for various distances, glabrous or pubescent. 38a. Corolla xb DPA or orange, the mouth > 5 mm across; calyx 5 mm long; fruit white; petioles > 2 ст long: curd cultivated exotic species Cestrum au- rantiacum 38b. Corolla dps uous, green, yellow, or whitish purple, the mouth < 5 mm across; calyx > 5 mm long; fruit purple-black; petioles < 1 ст long; native species. 39a. Minor leaves abundant, often ida AAA stamens barbate at insertion NN Cestrum mariquitense 39b. 2” leaves wanting; stamens glabrous. or pubesc ent but not at insertion 5 mm long; flowers and fruits half-enfolded lengthwise in a persistent leafy bract; filaments glabrous; leaves mos aA < 3 mm wide а ———— —— rum пеи 40b. Calyx < 5 mm long; lacking persistent bracts; ы. mostly bescent. 4la. Calyx glabrous; leaves mostly > 3 mm wide. 42a. Corolla < 26 mm long: Појина species > 250€ БСН Ces- trum ruizteranianum 42b. Corolla > 26 mm long; Guayanan species < 500 n 43a. Inflorescence axes slender, glabrate Dum acumi- natissimum 43b. Inflorescence axes stout, tomentose Cestrum tubulos- um 41b. Calyx puberulent. 44a. Corolla > с. mm long; corolla lobes > 5 mm long; Gua апап species < 500 m estrum en 44b. Corolla < 26 mm 1 corolla lobes < 5 mm long; central Venezuela, > 1000m |... Cestrum orm - ~ 1. Cestrum acuminatissimum Dunal, in A. DC.. . Loreto: La Victoria on the Amazon river, Wil- Prodr. 13(1): 627. 1852. TYPE: French Guiana liams 3129 (holotype, F). “circa Cayennam," Leprieur s.n. (holotype, G-DC Shrub 1.5-3 m tall, not seen, = IDC microfiche, = F photo 6907). Cestrum loretense Francey, Candollea 6: 225. 1935. TYPE: sometimes sprawling, branches terete, striate, nodes of leaves and inflo- rescences thickened; pubescence of simple, monil- Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 281 iform ascending and crumpled hairs. Leaves nar- rowly ovate or elliptical, 8-16(-18) apically short-acuminate, acute or obtuse and then orming a narrow apicule, the base narrowly cu- neate, truncate or subtruncate, the margins slightly revolute, papery-coriaceous, dark green above, light green beneath, glabrous on both sides but sometimes with minute trichomes on the veins be- neath, the veins impressed above, prominent beneath, ascending, 6-12 on each side; petiole 3-5(-10) mm long, the base somewhat thickened; minor leaves wanting. Inflorescences short axillary racemes, axes 5-10 mm long, 1.5-2 mm thick, tomentose with ascending, curved hairs; bracts 1-2.5 mm long, fo- liaceous, subulate, yellowish or reddish pubescent. Flowers apparently nocturnal, 28-32 mm long; ped- icels 0.5 mm or obsolete; bracteoles narrow, tomen- tose; calyx tubular, 3.5-5.5 X 1.5-3 mm, the tube 5-5 mm long, glabrous, the teeth triangular- acute, 0.5 mm long, ciliate and tufted, the costas and lateral veins salient, the 5 costas thickened upward; corolla pale greenish yellow, 28-31 mm long, the tube narrow, very gradually expanded up- ward, slightly contracted at the throat, mouth 2-2.5 mm wide, the lobes narrowly acuminate, 4—7.5 m long, sometimes sparingly pubescent, ciliolate; sta- mens 20.5-24.5 mm long, filaments adnate for 18- 22 mm, pilose to 1 mm or 8-15 mm above the base, insertion straight, smooth, free 0.5-3 mm, anthers 0.5—0.8 mm across; ovary ellipsoidal, 1— .9 X 1 mm, glabrous, disk conspicuous, ovules 10, style 20-24.5 mm long, filiform, puberulent just below the stigma, the stigma capitate, included. Fruit dark purple, ellipsoidal, 10-13(-16) x 6-9 mm, with a thin pericarp; fruiting calyx hardly ac- crescent; seeds 3—5, dark brown, 4—7 mm long. [Francey 6: 314.] Figure 3. 7 cm, orbicular, Cestrum acuminatissimum is distinguished by a number of features that tend to overlap with other species, notably C. megalophyllum. The longer flowers, pubescent stamen insertion, and matte ap- pearance of the upper sides of the revolute leaves are good recognition features. The corolla is quite variable in length but is usually more than 28 mm long with the lobes 4—7.5 mm long. The inflores- cence axes are notably slender. Distribution (Fig. 4). Amazonas, Apure, Bari- nas, and Táchira. Gallery forests along riverbanks; 100 to 300 m. Also in French Guiana, Colombia, Peru, and Brazil. Phenology. Collected in flower between No- vember and March, and in fruit mostly in March and May. Common names and uses. Веесоіб (Pumé lan- guage), Derello, Mecla. Mixed with other herbs, plants are used for stunning fish (Gragson & Grag- son 48, Representative specimens seen. VENEZUELA. Ama- zonas: Maroa—Yavita road, 5.5 km from Maroa port, Clark 6948 (MO, NY); 20 km $ of confluence of Río Negro and Brazo Casiquiare, Liesner 6877 (MO, VEN); 4 km 5 de Solano, Morillo & Hasegawa 5026 (MY, VEN). Apure: Distrito Pedro Camejo, main road S of Paso de San Pablo to Río Cinaruco, Davidse & González 15984 (MO, VEN); Reserva Forestal San Camilo, quebrada de El Dique, SW de San Camilo (El Nula), Steyermark et al. 101530 (US, VEN). Barinas: 5 km SW of dam site on Río Caparo, 31 km ESE of Santa Bárbara, Liesner & González 9311 (MO, VEN), 9322 (MO, VEN). Táchira: E of La Fundación around Represa Dorada, Liesner & González 10412 (MO, VEN); Montaña Guafitas, NW of El Piñal, Steyermark et al. 119545 (MO, VEN) 2. Cestrum alternifolium (Jacq.) O. E. Schulz, in Urb., Symb. Antill. 6: 270. 1909. Ixora al- ternifolia Jacq., Enum. Syst. Pl. 12. 1760. YPE: Cultivated Europe, seed from Marti- nique, Jacquin s.n. (lectotype, here designated, W not seen, = F photo 33021). Chiococca alternifolia L., Syst. Nat. ed. 12, 2: 165. 1767. Based on Chiococca scandens sarmentis tenuissimis ivist. at. Hist. Jamaica Cestrum confertum Miller, Gard. Dict. ed. 8, Cestrum no. : not HEE Cestrum bs sie) Pl. 2: 206. 1771. TYPE: ae E ‘LINN 258.2 not seen, = IDC 1; ). ene po Jacq., Pl. Hort. Schoenbr. 3: 42, pl. 327. Caracas, plate 327 in Jac- 1798 (lectotype, pss designated). бый .. Dunal, A. DC., Prodr. 13(1): 662. 1852. TYPE: Puls n Кобан! (holo- type, G-DC not seen, — IDC microfiche, — F photo 6896) Cestrum м мит Dunal, іп A. DC., Prodr. 13(1): TYPE: Guadeloupe, Bertero s.n. (holo- ram not seen, = IDC microfiche, = F photo 34066). Cestrum 22. var. mithanthum 0. E. Schulz, іп b., jill. 6: 273. 1909. TYPE: Venezuela. Nueva о Margarita ad El Valle, J. А. Johnston 285 (holotype, GH) Shrub 3-4 m tall, much branched, the branching irregular, often wand-like or pendulous, twigs green, minutely pubescent, glabrescent, the tips of- ten sharp, mature stems often whitish or yellowish, lenticellate; pubescence of simple hairs. Leaves malodorous, ovate, sometimes elliptical or narrowly elliptical, (1.5-)3.7-12.5 X 1.5-4.5 cm, attenuate toward the apex, the apex itself obtuse, rounded 282 Annals of the Missouri Botanical Garden Figure 3. Cestrum acuminatissimum.—A opened to show sta ov ДӘ», HT Vp} < | д | lowering and fruiting branches.—B. Flower.—C. Calyx.—D. Corolla ly | stamens and style. А, upper section, B, С after Davidse & González 21967 (VEN). А, lower section after Steyermark 101530 (VEN). toward the base, often narrowly cuneate, not revo- lute, membranous or firmly papery, both sides pale green, lamina and veins puberulent on both sides, sometimes glabrescent, veins 5—10 on each side, ascending; petiole 5-11 mm long, pilose; minor leaves sometimes present, 14—17 X 5-10 mm, their petioles 3.5 mm long. Inflorescences mass-blooming, terminal and axillary, few(7—10)-flowered, congest- ed fascicles or umbels mostly near the ends of leafy branches, axes 5—10 cm long, slender, tomentose; peduncles 1-2 mm, thickened, unbranched. Flow- ers crepuscular and nocturnal, fragrant, 26-33 mm Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 283 Cestrum acuminatissimum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. Figure 4. long, pedicels obsolete; bracteoles linear, 1.5-3 mm long, glabrate; calyx cupular, 2.5-3 mm long, basally narrowed and stipe-like, pilose or glabrous outside, the tube 2-4 mm long, teeth 0.5-1 mm long, costate, ciliolate and tufted; corolla white or yellowish green, sometimes with slight purplish col- oration on the tube, the tube contracted just above the ovary, then slightly and gradually expanded up- ward, abruptly expanded just below the apex and around the anthers, slightly contracted at the mouth, 24-32 mm long, glabrous, mouth 3-4 mm wide, lobes narrowly triangular, 4—6 mm long, cil- iolate, the folds puberulent; stamens equal, 20—22 mm long, filaments adnate for 18-23 mm, insertion straight, smooth, glabrous, free part 0.5-2 mm, ап- thers orbicular, 0.5 mm across, included; ovary l- 1.5 mm across, glabrous, disk conspicuous, ovules 6-10, style 18-22 mm long, the apical 2 mm pu- berulent, stigma capitate, included in the corolla. ruit maturing through violet to purple-black, shiny, ellipsoidal, 10-12 х 5—8 mm, the pulp 1.5 mm thick, spongy, whitish; fruiting calyx hardly ac- crescent; seeds 4-8, dark brown, 4.5-5.5 mm long. [Francey 6: 211.] Figures 2C, 5. Cestrum alternifolium may be recognized by its umbel-like inflorescences with sessile flowers and the slender corollas that are expanded into a small bulb around the anthers. This species resembles C. mariquitense, which tends to be a larger, leafier tree of more mesic regions. The absence of pubescence on the filaments distinguishes it from C. mariqui- tense. Although Jacquin cited Plumier (Pl. Amer. 7: 150, plate 157, fig. 1, 17. 1758), suggesting that the name /хоға alternifolia was based on the plate of Plumier, neither the plate nor one-line descrip- tion with it provide the details found in Jacquin's own description. Hence we select as lectotype the specimen at W from the garden at Schoenbrunn where Jacquin worked. We think that this specimen was grown from Jacquin's own collections in Mar- tinique, where he said the plant was found, and was probably the material from which the plate was made. Cestrum confertum Miller was placed in syn- onymy by Francey (1935: 211). Plants of this species are often associated with ants. The flowers often begin to open and emit fra- grance from about an hour before sunset to an hour after sunrise. Nectar is scant or absent in this spe- cies. Pollination by hawkmoths (Sphingidae) was reported by Haber and Frankie (1989) and White et al. (1994). Distribution (Fig. 6). Aragua, Bolívar, Falcón, Guárico, Lara, Mérida, Miranda, Nueva Esparta, Portuguesa, Sucre, Táchira, Trujillo, Zulia, and the Distrito Federal, as well as on the Island of Testigo Grande. Dry coastal scrub, savannas, riverbanks, deciduous and semideciduous forests below 1300 m and in cloud forests around 2000 m. Also oc- curring from Mexico to northern South America and in the Antilles. Phenology. The flowering season seems to de- pend on the climatic patterns of the particular veg- etation formations where the species occurs, thus varying from place to place. Common names. | Clavito, Dama de Noche, Fru- to de Culebra, Puta de Noche, Putica de Noche, Tapacamino, Tinte, Uvito Gallinero. Representative specimens seen. VENEZUELA. Ara- gua: Colonia Tovar, Allart 475 (NY, US, VEN). Bolívar: escarpment E of Miamo leading to Hato Nuria, Altiplan- icie de Nuria, Steyermark 88525 (NY, VEN). Falcón: Cerro Santa Ana, van der Werff & Wingfield 3107 (MY). Guárico: Estación Biológica de Los Llanos, Ramirez 234 (MY, NY), 2148 (MY, NY). Lara: Barquisimeto, Saer 15 (US, VEN). Mérida: Carretera Ejido-Las González, Ruiz- 284 Annals of the Missouri Botanical Garden “руу 122 675% LC A Y GENE RTP | | Figure 5. Cestrum alternifolium.—A. Branch with flowers and fruit. —B. Flower.—C. Style and stigma.—D. Corolla opened to show stamens. After Benítez 1809 (MY). Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 285 Figure 6. Cestrum alternifolium.—A. Distribution in Venezuela.—B. Representative occurrences outside of enezuela. Terán & R. Gallardo 12641 (MERF, MY). Miranda: hills between Carenero and Chirimena, Steyermark & Bunting 102318 (VEN). Nueva Esparta: El Espinal-La Guardia, NW de Margarita, Benítez 1609 (MY). Portuguesa: El Guasare, C 1ano, 123025 (MO, VEN, VZU); Репја, Tejera 133 (US). Dis- trito Federal: Jardín Botánico, Caracas, Nee 17. VEN, WIS). Archipiélago Los Testigos: Testigo Grande, Fernández et al. 200 (CAR, MY, PORT). . U. Skinner, Chimalapa, Guate- mala, not located. Cestrum pedunculare Dunal, in A. DC., Prodr. 13(1): 618. 852. TY Mexico. Pavón s.n. (holotype, С, = К photo 34132). Cestrum chaculanum Loes., Verh. Bot. Vereins Prov. Bran- denburg 65: 97. 1923. Cestrum aurantiacum var. chaculanum (Loes.) Francey, Candollea 6: 104. 1935. TYPE: Guatemala. Seler 2836 (holotype, B de- stroyed, — F photo 2972). Cestrum paucinervium Francey, Candollea 6: 101. 1935. TYPE: Guatemala. Quiché: San Miguel Uspantán, Heyde & Lux 3135 (holotype, B destroyed). Shrub or tree 1—3 m tall, branched, older trunks gnarled, conspicuously lenticellate, stems terete, flexible, soon glabrate, pubescence of reduced, glandular, perhaps branched, crumpled hairs. Leaves sometimes malodorous, narrowly ovate, 7.5-10 X 4.5—6.8 cm, apically acuminate, basally attenuate, margins undulating, sometimes appear- ing ciliolate, subcoriaceous to membranous, matte bright green, glabrous, veins 7-9 on each side, strongly ascending, slightly sunken above, main veins elevated beneath, minor veins plane, retic- ulate, drying conspicuous; petiole canaliculate, 2.3-3 cm long, glabrous; minor leaves wanting. Inflorescences showy, axillary and terminal, mostly emergent from the leaves near the branch ends, lax racemes 1.5-3 cm long, axes 2.5-6 cm long; bracts occasional, foliaceous, 5-10 X 15 mm, pu- bescent; peduncles mostly short, occasionally to 3 cm long, slightly longer in fruit, tomentose, clus- ters of 2—6 sessile or subsessile flowers separated by a 2-6-mm-long rachis. Flowers diurnal, un- scented, 25-29 mm long, buds with calyx teeth bent out, sessile in groups of 2—3; bracteoles lin- ear, 5-8 mm long, glabrate; calyx tubular, 5.5-7 X 2-3 mm, glabrous outside, the veins conspic- uous, pilose inside, tube ca. 5 mm long, teeth nar- rowly deltoid or subulate, 0.5-1 mm long, the tips subulate, often as long as the tube; corolla bright yellow to orange, 26-28 mm long, glabrous, tube 21-23 mm long, expanding gradually upward, mouth (2-)3-4.5(-5.5) mm wide, lobes 5 mm long, narrowly ovate, apically mucronulate; stamens 17-19 mm long, filaments adnate for 13-15 mm, pubescent to 10-12.5 mm from the base, insertion straight, smooth, tumid, geniculate, slightly den- ticulate, glabrous or sparingly pilose, free part 4— 5 mm; ovary globose 1-1.5 mm across, glabrous, disk inconspicuous, ovules 12-13, style 18-19 mm long, glabrous, stigma capitate, slightly ex- serted. Fruits numerous per inflorescence, white, ovoid, 8.5-10 x 5-7.5 mm, juicy; fruiting calyx often accrescent, 6-9 mm long, sometimes split- ting at the sinuses; seeds 5—9 per fruit, bright dark brown, 3.5—6 mm long. [Francey 6: 102.] Figure 7. This species of Cestrum is easily recognized in Venezuela by its showy orange or yellow flowers and white fruits. This is the only species of section Ha- brothamnus recorded from Venezuela. The flowers of Cestrum aurantiacum are unscent- 286 Annals of the Missouri Botanical Garden re 7. Cestrum aurantiacum.—A. Flowering branch.—B. Corollas with lobes open and closed.— C. Corolla opened to show stamens and style. After Ernst s.n. (HBG). ed and open night and day. The corolla lobes are Steudel (1840) listed the name Cestrum auran- strongly reflexed or recurved when open. tiacum but did not provide a description. Chromosome numbers for this species have been reported as n — 8, 2n — 16 (Dyer, 1963; Sharma Distribution. Occasionally cultivated in tropical & Sharma, 1958; Madhavadian, 1968; Berg & gardens for its showy flowers, this species is native Greilhuber, 1993b). to Nicaragua and Guatemala. Although not stated Volume 85, Number 2 1998 Benítez & D'Arcy 287 Cestrum and Sessea in Venezuela on the label, the sole Venezuelan collection was undoubtedly from a cultivated plant. Specimen seen. VENEZUELA. Lara: Barquisimeto, montium Coro, Ernst s.n. (HBG). 4. Cestrum bigibbosum Pittier, J. Wash. Acad. Sci. 22: 35. 1932. TYPE: Venezuela. Between El Aguacatal and Alto del Cogollal, 1500 m, Pittier 9245 (holotype, VEN; isotype, US). Cestrum laetum F зы Candollea 6: 378. 1935. TYPE: enezuela. Ara olonia Tovar, 1854, Fendler 955 (holotype, NY: ‘isotypes, GH, MO). Cestrum pumilum Francey, Candollea 6: 373. 1935. E: Colombia. Santander: Eastern Cordillera, vi- nity of Las Vegas, 2600-3000 m, Killip € Smith 15962 (holotype, NY). Cestrum umbrosum Francey, Candollea 6: 375. 1935. TYPE: Venezuela. Moritz 348 (holotype, W not seen, = F photo 33043). Cestrum venezuelense ан Candollea 6: 377. 1935. TYPE: Venezuela. Mérida: Moritz 212b (neotype, here designated, “Жан Weak, sparingly branched shrub or small tree 2— 6 m tall, sometimes a solitary unbranched wand- like stem, 1-3 cm DBH, branches arching or erect, striate, young branches pubescent, young parts and inflorescences with a faint dark purple color; pu- bescence of simple, moniliform ascending an crumpled hairs. Leaves elliptical or ovate, some- times narrow, (7-)13-26 X (2-)5-10 cm, apex acute or acuminate, base rounded or attenuate, sometimes unequal or slightly arching downward, margin revolute, membranous to subcoriaceous, dark green and shiny, lighter beneath, glabrous, veins 5-14 on each side, ascending at an angle of 50-65”, somewhat irregular, major veins slightly impressed above, elevated beneath; petiole canalic- ulate, 1-3 cm long, glabrous, often curving and twisting depending on orientation of the branch; minor leaves wanting. Inflorescences large, pendu- lous, terminal or axillary panicles, sometimes short axillary racemes; axes dark purple, 10-25 cm long with 1—10 branches, peduncles 8-25 mm long, 0.3—0.7 mm thick, bracts foliaceous, ovate, becom- ing narrower upward, glabrous, 6—40 X mm wide. Flowers nocturnal?, 25-30 mm long; pedicels 0.5-1 mm; bracteoles linear, 1.5-3 mm long, glan- dular pilose; calyx cupular, basally narrowed into an indistinct stipe, 3.5-7 X 2 mm, firmly membra- nous, often 5-costate, glabrous, tube 2.5-6 mm long, teeth 1 mm long, ciliate, tufted; corolla green- ish white, yellowish, or dark olive-green, the lobes pale yellow, 23-29 mm long, narrowly funnelform, the tube gradually expanded upward, mouth 2.5-3 X 1-2 mm, 5 lobes narrowly triangular or oblong, 4—8.5 mm long, apically acute or obtuse; stamens 17-23 mm long, filaments adnate for 13-20 mm, bigibbous, insertion 2 mm long and free, sometimes tumid and sparingly pilose, free part 3-3.6 mm, anthers orbicular, 0.5—0.8 mm across, the surface conspicuously rameniferous; ovary ellipsoidal, 1 X 0.8 mm, with minute papillae near the top, stipitate, the disk yellow, conspicuous, style 15-22 mm long, pilose for 3 mm below the stigma, exserted 1.5 mm, stigma subcapitate. Fruit often in dense, pendulous or sometimes arcuate-ascending racemes, dark pur- ple and shiny, subglobose or ovoid; fruiting calyx hardly accrescent, wrinkled, broadly cupular, in- side with 2-4 glandular bands that extend to the base of the teeth, the teeth with thickened margins; seeds 4—11 per fruit, yellow, 4.5—6.5 mm long. [Francey 6: 376.] Figures 1B, 8. This species typically occurs as a slender treelet with reclining branch ends and leaves that diminish in size near the branch apices. However, many ex- amples, perhaps damaged plants, are short shrubs with one or two stiff, inflorescence-bearing branch- es and larger than normal leaves. In this species, the flowers are inconspicuous and the fruits usually few. The staminal insertion usually bears one or two tooth-like emergences but is sometimes smooth Some plants greatly resemble C. salicifolium, hav- ing narrow leaves and dangling inflorescences, but the flowers are sessile on an often caducous sub- tending bract and not pedicellate as in C. salicifol- um. The collection here designated as the neotype of C. venezuelense, Moritz 212b (BM), was not seen by Francey, and we have not seen another collection of this number among material from G-DC, HAL, or HBG, where Francey borrowed material for his revision. However, this specimen agrees with Francey's description and is a close match for spec- imens of Moritz 212 (HBG), one of which was an- notated by Francey as C. venezuelense but was not cited in his protologue. Moritz 212 (B — F photo 2985) was a different species (C. scandens Vahl (as C. laxiflorum Dunal)), and is different in HBG (C. venezuelense) and G-DC (C. scandens Vahl (as C. laxiflorum Dunal)). Until an example of Moritz 212b that was annotated by Francey is located, we con- sider the specimen at BM to be acceptable as a neotype. Francey also cited a specimen of Karsten from Colonia Tovar, which we have not located. Distribution (Fig. 9). Aragua, Barinas, Carabo- bo, Cojedes, Falcón, Lara, Mérida, Miranda, Mon- agas, Portuguesa, Sucre, Táchira, Trujillo, Yaracuy, and the Distrito Federal. Evergreen cloud forests of low stature (9-11 m) in shade or in natural wood- lands; 1200—2200 m. Also in Colombia. ~. 288 Annals of the Missouri Botanical Garden ss EL 19 AN о а 5 ст р р о 4 тт у " C ааа] ) Hat | А an м 4 О 2 mm | Figure 8. Cestrum bigibbosum.—A. style.—D. Stamen. After Badillo 6629 ( Flowering branch.—B. Closed flower.—C. Flower opened to show stamens and MY). Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 289 A igure 9. Cestrum bigibbosum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela Phenology. Maximum flowering is in January, diminishing to May. Fruiting is mainly from March to August. Common names. Mata Perro, Uvito. Representative specimens seen. VENEZUELA. Ara- gua: Carr aracay—Choronf, después de las Moro- odas Benítez & Rojas 4998 (MY). Barinas: road from Altamira to Santo Domingo, van der Werff & Ortiz 5859 (MO, VEN). Carabobo: without other locality, Funck & Schlim 627 (G). Cojedes: Cerro Azul, fila La Blanquera, Delascio 4118 (CAR). Falcón: El Chorro entre La Chapa Uría, Benítez et al. 5149 (MY). Lara: Arriba de Sanare, Badillo 6693 (MY). Mérida: Carretera El Celoso—Las Ме- sas, Benítez & Rojas 4811 (MY). Miranda: Los Guaya- bitos, Baruta, Aristeguieta 2291 (VEN). Monagas: slopes of Cerro Negro above La Sabana de Las Piedras, NW of Caripe, Steyermark 61844 (F, VEN). Portuguesa: 21 km S de Biscucuy, El Rodeo de Santa Lucía, Benítez et al. 4287 (MY). Sucre: foothills of Cerro Turumiquire, SW of Guariglia 11658 (VEN). Trujillo: between La Playa, SW of Carache and Potreritos de Cendé, Dorr et al. 5099 (VEN). Yaracuy: Carretera Nirgua-La Chapa, Benítez et al. 5097 (MY). Distrito Federal: Carretera Caracas—Co- lonia Tovar, Meier 1211 (MY, VEN). 5. Cestrum buxifolium Kunth, in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 57. 1818. TYPE: Regni Novo-Granatensis, near sanctuary of Montserrat, 1650 hex, Humboldt & Bonpland s.n. (holotype, P-Bonpl., = IDC microfiche). Cestrum la eco Willd., in Roem. & Schult., Syst. Veg. 4: 808. 1819. TYPE: Colombia. Gulleator unknown 2. B-W 4460 not seen, Te microfiche). Cestrum melanochloranthum Dunal, DC., Prodr. 13(1): 622. 1852. SYNTYPES: Colombia. Santander Killip & Smith 15691 (NY), Killip & Smith 18215 (NY), Holton 571 (G? not seen).Venezuela. Mérida: around Portachuelo, Funck & Schlim 1264 (G-DC not seen, = IDC microfiche, = F photo 6911; MO, P Cestrum parvifolium var. venezuelense Francey, Candollea 6: 330. 1935. TYPE: Venezuela. Mérida: Chachopi- to, near San Rafael, Pittier 13210 (holotype, VEN; isotypes, B prar Se F, MO, US). Cestrum cuneatu y, Candollea 6: 326. 1935. SYN- PES: Colombie Linden (Funck & Schlim) 1645 (B not seen, G n n, = photo F 28356, BR not seen, MO, P); ЕН “Troll 3 3587 (6? not seen); Nor- te de Santander, Ir & Smith 1 2 (МҮ). Vene- zuela. Mérida: chopo, Linden 363 (G not seen). Ecuador. aime; André 897 das Shrub to 3 m tall, often dwarfed and 40—60 cm tall or prostate, irregularly branched, the branches often ascending, young branches pubescent, young shoots terete, often dark purple, mature branches dark yellowish and tomentose; pubescence of branched trichomes. Leaves ascending, imbricate and facing the stems, elliptical, 2-3(-5) X 1.1(-1.6) cm, apex obtuse, base attenuate or cu- neate, margin slightly revolute, coriaceous to char- taceous, both sides bright green and shiny, glabrous, veins 4—10 on each side, above deeply furrowed, only the principal veins impressed be- neath; petiole dark purple, 1.5-3 mm long, spar- ingly puberulent; minor leaves 4—5 X 1.5—4 mm, sessile. Inflorescences dense, axillary and terminal, axes and pedicels dark purple, axes 3.5 cm long, peduncles 1 cm long, bracts foliaceous, 1.5-5 X 0.5-1.5 mm, pilose. Flowers not closing, fragrance crepuscular and nocturnal, 16—25 mm long; sessile, calyx cupular, 3. X 2 mm, coriaceous, 5-cos- tate, glabrous, the tube 3-5 mm long, the 5 teeth triangular, 0.5-1.5 mm long, ciliolate, the sutures pilose; corolla 15-24 mm long, tube dark purple to greenish yellow, ampliate, slightly contracted above the ovary, nearly cylindrical to about halfway up. then expanded and again cylindrical, glabrous, mouth 3—4 (2.5—3) mm across, the lobes triangular, dark purple, 2-5 mm long, the folds yellowish green, pilose, reflexed at anthesis; stamens 11-19 290 Annals of the Missouri Botanical Garden mm long, filaments green, adnate for 4.5-7.5 mm, insertion geniculate-tumid, glabrous or sparingly pubescent, free part 5-9 mm, anthers dark brown, orbicular, 0.5 mm across; ovary lobed, 1-1.5 mm long, disk inconspicuous, ovules 7-13, style pur- plish, 13.5-19.5 mm long, slightly sunken in the ovary apex, puberulent 1.5 mm below the stigma, stigma capitate, bright green 0.5—1 mm across, ex- serted 0.5-1 mm from the anthers. Fruit in umbel- late clusters, dark purple to black, narrowly ovoid, 0.9-1.2 X 0.9-1.1 cm, smooth, the mesocarp dark purple, pulpy, staining fingers and paper deep pur- ple; seeds 2-8 per fruit, light brown, 5-8 mm long. Figures 2B, 10. Cestrum buxifolium often occurs on paramos as a dwarf shrub, flowering and fruiting when less than 60 cm tall with irregular branching and small leaves. In more sheltered places, it may be a shrub to 3 m tall, and such specimens are separated by the characters noted in the key. Distribution (Fig. 11). Andean regions of Lara, Mérida, Táchira, and Trujillo. Rocky exposed slopes, dry pastures, and around watercourses in dwarf cloud forests and paramos; 2650—4000 m Also in Colombia and Venezuela. Phenology. Flowering is throughout the year, more plentiful from October to May. The fruiting maximum is in April and May. Common names and uses. Chupa Sol, Chon- galé, Chungagué, Chungalé, Fruta Negra, fiu- ngagué, Uvito. The pulp of the fruit is used as writ- ing ink (Benítez de Rojas et al. 4674, MY; M. López del Pozo 477, 848, 868, MYF) Representative specimens seen. VENEZUELA. Lara: Without other locality, Burandt & Garófalo V0596 (MY). Mérida: Laguna Negra, SE of Laguna de Mucubají, Bar- clay & Juajibioy 9754 (MO); La Culata, Río Mucujün, to 15 km NE of Mérida, D'Arcy et al. 18236 (MO, MY); Sierra de La Culata, Ruiz-Terán 6911 (MERF). Táchira: Páramo de Tamá, frontera Colombo- Venezolana, Steyer- mark et al. 98775 (MY, US, VEN). Trujillo: Páramos de El Jabón-El Turmal, 15 km al E de Carache, Ruiz-Terán ipez- Figueiras 949 (MERF). P Cestrum cuneifolium Francey, Candollea 7: 60. 1936. TYPE: Colombia. Santander: East- ern Cordillera, eastern slope of Páramo del Hatico, from Toledo to Pamplona, 2900 m, Kil- lip & Smith 20590 (lectotype, here designated, NY). Shrub or tree 2.5-7 m tall, 3.5 cm DBH, branched in the upper half, branches terete; pu- bescence of reduced simple, sometimes gland- tipped and perhaps branched hairs. Leaves solitary, narrowly obovate, 6-12 X 2—4 cm, apically short- acuminate or obtuse, basally cuneate and fine-de- current on the petiole, margins revolute, coria- ceous, mostly drying reddish brown, glabrous, veins 7—9 on each side, parallel, arcuate, branching and anastomosing to form a looping, partial submarginal vein, veins elevated beneath; petiole 5-10 mm long, glabrous; minor leaves wanting. Inflorescences short, axillary racemes of 5—8 flowers; peduncles 5 mm long. Flowers nocturnal?, 18-22 mm long, ped- icels to 1.5 mm long, bracteoles foliaceous, 1.5-2.5 mm long, puberulent, caducous; calyx tubular, 4—5 X 2-2.5 mm, the veins inconspicuous, tube 3—5 mm long, glabrous outside, with minute hairs on the upper half within, teeth ca. 1 mm long, mi- nutely ciliolate and tufted; corolla purplish and yel- lowish, often drying yellowish with darker tips, 16— 19 mm long, tube 13-16 mm long, slightly con- tracted above the ovary, expanding abruptly on emerging from the calyx, expanding gradually up- ward and appearing sub-cylindrical or clavate, mouth 3—4 (2.5-3) mm wide, lobes narrowly tri- angular, 2.5-5 mm long, ciliate, the folds tomen- tose; stamens 10.5—14.5 mm long, filaments adnate or 5.5—8 mm, pilose to 2-2.5 mm from the base, insertion geniculate, tumid, slightly pilose, free part 6—8 mm; ovary 1-2 mm across, glabrous, disk con- spicuous, 0.5 mm long, ovules 3—5, style 13-14 mm long, papillose 2.5 mm below the stigma, stig- ma 0.5 mm long, bilobed, exserted 1 mm. Fruits subglobose, 6—8.5 X 6.7 mm wide; fruiting calyx almost unchanged; seeds 1 per fruit, 6—6.5 X 4.5— 6.5 mm wide. [Francey 7: 28.] Figure 12. This species is similar to Cestrum imbricatum Rusby, differing conspicuously in its smaller, crowded leaves. The calyx and corolla in C. imbri- catum are often purple rather than green as in C. cuneifolium. Distribution (Fig. 13). Western Venezuela in the Andean states of Mérida and Táchira. In cloud forest; 2200-2900 m. Also in eastern Colombia. Phenology. Collected in flower in March and November. аети kar imens seen. VENEZUELA. Méri- . Bernardi et al. 13056 (NY), 17200 da: La (MO, on San Eusebio. Meier et al. 531 (MY, Pipe Tachira: Pico de Vela and Buena Vista Charpin & Jac- quemoud 13146 (MO, NY); Delicias—Villa Paez, Morillo & Garcta 11385 (MERF, MY). 7. Cestrum diurnum L., Sp. Pl. 191. 1753. TYPE: Cuba. Hortus СШ нәни (lectotype, here designated, LINN 258.4). Benítez & D'Arcy 291 Volume 85, Number 2 1998 Cestrum and Sessea in Venezuela A D О 2 ст a i ga x О 4 b, C EUR MER ONU "n А Mala м. О 2 77 d 1 j mm Figure 10. Cestrum buxifolium.—A. Flowering branch.—B. Flower.—C. Flower opened to show stamens and pis- til.—D. Stamen. After Tamayo 4367 (VEN). 292 Annals of the Missouri Botanical Garden Figure 11. Cestrum buxifolium.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela dd ч Jacq., Pl. Hort. Schoenbr. 3: 44, pl. 8. Cestrum diurnum var. fastigiatum (Jacq.) 1 іп Fournet, Fl. Illustr. Guad. Mart. 12 1978. TYPE: provenance unknown, plate 330 i in ren quin, 1798 (lectotype, here aces 1). Cestrum odontospermum Jacq., rt. Schoenbr. 3: 44, 1. 1798. Cestrum bo var. 2. Card? O. E. Schulz, in Urb., Symb. Anull. 6 E: provenance unknown, plate 331 in js quin, 11798 (lectotype, here э» d). Cestrum tinctorium Jacq., Pl. . Schoenbr. 3: 45 pl. 332. 8. Cestrum uS var. y tinctorium (Jacq. . Gómez, Anales Hist. Nat. 23: 269. 1894. TYPE: from Caracas, plate 332 in Jacquin, 1798 (lectotype, here designated). Cestrum venenatum Mill., Gard. Dict. 16 Apr. 1768, non C. venenatum Burm. f. (1 y m var. venenatum ., Symb. Апі. 6: 263. 1909. TYPE: Jamaica. Houston s.n. (BM? not seen). Cestrum vespertinum Lunan, Hort. Jamaic. 2: 78. 1814 non L. (1771 Cestrum laurifolium Fawc., Jamaic. Bull. 11: 7. 1889, non Cestrum laurifolium UHerit. (1788). TYPE: Jamaica. Fawcett 660 (not seen). кый ed. 8, Cestrum no. f. (1 Mar.- Shrub or tree 2-6 m high, crown conical, trunk erect, branches light green, terete; pubescence of erect and crinkly, white, simple hairs, glabrate ex- cept on calyces and corolla apices. Leaves ovate or elliptical, 4.5-7.5 X 1.5-4.5 cm, apically obtuse or acute, basally narrowly acute to rounded, slightly asymmetrical, gradually attenuate above the mid- dle, firmly membranous or subcoriaceous, bright shiny green above, lighter beneath, glabrous on both sides, veins 6-7 on each side, main veins el- evated above, evident and whitish beneath; petiole terete, 5-10 mm long; minor leaves mostly rotund, 10-12 X 4—7 mm, sessile, caducous. Inflorescences axillary and terminal, pedunculate congested spi- cate cymes, racemes, or umbels, peduncles green or tan, elongate, to 9 cm long, bracts foliaceous, 3 X 1.5 mm, pubescent with branched or simple hairs. Flowers open and fragrant day and night, 12— 17 mm long, often 6-merous, pedicels obsolete; ca- lyx light green, cupular, 2.5—3.5 X 1.5-2 mm, cos- tate, the tube 2-3 mm long, glabrous, teeth less than 0.5 mm long, ciliolate; corolla funnelform, ob- conical, 11-16 mm long, the tube evenly expanded toward the top, mouth 3-3.5 mm wide, the 5-6 lobes rounded, 1.5 mm long, recurved at anthesis, the folds puberulent; stamens 8-11 mm long, fila- ments adnate for 7.5-10 mm, pilose 2.5-3 mm from the base, insertion straight, smooth, 0.5—1 mm free, anthers cordiform, 0.5-0.8 mm long; ovary oblong, ] mm long, seated in a conspicuous disk, style white, 8.5-10 mm long, pilose 2 mm below the stig- ma, stigma dark green, capitate, slightly exserted. Fruits purple-black, subglobose, 11-12 X 8 mm; seeds 13-14, dark brown, 2.5-3.5 mm long. [Fran- cey 6: 284.] Figure 14. Cestrum diurnum is distinctive in its bright white flowers with 5—7 strongly recurved corolla lobes, and in its often slightly bluish foliage. The plate (332 in Jacquin, 1798) that typifies Cestrum tinctorium Jacq. was prepared from a plant cultivated from seeds from Caracas, Venezuela. See D'Arcy (1970) for a discussion of Jacquin's career and typification of his names. Cestrum vespertinum Lunan is based on Jasminum laurinis folüs..., Sloane Voy. Jamaica 2: 96, pl. 204, f. 2. 1725, and on F 2. follis oblongo-ovatis..., Browne Civ. ist. . Jamaica 178. 1756. This name was 5. іп пе synonymy of this species by Егапсеу (1935: 284). Cestrum laurifolium Fawc. was placed in synonymy by Francey (1935: 284). iurnum var. venenatum (Mill.) O. E. Schulz was placed in synonymy by Francey (1935: Chromosome numbers of this species have been reported as n = 8 (СШ, 1972), the normal comple- Cestrum Volume 85, Number 2 Benítez & D'Arcy 1998 Cestrum and Sessea in Venezuela “у AS AS 557 SANA Figure 12. Cestrum cuneifolium.—A. Branch with flowers and fruit.—B. Flower.—C. Corolla opened to show sta- mens and style. After Morillo & Garcia 11385 (MY). 294 Annals of the Missouri Botanical Garden Figure 13. Cestrum cuneifolium. Distribution in Ven- uela. ment in the genus. There is also a curious, older report of п = 15, 16 (Sharma & Sharma, 1957, 1958). Distribution (Fig. 15). Cultivated for ornament and sparingly naturalized in Venezuela. Aragua, Carabobo, Falcón, Mérida, Miranda, Sucre, Táchi- ra, and the Distrito Federal. In sun and shade; to 1500 m. Also in Florida and the Antilles and in- frequent on Caribbean coasts of Mexico. Perhaps native to the Greater Antilles. Phenology. Flowering in irregular spurts 3-5 times a year. Flowers stay open night and day. Common names and toxicity. Miel. Other com- mon names referring to night-blooming probably re- fer to Cestrum nocturnum. The species has been implicated in deaths of horses (Krook et al., 1975). Representative specimens seen. VENEZUELA. Ara- a: Maracay, Universidad Central de Venezuela, Benítez 1556 (MY). Carabobo: Canoabo, cerca de la Universidad Francisco de Miranda, Benítez et al. 5162 (MY). Falcón: Coro, Plaza Manaure, Wingfield 5116 (CORO, MY). Mé- rida: Plaza Bolívar de Ejido, Ruiz-Terán & S. López-Pa- lacios 6113 (MERF, MY). Miranda: Petare, Elías 239 (CAR). Sucre: Cultivated, Eulogio Mago, Valley of Co- collar, Steyermark 62434 (MY, VEN). Táchira: Plaza de Lobatera, Rutz-Terán 3606 (MER). Distrito Federal: Jar- dines de Caracas, Lasser 3533 (MY, VEN). 8. Cestrum glabrescens (C. V. Morton) Steyerm. & Maguire, Mem. New York Bot. Gard. 17(1): 463. 1967. Cestrum tenuiflorum var. glabres- cens C. V. Morton, Bull. Torrey Bot. Club 58: 466. 1931. TYPE: Venezuela. Amazonas: Agúita, Mount Duida, Tate 885 (holotype, US). Shrubs 1.5—4 m tall, adult stems glabrous, young branches puberulent; pubescence of simple an perhaps branched, moniliform ascending and crum- pled hairs, most parts soon glabrescent. Leaves el- liptical or ovate, sometimes narrow, (7-)9-16(-21) X 3—6(—8) cm, apex acute, acuminate or abruptly acuminate, base rounded or short-cuneate, margins plane or slightly revolute, firmly membranous, pa- pery or subcoriaceous, dark green above, lighter beneath, glabrous, veins 6-13 on each side, as- cending, arcuate or bending toward the tips, im- pressed above, salient beneath; petiole 0.5—1.8 cm long, glabrous; minor leaves wanting; bracts want- ing or caducous. Inflorescences axillary spikes or racemes, sometimes forming paniculate masses, 2.5-3 cm long, axes glabrescent. Flowers noctur- nal?, 23-28 mm long, pedicels 0.5-2.5 mm long, tomentose; bracteoles linear, 3-6 mm long, tomentose; calyx cupular, 3-5 1-1.5 mm, costate, pubescent outside, glabrous within, the tube mm long, the 5-6 teeth unequal, del- toid, 0.5-1 mm long, ciliolate; corolla yellowish green with purplish marks on the tube and lobes, 22-27 mm long, the tube gradually expanded up- ward, mouth 2 mm wide, lobes narrowly ovate, 3— 4.5 mm long; stamens 14—19.5 mm long, filaments glabrate, adnate for 13-18 mm, insertion straight, smooth, free part 1-2 mm, the anthers orbicular, 0.3—0.5 mm across; ovary lobed, 1.5-1.8 mm across, glabrous, ovules 10-13, style 15-20 mm long, glabrous, papillose below the stigma, the stig- ma bilobed. Fruit purple, subglobose, 8-11 х 5-8 mm, the pericarp thick; seeds 3-4, olive-colored, 6—8 mm long (Steyermark 93274). Figure 16 This species resembles Cestrum latifolium, but the leaves are glabrous and the flowers are longer. When dry, the membranous leaves of C. glabrescens are often shiny with yellowish veins. The inflores- cences are usually reduced on relatively thick stems, but sometimes they are enlarged into race- mose or spicate subterminal clusters. Distribution (Fig. 17). Tepuis (inselbergs) of Amazonas and Bolívar. Associated with low-grow- ing plants along riverine woods and hills in cloud- forest (2); 870—1900 m. Apparently endemic. Phenology. Collected in flower from October to June, and in fruit in May and June. Representative specimens seen. VENEZUELA. Ama- zonas: Dept. Atabapo, slope of cerro Marahuaca, Liesner (NY); Caño Mojado, Chimantá-Massif, Torono—tepui, Wur- dack & Steyermark 1083 (F, NY, VEN). 9. Cestrum humboldtii Francey, Candollea 6: 393. 1935. TYPE: Peru. Pampayacu, hacienda at mouth of Chincao Río, 1050 m, Macbride 5129 (holotype, F). Volume 85, Number 2 Benítez & D'Arcy 295 1998 Cestrum and Sessea in Venezuela N | GT { с ӘС eile hos 45 S У ЖЕ ғы Ду. N N p 2 = пар; далы наным... E ЧИ eoa IN ну r к 4% Мед бын eres P бру S i PEN ПО УРАНА Т ОЛДУ a Ud eA din lio А E Ahorn trum diurnum.—A. Branch with flowers and fruit. —B. Flower.—C. Corolla opened to show stamens Figure 14. Ces and style. After Cawz 12 (MY). 296 Annals of the Missouri Botanical Garden 15. Cestrum diurnum.—A. Distribution іп Venezuela.—B. Representative occurrences outside of Venezuela. = cou var. calycinum Francey, Candollea 6: TYPE: Peru. Muna, trail to Tambo de 5. ft., Macbride 4332 (holotype, | Cesrum КОШ var. tenuiflorum Francey, Candollea 6: 935. TYPE: Colombia. 7 Monte de Fu- айкыш Triana s.n. (holotype, G-DC, = IDC mi- crofiche, — F photo 28363) | Shrub or tree 4-13(-20) m tall, branches terete, densely gray pubescent, the dark color of the stems visible beneath them; pubescence of branched, stellate and simple hairs. Leaves ovate or elliptical, -27(-45) X 4.5-11 cm, apically attenuate, base cuneate, occasionally long-cuneate, sometimes somewhat oblique, firmly membranous to coria- ceous, sometimes rugose, above drying dark brown, beneath dark green, veins 9-16 on each side, most- ly parallel, ascending, and moderately arcuate, the major veins impressed and all veins pubescent, the major veins black, reticulate veined beneath; peti- ole canaliculate, purple, 1.5—5.5 cm long, puberu- lent; minor leaves present on young branches, 3 X .5 cm, subsessile. Inflorescences axillary panicles, axes basally branched, 2.5—5(-7) cm long, some- times lax and arching out from the stems, pedun- cles 2-4 mm long, bracts foliaceous, 18 X 5 mm; bracteoles linear, 6-8 mm long. Flowers fragrant, 16-21 mm long; pedicels 0.5-1.5 mm long; calyx mostly tubular, 3,5-6 X 2-2.5 mm, costate, sub- coriaceous, sometimes densely tomentose outside, especially near the base, lanose-pubescent within, the tube 2.54.5 mm long, the teeth 1-1.5 mm long, ciliate, the tips pilose; corolla whitish, yellow- ish, or light purple, 14-20 mm long, the tube con- tracted above the ovary, slightly expanded upward, contracted below the limb and appearing clavate, glabrous outside, mouth 2-2.5 (2.5-3.5) mm wide, the lobes 2-4 mm long, ciliolate, the folds puber- ulent; stamens 11—14 mm long, filaments adnate for mm, the insertion 1-2 mm long, geniculate- tumid, pilose with dispersed hairs, free part 3.5—5 mm, anthers rounded, 0.5 mm across, dark brown; ovary ellipsoid, 0.5-0.7 mm across, glabrous, disk conspicuous, style 10.5—14.5 mm long, puberulent 2 mm below the stigma, stigma capitate, exserted 1 mm. Fruit dark purple to almost black, globose or elliptical, 4–6 mm across, slightly stipitate, peri- carp slightly thickened, the calyx slightly accres- cent and glabrescent; seeds 4—6, dark brown to al- most black, 2-5 mm long, embryo white, epicotyl 0.5 X 0.5 mm, hypocotyl 1 X 0.5 mm. [Francey 6: 393.] Figure 18. This species is recognizable by its large stature, large leaves, and uneven, often floccose whitish pu- bescence. Except for the pubescence, the leaves mostly dry dark. Most of the collections from Ven- ezuela have glabrate calyces. Cestrum humboldtii is apparently related to C. tomentosum. While plants of the two species seem amply distinct in Venezue- la, some collections from Ecuador suggest a close relationship and are easily confused. Distribution (Fig. 19). Mérida, Táchira, and Trujillo. Cloud forests; 1000—2500 m. Also in the Andes of Colombia, Ecuador, and Pera. Phenology. The species has been collected in flower from January to August and in fruit from August to December. Common name. Uvito. Representative specimens seen. VENEZUELA. Méri- da: Puente de la Quebrada del Plan hasta Los Magos. Municipio Aricagua, Bernardi 6194 (MY, NY, P). Táchi- ra: Parque Tamá, zona de Buena Vista, Morillo & García 11475 (MERF, MY). Trujillo: Vía La Morita, Benítez et al. 3783 (MY). Volume 85, Number 2 1998 Benítez & D'Arc Cestrum and Sessea in Venezuela ke > 297 ТУ » ФУ > Jet Dm Y $9 A SS 22. O N "| STA EN 2 СУДА mee 7 М, Y Á Y V т, V (AS \ W KZ а! қ ON fs LA \ Y v [^ N МА + ДАР» “Әле қа O Л Quo СА 4; с Му \ — Же / LEN (IEE қ TAR 77,2 CH У А0 % [ N Ў [с ете ONZAS d UAI АРД уд Ao Figure 16. Cestrum glabrescens.—A. Branch with f mens and style. After Лее 30689 (VEN). lowers and fruit. —B. Flower.—C. Corolla opened to show sta- 298 Annals of the Missouri Botanical Garden Figure 17. Cestrum glabrescens. Distribution in Ven- ezuela. 10. Cestrum imbricatum Rusby, Descr. 5. Amer. Pl. 119. 1920. SYNTYPES: Colombia. Santa Marta: rare on the extreme top of the San Lorenzo ridge, 7000 ft., H. H. Smith 1896 (CM, NY). Shrub 2—3 m tall, branches terete, densely leafy, glabrous; pubescence of reduced simple hairs. Leaves narrowly ovate, 5-8 X 2-2.5 cm, apically short-acuminate, basally narrowed and decurrent on the pedicel, margin subrevolute, coriaceous and rigid, dark green above, dull green beneath, gla- brous, lateral veins 6—8 on each side, main veins elevated beneath, petiole drying dark brown, 5-10 mm long; minor leaves wanting. Inflorescences ax- illary racemes 1-2 cm long, 8-10-flowered, pedun- cles 4—5 mm long, glabrate. Flowers 12-21 mm long, sessile, bracteoles drying almost black, linear, 3.5 mm long, sparingly pilose; calyx tubular, 3-4.5 X 1.5-2.5 mm, glabrous, tube 2.5-4 mm long, teeth deltoid, 0.5 mm long, ciliolate and tufted; co- rolla purple, 10-19 mm long, tube 13 mm long, slightly compressed at the ovary, expanding grad- ually, mouth 2.5-3.5 mm wide, lobes 2.5 mm long, folds pilose; stamens 13-14 mm long, filaments gla- brous, adnate for 7.5 mm, insertion geniculate, tu- mid, slightly granular, free part 5.5 mm; ovary lobed, 1 mm across, disk conspicuous, ovules 8, style 11-15 mm long, stigma lobed, 1 mm long. Fruit not seen. [Francey 6: 381.] Figure 20. This species is similar to and may include Ces- trum cuneifolium Francey, from which it differs mainly in its conspicuously smaller and more uni- form leaves. Because specimens from Venezuela lack fully developed flowers, much of the above description was made from the type collection, which is from Colombia. Distribution (Fig. 21). Táchira. 2350—2500 m. Also in Colombia at 2100 m Phenology. Collected in flower in May (Colom- bia) and November. Additional specimens seen. VENEZUELA. Táchira: Tierra Negra, cabeceras del Río Quinimarí, Steyermark ra, 101047 (MO, US, VEN). 11. Cestrum jaramillanum Benítez & D'Arcy, Novon 5: 311. 1995; Phytologia 81: 382. 1996. TYPE: Ecuador. Pichincha: Centenela, Mon- tañas de Па, 12 km E of Patricia Pilar, virgin rain forest, 550—650 m, 79?19'W, 0?34'S, Løjtnant & Molau 15835 (holotype, AAU; iso- type, GB). Shrub or small tree 2-3 m tall, branched, the branches slender, at first compressed, sometimes drying grooved, later terete, tomentulose, the inter- nodes 3—4.5 cm long; pubescence of reduced, sim- ple, crinkled, ascending hairs. Leaves ovate, 10—14 7.5 cm, apically short-acuminate, basally rounded, truncate, or subcordate, the margin plane, membranous or chartaceous, glabrous above on emerging, the basal half beneath tomentulose on emerging, glabrescent, veins 7—8 on each side, as- cending, distally arcuate or looping and forming a submarginal vein 0.5-1 cm from the margin, veins inconspicuous above, the costa and major veins slightly reddish, somewhat elevated and finely pu- berulent beneath; petiole 5-10 mm long, finely pu- bescent; minor leaves ovate, 7 X 1.5 mm, sessile, glabrate. Inflorescences axillary; peduncle dark, 1— 15 mm long; bracts foliaceous, 30 X 10 mm wide. Flowers nocturnal?, 25—34 mm long, sessile; brac- teoles foliaceous, narrowly lanceolate, 10 X 3. mm, acuminate, glabrate or pubescent on both sides, persistent; calyx tubular, 4 X 2.5 mm, drying stramineous, basally rugulose, glabrate, tube 3-3.5 mm long, the teeth deltoid to broadly acuminate, 1.5 mm long, faintly 5-costate; corolla pale green, 29-30 mm, the tube 26 mm long, gradually ex- panded upward, the mouth not contracted, 2.5-3 mm wide, 5-lobed, the lobes 6—6.5 Х 1 mm, ob- long, the apex acute, the pleated margin short-pi- lose; stamens 22-23 mm long, the filaments adnate for 21 mm, insertion straight, smooth, glabrous, free part 1 mm, anthers orbicular, 1 mm across; ovary 0.7 mm across, slightly rugose, glabrous, ovules 5— 6, style 23 mm long, stigma slightly bilobed, in- cluded. Fruit ovoid, 8-10 X 6—7 mm, the pericarp thin, about equaling and loosely enclosed in the bract; fruiting peduncle 15 mm long; fruiting brac- teole partially enveloping the fruit laterally; seeds 3-4, dark brown, 5-6.5 mm long. Figure 22. Volume 85, Number 2 Benítez & D'Arcy 299 1998 Cestrum and Sessea in Venezuela A X о у S à © У. QE e, = 4 A Ko „карме, ————À — — = à Es cilisis | Figure 18. Cestrum humboldti.—A. Flowering branch.—B. Fruiting branch.—C. Flower.—D. Corolla opened to show stamens.—E. Style.—EF. Trichomes. After Benítez 4870 (MY). This species is distinct in its folded bracts, which — ezuelan report is based is in fruit and is identified half envelop the flower and fruit. The inflorescences — with this species with some hesitation. 4 congested at the enda > pokeci tiak appear ta Distribution (Fig. 23). Distrito Federal; 1000- elongate in age, leaving one or two fruits at the apex. ; 1300 m. Also in Ecuador. These mature peduncles are usually shorter than the neighboring leaves. The specimen on which the Ven- Phenology. Flowering and fruiting in November. Annals of the Missouri Botanical Garden 19. Cestrum humboldtii.—A. Distribution "igure n Venezuela.—B. Representative occurrences outside of Venezuela Specimens seen. VENEZUELA. Distrito 2. ral: Cerro Naiguatá, rag het neds 92147 (MY, NY, P, US, VEN): Fila Las Delicias, arriba de Naiguatá, 1000 m ps Bun- ting & Manara 2092 (MY). 12. Cestrum latifolium Lam., Tabl. Encycl. 2: 5. 1794. TYPE: [Trinidad?] (holotype, P-LA). е Мека 22. a Humb., Bonpl. & Kunth, 818. b latifolium var. tenuiflorum duni) 0. Y Schulz, in Я 27 . vu te Duidae, near district Orinoco, Humboldt & Bonpland 1017 (Р, = F photo Cestrum floribundum Willd. ex Roem. & Schult., Syst. e 9. TYPE: Brazil. Hoffmannsegg s.n. (holotype, B- Ww 4. not seen, = IDC microfiche, = F photo 3019). Cestrum ovatum. Willd. ex Roem. & Schult., Syst. Veg. 4: 07. 1819. TYPE: Venezuela. Near Esmeraldas, n (holotype, B-W 4439 not seen, = IDC mi- e). Саа. onm Sieber ex Sendt., in Mart., Fl. Bras. 10: 210. 1846, non C. hirtum Sw. (1788). TYPE: Mar- tinique. Sieber 81 (MO). Cestrum poeppigii Sendtn., in Mart., Fl. Bras. 10: 210. 1846. TYPE: Brazil. Poeppig 2979 (lectotype, here designated, HAL, F E 6905, G-DC not seen, — IDC — F photo 33037). Cestrum 5 [albo-punctatum| Dunal, іп А. Prodr. 13(1): 635. 1852. TYPE: Brazil (erro- еды 1. to Peru). Poeppig 2979 (holotype, G-DC not seen, = IDC microfiche; isotypes, HAL, G not seen, = F photo 6905, W not seen, = F shoto 33037). Cestrum 1. Dunal, in A. DC., Prodr. 13(1): 6 1852. T inidad. Sieber 143 (holotype, G- Dc isolec totypes, G not seen, = C microfiche, W ‹ = IDC n nicrofiche; isotype, MO). Cestrum ve indi Dunal, in A. DC., Prodr. 13(1): 634. vim PYPE: French Guiana. жени 250 (holotype, G- DC not seen, = IDC microfiche, = F pho to 6906). Cestrum riales var. latifolium Dunal, in A. DC., Prodr. 13(1): 634. 1852. TYPE: French Guiana, col- lector not indicated [annotated by Franc ey] ( (holo- type, G-DC not seen, = IDC microfiche, = F photo 23176). Cestrum prieuret Dunal, A. DC., Prodr. 13(1): 635. 1852. М РЕ: edet Leprieur Г ІШІ, G-DC not seen, = IDC microfiche, = idis) Cestrum пе cus Beurl., мо, Vete Acad. . 40: 140. 1854. TYPE: Panama. кй, Billberg s.n. (holotype, 5, = MO phot Cestrum vespertinum Griseb., Fl. Brit. W. ЈЕ . 1862. Shrub or tree 1.5—7 m tall, 10-12 cm DBH, bark grayish, smooth, branches terete, elongate, some- times decumbent or lianoid, sprawling, young parts densely pubescent, glabrescent; pubescence of sim- ple hairs. Leaves malodorous, ovate or elliptical, cm, apically acuminate, the very tip acute or acuminate, round- ed at the base, membranous to firmly membranous, dark green above, light green beneath, both sides sometimes oblique, 5-11 X 2.5-6.5 pubescent, more so beneath and on the major veins, lateral veins 5—7 on each side, mostly strongly as- cending; petiole 0.8-1.5 cm long, pilose; minor leaves wanting. /nflorescence mass-blooming, axil- lary or terminal, of short racemes or short-branched panicles, many-flowered, peduncles 0.7-1.5 cm long, branched, pubescent. Flowers crepuscular or nocturnal, fragrant, nectar scant or wanting, 15-2 mm long, pedicels 0.5-0.8 mm long, pubescent, bracteoles arcuate, 1.5-2 mm long, puberulent; ca- lyx membranous, 2-2.5 X 1 mm, 5-costate, spar- ingly pilose outside, glabrous inside, the tube 1.5— 2 mm long, the teeth short, 0.5 mm long; corolla greenish white, 15.5-18 mm long, tube narrow, slightly contracted above the ovary then gradually expanded upward, mouth 1.5 mm wide, the lobes sometimes light purple, narrowly deltoid, 2-4 mm long, ciliate and the folds pilose; stamens equal, 11-14 mm long, filaments adnate for 9.5-12 mm, insertion straight, smooth, adnate, free part 2-2.5 mm, pilose 2-2.5 mm below the insertion and at Volume 85, Number 2 Benítez & D'Arcy 301 1998 Cestrum and Sessea in Venezuela = DKS 4 SA aW 62-52 = ae NA, | P = 52 SS ты - ы |84442 # Figure 20. Cestrum imbricatum.—A. Flowering branch.—B. Corolla opened to show stamens and style.—C. Flower. After Steyermark & Dunsterville 101047 (MY). 302 Annals of the Missouri Botanical Garden 3, Figure 21. Cestrum imbricatum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. the base, anthers suborbicular, 0.5 mm across; ova- ry 0.7-1 X 0.5-1 mm, glabrous, disk inconspicu- ous, ovules 5-7, style 13-13.5 mm long, about equaling the stamens, stigma subcapitate, included. Fruit several per inflorescence, obovoid, 6-10 X 4—8 mm, bluish purple to almost black, ain me- socarp white; fruiting calyx slightly accrescent, the limb and lobes spreading; seeds 2—6 per fruit, dark brown, 3—4.5 mm long (Aristeguieta et al. 7260); embryo white. [Francey 6: 289.] Figures 2A, 24. This species may be recognized by its sparingly pubescent leaf undersides and filament pubescence found well below the point of insertion. The unpublished name “Cestrum latifolium var. genuinum Stehlé" was applied by Fournet (1978: 82) in a sense of including C. chloranthum, C. hirtum, C. vespertinum, C. nocturnum, and perhaps C. latifolium in a single taxon. This may have been Stehlé's way of indicating a typical variety, or it may be a form taxon which has little meaning, as its name was not validly published in conformity with the ICBN (which has required clear designation of a type since 1958). Fournet employed this varietal concept for plants from the French Antilles. Ces- trum vespertinum Griseb. was placed in synonymy by Francey (1935: 289) Distribution (Fig. 25). All states except Lara, Miranda, and Nueva Esparta. Widely distributed in deciduous, semideciduous, riverine, and gallery forests, on savannas and secondary vegetation; 50— 1200 m. Also in Nicaragua, Costa Rica, Panama, the Antilles, Colombia, Guyana, Surinam, French Guiana, and Brazil. Phenology. Flowering and fruiting sporadically throughout the year, mainly from January to May, and most heavily in April. The species blooms noc- turnally in very fragrant masses for 1-3 days at a time. Common names and uses. Barriga de Sapo, Bello de Noche, Bonita de Noche, Cazabe, Ciruelil- la, Ciruelo de Monte, Clavito, Coral Blanco, Hoja privada, Huele de Noche, Ke Tipen (Panare lan- guage), Mortifio, Palo hediondo, Quasimillo, Rabo Pelado, Tepuru, Uvito. Used in popular medicine against mange (Ruiz-Terán 2878, MER, MERF). Representative specimens seen. VENEZUELA. Ama- curo: between Tucupita and Las Mulas, Steyermark et al. 114590 (MO, NY, VEN). Amazonas: Hío Siapa, Gutiérrez 225 (TFAV). Anzoátegui: between San Durrial and Mata Redonda, Davidse & González 19999 (MO, VEN). Apure: P. Nacional Santos Luzardo, Ruiz et al. 4501 (MY). Ara- gua: entre La Victoria y Colonia Tovar, Вепйег 563 (MY). Barinas: Río Zulia, Santa Bárbara de Barinas, Valverde & Peña 1061 (MER, MY). Bolívar: entre Villa Lola y Río Grande, Fernández 2671 (MY). Carabobo: Hacienda La Cumaca, Municipio Valencia, Benítez et al. 5159 (MY). m x entre Manrique y Tierra Caliente, Benítez 2169 MY). Falcón: g and cerro Montero, Agostini Ур 1171 EN). Guárico: Estación Biológica de Calabozo, Aristeguieta 5041 (VEN). Mérida: La F ida, SE de San a de Caparo, Aymard et al. 4496 MY). nagas: Río iun e 2 km SSW of Jusepín, Pursell et al. 8432 (VEN). Portuguesa: Fundo El Chap- arral, Río Portuguesa, Aymard & Cuello 5591 (MO, MY, PORT). Sucre: P. Nacional Península de Paria, Río Gran- de Arriba, Benítez et ix 5120 (MY). Táchira: Granja Na- ren, cerca de La Fría, Benítez de Rojas & Rojas 4756 (MY). Tryjillo: Cerro Gordo, Steyermark & Espinoza 111662 (F, NY, US, VEN). Yaracuy: Río Yurubí, Laber- пађб, Delascio & López 2555 (CAR, VEN). Zulia: Casi- dem sector Las Cruces, Bunting & Alfonzo 7294 (MO, EN). Distrito Federal: Cuenca del Río Macarao, Mon- tes 61 (УЕМ) 3: < 13. Cestrum lindenii Рипа], іп A. DC., Prodr. 13(1): 611. 1852. TYPE: Venezuela. Trujillo: 6000 hex, Linden (Funck & Schlim) 784 (ho- lotype, G-DC not seen, = IDC microfiche, = F photo 6912; isotype, BM). Volume 85, Number 2 Benítez & D'Arcy 303 1998 Cestrum and Sessea in Venezuela ES, Ss Set oe) UE S (+1013 Figure 22. Cestrum jaramillanum.—A. Fruiting branch.—B, C. Flowers.—D. Corolla opened to show stamens and style. After Løjtnant & Molau 1585 (AAU). 304 Annals of the Missouri Botanical Garden Figur ге 23. Distribution is Cestrum species. Distribu- tion in Venezuela of four species with restricted occur- rence. Solid square ae ie tillettii. Open square = Cisisum jarama, Circle — Cestrum pariense. Star — Cestrum ruizteranianum. Cestrum amplum Pittier, J. Wash. Acad. Sci. 22: 35. 1932. TYPE: Venezuela. Mérida: A. Jahn 1075 (holotype. VEN; isotype, US). Cestrum dubium Pittier, J. Wash. Acad. Sci. 22: 30. 1932, non Steud. (1843). Cestrum costanensis Steyerm., Acta Bot. Venez. 3: 212. 1968. TYPE: Venezuela. . Federal: Los Venados de Galipán. 1500— 1800 m, Е. Pittier 166 (holotype, VEN). Cestrum um var. grandifolium des еу, 381. 1935. TYPE: Venezuela. А var, Fendler 954 (holotype, NY). Shrub or tree 4-15 m tall, 12-14(-30) cm DBH, the crown little branched, the bark dark and nearly smooth, the branches terete-furrowed, glabrous twigs sometimes stout, often drying blackish; pu- bescence of simple hairs. Leaves strongly malodor- ous, narrowly elliptical, 6–13(– 16) х apically acute, obtuse or short-cuspidate, the base cuneate or acute, sometimes slightly decurrent on the petiole, articulated at the base, the margin slightly revolute and slightly folded, plane or un- dulate, thick-coriaceous or subcoriaceous, dark green and shiny above, light green beneath, gla- brous, veins 7-13 on each side, somewhat ascend- ing and parallel, elevated, major veins impressed above, elevated beneath; petiole dark purple, dry- ing black, canaliculate, 4-12(-15) mm long; minor leaves glabrous, 5-14 х 3-3.5 mm, subsessile. In- florescences lax terminal panicles, axes trigonal, dark purple, 5-8 mm long, glabrous. Flowers closed but fragrant at midday, 1 mm long, sessile; calyx dark purple, 4.5—7.5 X 2-3.5 mm, thick, cos- tate, tubular, the tube 3.5-6 mm long, 5-toothed, glabrous outside, the teeth 1–1.5 mm long, pilose within, ciliolate and the folds puberulent; corolla dark purple with bright yellow lobes, 17-22 mm Candollea 6: ragua: Colonia To- cm, long, tube contracted around the ovary then grad- ually expanded to the top, glabrous, mouth (2.5—)3—4 mm wide, the lobes inflexed, triangular, 3.5-5 mm long, apex acute; stamens 12-17 mm long, filament adnate for 5—10.5 insertion geniculate-tumid and denticulate, pilose, free part 5-7.5 mm, anthers orbicular, 0.5-1 mm across; ovary dark purple, obovoid or subglobose, 0.6-2.5 X style green, 9.5- 14.5 mm long, apically pilose, stigma bright green, capitate, 0.5 mm long. Fruit purple-black, ovoid or globose, 8-12 X 6–9 mm, the pulp purple; fruiting calyx cupular, dark purple; seeds 5—6 per fruit, ae brown, 6.5—7.5 mm long, drop-like. [Francey 7: 14.] Figure 26. mm, adnate portion pilose, the X 0.5-2 mm, glabrous, This upland species has clavate corollas with mouths mostly 3—4 mm wide, wider than most spe- cies in Venezuela and suggestive of pollination by birds. The combination of broad corollas and nu- merous veins on the leaves is diagnostic among the Venezuelan species of Cestrum. "Cestrum glabrum Klotzsch & Karsten" is an un- published manuscript name that identifies a Colom- bian specimen (Karsten 58) represented by the photo F 2979. The specimen represented is C. lin- denii. Distribution (Fig. 27). Aragua, Mérida, Miran- da, Táchira, and Trujillo. Dense cloud forests and in dwarf forests, 1600—3000 m. Also in Colombia. Phenology. Collected in flower and fruit from March to January, but not collected in February. Common names. Borrachero negro, Cafecito, Laurel, Tábano, Verdecito. Representative specimens examined. VENEZUELA. Aragua: Colonia 4. Tamayo et al. 2504 (VEN). Mé- rida: La Culata, D'Arcy & 4. 18258 (MO, МҮ). Mir- em La pica ae 1-1 Silla, Meier 3180 (MY, М). Táchira: Р. Nac ‘ional ve Páramos, Benítez & Ro- jas 24 (MY). binae Arriba de La Puerta, hacia el páramo Los Laureles, Bono 5891 (MY, VEN). Distrito F : ederal: төлі Aristeguieta 792 (VEN) 14. Cestrum mariquitense Kunth, in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 57. 1818. TYPE: Colombia: near Santa Ana and Mari- quita, 550 hex., Humboldt (holotype, P-Bonpl., — [DC microfiche). Cestrum bogotense Willd. ex Коет. & Schult., Syst. Veg. : 807. 1819. TYPE: Colombia. Bogotá, Humboldt (holotype, B- W 4454 not seen, = IDC microfiche). Cestrum bogotense var. latifolium Francey, Candollea 6: 211. 1935. Cestrum mariquitense var. е (Frances) Standl. & C. V. Morton, Field . Nat. Hist. Bot., Ser. 18: 1049. 1938. ~ TYPE: Costa ; Rie 'a. Volume 85, Number 2 Benítez & D'Arcy 305 1998 Cestrum and Sessea in Venezuela 5 N R enn e w 227 қар “ 7р WA Ys SUI E, Figure 24. Cestrum latifolium.—A. Flowering branch.—B. Flowers and fruits.—C. Flower.— D. Corolla opened to show stamens and style. After Benítez 92 (MY). San José: El General, 600 m, Pittier 10509 (holo- | of simple, moniliform, sometimes collapsing, some- type, BR not seen; isotype, times gland-tipped, ascending hairs. Leaves often Shrub 2—4 m tall, leafy, much branched, nodes crowded, membranous, elliptical or ovate, 3.5-9 х many-branched, stems striate, pilose; pubescence 1.5-4 cm, apically acute or acuminate, basally 306 Annals of the Missouri Botanical Garden w* 100° [^d Су 10° [ so* С ~ — ~ 10° e, ~ o° “г: “” EN B 4 n 20° w Су “a sw Figure 25. Cestrum latifolium.—A. Distribution in . Representative occurrences outside of acute, obtuse or acuminate, light green, glabrous, lateral veins 5-8, ascending, the main veins spar- ingly pilose beneath; petioles 3-7 mm long, pilose; minor leaves 10-20 X 5—6 mm. Inflorescences ter- minal ог axillary, few-flowered, peduncles unbranched, 10 mm long, pilose, hairs sometimes gland-tipped, bracteoles foliaceous, 3-5 mm long, puberulent. Flowers crepuscular or nocturnal, fra- grant, 29-35 mm long, sessile; calyx cupular, 3.3- 9.9 X 1.5-3 mm, inconspicuously costate, tube 3— 4 mm long, glabrous outside and in, the teeth 0.3— 1.5 mm long, triangular or slightly acuminate, cil- iolate, apically pubescent; corolla creamy white or yellow green, 28-35 mm long, tube narrow, grad- ually expanded upward, obconic-cylindrical, mouth 1.5-2.5 mm, the 5 lobes 4—7 mm long, ciliolate, the folds puberulent; stamens 25-28 mm long, fil- aments adnate for 20-26 mm, pilose basally and just below the insertion, insertion pilose-barbate, straight, slightly denticulate, free part 1. mm, glabrous, anthers orbicular, 1 mm across; ovary glo- Бозе, 0.5-1 mm across, glabrous, disk inconspic- uous, style 20-28 mm long, puberulent 1.5-3 mm below the stigma, stigma capitate, 0.5 mm across, exserted 0.5 mm. Fruit shiny dark purple, inside purplish white, ovoid, 10-14 mm long, mm wide with a small apical protuberance, the pulp 2 mm thick, not juicy; seeds 5—9, dark brown to al- most black, 4—6 mm long. Figure 28. Cestrum mariquitense is distinguished by its abundant, congested minor leaves and by the bar- bate stamen insertions. Additionally, herbarium specimens of this species may be recognized by the blackish leaves with whitish, irregular pubescence. Distribution (Fig. 29). Barinas, Lara, Mérida, Portuguesa, and Táchira. At edges of semideci- duous, riverine, and cloud forests; 400—1800 m. Also in Costa Rica and Colombi Phenology. Flowering is from March to Sep- tember with a peak in June. Jazmin de Monte, Rudo, Ca- Common names. fecillo. е specimens seen. VENEZUELA. a Altamira, Quintero & Ricardi 1524 (MER). pos 378 (MY). Mé guá, cung A 56322 (F, MY, VEN). of Biscucuy, egro y Puente "Salom. Badillo et al. 7848 ( 15. Cestrum megalophyllum Dunal, in A. DC., Prodr. 13(1): 638. 1852. TYPE: Trinidad. Sie- ber 176 (lectotype, here designated, G-DC, — IDC microfiche, — F photo 33963; isolectoty- ). Cestrum clausseni Dunal, in A. M Prodr. 13(1): 637. 1852. T ҮРЕ: Brazil. Minas Gerais: Claussen 446 Sis totype, sio ar di by D' ye (1974: 606), MPU; solectotypes, бини, т Dinah: in A. DC., Prodr. 13(1): 640. 1852. TYPE: Cultivated in Spain, Faucher s.n. (ho- . G not seen). Cestrum beni | Angelsh., Repert. Spec. Nov. Regni Уе 7: 248. . TYPE: Bolivia. Bang 1634 xe designated by D'Arcy (1974: 606), MO). Shrub or tree 1.5—8 m tall, 6-8 cm DBH, the trunk erect or somewhat arching, bark smooth and ray-green, young branches and emerging growth puberulent; pubescence of scruffy, simple, some- times glandular, and sometimes branched hairs. Leaves obovate to ellipitical, * upfolded from the median vein, 10.5-27(-34) х 3.6-8(-12) cm, api- cally acute or attenuate, the base cuneate and somewhat decurrent on the petiole, margin slightly revolute, coriaceous or subcoriaceous, thick, dark green and opaque above, paler and silvery beneath, Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 307 Figure 26. Cestrum lindenii. —A. Flowering branch.—B. D. Fruits. After Benítez 4740 (MY). often drying dark, laminas glabrous on both sides, veins 7-12 on each side, sometimes variably spaced, strongly ascending, the main veins and lat- eral veins impressed above, elevated beneath and sometimes with a few hairs, minor venation plane beneath and drying light-colored; petiole dark pur- ple, 1.2-2(-3) cm long, canaliculate, slightly flex- ible, swollen at the base, glabrous; minor leaves Flower.—C. Corolla opened to show stamens and style.— wanting. /nflorescences axillary, groups of small fas- cicles or short pedunculate clusters, axes 3-5 mm long, generally unbranched, puberulent. Flowers nocturnal, faintly fragrant, 18-22 mm long, buds white with a purplish tinge, sessile or subsessile; bracteoles linear, slightly arcuate, 1 mm long, pu- berulent; calyx purple, 2.8-3.6 X 1-1.5 mm, firmly membranous, 5-costate, the tube 2-2.8 mm long, 308 Annals of the Missouri Botanical Garden Figure 27. ezuela.—B. Representative occurrences outside of Vene- zuela. Cestrum lindeni.—A. Distribution in Ven- sparingly pilose outside, irregularly 5-dentate, teeth 0.8 mm long, ciliolate and tufted; corolla greenish white or pale green, 14—20 mm long, tube narrow, expanding gradually upward, slightly contracted around the ovary and again at the mouth, mouth 1– 1.5 mm wide, 5-lobate, the lobes white or purplish, narrowly ovate, apically acute, the folds pilose out- side, 2.5-3.5 mm long, reflexed or spreading; sta- mens equal, 11-14 mm long, filaments adnate for 7-11 mm, basally pubescent, insertion mostly straight, sometimes geniculate, tumid, pubescent, free part 2-4 mm, anthers 0.2-0.4 mm across; ova- ry 0.5-1 X 8 mm, glabrous, ovules 2—3, style 10— 14 mm long, sparingly pilose below the stigma, stig- ma capitate, 0.5 mm across. Fruit borne on older woody branches, purple-black, ellipsoidal and ob- ovoid, juicy; seeds 1-3 per fruit, dark brown, 3- mm long. [Francey 6: 312.] Figure 30. This species typically has large firm leaves that often dry dark, especially the costa of the leaf un- dersides and the petioles. Some specimens of Ces- trum lindenii have similar-appearing dark petioles and major veins, but the minor venation beneath is not light-colored as in C. megalophyllum. The name Cestrum schwenckiiflorum Dammer was used by Ule (1908: 401) for a specimen labeled "Peru. Iquitos, July 1902, Ule 6240" deposited at HBG and perhaps other places. Cestrum faucheri Dunal was placed in synonymy by Francey (1935: 12). Distribution (Fig. 31). Amazonas, Aragua, Bar- inas, Bolívar, Carabobo, Cojedes, Falcón, Lara, Mé- rida, Miranda, Nueva Esparta, Portuguesa, Sucre, Táchira, Trujillo, Yaracuy, Zulia, and the Distrito Federal. Shady, gallery forests; 30—600 m and in primary and secondary cloud forests, especially in very moist and shady sites; 1200—1500 m. Also in Mexico, Central America, the Antilles, Colombia, Ecuador, Peru, Bolivia, and Brazil. Phenology. Flowering is mainly from November to April, with the maximum in January. Fruiting is during the dry season, from January to May, with the maximum in March and April. While some- times seen full of flowers, plants often produce only a few flowers at a time. Bella de Noche. Common name. VENEZUELA. Ama- Representative specimens seen. quera, NE de La Sierra, Delascio 7579 (VEN). Falcón: Sierra de San Luis, arriba de Santa María, van der Werff et al. 3209 (WIS). Lara: Laguna Negra, loma de Los Nar- jos, S de Río Claro, Steyermark et al. 111522 (US, VEN). Mérida: La Isla, Jají, López-Palacios 1886 (MO, US, VEN). Miranda: El Guapo, Aristeguieta 4009 (MO, EN). Nueva Esparta: Cerro Copey, Hoyos & Delascio esa: Quebrada Cuchilla Alta, Stergios et al. 6626 (MO, MY, NY, PORT). Sucre: Río Grande Arriba hasta La Pava, Benúez et al. 5124 (MY). Táchira: Quebrada La Buenañita, Las Coloradas, Benítez & Rojas 5046 (MY). Trujillo: 1 km W of Guaranecal, Liesner et al. 12836 (MO, VEN). Yaracuy: Cerro Negro, Río Cocorotico, Steyermark & Wessels Boer 100397 (MO, US, VEN). Zulia: Río Guasare, Serranía de Perijá, Gentry 41165 (MO, NY). Distrito Federal: between Colonia To- var and Carayaca, 2.3 km below the junction with Colonia Tovar—Caracas road, Croat 54467 (MO, NY, VEN). 16. Cestrum microcalyx Francey, Candollea 6: 301. 1935. TYPE: Colombia. Prov. Túquenes, 3000 m, Triana 2295 (holotype, G-DC, = IDC microfiche; isotype, P). Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 309 2 cm to show stamens. After Benítez 4901 cm Figure 28. Cestrum mariquitense.—A. Branch with flowers and fruit. —B. Flower.—C. Style.—D. Corolla opened Y). = = 310 Annals of the Missouri Botanical Garden Figure 29. Cestrum mariquitense.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. Cestrum silvaticum Francey, Candollea 6: 316. 1935, non C. sylvaticum Dunal (1852). TYPE: Peru. Junín: Pi- chis trail, Yapas, 1300-1600 m, Killip & Smith 25483 (lectotype, here designated, F; isolectotypes, N Cestrum standleyi Francey, Candollea 6: 249. 1935. PE: Costa Rica. San José: Zurquí, 2000-2500 m, Standley & Valerio 48082 (holotype, F). Cestrum tenuissimum Franc SYNTYPES: Bolivia. Mapiri, 700 m, Troll 2752 (B destroyed, G not seen, F); Buchtien 32 (B, G, LAU, none seen, B — F photo 28375, MO). Shrub or tree to 6 m tall, much branched from 1.5 mm above the base, 5-15 cm DBH, branches bright grayish green, young branches puberulent; pubescence scant, of reduced, simple, mostly gland-tipped hairs. Leaves narrowly elliptical, 8-17 X 2-5.5 cm, apically acuminate, the tips sometimes arching, basally obtuse, or acute, sometimes slight- ly oblique, margin mostly plane, firmly membra- nous, dark green and lustrous above, lighter be- neath, glabrous, veins 6—11 on each side, often ill-spaced, ascending, major veins not impressed, elevated beneath, minor venation inconspicuous on both sides; petioles 7-12 mm long; minor leaves generally wanting. Inflorescences axillary, short, lax, few(-7)-flowered racemes, axes 2-12 mm long. Flowers nocturnal, 18-23 mm long; sessile, insert- ed along the rachis leaving a cicatrice when fallen; bracteoles to 1 mm long, filiform, caducous; calyx tubular-urceolate, 2-3 mm long, inconspicuously costate, tube 1.5-2.5 mm long, glabrous, the teeth 0.5 mm long, unequal, narrowly triangular, curving outward, mostly glabrate, ciliate; corolla light yel- lowish green, 19-22 mm long, cylindrical, the tube slender, slightly contracted at the ovary, then hardly expanded upward, expanded apically around the anthers and contracted at the mouth, the mouth 1.5-2 mm across, the lobes narrowly triangular, 3- 4.5 mm long, ciliate and puberulent on the folds, otherwise glabrous; stamens 15.5-17.5 mm long, filaments adnate for 12-15 mm, pilose 1-2 mm be- low the insertion, insertion straight, smooth, free part 2-3.5 mm, anthers orbicular, 0.5 mm across; ovary globose, 0.5-0.8 mm across, disk inconspic- uous, ovules 7—8, style 16-17 mm long, laxly pu- berulent near the apex, exceeding the stamens by 0.5 mm. Fruit dark purple, globose, 4—10 mm across, with little pulp or juice; seeds black, 2—3, 9 mm long. Figure 32. This species may be confused with Cestrum ra- cemosum or C. megalophyllum, but it has fewer, more membranous leaves with fewer veins than the former and smaller leaves than the latter. Distribution (Fig. 33). Aragua, Táchira, and the Distrito Federal. Restricted to cloud forests; 800— 1700 m. Also in Nicaragua, Costa Rica, Panama, Colombia, Ecuador, Peru, and Bolivia. Phenology. Collected in flower in May and June, especially the latter, and in fruit in October. Common names. Tapa Camino. Representative specimens seen. VENEZUELA. Ara- a: do, carretera Maracay-Choroní, Benítez & Rojas 3994 (MY). Táchira: Cerro de Cuite, quebrada La Colorada, Steyermark et al. 119733 (MO, NY). Distrito Federal: Carretera Colonia Tovar-Puerto Cruz, Trujillo 15831 (MY). 17. Cestrum neblinense D'Arcy & Benítez, Ann. Missouri Bot. Gard. 77: 206. 1990. TYPE: Venezuela. Amazonas: Dept. Río Negro, cerro de La Neblina, 0%51'N, 65%57'W, 700 m, Lies- ner 16661 (holotype, MO; isotypes, MY, VEN). Shrub 1-2 m tall, branches reddish, tomentose, hairs curved, ascending, glabrescent, the base of the internodes and inflorescences thickened; pubes- Volume 85, Number 2 1998 Benítez & D'Arcy 311 Cestrum and Sessea in Venezuela ” >” Figure 30. Cestrum megalophyllum.—A. Flowering branch.—B. Flower.—C. Corolla opened to show stamens.—D. Twig with fruits. After Benítez 3617 (MY). cence of small, simple, coarse, multicellular yellow- ish hairs, often reduced. Leaves linear or narrowly ovate, sometimes slightly oblique, 5-10 X 0.5-2.5 cm, attenuate above the middle, apically obtuse, ba- sal third attenuate, margin slightly revolute, subcor- iaceous or papery, bright green above, lighter be- neath, veins 4—6 on each side, strongly ascending, the major veins sunken, venation elevated beneath; petioles 2-4 mm long, slender; minor leaves want- ing. Inflorescences of 1-3 flowers grouped in the leaf 312 Annals of the Missouri Botanical Garden 3, Figure 31. Cestrum megalophyllum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. axils or terminal, peduncles tomentose, 2-5 mm long, bracts 4-5 mm long; bracteoles 2 mm long. Flowers 27-30 mm long, sessile or on pedicels 1.5 mm long, bracteoles linear, ca 3 mm long, pubes- cent; calyx cupular, 2.5-3.5 X 2.5 mm, 5-costate, glabrate, tube 2.2-3 mm long, 5-toothed, the teeth undulate, ciliolate, 0.3-0.5 mm long; corolla white, 25-27 mm long, tube narrow, slightly expanded up- ward and at the throat, mouth 3.5 mm wide, the lobes narrowly elliptic, ciliolate, 2.5-3.5 mm long; stamens subequal, 17-19 mm long, filaments adnate for 16-17.5 mm, pilose at the base, insertion straight, smooth, free part 1.5-2 mm, anthers subor- bicular, 0.8 mm across; ovary globose, 1 mm across, style 19 mm long, moderately pilose for 10 mm be- low the stigma, stigma capitate. Fruit ellipsoidal or obovoid, 9 mm long, 6.5-7 mm wide; fruiting calyx ca. б mm long; seeds 7, 5-7 mm long. Figure 34. This species is distinctive in its coriaceous, nar- row leaves and abbreviated inflorescences. The col- ors of the fruit and seeds are unknown. Distribution (Fig. 35). Amazonas. Evergreen cloud forests in the Cerro La Neblina, partly flood- ed forests along the River Yatúa in sandy soil; 780— 2200 m. Apparently endemic. Phenology. Collected in flower and fruit in arch. Additional specimens seen. VENEZUELA. Amazo- nas: rocky beaches, Cañon Grande, SSW Cumbre Camp, Río Yatúa, cerro La Neblina, Maguire et al. 42500 (MO, , US). 18. Cestrum nocturnum L., Sp. Pl. 1: 191. 1753. TYPE: Jamaica. Pl. Hortus Cliffortianus (LINN 258.1), fide Howard, Fl. Lesser Antilles 6: 276. 1989, Cestrum suberosum Jacq., Pl. Hort. Schoenbr. 4: 26, pl. 52. 1798. TYPE: from Caracas, pl. 452 in Jacquin, 1798 (lectotype, here designated). Shrub 2—3 mm tall, branches angular, sprawling, leafy, olive or bluish green, lenticellulate, emerging growth puberulent; pubescence of simple hairs. Leaves narrowly ovate to ovate, 8-13 X 24.5 ст, evenly attenuate from the middle, the tip acute, ba- sally obtuse or narrowly cuneate, membranous to coriaceous, dark shiny green above, lighter be- neath, veins 7—8 on each side, ascending, elevated beneath; petioles 1-2 cm long, slender, glabrous; minor leaves wanting. /nflorescences axillary and terminal, many-flowered racemes or panicles, axes glabrous, 3-8 cm long; peduncle 5-10 mm long, glabrous. Flowers nocturnal, heavily fragrant, yel- lowish or greenish white, 21-27 mm long, buds sometimes with a slight violet tint, pedicels 0.5 mm long; bracteoles 2, one foliaceous, 3.5 mm long and puberulent, the other linear, pilose 2.5 mm long; calyx cupular, 2.5-3.5 X 1-1.5 mm, glabrous out- side, costate, tube 2—4.5 mm long, 5—6-toothed, teeth 0.5-1 mm long, ciliolate and minutely tufted; corolla 20-30 mm long, the tube gradually expand- ed upward, slightly contracted beneath the ovary, 5-lobed, mouth 2.5 mm wide, lobes 2-5 mm long, puberulent, apically obtuse or slightly apiculate; stamens inserted equally, 13-15.5 mm long, fila- ments adnate 10—12 mm, mostly pilose from the base to the middle of the adnate portion, the in- sertion straight or geniculate, bidentate, glabrous, free part 2.5-3 mm; anthers included; ovary 0.5-1 mm long, the disk yellow, about as long as the ova- ry, clearly delimited, ovules 7—9, style exserted 1 mm, puberulent 2 mm below the stigma, stigma bi- lobed. Fruit white, sublustrous, globose, 5-9 mm across, mesocarp white, spongy-granular, the pla- centa green, juicy; fruiting calyx slightly accres- cent; seeds 1-3 per fruit, black, ovoid, 2.5-6 mm Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 313 Q = == ~ о Cestrum microcalyx.—A. Branch with flowers and fruit. —B. Flower.—C. Corolla opened to show stamens ). Figure 32 est and style. After Benítez 3093 (MY long, abortive ovules dull yellow-orange. Figure 36; Nee, 1986: 55. Cestrum nocturnum is cultivated as an ornamen- tal for the nocturnal fragrance of its flowers. The greenish flowers have small, glabrous calyces, and the axes are slender, although the inflorescence structure varies greatly. Herbarium specimens are superficially much like other species, but the two small teeth at the stamen insertion are a good rec- ognition character. Chromosome numbers of this species have been reported as л = 8, 2n = 16, the normal comple- ment for the genus (Darlington & Wylie, 1955). Distribution (Fig. 37). Aragua, Mérida, Miran- 314 Annals of the Missouri Botanical Garden Figure 33. Venezuela. —B. Representative occurrences outside of uela. Vene Cestrum microcalyx.—A. Distribution in da, Monagas, Sucre, Táchira, and the Distrito Fed- eral. Widely spontaneous and naturalized; sea level to 1300 m. Also in Mexico, the Antilles, and Cen- tral America. Perhaps native to Central America. Phenology. Flowering occurs in short or long spurts or almost continuously throughout the year but is heaviest in June. Flowering periodicity was also reported by Sachs (1985). Fruiting is sporadic, especially in the second half of the year. Common names. Bella de Noche, Dama de No- che. Representative specimens seen. VENEZUELA. Ara- Maturín, Barrios 14 (MY). Sucre: Chamariapo, 6 km de Cariaco, Ruíz-Terán & López-Pala- cios 9975 (MY). Táchira: Distrito Cárdenas, Río Torbes, Bono 5043 (MY). Distrito Federal: Caracas, Lasser 3469 (VEN). 19. Cestrum olivaceum Francey, Candollea 6: 129. 1935. TYPE: Colombia. Santander: east- ern cordillera, vicinity of Charta, 2000-2600 m, Killip & Smith 18917 (lectotype, here des- ignated, NY). Shrub 3—4 m tall, branches climbing, terete, to- mentose; pubescence of stellate hairs. Leaves de- scending or disposed vertically, ovate, 5-11 х 3.5- 6.5 cm, attenuate from the lower third upward, apically acute, basally rounded, margins mostly drying slightly revolute, densely rugose above and beneath, subcoriaceous, dark olive green above, lighter beneath, pubescent, more so beneath, veins on each side, elevated beneath; petiole 0.5—1 cm long, tomentose; minor leaves wanting. Inflores- cences short, axillary, condensed racemes, axes 1.5— 2.5 cm long; bracts filiform, 5.5 mm long, pubes- cent. Flowers nocturnal?, 25-26.5 mm long, sessile; calyx green, 5—7 X 3.5—4 mm long, tubular, stellate pubescent outside, glabrous inside, tube long, 4—5-dentate, teeth 1.5 mm long, obtuse or acute; corolla lilac, apically with whitish and pur- plish markings, funnelform, 24-26 mm long, pu- bescent outside, especially above the middle, tube contracted below and above the ovary, mouth 4 mm wide, lobes 4—6.5 mm long, deltoid; stamens 17— 19.5 mm long, filaments glabrous, adnate for 10— 12 mm, insertion geniculate-tumid, free part 7-8.5 mm, anthers 0.7 mm long, orbicular; ovary subglo- bose, 1.5 mm across, glabrous, ovules 9-12 mm, style 19-20 mm long, filiform, papillose below the stigma, stigma capitate, included. Fruit ovoid, 8-9 X 6—7 mm wide; seeds 5, 4—5 mm long. Figure 38. Cestrum olivaceum is distinctive in its dense overall brownish yellow pubescence and lilac flow- ers. Distribution (Fig. 39). Táchira. Cloud forests along ravines and disturbed forest margins; 2150 to 2450 m. Also in the Department of Santander, Colombia. Phenology. Of the two collections seen, one was in flower in April and the other in fruit in July. ааа Қ бесік Parque Тата, zona de Buena Vista, km arriba de San Vicente de La Revancha, Morillo & García jj (MERF, MY); Quebrada Azul, S of El Reposo, 14 km SE of De- licias, Steyermark & 4 118501 (МО, УЕМ). Additional specimens seen. 20. Cestrum pariense Steyerm., Acta Bot. Ve- nez. 1(2): 62. 1966. TYPE: Venezuela. Sucre: Cerro Patao, N de Puerto de Hierro, NE de Güiria, 850 m, Steyermark & Agostini 91026 (holotype, VEN; isotype, US). Volume 85, Number 2 Benítez & D'Arc 1998 Cestrum and Sessea in Venezuela 315 S AO "A | ^ SARA | SS IN) Z \ || Wy 2 W / 7 У Q / - к "E „=й шее „295 4 = p Rm 21 d „ЖУ ЖЫ EM -— pe \ \ 5 ст \ Figure 34. Cestrum neblinense.—A. Branch with fruit. —B. Detail of leaf venation.—C. Flower.—D. Style and stigma.—E. Corolla opened to show stamens. After Liesner 1661 (MY) 316 Annals of the us Botanical Garden Figure 35. Distribution of two localized species of Cestrum. Circle = Cestrum neblinense. Star = Cestrum schulzianum Shrub or subshrub 0.6-2.5 m tall, stems fleshy, young parts pilose, glabrescent; pubescence of re- duced, crinkled, simple and branched hairs. Leaves elliptical, 10-25 х 6-10 cm, apically acute or short-acuminate, basally cuneate, firmly membra- nous, sometimes fleshy or coriaceous, dark green above, pale beneath, glabrous, veins 6-9 on each side, salient beneath; petiole purple, 1—3.5 cm long, puberulous; minor leaves wanting. /nflores- cences axillary, peduncles solitary, 2.5—4 cm long, l-4-flowered, pendulous or erect, elongated іп fruit, puberulent; bracts foliaceous, folded upward along the costa and covering the sides of the calyx, pilose, especially along the costa beneath, 7-10.5 mm long, (-10.5) mm wide. Flowers nocturnal, 29-31 mm long, sessile; calyx cupular, 3-3.5 X 2 mm, sparingly pilose, inconspicuously 5-costate, 2.5-3 mm long, 5-toothed, the teeth 0.5 mm long, narrowly triangular, ciliate, tufted; corolla greenish white, 28-31 mm long, glabrous, subfun- nelform, tube contracted around the ovary, abruptly expanded toward the apex, 5-lobed, mouth 2.5 mm wide, lobes 6.5-8 X 1.5-2.5 mm, ciliolate, folds puberulent; stamens 16—17 mm long, filaments gla- brous, adnate for 13.5—14 mm, 1-1.5 mm free, in- sertion straight, smooth, anthers suborbicular, 0.8 mm across; ovary subglobose, 1-1.5 mm across, glabrous, ovules 6-8, style 15.5-16.5 mm long, ex- serted 0.5 mm, papillose 1 mm below the stigma then short-pilose 2 mm further down, stigma bi- lobed, 1 mm long. Fruit dark purple, compressed- globose, 14-15 X 10-13 mm wide; seeds 7, 6.5 mm long. Figure 40. This species is similar to Cestrum bigibbosum in its habitat and in the length of its corolla lobes. Small (to 1 m) plants are similar in aspect. How- ever, flowers of C. pariense are elevated on slender peduncles and have much smaller calyces. Distribution (Fig. 23). Sucre. Evergreen forests and river banks; 600—1400 m. Apparently endemic to the Paria Peninsula. Phenology. Collected in flower in July and in fruit in February and March. VENEZULEA. Sucre: P. Additional specimens seen. Nacional Peníns ы "9 ы ж ы =: Е z, È n. jM © = 15%) d = Ф NE de Pali Foe 3154 (MY); Cerro de "NE de lrapa, Steyermark 95075 (VEN), 95076 (US); Río Tacarigua and headwaters of Río Tacarigua, E of Cerro Humo, N of Río Grande Arriba, Steyermark et al. 121595 (MO, VEN). 21. Cestrum petiolare Kunth, in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 58. 1818. Sessea petiolaris (Kunth) Spreng., Syst. Veg. 1: 584. 1825. SYNTYPES: Peru. Between Ayavaca - s 800 hex, DET 1409 (P-Bonpl. een, — microfiche 63-1-4, — F photo e B destroyed, = F photo 3020). Cestrum rer Willd. re o = Schult., Syst. e. 807. . TYPE: rid. Humboldt s Ac a E Е 4438 not seen, = IDC uia те) Cestrum caloneurum Pittier, J. Wash. Аса ci. 22: 31. 1932. TYPE: Venezuela. Aragua: Co ee Төл 22. т, аны 10045 (holotype, VEN; pes, G-DC, Cestrum Airis Pittier, J. Wash. Acad. Sci 932. TYPE: l. 1 : Venezuela. Aragua: . Tovar. 1800-2000 m, Allart 480 (holotype, VEN). Shrub or small tree 4-5(-8) m tall, 5 cm DBH, bark smooth, dark green to gray with prominent lenticels, branches flexuous, young parts often pu- berulent; pubescence of dendritic white or yellow hairs. Leaves malodorous, „РА revolute, narrowly elliptical or narrowly ovate, sometimes oblique, (7-)16-22(-29) х (8. 5-)5-8(-11) cm, api- cally acute, often shortly and narrowly prolonged, asally obtuse or narrowly cuneate, sometimes al- most truncate, firmly papery to subcoriaceous, above yellowish green, matte, beneath paler, some- what shiny, veins 14-21(-26) on each side, parallel and ascending near the margins, sunken above, some pustular, purplish beneath, and the minor ve- nation mostly elevated, often appearing pulveru- lent; petiole purplish on young leaves, green when mature, flexuous, canaliculate, 2—4.5 cm X 1.2-1.5 5—8 mm, sessile, am- plexicaulous, sometimes falcate, pointed, glabrous, often caducous, mm; minor leaves 7-15 X sometimes wanting on mature branches. Inflorescences axillary and terminal, form- Volume 85, Number 2 Benítez & D'Arcy 317 1998 Cestrum and Sessea in Venezuela — -— | | Ыр ci E а 3 Figure 36. Cestrum nocturnum.—A. Flowering branch.—B. ‘Twig with fruits.—C. Flower.—D. Stamen.—E. Corolla opened to show stamens. After Benítez 1034 (MY) 318 Annals of the Missouri Botanical Garden EM Figure 37. Cestrum nocturnum.—A. Distribution in со —B. Representative occurrences outside of Venezuela. ing lax, paniculate clusters between the branches, sometimes enlarged panicles 5-8(-13) cm long, flexuous, axes angular and pulverulent. Flowers di- urnal, fragrant, 20-24 mm long, sessile, bracteoles linear-subulate, deciduous, 5 mm long; calyx bright green, tubular, 7-10.5 X 3.5-4 mm, inconspicu- ously 5-costate, coriaceous, thick, glabrate outside and puberulent inside, especially on the veins with- in, tube 6–9 mm long, 3—5-toothed, teeth unequal, 1-1.5 mm long, ciliolate and apically pubescent; corolla yellow, yellowish green, or greenish white, 18-23 mm long, clavate, tube obconical, gradually expanded upward, barely contracted below the ova- ry and lobes, glabrous, mouth 5 mm wide, the lobes narrowly ovate, ciliolate, apically mucro- nate, 2.5—4.5 mm long; stamens 12-16 mm long, filaments adnate for 5.5-9.5 mm, pilose 2-5 mm from the base or to the insertion, insertion genic- ulate-tumid, pilose, 1.5-2.5 mm long, free part 4.5— 9.5 mm, anthers orbicular, 0.5 mm across; ovary ovoid or subglobose, 1-2 X 1.5-2 mm, glabrous, disk conspicuous, ovules 8-32, style 14—18 mm long, bright dark green, glabrous, stigma bright green, capitate, 0.5 mm long, slightly exserted. Fruit narrowly ellipsoidal, dark purple, 13-25 X 5-12 mm, pulp whitish, 1.7 mm thick; fruiting ca- lyx enclosing about half of the fruit; seeds (6-)18- 31 per fruit, brownish yellow, 2.5-3 mm long (Croat 54936), embryo white, straight. Figures 1A, 41. Cestrum petiolare, with its broad corollas and nu- merous veins on the leaves, resembles C. lindenii, but it differs in its pubescence of dendritic rather than simple hairs on twigs and leaf undersides, and in usually having much larger calyces. This species commonly has many more ovules than any other species in the genus. Francey (1935: 390) referred the unpublished name “Cestrum moritzianum Klotzsch & Karsten,” used on a manuscript in Berlin, to this species. Distribution (Fig. 42). Aragua, Mérida, Miran- da, Táchira, Trujillo, and the Distrito Federal. Tran- sitional cloud forests, dwarf cloud forests, and near ravines on paramo; 1120-3200 m. Also in Colom- bia and Peru. Phenology. Flowering year-round except in July, mainly from January to April and in August. Fruiting is year-round. mmon names. Com Borrachero, Tabacón, Titiera, Uvito de árbol. VENEZUELA. Ara- ШЕ ылар саја imens = nié БН Б E s, López-Figueiras 8754 ( de Naiguatá, el fondo de la quebrada Rancho Grande, Meier 3330 (MY, VEN). Táchira: Carretera Seboruco—El Suspiro, Benítez & Rojas 4746 (MY). Trujillo: Carretera rep de Aristeguieta & Medina 3689 (NY, US, rito Federal: entre La Rosita y El Porta- ee Hanes 1438 (VEN). 22. Cestrum potaliifolium Dunal, in A. DC. Prodr. 13(1): 638. 1852 кол | E: Venezuela. Aragua: Colonia Tovar, Mo- ritz 824 (holotype, G-DC, = IDC microfiche; isotypes, B destroyed, = F photo 2991, BM). Cestrum tovarense Francey, Candollea 6: 388. 1935. TYPE: Venezuela. Aragua: Colonia Tovar, Fendler 962 (holotype, NY; isotype, GH). Shrub or tree to 5 m tall, 6 cm DBH, stems straight, slightly branched, ridged and grooved, with dispersed glands, young growth pulverulent or puberulent with a purplish hue; pubescence of sim- ple reduced hairs. Leaves inodorous, ovate or ellip- tical, rarely obovate, 10-28 X 5-12 cm, apically acute, basally unequal or rounded, rarely narrowly Volume 85, Number 2 1998 Benítez & D'Arcy 319 Cestrum and Sessea in Venezuela м 667 > WE D еу i orbe Figure 38. and style. After Morillo & Garcia 11478 (MY). cuneate, the margin sometimes retracted between the main lateral veins and appearing dentate or erose, slightly revolute, lamina coriaceous or sub- coriaceous (papery when living), bright dark green above, pale yellowish green beneath, drying yellow, glabrate, sometimes with scarce minute glands be- neath, veins 6-10 on each side, ascending, slightly prominent above, elevated beneath, and the minor Cestrum olivaceum.—A. Branch with flowers and fruit. —B. Flower.—C. Corolla opened to show stamens venation reticulate, inconspicuous when alive; pet- ioles flat-topped, distally canaliculate, flexible in mature leaves, 1.2-2.5 cm long, glabrous, the in- sertion at the stem often expanded into a rounded pillow-like form; minor leaves wanting. /nflores- cences axillary, clustered. Flowers 20-24 mm long, sessile or with pedicels 0.5 mm long, bracteoles linear, arching, 24.5 X 0.5-1 mm, puberulent; ca- 320 Annals of the Missouri Botanical Garden Figure 39. Cestrum olivaceum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. lyx cupular, 4-5.5 X 2 mm, slightly coriaceous, rugose, glabrous or sparingly puberulent, tube 3- 4.5 mm long, 5-dentate, teeth narrowly ovate, 1—2 mm long, apically cuspidate, pilose and with sparse dispersed glands; corolla greenish white or pale yellow, 18-23 mm long, narrowly funnelform, tube gradually expanded rein slightly contracted at the ovary, mouth 2-2.5 mm wide, lobes 3.5—4.5 mm long, narrowly ovate, "ud obtuse; stamens .5-15.8 mm long, filaments adnate for 9-11 mm, sparingly pilose from the base to the insertion, straight, barbate, free part 4—5 mm, anthers subor- bicular, 0.5 mm across; ovary globose or elliptical, 0.8-1 mm across, glabrous, disk conspicuous, style 12-15 mm long, exserted 0.5 mm, stigma capitate. Fruit purple, obovoid to ellipsoid, 8-12 х 7-9 mm wide; seeds 6-8, dark brown, 7-7.5 mm long. [Francey 7: 65.] Figure 43. Cestrum potalüfolium is similar to C. megalo- phyllum but is distinct in its often yellowish leaves and in the expansion of the stem around many of the petiole bases into a conspicuous, often pillow- like ridge. n the original publication, Dunal erroneously cited the type collection as being from Colombia. Distribution (Fig. 44). Aragua and the Distrito Federal. Endemic to Venezuela. Found in ever- green cloud forests; 1000 to 2000 m Collected in flower from June to October, mainly in June. Fruiting collections have Phenology. been made from September to January, mainly in January. Representative specimens seen. VENEZUELA. Ara- gua: entre Las Marochas y Choroní, 2. et al. 4912 (MO, MY); Portachuelo forest, Wood 3 EN). Distrito Federal: E of Junquito, Steyermark с (MY, VEN). 23. Cestrum racemosum Ruiz & Pav., Fl. Pe- ruv. 2: 29, pl. 154. 1799. TYPE: Peru. Chin- chao and Macora, Ruiz s.n. (holotype, F). Cestrum mathew sii Dunal, in A. DC., Prodr. 13(1): 637. '"NTYPES: Peru. Chachapoyas: Mathews s.n. PU, G, neither seen, US). Cestrum panamense Standl., J. Wash. Acad. Sci. 15: 460. 1925. Cestrum racemosum var. panamense Stand Franc еу. Candollea 6: 274. 1935. TYPE: Panama. ía, Standley 28042 (holotype, US). Cestrum grande Pittier, J. Wash. Acad. Sci. 22: 32. 1932. estrum 2 var. de (Pittier) Francey, C andolle a 6: 275. 1935. TYPE: Venezuela. Distrito Federal: Curucutí, 400 m, on old road from Caracas to La Guaira, Pittier 10393 (holotype, VEN; isotypes, GH, NY, US). Cestrum racemosum var. bolivianum Francey, 274. 1935. SYNTYPES: Bolivia. S Bosques del Fraile, Buenavista, 450 7259 (B not seen, MO, S not seen). Trees 6-20(-25) m tall, trunk 12(-60) cm DBH, grayish green, straight, branching with a narrow crown, wood soft and whitish, =o “= Y бо 22 5 Candollea 6: ig Cruz: 1 Steinbach stems lenticellular, young growth sparingly puberulent; pubescence of simple, white, moniliform ascending and crumple hairs. Leaves malodorous, spreading, ovate 11—19 (-22) X 2.5-7(-9) cm, apically acute or acuminate, base rounded, membranous, slightly sticky to touch, matte dark green above, pale yellowish green beneath, glabrate with a few scattered hairs above, when mature glabrous and often lenticellulate be- neath, veins (10-)15-22(-27) on each side, equally spaced and arcuate-ascending at (62°—)70°(— 75°), prominent beneath; petioles canaliculate, 0.7—2 cm long; minor leaves absent. Inflorescences mostly ax- illary clusters of short racemes, peduncles 1.5-2.5 cm long, sparingly pilose. Flowers greenish white or yellowish green, the apex sometimes slightly pur- ple, 14-20 mm long; pedicels 1-2 mm long, pu- bescent; bracteoles 1–2.7 mm long, linear, pubes- Volume 85, Number 2 Benítez & D'Arcy 321 1998 Cestrum and Sessea in Venezuela C Q = y у : X Ха” D V | Е : і c. ( НЕ : : : + EEN | | “де N * E : : ; ; ; 5 / MAE if 2 Figure 40. Cestrum pariense.—A. Branch with flowers and fruit. —B. Fully opened (night) flower.—C. Corolla opened to show stamens and style.—D. Closed (day) flower. After Benítez 5130 (MY). cent; calyx cupular, 1.7-3.6 X 1-1.5 mm, basally lobes triangular, = acute apically, folds pilose, 2— narrowed into a stipe, 5—6-costate, membranous, 3.5 mm long; stamens 10.5-12.5 mm long, fila- pilose outside, tube 1.5-2.8 mm long, 5-dentate, ments adnate for 8-12 mm, with sparse hairs from teeth 0.2-0.8 mm long, pilose outside on the veins the base to 1.5-2.5 mm below the insertion, the and the tips; corolla narrowly funnelform, 13-18 base glabrous or pubescent, insertion straight, mm long, tube gradually expanded upward, slightly smooth, free part 1.5-2.5 mm, anthers spherical, contracted below the ovary, mouth 1.5 mm wide, 0.2-0.5 mm across; ovary 1-1.5 X 0.5-0.8 mm, 322 Annals of the Missouri Botanical Garden 4 4 DIH 6 У f; ја „ТУ? 2224 | ҰТО” ИЕ ТҮ, 2 => po И а” LES AD AE чује. e 41. Cestrum petiolare.—A. Branch with flowers and fruit.—B. Flower.—C. Corolla opened to show stamens. After Benítez 4739 (MY). glabrous, disk inconspicuous, ovules 4—7, style 9- — (after Greenman 5218), embryo white, 1.5 mm long. 12.5 mm long, papillose below the stigma, stigma Figure 45; D'Arcy, 1974: 609, figure 5. capitate, included. Fruit purple, globose, 5-6 mm long, 3-5 mm wide; fruiting calyx accrescent and Cestrum racemosum may be recognized by its cupulate; seeds 3-7 per fruit, brown, 3 mm long usually arborescent stature and usually narrow Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 323 3, е, Figure 42. Cestrum petiolare.—A. Distribution in Venezuela.—B. Representative occurrences outside of uela. Venez leaves with numerous, evenly spaced veins, which are salient on the leaf undersides. Distribution (Fig. 46). Amazonas, Anzoátegui, Aragua, Barinas, Falcón, Lara, Mérida, Miranda, Portuguesa, Táchira, Yaracuy, Zulia, and the Dis- trito Federal. Evergreen forests, secondary forests, and gallery forests; 4 00 m. Also occurring from Mexico to Bolivia. Phenology. Flowering and fruiting year-round with maxima in the first half of the year. Common names and uses. Used to shade coffee plants, sometimes cultivated in parks and gardens. Representative specimens seen. VENEZUELA. Ama- zonas: Cerro de la Neblina, Ewel 195 (MY). Anzoátegui: Río Maravilla, E of Bergantín, lona i 61711 (MY, a: Carretera Maracay—Choroní, Benítez & Ж © = = E (b. > = z @ a o Paso de Angostura, represa de Yacambú la Quebrada Honda con el Río Yacambú, Жете ав & Carrefio Espinoza 108771 (VEN). Mérida: entre San Ja- cinto y Tienditas del Chama, Quintero 526 (MER); La Punta, Ricardi & Salcedo 5744 (MER), 5756 (MER). Mir- udtegui, 71 53 (VEN). : La Petrolea, carretera Rubio-San Vicente de la рану Вепйег et al. 4864 (MY). Yar- n Caracas and La Guaira, Fendler 961 (GH). 24. Cestrum reflexum Sendtn., in Mart., Fl. Bras. 10: 218. 1846. TYPE: Bolivia [Brazil]. Chiquitos, Orbigny 659 (lectotype, here des- ignated, P). Cestrum А Britton, Mem. Leer bæ Club. 6: 92. ». TYPE: Bolivia. La Paz: пне үзе go iia (MO, Cestrum reflexum m Francey, Candollea 6: 267. 1935. SYNTYPES: “Bolivia, Santa Cruz: bosque de Buenavista, 450 m, Steinbach 1480 (B dites 3216 (B destroyed), 6172 (MO, B destroyed, S not seen), 7162 (MO, B destroyed, S not seen). Climbing shrub 2-3 mm tall, branches flexuous, puberulent; pubescence of simple and sparingly branched, often cobwebby hairs. Leaves often dry- ing grayish, ovate or narrowly ovate, 4.5-11 X 2- 6 cm, apically acuminate, the tip acute or obtuse, basally rounded, sometimes short-decurrent on the petiole, membranous, shiny green and puberulent on both sides, 7—8 veins on each side, minor veins impressed above; petiole 1-1.3 cm long, at inser- tion slightly bent and hooked, slightly thickened and densely tomentose; minor leaves not seen. In- florescences leafy axillary or terminal racemes and panicles, axes 3-5 cm long, pubescent; bracts fo- liaceous, reduced upward. Flowers whitish or yel- lowish, 25-29 mm long, pedicels obsolete, bracte- oles 3 mm long, pilose; calyx cupular, 2 5X 1.5 mm, inconspicuously costate, pilose at the level of the teeth, tube 1.5—3.5 mm long, the teeth 1 mm long, narrowly triangular, + reflexed, pilose inside and out, ciliolate, tufted; corolla 24-27 mm long, glabrous, tube contracted below the ovary, cylin- drical, suddenly expanded apically, mouth 1.5-2 mm wide, lobes 4—6 mm long, narrowly ovate, folds and margin puberulent; stamens 19-19.5 mm long, filaments adnate for 18.5 mm, glabrous, insertion straight, smooth, free part 0.5-1 mm, glabrous, an- thers rotund, 0.5 mm across; ovary 0.5-1 mm long, glabrous, disk conspicuous, 0.5 mm long, ovules 10-14, style 19-19.5 mm long, stigma bilobate, ex- serted 1 mm. Fruit dark purple, ellipsoidal, 6.5— 7(-10) X 5 mm wide; fruiting calyx often drying dark, sometimes flaring; seeds 6—14, light brown, Annals of th Missouri Botanical Garden 7 67 7 = 972 A 7, Г, qu EN <> У +. MSS ж 54 pe ам Sy T S > A Lite Figure 43. Cestrum potaliifolium.—A. Fruiting branch.—B. Flowering branch.—C. Flower.—D. show stamens. After Steyermark 57006 (VEN). Corolla opened to 44.5 mm long. [Francey 6: 265.] Figure 47; D’Arcy, 1974: 611, figure 6. small but persistent, foliaceous bracts. The plant is a scrambler. Herbarium specimens of Cestrum reflexum are often best recognized by their grayish color and Distribution (Fig. 44). Bolívar; 290 m. Also in Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 325 o Figure 44. Distribution of two localized species of Cestrum. Solid square = Cestrum potaliifolium. Open square — Cestrum reflexum. Nicaragua, Costa Rica, Panama, Colombia, Ecua- dor, Peru, Bolivia, and Brazil. The species may be native to Brazil. The sole collection from Venezuela was flowering in February. Outside of Venezuela, Cestrum reflexum is an un- common species found from 50 to 300 m in gallery forest, rainforests, and disturbed areas. Phenology. Specimens seen. VENEZUELA. Bolívar: Camino a la estación Magdalena, El Palmar, NE, Bernardi 7127 (MER, VEN). 25. Cestrum ruizteranianum Benítez & D'Arcy, Novon 5: 313. 1995. TYPE: Venezuela. Méri- da: Distrito Rangel, trail from La Negrita downstream towards Puente de La Escalera, montane cloud forest, 2550-2950 m, Luteyn et al. 6171 (holotype, NY; isotypes, MY, VEN). Shrub 1—4 m tall, erect, branched, young stems angular, terete when mature, scurfy pubescent; pu- bescence of dark, crinkled, perhaps branched hairs. Leaves narrowly elliptical to elliptical, 4-6 X 1.5-3 cm, apically acute or obtuse, basally obtuse, margins slightly revolute, firmly membranous, glabrous on both sides except for some scurfy hairs on minor veins, the major veins impressed above, elevated be- neath, 8-10 on each side, ascending, looping and uniting near the margins; petioles 6—8 mm long, slender, inrolled, tomentulose above; minor leaves sometimes present, ovate, 3 mm, with petioles 0.5-1 mm long. /nflorescences axillary ra- cemes, sometimes appearing as terminal panicles, 2.56 cm long; peduncles 0.7-5 cm long; bracts 1.5-2.5 mm, linear. Flowers 22-26 mm long, pedicels obsolete; bracteoles linear, 3 mm long, sparingly pubescent, caducous; calyx drying dark brown, tubular, —3.5 mm, faintly striate, thick, glandular and with sparse hairs outside, pu- bescent within and with glandular hairs halfway down, 5-toothed, the teeth 2 X 2 mm; corolla pale green, purplish outside, pale yellowish green inside, 20—26 mm long, exserted ca. 19 mm from the calyx, the tube 17-20 mm long, 3-3.5 mm at its widest, contracted around the ovary and then gradually ex- panded upward, the throat not constricted, mouth mm wide, lobes 3-5 X 1.5-2 mm, narrowly triangular-acuminate, sometimes sparingly pubescent, the folds tomentose; stamens 15.5-19 mm long, ad- nate for 7-9 mm, the adnate portion pubescent for the basal 3—4 mm, insertion geniculate-tumid, 1.5— 2.5 mm long, free part 7.5-9.5 mm, anthers orbic- ular, 1 mm across; ovary lobed, 0.7 mm across, glabrous, ovules 16-18, papillose 4—5 mm below the stigma, exceeding the stamens by 1.5 mm, style 15-19 mm, stigma subcapitate, slightly bilobed, in- cluded. Fruit unknown. Figure 48. Cestrum ruizteranianum is very like Cestrum lin- denii but differs in its uniformly smaller leaves, arger flowers, and glandular calyces. Distribution (Fig. 23). Mérida. Montane cloud forest; 2950—2550 m. Endemic. Phenology. Collected in flower in November. Additional specimens seen. VENEZUELA. Mérida: Gavidia, Dist. Rangel, Ruiz-Terán et al. 16171 (MERF, MY), 16154 (MY). 26. Cestrum salicifolium Jacq., Pl. Hort. Schoenbr. 3: 42, pl. 326. 1798. TYPE: from "Caracas," pl. 326 in Jacquin, 1798 (lecto- type, here designated). Cestrum salicifolium var. angustifolium Dunal, in A. DC., Prodr. 13(1): 670. 1852. TYPE: cultivated in Europe as C. salicifolium in herb. Requien (MPU not seen). Shrub or tree 1.5—5 m tall, trunk very slender, sometimes arching, 5-8 cm DBH, branching high on the trunk, branches narrow, often hanging, pur- ple when young, twigs slender, purplish, the epi- dermis longitudinally striate; pubescence of simple, multicellular hairs, evident only on bracts and per- haps emerging growth, plants otherwise glabrous. Leaves narrowly elliptical, 7-15 X 0.8—3 cm, nar- rowing upward, base obtuse or acute and + decur- rent on the petiole, margin slightly revolute, mem- branous to subcoriaceous, glabrous, veins 16—18 (-30), arising nearly perpendicular to the costa, ap- pearing straight and evenly spaced, furcating near the margin and forming a partial, undulating sub- marginal vein, reticulate venation plane above, 326 Annals of the Missouri Botanical Garden aes le ~, Figure 45. Cestrum racemosum.—A. Branch with flowers and fruits.—B. Flower.—C. Corolla opened to show sta- ON). mens. After Morillo & Manara 2135 (VEN) mostly not evident, the costa and sometimes major lateral veins elevated beneath; petiole flat-topped, distally canaliculate, 0.4—1 cm long, often curving and twisting depending on orientation of the stem, leaf scars discoid; minor leaves wanting. /nflores- cences axillary and terminal, lax panicles of ra- cemes, 10-12 (5.5-9) cm long, not leafy; peduncles purple, 1-2 cm long, 0.5-1 mm thick, glabrous, unbranched or 2-3-branched, ultimate segments resembling the pedicels, 1-7 mm long, the distal Volume 85, Number 2 1998 Benítez & D'Arcy 327 Cestrum and Sessea in Venezuela ones usually shortest; bracts linear, 2-6 mm long, inserted at base of the rachis, the branches, and along the peduncle and rachis branches. Flowers 25-32 mm long, buds purple, pedicels 0.5-5 mm long, articulated, bracteoles filiform, 1-5 mm long, puberulent with weak caducous simple hairs that dry appearing moniliform; calyx tubular, 3.5-5 X 2 mm, subcoriaceous, wrinkled, venation obscure or evident on the lobes, tube 3—4 mm long, 5-den- tate, pilose, teeth deltoid, 0.5-0.8 mm long, mi- nutely ciliate and tufted; corolla greenish white, 26—31 mm long, funnelform, tube slender, slightly contracted above the ovary, then gradually expand- ed toward the apex, (10-)13-15 mm long, mouth 1.5-2 mm wide, teeth narrowly acute, 4-6(-8) mm long, ciliate, folds pubescent; stamens inserted 0.5 m apart, 18-21 mm long, filaments adnate for 14— 20 mm, glabrous, insertion free 0.5 mm, straight, pilose, gibbose (with a tooth 1 mm long), free part mm, anthers globose, 0.5 mm across; ovary 0.6-1 mm across, papillose, ovules 8-10, style 18- 21 mm long, papillose below the stigma, stigma subcapitate, exserted 0.5 mm. Fruit dark purple, ellipsoidal, (8-)10-15 X 6.5-10 mm, stalked, pulp fleshy; fruiting calyx flaring, conspicuously wrin- kled, not splitting; seeds 6-9, dark brown, 3.5—5 m long. [Francey 6: 359.] Figure 49. Cestrum salicifolium is a slender treelet with arching crown and branches and narrow, membra- nous leaves. The inflorescences are open and pen- dent. Specimens are often very like narrow-leaved examples of C. bigibbosum, but the bracteole (or articulation) along the flower stalk is distinctive. With its saliciform leaves and aspect, this spe- cies appears to be a rheophyte, adapted to inun- dation in periodic torrents that flood narrow water- courses. А specimen of Cestrum salicifolium, labeled as having been collected by Sintenis in November 1886 in open woods at Bayamon, Puerto Rico, is deposited at Hamburg (HBG). That the species is otherwise known only from Venezuela, where it is confined to narrow ravines, casts doubt on the prov- enance of this specimen. Although Sintenis is not known to have visited Venezuela, the specimen is labeled *ex Herbario Reineck," a possible setting for a mix-up in label data. Cestrum salicifolium var. angustifolium Dunal was placed in synonymy of Francey (1935: 359). Distribution (Fig. 50). Carabobo, Miranda, and the Distrito Federal. Cloud forests and ravines; 00-1800 m. Probably endemic to the Caribbean region of Venezuela. Phenology. Flowering takes place mainly in November and December and fruiting in December and April, but some flowering takes place in other months. Flowers are open at night, with the corolla lobes spreading, and are closed during the day. hey are strongly scented when open. Nectar was not detectable. Representative specimens seen. VENEZUELA. Ara- gua: Carretera Магасау-Сһогопі, Benítez et al. 4907 (MY). Carabobo: Cuenca hidrográfica del Río Morón, carretera hacia La Justa, Díaz 526 (MO). Miranda: Que- brada de las Comadres cerca de Las Mostazas, Allart 254 NY, US, VEN). Distrito Federal: Cordillera del Avila above Caracas, Steyermark 55008 (MY, VEN). ~ 27. Cestrum scandens Vahl, Eclog. Amer. 1: 24. 1797. TYPE: Colombia. Santa Marta: von Rohr s.n. (holotype, C not seen, = IDC microfiche, = F photo 22927). Cestrum paniculatum Kunth, in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 62. 1818. TYPE: Venezuela. Dis- trito Federal: banks of river Guayre near Caracas, INOV. 328 Annals of the Missouri Botanical Garden ) C| Di; ге 1 ст 2 | E Figure 47. Cestrum reflexum.—A. Branch with flowers and fruits.—B. Flower.—C. Style.—D. Corolla opened to show stamens. After Bernardi 7127 (MER). Volume 85, Number 2 1998 Benítez & D'Arcy 329 Cestrum and Sessea in Venezuela М RÀ Figure 4 8. Cestr > style. After Luteyn 6171 (NY). alt. 420 hex, Humboldt s.n. (holotype, B-W 4453, = IDC microfiche, = F photo 2989 Cestrum laxiflorum Dunal, in A. DC., Prodr. 13(1): 655. 1852. TYPE: Venezuela. Moritz 212 (holotype, G- C, = IDC microfiche; isotypes, B destroyed, = F photo 2985, BM). Cestrum scandens var. terminale Dunal, in A. DC., Prodr. 1): 665. 1852. Cestrum terminale (Dunal) Pittier, — strum ruizteranianum.—A. Flowering branch.—B. Flower.—C. Corolla opened to show stamens and N J. Wash. Acad. Sci. 22: 33. 1932. TYPE: Colombia. Santa Marta: Bertero s.n. (holotype, G-DC, = IDC microfiche). Cestrum perilambanon Loes., Verh. . Vereins Prov. Brandenburg 65: 98. 1923. TYPE: Guatemala. Seler 3381 (B? destroyed; F, fragment). Climbing shrub 2.5—4 m tall, main stem erect, 330 Annals of the Missouri Botanical Garden VOI ум TL — 9! —— aco LAL is | LZ “тт = == Figure 49. Cestrum salicifolium.—A. Flowering branch.—B. Flower.—C. Corolla opened to show stamens.—D. Fruits. After Manara 65436 (MY). Volume 85, Number 2 Benítez & D'Arcy 331 1998 Cestrum and Sessea in Venezuela ай w d ENSE. PER. in C. reflexum the peduncle forms the whole stalk with the bracteole at the top. e | Duplicates of Moritz 212 at herbaria not noted m - " _ above are other species. See note pertaining to Ces- Figure 50. Cestrum salicifolium. Distribution in Ven- uela. branches high-climbing or decumbent, terete, most- ly glabrous; pubescence of reduced simple hairs. Leaves ovate, 5.5—13 X 2.4–8 cm, apically acute or acuminate, basally rounded, margin plane or slight- ly revolute, membranous or subcoriaceous, shiny on both sides, glabrate, veins 6—9 on each side, some- times puberulent; petioles canaliculate, 7-17 mm long, glabrate; minor leaves wanting. Inflorescences axillary or terminal, many-flowered, lax, scandent compound racemes, often dangling, axes 8-15 cm long; peduncles ca. 1 mm long, bracts s dn 1.5 cm long. Flowers nocturnal, fragrant, 25-35 m ав pedicels 1.5-3 mm long, glabrous; bad linear, 0.5-3 mm long; calyx 3.5-5.3 X .5 mm, costate and rugose, glabrous outside, pilose within, tube 3-3.5 mm long, teeth 0.5-1.8 ciliate, especially within; corolla yellowish white or pale green, sometimes with purple areas outside, 23-34 mm long, narrowly funnelform, tube narrow, mouth 2-3.5 mm wide, lobes pilose, 6-8 mm long; stamens 17-20 mm long, filaments adnate for 15— 19 mm, sparingly pilose 1-3 mm above the base, insertion straight, smooth, free part 0.5-1.2 mm, anthers rotund, 0.5-1 mm across; ovary 1-1.2 X 0.5 mm, glabrous, disk inconspicuous, ovules 8—9, style 16-21 mm long, pilose toward the apex, stig- ma dilated or subcapitate. Fruit color unknown, ob- ovoid, 11-12 X 7.5—8 mm wide; seeds 6—8, brown, 5—5.5 mm long (after Nee 3627). [Francey 6: 268.] Figure mm long, erect, In this species and in Cestrum reflexum, the pe- duncle and pedicels appear as a single continuous stalk, their identities delimited by an articulation and bracteole. In Cestrum scandens, the basal, pe- duncular portion is only about 1 mm long, and the pedicel base continues about 1 mm further, while trum venezuelense under C. bigibbosum (above). Distribution (Fig. 52). Aragua, Barinas, Cara- bobo, Falcón, Lara, Miranda, Yaracuy, Zulia, and the Distrito Federal. Gallery forests, and deciduous, semideciduous, and low evergreen forests; 60-1200 m. Also in Mexico, all Central American countries, and Colombia. Phenology. Flowering is from November to pril and fruiting from January to April, with max- ima in Marc Common names. Iguanito Blanco. Representative specimens seen. VENEZUELA. Ara- gua: El Limón, Ferrari 221 (MY). Barinas: Isla Mapora, Reserva Forestal de Caparo, Hernández 1189 (MER). Car- bo: Canoabo, Trujillo 6147 (MY). W - E 5 e. 5 M ~ 8 US, VEN). Yaracuy: Quebrada Berracón, 3 km де AI- barico, la carretera hacia Aroa, Manara & Vera s.n. (MY- 28669). Zulia: Vía entre El Pensado y Las Tres Marías, Bunting & Arboleda 8726 (MO, VZU). Distrito Federal: El Valle, Arteaga 251 (CAR 28. Cestrum schulzianum Francey, Candollea 6: 272. 1935. TYPE: Venezuela. Amazonas: near San Carlos de Río Negro, Spruce 2974 (syn- types, BR not seen, G not seen, — F photo 28372, NY, W not seen). Shrub 1.5-3 m tall, young stems pubescent, branches ridged, leaf scars slightly enlarged; pu- bescence of simple, moniliform ascending and crumpled hairs. Leaves narrowly ovate or narrowly elliptical, sometimes slightly asymmetric, 12-16 X 2.5—4 cm, apically long attenuate, basally acute or slightly cuneate, margin slightly revolute, subcor- iaceous, undulate, glabrate, sometimes with fine pubescence on the main veins, veins 6-15 on each side, arising at (60°-)70°(—75°); minor leaves not evident. Inflorescences short axillary or terminal ra- cemes, peduncles 3-5 mm long, pubescent with very small ascending hairs; bracts 1-3 mm long, foliaceous, pubescent. Flowers nocturnal, fragrant, 28-32 mm long, pedicels 1 mm long or obsolete; calyx tubular, 4—5.5 X 3–3.5 mm, puberulent when young, glabrescent, slightly zygomorphic, costate, the veins salient, tube 3—4 mm long, teeth 1–1.5 mm long, ciliate; corolla pale yellow-green, 27—31 mm long, narrowly funnelform, tube gradually ex- panded toward the apex, contracted below the limb Missouri Botanical Garden Annals of the 332 ~ n, =- е, — —— —À — ——— ——— ——— usi енем Нн ДИП T MR AUT TTE / - DOE B ДО ий ICE ME AUI E ИҢЕ йг ERIS TIRAGE Udo acidum A —— ot Ius VE улице мраве април EDI E EN A лас > aem im i ee TAA. VM Mi iuh C рит ee pma pd m OS C E DY 9g | ^N hs. b Cestrum scandens.—A. Branch with flowers and fruit. —B. Flower.—C. Corolla opened to show stamens. After Trujillo 6147 (MY). Figure 51. Volume 85, Number 2 1998 Benítez & D'Arcy 333 Cestrum and Sessea in Venezuela т e, m 8, В d 2° 20° Су “ Су во“ 0 ° 7 P e . “ ГЫ e e A ҒЫ e 52. Cestrum scandens.—A. Distribution in Representative occurrences outside of Venezuela.—B. ezuela. and then expanded, mouth 1-2 mm wide, lobes 5— 7.5 mm long, very narrowly ovate, acuminate, pu- bescent outside; stamens 18. m long, fila- ments adnate for 17-19 mm, base pilose to 13-15 m up, insertion straight, smooth, free part 0.5-3 mm, anthers globose, 0.5 mm across; ovary 1-1.5 mm across, disk inconspicuous, glabrous, ovules 7— 8, style 18-22 mm long, filiform, papillose below the stigma, stigma subcapitate, exserted 0.5 mm. Fruit dark purple, subglobose, contracted at the base, 11-12 х 8-9 mm, pericarp thin; seeds 5—7, brown, 5—6 mm long. [Francey 6: 272.] Figure 53. This species is recognizable by its lanceolate, often narrow, firm leaves with even venation nearly perpendicular to the midvein, and its large corolla lobes. The flowers tend to be aggregated in dense, several-flowered clusters. Distribution (Fig. 35). Amazonas. Evergreen rainforests and “morichales” groves occurring on sandy substrates with high wa- (successional palm ter table); 100 to 400 m elevation. Apparently en- mic. Phenology. Flowering specimens have been seen from April and May. Fruiting data are lacking. Common names. СајесШо hoja fina. АЙ абалы specimens seen. VENEZUELA. Ama- zonas: San Carlos de Río Negro, Liesner 7134 (MO, VEN); Río D рЫ Chapazó Solano, Morillo et a. 04 (MY, VEN); Rí la desembocadura de Y Rio Casiquiare y San Carlos de Rio Negro, Morillo et al. 4040 (VEN); Rio N Paleta y El Сайо de la División, W y S « Morillo et al. 4105 (MY, VEN); San Carlos de Río Negro, Stergios & Aymard 7311 (PORT); Río Casiquiare, entre la boca y la piedra Guachapita, Stergios & Aymard 7358 (MO); Bajo Casiquiare, entre la boca del Pasimoni y Porv- enir, Stergios & Aymard 7606 (PORT). 29. Cestrum strigilatum Ruiz & Pav., Fl. Peruv. 2: 29, pl. 156. 1799. SYNTYPES: Peru. Po- zuzo, Chinchao and Cuchero, Ruiz & Pavon s.n. (B not seen, — F photo 18394, HAL). Сезігит calycinum Kunth, in Humb., Bonpl. & Kunth, Nov. Gen. Sp. 3: 58. 1815-1816. Cestrum strigila- ч һех, Ай: d s.n. (holotype, B-W 4461 not een, icrofiche, — F photo 2998). Pu viridiflorum Ноок., Bot. Mag. pl. 4022. 1843. TYPE: Brazil. Porto Alegre. Tweedie s.n. (holotype, Cestrum cancellatum Dunal, in A. DC., Prodr. 13 (1): 657. 2. TYPE: Peru. Poeppig 3080 (holotype, G-DC, = IDC microfiche; isotypes, B destroyed, = F photo 2969, F). Cestrum unibracteatum var. B brachystachys Dunal, in A. DC., Prodr. 13(1): 657. 1852. TYPE: Brazil. Circa Cube ubique da Silva Manso M а G-DC, = [DC microfiche, = F E 23 Cestrum unbrateatum Dunal, in b Prodr. 13(1): 656. 1852. SYNTYPES: Peru. de hero, Dombey s.n. — [DC microfiche, — F photo 6899; MPU not séen); Poeppig 96 Dd ois as "C. longifol- Ruiz av. " not se Cestrum lundianum Dunal, in A. DC. Prodr 13 (1): 658. | 5 TYPE: js Sancti Pauli, Lund 34 (holo- e, G-DC, DC microfiche, — F photo 6898). fan 2 var. laxiflorum Kuntze, Revis. Gen. l. 3 (2): 220. 1898. SYNTYPES: Argentina. Oran, Lorentz £ Hieronymus s.n. (NY); Bolivia. Jungas [Juntas] collector un n own. Cestrum а Rusby, Bull. New York Bot. Gard. 4: 425. 5. TYPE: Bolivia. Bang 2516 (holotype, NY not s ee Sessea bue Rusby, Bull. New York Bot. Gard. 8 (28): 119. 1912. TYPE: auda Apolo, 4800 ft., Williams 2449 (NY not see Cestrum calycinum var. AN Francey, Candollea 6: 2. 1935. SYNTYPES: Paraguay. Villa Encarna- ción, Bettfreund 131 (place of deposit not indicated); es ntina. how p Niederlein 268b (С not — F photo 2998). 334 Annals of the Missouri Botanical Garden A | : ee E DNS zA A 1: hi ы: IM HE ы: ы | E: Figure 53. Cestrum schulzianum.—A. Flowering branch.—B. Flower.—C. Corolla opened to show stamens.—D. igure Style and stigma. After Morillo et al. 4004 (VEN) Cestrum strigilatum var. tenuiflorum Етапсеу, Candollea 6: 144. 1935. TYPE: Ecuador. Balao, Eggers 14274 (lectotype, here designated, W; isolectotype, US frag- ment). Cestrum aristeguietae Steyerm., Acta Bot. Venez. 6: 86. 1971. TYPE: Venezuela. Carabobo: along Río San Gean, 2 km below planta eléctrica, S of Borburata, 350 m, Steyermark & Steyermark 95463 (holotype. VEN; isotypes, NY, P, US). Sprawling or climbing shrub or tree, 2—4 m tall, 10 cm DBH, much branched, branches elongate, tomentose; pubescence of dendritic and stellate hairs. Leaves usually not malodorous, ovate to nar- rowly ovate, 11-24,5 X 5.5-11 ст, apically atten- uate, short-acuminate, base rounded or truncate, margin revolute in firmer leaves, membranous to subcoriaceous, above bright green, gray-green be- neath, pubescent, especially on the main veins, lamina glabrescent, sometimes appearing pustular, Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 335 tomentose beneath, veins 5—8, prominent on each side, veins sunken above; petiole terete, 0.9-1.5 cm long, tomentose; minor leaves wanting. /nflores- cences simple axillary or terminal racemes, some- times short, lateral racemes, peduncles 1.5—4 cm long, pubescence dendritic or stellate, bracts linear, 6—7 mm long, stellate pubescent. Flowers diurnal, fragrant?, 30-36 mm long, sessile; calyx firmly tu- bular, 8-13.5 X 3-3.5 mm, 5-costate, the costas salient, membranous, densely stellate pubescent outside, glabrous within, tube 5-7 mm long, 5- toothed, teeth cuspidate, unequal, 3—6.5 mm long; corolla pale green to white, 29-35 mm long, паг- rowly funnelform, tube slightly curved, slightly con- tracted below the ovary, gradually expanded up- ward, tomentose, especially in the upper half, mouth 2.5-3 mm wide, the lobes narrowly ellipti- cal, 5-8 mm long, apically subacute, folds stellate pubescent; stamens equal, 20-25.8 mm long, fila- ments glabrous, adnate for 18.5-24.5 mm, insertion straight, smooth, free part 1-1.2 mm, anthers spherical; ovary 1 X 0.8 mm, glabrous, disk incon- spicuous, ovules 4-6, style green, 24-25 mm long, hirsute below the stigma, stigma green, capitate. Fruit maturing from white to purple, ellipsoidal, 6— 10 X 3-5 mm wide; fruiting calyx accrescent, 14— 20 mm long, often splitting along one side; seeds 3—7, light brown, 5—5.5 mm long. [Francey 6: 137, 142.] Figure 54; Benítez de Rojas, 1974: 89, figure 22, as C. aristeguietae. Cestrum strigilatum 18 distinct in its uniform pu- bescence, large, almost spathaceous fruiting caly- ces, and slender, pubescent corollas. The unpublished name “Cestrum longifolium” was used by Ruiz and Pavón on a specimen of C. strigilatum from Peru that was later seen by Dunal (1852: 657). Cestrum impressum Rusby was placed in synonymy by Francey (1935: 137). Sessea rugosa 1 was placed in synonymy by Francey (1935: 142). The chromosome number for this species was re- ported as 2n = 16 by Berg and Greilhuber (19934). Distribution (Fig. 55). Amazonas, Aragua, Car- abobo, and Táchira. Evergreen moist riverine and gallery forests; 300-1600 m. Also in Costa Rica, Panama, Colombia, Ecuador, Peru, Brazil, Bolivia, Paraguay, and northern Argentina. Phenology. Collected in flower from January to April and in fruit in March and April. Representative specimens seen. VENEZUELA. Ama- zonas: | Montserrat, Alto Orinoco, Croizat 666 (NY). Aragua: P. Nacional Henri Pittier, Benítez & Aguil- era 4691 (MY). Carabobo: Borburata, Aristeguieta 1165 (MO, NY, VEN). Táchira: Palo Grande, Distrito Lobatera, Benítez de Rojas 1269 (GH, MY). 30. Cestrum tillettii Benítez & D'Arcy, Novon 5: 315. 1995. TYPE: Venezuela. Zulia: headwa- ters of Río Guasare, Distrito Perijá, Sierra de Perijá, Serranía de Valledupar, environs of Campamento Frontera V, along international boundary 2700-3300 m, 10?23'07.8"N, 72*52'42.5"W, Tillett 747-1021 (holotype, MY; isotypes, AAU, MO, MYF, VEN). Tree 2—3 m tall; stems brown, glabrous, striate and scarred; pubescence of reduced, simple, glan- dular hairs. Leaves narrowly elliptical, 4-6 X l- 1.7 cm, slightly acute apically, the tip obtuse, ba- sally narrowly cuneate, margin conspicuously rev- olute, firm, subcoriaceous, dark green, shiny above, dull beneath, glabrous, veins 6—8 on each side, the major veins sunken above, salient beneath, minor veins impressed beneath; petiole canaliculate, 3—7 mm long, glabrous; minor leaves wanting. /nflores- cences axillary, congested, the main axes tomentose, bright matte green, 1.5-3 cm long, the peduncles 5-7 mm long, pubescent, thickened, with circular scars from fallen flowers, flowers few per node. Flowers nocturnal?, 15-19 mm long, sessile (ped- icels obsolete) with faint, sweet fragrance during the day; calyx yellow-green, flushed distally with dark purple, tubular, thick, costate, the costas es- pecially conspicuous distally, slightly pubescent outside with hairs slightly thickened near the base, 4.5—5.5 X 2-2.5 mm, 5-toothed, the teeth 1.5-2.5 mm long, narrowly triangular, the apex pubescent; corolla yellow-green and purplish, drying light with darker lobes, clavate, 15—19 mm long, tube 13-16 mm long, slightly expanded toward the apex, the throat not noticeably contracted, mouth ca. 1.5-2 mm wide, lobes 2-2.5 mm long, the folded margins puberulent; stamens 10—13 mm long; filaments white, adnate for 6-9 mm, the insertion geniculate- tumid, slightly pilose, 1 mm long, free part 4.5 mm, anthers brown, spherical, 0.5 mm across; ovary ovoid, glabrous, smooth, 1 mm across, ovules 5, style 12-13.5 mm long, papillose 1-2 mm below the stigma, exceeding the stamens by 1.5 mm, stig- ma green. Fruit shiny blue-black, ovoid, 9 X 6 mm, pericarp thick, opaque; seeds 6—7 per fruit, brown, 3-3.5 mm long. Figure Cestrum tillettii has more or less congested small leaves of uniform appearance on relatively thick, rough branches. The congested flowers are situated among the leaves and close to the stems. Distribution (Fig. 23). Zulia. Slopes on lime- 336 Annals of the Missouri Botanical Garden ч МОМ eene qoos a ) pue ! Figure 54. 2. e —a. Flowering branch.—b. Fruit enclosed in persistent calyx.—c. Corolla opened to show stamens.—d. e. Seed opened to show embryo. After Benítez 1269 (NY). Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 337 A Figure 55. Cestrum strigilatum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. stone; 2700—3620 m. Restricted to the western bor- der between Venezuela and Colombia. Phenology. Specimens seen were collected in July and were in flower. other collections seen. VENEZUELA. Zulia: Sierra de Perijá, Tillett & Kónig 747-929 (МУ, V E ~ headwaters of Río Guasare, Wood & Berry 88 (MO, VEN). 31. Cestrum tomentosum L. f., Suppl. Pl. 150. 1782. SYNTYPES: Colombia. Mutis 94 (LINN. 258.6, — IDC microfiche), 95 (LINN. 258.7, — [DC microfiche). Cestrum hirsutum Jacq.. Pl. Hort. Shoenbr. 3: 41, pl. 324. 1798. TYPE: West Indies. Cultivated hortus Schoen- brunensis (holotype, *; isotype, B-W 4449, = IDC microfiche, — F photo 3 ig Cestrum 2. Ruiz & Pav., КІ. Peruv. 2: 30, pl 157. 9. TYPE: Peru. Пен ан Сатапа, ык arenosis, v Ruin (lectotype, here V фаш G not seen, = F photo 2984; isolectotyy Cestrum eee. M. gre ens & Galeotti, Bull. Acad. Roy. Sci. Bruxelles 12: 146. 1845. TYPE: Mexico. Colon- ie de Mirador, 3000 pieds, Galeotti 1208 (holotype, BR not seen; isotypes, G, Y, US, W Cestrum moritzii Dunal, in A. DC., Prodr. 13(1): 619. 1852. TYPE: Venezuela. Caracas, е 309 (holo- IDC microfiche, — F photo 2988; B duae. BM, HBG, MO). Cestrum miersianum Wedd., Chlor. Andina 2: 97. 1859. TYPE: Colombia. Sierra Nevada de Santa Marta, 3300 m, Linden 1615 (holotype, G-DC, — IDC mi- crofiche). шш end — Wash. 1 = = ње o uela. Mérida: a u- мене, 0 m, Pittier 12919 (holotype, VEN: iso- type, Cestrum a Pittier, J. Wash. Acad. Sci. 22: 36 1932. TYPE: Venezuela. Vecinidades de Mérida, 1700 m, Pittier 12858 (holotype, VEN; isotype, US). Cestrum miersianum Pittier, J. Wash. Acad. Sci. 22: 37 1932, non Wedd. (1857). Cestrum neomiersianum Benítez, iae Fac. Aen. (Maracay) 7: 90. 1974. TYPE: Venezuela. MESS Mucu- chíes, изо m, А. Jahn 767 (holotype, VEN; isoty- pes. MO, ?NY not seen, Cestrum отете Francey, РИТА б: 169. 1935. TYPE: Ecuador. Tungurahua: vicinity of Ambato, Расћапо 75 (holotype, NY; isotype, US fragment). Cestrum densi orum Francey, Candollea 6: 195. 1935. TYPE: Venezuela. Mérida: between Chacho po and Timotes, Pittier 13294 (lectotype, гак аг һу Benítez & D'Arcy (1995: 317), NY; isolectotypes, F, O, US, N). Cestrum densiflorum var. puberulum Francey, Candollea 6: 196. 1935. TYPE: Venezuela. Trujillo: San Pablo de endoza, Pittier 13323 (syntypes, F, MO, US, VEN). байы, sesseoides Francey, Candollea 6: 395. 1935. TYPE: Colombia. Santander: eastern cordillera be- tween El Robel and Tona, 1500-1900 m, Killip & Smith 19423 (NY). Cestrum прерије Zucc. ex Francey, СапдоПеа 6: 191. 1935. SYNTYPES: Eo Berlin and Mu- nich (B not seen, BR no n). Shrub or tree 1—8 m tall, much branched, stems terete, flexible, tomentose, lenticellate, sometimes subscandent or sprawling, branches often arching or horizontal, brownish tomentose with a yellowish cream hue; pubescence of mostly sessile, branched and some stellate hairs. Leaves malodorous, ovate to elliptical, 4-17 X 2-7.5 cm, apically obtuse, cute or acuminate, basally truncate, rounded or obtuse, narrowly cuneate, margins slightly revolute, sometimes appearing ciliolate, subcoriaceous to membranous, dark green, glabrescent above and then shiny, yellow green, softly tomentose beneath, veins (4-)7-8(-12) on each side, evenly spaced, slightly elevated beneath; petiole canaliculate, 5— 12 mm long, tomentose; minor leaves sometimes present, rotund, to 2 cm long, subsessile, some- times persistent. /nflorescences axillary and termi- nal, mostly among the leaves near the branch ends, many- or few-flowered, congested fascicles, ra- cemes or spikes, much shorter than the leaves, to 6 cm long, sometimes appearing paniculate; bracts Annals of the Missouri Botanical Garden 5 cm Figure 56. Cestrum tillettii.—A. Flowering branch.— B. EN). D. Fruits. Based on Tillett & König 747-929 (VEN) ovate or lanceolate, foliaceous, 5-20 X 3-8 mm, pubescent; peduncles mostly short, occasionally to 3 cm long, slightly longer in fruit, tomentose, clus- ters of 2—6 sessile flowers separated by а 2-6-mm- long rachis. Flowers nocturnal; 12-21 mm long, Flower.—C. Corolla opened to show stamens and style.— buds with dark purple lobes; sessile; bracteoles lin- ear, rarely narrowly ovate, to 12 mm long, mostly tomentose, caducous; calyx cupular, campanulate, or tubular, 3—6.5 X 1-3 mm, outside tomentose except sometimes near the base, the veins some- Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 339 times inconspicuous beneath the pubescence, gla- brescent, inside glabrous, tube 3-5 mm long, teeth narrowly deltoid or obtuse, 0.5-2.5 mm long, nar- rowly triangular, ciliolate, minutely tufted; corolla greenish, whitish, or yellowish, dark purple outside, often drying dark and with 15 fine nerves evident, yellow-green within, 10—19 mm long, glabrate out- side, glabrous within, tube 7-9(-15) mm long, ba- sally slender, expanding gradually or about 2/3 way up, mouth 2.5—4.5 mm wi inside, dark brown or dark purple outside, often colored differently from the tube, 1.5-3.5 mm long, acute, ovate, often ciliolate; stamens inserted equally or 1.5 mm apart, 9-15 mm long, filaments adnate for 4.5-11 mm, mostly pubescent, insertion free 1-3 mm, geniculate, tumid, slightly denticu- late, glabrous or sparingly pilose, free part (2.5-) 4.5—5 mm; ovary bright green, lobed, 0.5-1 mm across, glabrous, disk 0.5 mm long, ovules 4—8, style 8-15 mm long, bright green 1 mm below the stigma, puberulent, stigma capitulate, included. Fruits 16—40 per inflorescence, violet, then purple- black, shiny, ovoid, 6—8 X 3—4 mm, juicy; fruiting calyx slightly accrescent, 5-6 mm, basally multi- nervate, slightly splitting irregularly at the sinuses, often glabrescent; seeds 2—4 per fruit, yellowish brown, 2.5-3.5 mm long. [Francey 6: 171.] Figures ot wide, lobes greenish white In Venezuela, Cestrum tomentosum is usually amply distinct in its overall pubescence and small, dense flower and fruit clusters. Some collections of C. humboldtii from Colombia and eastern Venezue- la have similar leaves, but these can be separated by their larger leaves and glabrate calyces. The name Cestrum verbascifolium Francey is placed in synonymy based on Francey’s (1935-1936) de- scription an Chromosome numbers for this species have been reported as 2n = 16 (Sharma & Sharma, 1958, as C. hirsutum; Nanda, 1962, as C. lanatum). Distribution (Fig. 58). Aragua, Lara, Mérida, Miranda, Monagas, Sucre, Táchira, Trujillo, and the Distrito Federal. Dwarf evergreen forests, ravine banks, dense thickets, roadsides, abandoned coffee plantations; 700 to 2800 m. Also occurring from northwest Sonora, Mexico, through all countries of Central America, and in Colombia, Ecuador, and Peru. Phenology. Flowering occasionally year-round but mostly from January to April. Fruiting is mainly April to May. Common names. Hedionda, Hediondo, Quesil- lo, Puta Vieja, Trompillo, Uvito. A гн specimens seen. VENEZUELA. Ara- Henri Pittier, between Ocumare de la Costa and El Mi rador, Benítez et al. 4876 (MY). Lara: anare, Montecarlo arriba, Badillo 6763 (MY). Mérida: 8 km E of Jají, D'Arcy et al. 18260 (MO, MY). Miranda: Paraíso, vertiente sur, en antiguos cafetales, 1465 m, Meier 3603 (МУ, VEN). Monagas: Praderas between Sa- запа de Las Piedras and las selvas de Cerro Negro, NW of Caripe, Ste) 1816 (F, MY, VEN). Suere: humid trail to кривац. Tate 24 (US), 25 (US). Táchira: entre Cordero y páramo El Zumbador, Romero 750 (MY). Tru- jillo: La Puerta, entre La Lagunita y la quebrada El Por- tachuelo, Ruiz-Terán & Dugarte 11999 (MERF, MY). Dis- trito Federal: Antímano, near Caracas, Archer 2991 JS). Y 32. Cestrum tubulosum Sendtn., in Mart., Fl. Bras. 10: 207. 1846. SYNTYPES: Brazil. Sáo Paulo: near city of Ytu, da Silva Manso 336 (BR); Campinas, Severin 168 (S not seen). Cestrum rojasianum Hassl., Керегі. Spec. Nov. Regni Veg. 9: 120. 1910. TYPE: Paraguay. Río Aquidaban, Ко- jas 10033 (holotype, G not seen, — F photo 28371 < Shrub or small tree to 2 m tall, branches terete, pubescent, especially toward the tips, internodes 10-15 mm long; pubescence of small, simple or sparingly branched, stout yellowish or reddish, as- cending, usually dense, curved, sometimes gland- tipped hairs. Leaves ovate or elliptical, strongly up- folded from the costa, 5-9 X 3—4.8 cm, attenuate upward from the middle, basally obtuse, rounded, slightly cordate, margins revolute, subcoriaceous, rigid, matte dark green above, lighter beneath, gla- brous above, beneath with few ascending, gland- tipped hairs, veins 5—9 on each side, elevated beneath, minor (5th order) venation forming well- developed areoles, less visible beneath; petioles 2— 6 mm long, margined above by leaf base; minor leaves wanting. Inflorescences few-flowered in the leaf axils; axes stout, short, densely pubescent with long, branched hairs. Flowers diurnal, strongly fra- grant, 25—32 mm long, sessile, bracteoles linear to narrowly ovate and foliaceous, 3-5 mm long, bescent; bracts ascending, resembling the leaves, caducous; calyx cupular, 3 2 mm, costate, glabrous outside, tube 3-4 mm long, 2.5 mm, teeth 0.5-2 mm long, slightly unequal, sutures rounded, ciliate, apically mucronate; corolla light green, 24— 31 mm ed upward, long, tubular, tube cylindrical and expand- slightly contracted. below the limb, mouth 3.5-5 mm wide, lobes narrowly ovate, api- cally tufted, sinuses basally ciliate, 4–6 mm long; stamens equal, 19.5-21.5 mm long, filaments ad- nate for 15-20 mm, basally pilose, straight, smooth, free part 1.5-2 mm, anthers spherical, 0.5 mm across; ovary globose, 1.2 mm across, glabrous, disk conspicuous, ovules 9-11, insertion 340 Annals of the Missouri Botanical Garden pU У; ay |» ка у> OG LY MES Sra ae d 5 dy бі, d А ае" A у E N BE > i ae ы NN Ue dc, Figure 57. Cestrum tomentosum.—A. Vegetative branch showing minor leaves.—B. Flower.—C. Flower beginning to open.—D. Corolla opened to show stamens.—E. Flowering branch.—F. Fruits. A, B, D, E, F after Morillo 3017 (VEN). С after Benítez 4667 (MY). Volume 85, Number 2 1998 Benítez & D'Arc Cestrum and Sessea in Venezuela 341 3, s o Figure 58. Cestrum tomentosum.—A. Distribution in Venezuela.—B. Representative occurrences outside of Venezuela. style filiform, 16-20 mm long, papillose below the stigma, stigma peltate, inconspicuously 4-lobed. Fruit green?, ovoid, 8-11 mm, pericarp thin; seeds 3—4, olive brown, 6–7.5 mm long. [Francey 6: 208.] Figure 59. Hairs on Cestrum tubulosum are much like those of C. neblinense, but denser, and the leaves are broader, mostly more than 2 cm wide. Our use of this name is based on a photo of the type of Cestrum rojasianum and Francey's (1935— 1936) placement of C. rojasianum in synonymy un- der C. tubulosum. Distribution (Fig. 60). Amazonas and Bolívar. Gallery forests; 100 to 300 m, and riverine woods associated with tepui vegetation; 1100 to 1400 m. Also in Paraguay and Brazil. Phenology. Тһе collections seen were flowering in January and March and fruiting in March. All remaining specimens seen. VENEZUELA. Ama- zonas: Dept. Atures, Río Corocoro, W of Serranía de Yu- tajé, Holst & Liesner 31734 (МО); Dept. Atabapo, Mara- huaca, Liesner 18464 (MO); Serraní utajé, Río Manapiare, Camp Yutajé, Maguire & Maguire 35094 (MO, NY, 05). Bolívar: Distrito Седећо, Serranía de Guanay, Río Parguaza, Huber 11033 (MY, MYF). Sessea Ruiz & Pav., Fl. Peruv. Prodr. 21. 1794. TYPE: Sessea stipulata Ruiz & Pav. Unarmed trees; pubescence of simple or branched hairs. Leaves simple, entire, pinnately nerved, mostly glabrate above; mostly short petio- late; minor leaves present or not. Inflorescences ax- illary and pseudoterminal, few- or many-flowered racemes, spikes, or cymes, often large and appear- ing paniculate; bracts often present. Flowers mostly 5-merous, pedicellate, bracteolate; calyx small, cu- pular or tubular, mostly shallowly lobed; corolla narrowly tubular, much exceeding the calyx, lobed, the lobes narrow, shorter than the tube, spreading or reflexed when open; stamens inserted in the co- rolla tube at similar levels, the insertion levels varying greatly in different species, the adnate por- tion mostly evident from the corolla base, the in- sertion variously pubescent or tumid, anthers small, situated together at the corolla mouth; ovary mostly shorter than the calyx, 2-locular, ovules 4-16, style slender, style capitate or variously lobed, small, in at least some species oblique on the style apex. Fruit a. narrowly ellipsoid or ovoid, terminally de- hiscent capsule; seeds varying in number, ovoid, surrounded by a hyaline wing; embryo straight. The name Sessaea Endlicher (1838: 668) is an orthographic variant of Sessea Ruiz & Pav. 1. Sessea corymbiflora Goudot ex Rich. Taylor & R. Phillips, Philos. Mag. Ann. Chem. 3: 132. 1828. TYPE: near Bogotá, Goudot 1 (lec- totype, designated by Benítez & D'Arcy (1993: 324), P; isolectotypes, K, G-DC not seen, — IDC microfiche, С). — Sessea corymbosa Miers, Hooker's J. Bot. Kew Gard. Misc. э: 156. 1846. TYPE: Bogotá ad Barro Blanco, Gou- dot 1 (holotype, K not seen; isotypes, G-DC not seen, = IDC microfiche, P. С). Cestrum atrovirens Dunal, in A. DC., Prodr. 13(1): 648. 1852. Sessea atrovirens (Dunal) B. D. Jacks., Index. Kew. 2: 892. 1895. TYPE: Ecuador [Peru]. Hartweg 1309 (holotype, G pes. B not seen, K not seen, Quito, — F photo 8573; isoty- Shrubs or trees to 6 m tall; unarmed, pubescence of reduced moniliform simple hairs to 0.5 mm long, glabrous on most parts; twigs angled from the pet- iole bases and often striate-furrowed. Leaves 11—15 per twig, not odorous, perennial, elliptical, occa- sionally ovate or obovate, mostly 9-13(-20) x 3- Annals of the Missouri Botanical Garden 342 SN dí Q Шш O X Р * МАУ | > | АНА E | AV / ^ MA ED LAT di Ко» and ering branch.—B. Flower.—C. Corolla opened to show stamens l rit (VEN). ‘igure 59. Cestrum tubulosum.—A. Flow tyle.—D. Fruiting branch. After Huber 3245 Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 343 —. п Representative occurrences outside of Figure 60. Cestrum tubulosum.—A. Distribution id Venezuela.—B. Venezuela. 5(-6) cm, apically obtuse or acute, sometimes slightly short-acuminate, basally obtuse, margins = il sometimes slightly revolute, coriaceous or membranous, costa drying dark above, elevated and excurrent beneath, glabrous or with occasional re- duced trichomes near the base, major veins 17-20, mostly evenly spaced 3—6 mm apart, arcuate at 30°—40° to the costa, plane above, finely reticulate beneath, drying reddish or slate gray; minor leaves mostly wanting, when present oblong, to 6 cm long and resembling the major leaves; petioles mostly slender, 5-25 mm long, slightly scurfy with reduced trichomes, drying darker than the leaf, especially at the base. Inflorescences lax, crowded, terminal corymbs to 6 cm long; peduncles and pedicels dry- ing dark, pedicels wanting, bracts few, scattered among the inflorescence and resembling reduced leaves; pedicels obsolete; bracteoles 1.5 mm long, narrowly ovate or linear, glabrate with sparse, re- duced, glandular-appearing trichomes, soon cadu- cous. Flowers numerous (ca. 83), crowded, malo- dorous, 20-22 mm long, sessile; calyx dark green, tubular-obconical, 5-6 X 4 mm, glabrous outside, the costas sometimes conspicuous, the teeth sinu- ate-deltoid 0.5 mm long, minutely ciliate, but not tufted apically, pubescent within; corolla green with purple areas, 14-21 mm long, exserted 9-13 mm from the calyx, tube 13-21 mm long, basally slen- der, 1 mm wide, expanding about % way up to 4— 5 mm wide, glabrous outside, the fine nerves in- conspicuous, glabrous within, mouth 2.5—3.5 mm wide, lobes 1-3 mm long, obtuse or rounded, gla- brous outside; stamen insertion levels subequal, fil- aments adnate for 6—7 mm, 6 mm free, the free insertion 1.5-2 mm, distal free portion 5 mm long, insertion tumid, a few minute hairs present just be- low the insertion, distal portion glabrous; style 12— 13 mm, glabrous, stigma unequally bilobate, the two lobes forming a mouth flanking the stigmatic surface; ovary glabrous, disk inconspicuous, ovules ruit a woody, apically dehiscent capsule, 6— 7 mm long, the valves 4, linear, 2.5 mm wide at the base; fruiting calyx slightly accrescent, 6—7 mm, enclosing the base of the capsule, splitting irregularly; seeds 5-12, appearing flat and 14—15 mm long overall, the seed body ellipsoidal, 3—4 х 1 mm, chestnut-brown, surrounded by a light green, membranous, minutely reticulate, oblong wing ex- tending 3—4 mm beyond each end of the seed and 0.25 mm on each side, the ends pointed or round- ed, sometimes with one or more narrow wings in another plane; embryo white, 2.5-3 mm long, the hypocotyl straight, terete, the epicotyl laminar, broadly elliptical, forming % the length of the em- bryo. Figure In the absence of fruit, Sessea corymbiflora is similar to Cestrum lindenii, with its large leaves with many veins and in the general appearance of its flowers. However, the flowers of S. corymbiflora are shorter, and the stigma is placed obliquely on the style. The spent capsules are persistent for sev- eral months, and from a distance the trees resemble arborescent Asteraceae. Distribution (Fig. 62). Mérida, Táchira, and Trujillo. Subparamos; 2200-2900 m. Also in Co- lombia and Ecuador. Phenology. Flowering collections have been seen from February, July, and September. All specimens seen. VENEZUELA. Mérida: Páramo Las Nieves, 48 km al sur de Estanquez, Benítez et al. 4839 CAR, MA, MO, MY, P. VEN). Táchira: P. Nacional Los Páramos, carretera Pregonero-El Portachuelo, Benítez et al. 4741 (Е, MER, МЕКЕ, MO, MY, NY, PORT, US, VEN). Trujillo: entre El Alto de Tuñame y Quebrada El Pajarito, Ruíz-Terán & López-Palacios 7552 (MERF, MY). = 344 Annals of the Missouri Botanical Garden EON = 5 cm Figure 61. Sessea corymbiflora. € омади. brane h.—A’. Infructescence.—B. Flower.—C. Corolla ке to show stamens.—D. Fruiting capsule.—E. Seeds. Seed showing body and wing.—G. Embryo.—H. Style. АН afte Ruiz-Terán 7552 (MERF). A' after Benítez 4741 pos Literature Cited ceedings of the 32nd New Zealand Weed & Pest Control Bins әгепсе. New Zealand Weed and Pest Control So- Andrade, S. O. 1960. Estudios sobre a toxicidade de d 'rston y New Zealand. Sessea brasiliensis. Arq. Inst. Biol. (S. Paulo) 27: m йе? с tojas, С. 1974. Los géneros de las Sola- 191-196 naceae ^ Vene 2 Ravinia Fac. Agron. (Maracay) Atkinsor С. С & Т. K. James. 1979. Preliminary study 7(3): 25-108. on is tol of Red Cestrum. Pp. 289-291 in Pro- ————& №. С. D'Arcy. 1993. Nomenclature of Sessea Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 345 3, ө А Figure 62. Sessea corymbiflora.—A. Distribution in Venezuela.—B. Representative. occurrences outside of Venezuela. corymbiflora (Solanaceae) and its occurrence in Vene- zuela. Novon 3: 324—327. & species of Cestrum (So- lanac Mu and synonymy did two widespread species. Novon 5: 311-317. & 98. пеш“ in Venezuela. /n M. Nee, D. Symon, M Jessup & J. Hawkes (editors), кете сеге ТУ: апсев іп Biology and Utilization. Royal Botanic _ Kew (in pre Bernardello, L. M., L. Galetto, J. Тоо Е. Eu gu 1994. Floral pe chemical composition of so cies fom Reserva Río Guajalito, Ecuador. pa A 26: 113-116. Berg, C. € J. Greilhuber. 1992. Cold-sensitive chromo- some regions and their relation to constitutive hetero- chromatin in Cestrum parqui (Solanaceae). Geome 35: 921-930. a. Cold-sensitive chromosome regions and кекенди in Cestrum (Solanaceae): C. ооа С. е M апа С a Р]. Syst. Evol. 185: — 993b. Cold-sensitive chromosome regions and TE in Cestrum aurantiacum (Solanaceae). Pl. Syst. Evol. 185: 259-273. . 1922. Zur Gattung Sessea. Repert. Spec. Nov. Regni Veg. 18: 199-215. Danert, S. 1958. Die Verzweigung der Solanaceen im re- productiven — Abh. Deutsch. Akad. Wiss. Ber- lin. Kl. Chem. 6: 1-292. D'Arcy, W. С. b Jacquin е some notes on their аро Taxon 19: 554—560 ——, 1974 Solanaceae. [п К. " Woodson et al., Flora 70 . Skel и The Biology and {койбошу of the Solanan, cademic Press, London. 997. Red-flowered cestrums and red- M hummingbirds. In M. Nee, D. Symon, J. Jessup & J. G. Hawkes (editors), к IV: Advances in Biology xd ин ~ The Royal Botanic Gardens, Kew (in fia eis C. D. & A. P. Wylie. 1955. Chromosome At- las of Flowering Plants, 2nd ed. Allen & Unwin, Lon- don Dunal, Е. 1852. Solanaceae. In A. De Candolle, Prodro- mus systematis naturalis regni vegetabilis 13(1): 1-741. Paris, France. Dyer, A. К. 1963. Allocyclic segments of chromosomes and the structural heterozygosity that they reveal. Chro- mosoma 13: 547—5 Eichler, A. 1875. es Construirt und Er- ü . Engelmann, Le и 5. L: а а Бали т 668. Fleming, T. H. . The Short- tailed Fruit Bat: A Study in Plant- с пне оци Univ. Chicago Press, Chi- ‘ago. ice J. 1978. Flore Illustrée des Phanérogames de Guadeloupe et = Martinique. Inst. Nat. Recherche 4. Übersicht über ` Gattung Sessea. No- tizbl. x [^ Berlin-Dahlem 8-990. 1936. Monographie du genre Cestrum L. Cute A . 46-398; 7: Freeman, С., W. Н. Reid, J. 5 [T R. Scogin. 1984. Similarity and apparetit convergence in the nectar-sugar composition « of some ы cpolfinated flowers. Bot н az : 2 Gill, L. S. 1972. Жилин numbers in West-Hima- aan bic 4 species II. Bull. Torrey Bot. Club 99: 2 G., n Schrauwen & Н. К. Linskens. 1984. mino acids and sugars in nectar, and yox putative evolutionary Significance Pl. Syst. Evol. 145: 55-77. Haber, W. A. & G. W. Frankie. 1989. A e hawk- moth Pan Costa Rican dry forest Sphingidae. Biotropica 21: 156-172. Jacquin, N. J. 1798. Volume 3. Plantarum rariorum horti caesarel schoenbrunnensis. Vienna. . H. Wasserman, J. N. Shively, A. H. Tashjian, T. D. Brokken & J. F. Morton. 1975. Hypercalcemia and calcinosis in Florida horses: Implication of the shrub, Cestrum diurnum, as the causative agent. Cornell Veterin. 65: 26. Madhavadian, P. 1968. Chromosome numbers in South Indian Solanaceae. Caryologia 21: 343-347. McLennan, M. W. & W. R. Kelly. 1984. Cestrum parqui (green cestrum) poisoning in cattle. Austral. Veterin. J. 61: 289-29]. Mehra, P. N. & K. S. Bawa. 1969. Chromosomal evolution in n v3 propo Evolution 23: 466-481. Nanda, P. ( Chromosome numbers Bel = trees and abria _ pe Bot. Soc. 41: 271- 346 Annals of the Missouri Botanical Garden Nee, M. 1980. каласе I. In Flora de Veracruz. Fasc. 49: 1-191. 2” Мас . Inv. en Mesue R. chloroplast бунет of the Solanaceae: 4. relation- i aracter evolution. Ann. Missouri Bot. Gard. 79: 346-3 uk ‚ R. E. Spangler, L. s «СА. Palmer. d " Phyloge ny ne acier: c 4. 1. of the Solanaceae based on chloroplast DNA. /n M ‚ D. Symon, J. Jessup & J. С. Hawkes (editors), Solanaceae IV: Advances in Biology and Utilization. The Royal Botanic Gardens, Kew (in үрге ess). Overland, L. 1960. Endogenous rhythm in ope i and odor of flowers of Cestrum nocturnum. Amer. 47: 7 Percival, M. 1965. Floral Biology. Pergamon Press, Ox- ford. Pittier, Н. 1932. Studies in Solanaceae. 1. The species of Cestrum collected in Venezuela up to 1930. J. Wash. Acad. Sci. 22: 25-37. Prema, T. Р. . Raghuramulu. 1994. Free vitamin D, metabolites in Cestrum diurnum leaves. Phytochemistry 37: 677-681. Procter, M., P. Yeo & A. in of 7 P pa Romeike, 996. The Natural History s, Portland, Oregon. УМ alkaloids Oc 'currence and systematic 2. іп angiosperms. Bot. Not. 131: 85-96. Ruiz 2. Т. 1993. La кан del polen de Cleome L. ización. Pittieria 21, Sachs, R. M. 1985. Cestrum. Pp n A. H. Hal- evy (editor), CRC Handbook of . Vol. 2. СКС Press, Boca Raton, Florida. Sharma, A.K. ФА. 5 a. 1958. Karyotype studies in estrum as an aid to taxonomy. Genetica 29: 83-100. Ule, E. 1908. Die Pflanzenformationen des Amazonas- Gebietes II. 4. ische Ergebnisse meiner in der Jahren 1900-1903 in Brasilien ws Peru unter- nommenen е Bot. Jahrb. Syst. 40: 398—443 Wasserman 974. Calcium alisurulion and calci- um- mining protein synthesis: Solanum malacoxylon re- es stronti ium inhibition. Science 183: 1092-1194. 978. The nature and mechanism of action of the calcinogenic principle of Solanum malacoxylon and Cestrum к and a comment on Trisetum flaves- cens. Pp. 545-553 in R. Е. Keeler, К. Н. van Kampen L. E 4. Effects of Poisonous Plants on Livestock. Academic Press, New York. White, R. H., R. D. Stevenson, R. R. Bennett, D. E. Cutler & W. A. Haber. 1994. Wavelength discrimination and the role of ultraviolet vision in the viy behavior of —435 hawkmoths. Biotropica 26: 427 Lisr OF SPECIES . Cestrum acuminatissimum Dunal . Cestrum alternifolium (Jac “ү E. Schulz | estrum aurantiacum Lindl. DN шм — Cs % - A 5 = 3 с- = ES X © - ы 3 5 > ч т 1-4 Francey 7. Cestrum diurnum 1.. 8. Cestr um pr acen (C. V. Morton) Steyerm. & Ma- © E ae humboldtii Francey 10. Cestrum imbricatum Rusby ГІ. Cestrum jaramillanum Benítez & D'Arcy 2. Cestrum latifolium Lam. 3. Cestrum lindenii Dunal 14. Cestrum mariquitense Kunth 15. Cestrum megalophyllum Dunal 16. Cestrum microcalyx Francey . Cestrum neblinense D'Arcy & Benítez 18. Cestrum nocturnum L 19. Cestrum olivaceum Francey 20. Cestrum pariense Steyerm. ; petiolar 22. Cestrum Dunal 23. emosum Ruiz & 24. Cestrum hoe m Sendtn. 25. Cestrum ruizteranianum Benítez & D'Arcy 26. Cestrum salicifolium Jacq. 27. Cestrum scandens Vahl 2 schulzianum Francey 29. Cestrum strigilatum Ruiz & Pav. 30. Cestrum tillettii Dh & D'Arcy 1 „з N ~ С ЫЗ P4 а 53 ры. = 9d = = = Рау. o РА <% = Е S со т E Вч 52 5 = 5 ‹ tomentosum L. 1. 32. Cestrum tubulosum S Ser dtn 33. Sessea corymbiflora Taylor & Phillips ~ ~ iv. е с = = INDEX TO EXSICCATAE Specimens are listed alphabetically by collector, fol- lowed by collector number and herbarium of deposit; the species is indicated by a number in parentheses corre- sponding to the number in the List of Species above. Agostini (МУ-56467) (16); 76 (VEN) (2). . ~ де Agostini 1171 (MY, VEN) (12). Agostini € Fariñas 90 MY, US, VEN) (4). Acosta: et al. 1187 (V А аі 8 (VEN) (7). Allart 254 (NY, US, VEN) (26); 475 . VEN) (2); 480 (VEN) (21). Alston: 6619 (MO, US) - '7098 (BM, US) (29); 7099 (US) (31). Archer 2986 (US) (27); 2987 (US) (27); 3090 (US) (12). Aristeguieta 792 (VEN) (13); 1667 (VE > | 5); 2291 (VEN (4); 2: (VEN) (5); 2525 (F, MER, М) (21); 2765 (VEN) (26); 2769 (US, VEN) (23); 2869 its VEN) (15); 3689 (VEN) (21); 3691 (MO, US, VEN) (31); 3703 (F, MO, US, VEN) (26); 4009 (MO, VEN) (15); 4179 (VEN) (12); 4465 (MO, VEN) (29); 5041 (VEN) (12); 5434 (VEN) (26); 5463 (MO, : 5962 (VEN) (31); 6464 (VEN) (12); 6649 (VEN) (7); 747 (VEN) (31); 748 (VEN) (21); 7757 (VEN) (31); 7812 (VEN) rcs 12460 (VZU) (2). Aristeguieta & Foldats 1504b (VE Aristeguieta & Huber 302 PN ieta et al. 1260 (E MER, MY, US, VEN) (12). Ar- 5-123 (VEN) (15). Arroyo & Ar- м кй 75- 100 (VE N) a 5). AES 107 (CAR) (31); 251 (CAR) (27). iuge 697 (MER, MY) (12). Aymard 2 (MER) js 524 (MY, PORT) (12); Tn (VEN) (12); 1951 (MY, PORT) (12); 4496 (MY) (12); 5536 (MO, MY, PORT, ы (12). Aymard & Cuello 3567 (MY, PORT) (12); 559 ron MO, PORT) (12). Aymard & Ortega 1343 (MY) de (12). Aymard et al. 2222 (MER, Y. PORT) (31); 2747 (MY, PORT) (12); 3741 wi Sonn a Badillo 1726 (MY) (31); 218 (VEN) (31); 5255 (MY) (21); 6629 (MY) (4); 6693 (MY) (4); 6763 (MY) (31); 6818 (MY) (5); 6836 (MY) (5); 6909 (MY) (21); 691 (VEN) (21): 6942 (MY) (21); 7058 (MY) (21); 7078 (MY) (13); 7085 (MY) (21); 720 (VEN) (2); 8004 (MY) (21). Badillo & Badillo 6782 (MY) (5); 7297 (MY) (31). Badillo et al. 7848 (MY) (14). Barclay & Juagibioy 51 (МО) (5). Ваг- сов (MY-67838) (15). Barrios 14 (M 18). Bascopé `- (MY) (31). Belandria 191 (MER) he ла (MER) (14); € Volume 85, Number 2 1998 Benítez & D'Arcy Cestrum and Sessea in Venezuela 347 (MER) (14). Benítez 87 (MY) (18); 91 (MY) ca 95 (MY) (12); 550 (MY) (12): 563 (MY) (12); 1033 (MY) (18); 1034 (MY) (18); 1211 (МТ) әді 1212 (МҮ) (12); 1213 (МҮ) : 1294 (МУ) (31): (МҮ) (26); 3902 (МҮ) (12); 3920 (MY) (12); 3955 5 (MY) 21); 3994 (MY) (16); 4159 (MY) (15); 4230 (MY) (15); 4239 en (12); 4281 hs (2); 4287 M (4); 4522 (МҮ) pee (21): 4740 (MY) (13 : ; MY) (13); 4860 (MY) (21): 4864 (MY) (23); 4870 (MY) (9); 4876 (MY) (31): 4877 (MY) (23); 4885 (MY) (15); 4895 (MY) (12): (MY) (12); 4897 (MY) (14); 4900 (MY) (23); 4901 an) (14): 4907 (MY) (26); 4909 (MY) (4); 4912 (MO, MY) (22); 4913 (MY) (4); 4915 (MY) (15); 4916 (MY) (2); 4997 (MY) (26). 4998 (MY) (4); 4999 (MY) (22); 5036 (MY) (14): 5037 (MY) (12); 5040 (MY) (14); 5046 (MY) (15); 5049 (MY) (14); 5051 (MY) (21); 5056 (MY) (13). Benítez 7 Aguilera 4691 (MY) (29). Benítez & Baldizán 4945 (MY (23). Benítez & Gimenez 4157 (MY) (15). Benítez & Otero 4604. (MY) (13); 4716 (MY) (31). Benítez & Rojas 3083 (MY) (12); 4719 (MY) (23); 4996 (MY) (23). Benítez et al. 3576 (MY) (4); 4110 (MY) (4); 4223 (MY) (15); 4607 sy (31); 4814 (MY) (5); 4741 (F. MER, МЕКЕ MO, MY, NY, PORT, US, VEN) (33); (CAR, MA, MO. MY. de d (33). Bernardi 313 (P) (13); 484 (MER, NY) ( 2045 (MER) pis 2079 (VEN) (5); 3329 (MER, MY) As ) ; 6183 (MER) (21); 6235 (ME RI (4); 7439 (MER) (23); 7 (15); 7760 (VEN) (5). Bernardi et al. Е Е 17200 (С, MO, NY) (6). Berry 3492 (MY) (2); 3663 (MO) (2); 379 (VEN) (2); 439 (VEN) (12); 4807 (MO, МУКЕ) (12); 49 (MO, VEN) (2); 501 (VEN) (2). Bevilacqua 163 (VEN) (2); 414 (VEN) (2). Birschel s.n. (GH) (18). Blanco 24 (VEN) (31). Bond et al. 101 (US) (12). Bono 4022 (MY (4); 5043 (MY) (18); 5082 (MY) (4); 5891 (MY, VEN) (13): | 13). jo a ege 6559 (МО, МҮК, NY) (12). diet 3352 (M Р, U, US, VEN) (5); 3384 (VEN) (21); 3568 (MER, P, SENS (31); 3614 (MER, MO, SN) : 3721 (MER, US, VEN) (14); 3732 (F, MER, US, VEN) (12); 3898 (F, MO, US, VEN) (12); 4029 (F, MER. MO, P, US, VEN) (12); 4072 (MER) (4); 4305 (MER, MO, US, VEN) (2); 4652 (F, MO, VEN) (5). Bri- ceño 40 (VEN) (18). Briceño & Adamo 717 (MY) (5). мах — м — л © л Ж Ta P = ~ Broadway 316 (US) (12); 707 (US) (12). Bruijn 1183 (MER, MO, US, VEN) (23). Bunting 5598 (MO, VZU) (12); 7747 (М0) (12); 8042 (MO, VEN, VZU) (12); 9418 12). Bunting & Alfonso 13278 (MY, VZU) MO, d (12). Bunting & Arboleda 8726 159 (VZU) (1: 2). Bunting & Fucci 6049 12). Bunting & Stod- (VZU) (27). Bunting et -— N c — кы N 7 aij E + 2 & (МО, VZU) dart 8907 (МО, VZU) (12); 9055 al. 11120 (MY, VEN) (15). Bunting + al. 12023 (МО) . Bunting et al. 12481 (MO, MY, VZU) (12). Bunting . 12800 (MY) (15). Bunting et al. 7345 (MO, VZU) (12). Bukendi V0049 (MY, UCOB) (4). Burandt & > У 0596 (MY) (5). Burkart Arturo 16396 (VEN) (27). Calvo 0003 (MY) (23). Camero (MY-85034) E Cár- denas de Guevara, L. 1707 (MY) (2). Cardona 31 (US, VEN) (12). Cardozo 1137 (MY) (4); 1225 (MY) (4); 1364 (MY) (4); 1515 (MY) (22); 1709 (MY) (4). Carnevali et al. AN) (27). Caro З (MY) (12). Casadiego & Campos | ') (14). Castillo 1961 (MY) (16); pt ^ № (15): 478 > EN) (12). Castroviejo et al. 411 (MA) (12). Cawz 12 (MY) (7). Chaffanjon s.n. (P) (12). м 33 (VEN) (12); 271 (US. VEN) (4). Chardon et al. 3204 (VIA) (12); 3222 М A) (29). Charpin & Jacquemoud (HAC-13440 С, MO) (21). Charpin et al. 13146 (MO, NY) (6). Clark 6948 (MO, dua (1). Colella & Morales 563 (MY) (15). Colma et al. 103 (CORO, MY, VEN) (2); 244 (CORO, MY, VEN) (2); 247 (VEN) (2); 248 (CORO, MY, VEN) (2); 351 (CORO, MY) (2). Cred 21406 (MO, VEN) (15); 54467 (MO. VEN) (1 Эв Wes (MO) (21); 60795 (MO, _ N) (4). Croizat "ap 12). Cumana 1274 (IRBR, MY) (12); 1574 (IRBR, A (12). Curran & Haman 988 pH ), US) (23 T ~ y Arcy « Benítez 18236 (MO) (5); 18258 (MO) (13); 18261 (MO) (21). Davidse & González 15984 (MO, VEN) (1): 16721 (MO) (15); 19999 (MO, VEN) (12); 21967 (MO, VEN) (1). De Martino et al. SPB 1120 (MYF) (12). De- ascio 4087 (CAR, VEN) (4); 4118 (CAR) (4); 7579 (VEN) (15); 7673 (CAR) (12); 9724 (VEN) (4). kowski 3145 (CAR) (12); Na (CAR) (12). Delascio & Delascio 5042 (CAR. VEN) (5). - & López 2555 (CAR, VEN) (12). A 25 & Velasquez 692 (CAR) (22); 693 (CAR) (22). Delascio et al. 13401 (VEN) (12); 15026 (VEN) (12); 5006 (CAR, VEN) (5). Delgado 236 (VEN) e bí az 526 (MO) (4). Diederichs 172 (VEN) (4); 303 ) (4). Dorr & Barnett 5163 (VEN) (5 Э). Dorr et al. mea (VE N) (23); 7900 (MY, NY, PORT) (13). Edwards & Roe 28 (MY) (4). Edwards et al. 99 (MY) 22); 107 (К, MY) (4); 360 (К, MER, МЕКЕ, MO, MY, NY. PORT. US, VEN) (29); 467 (K, MY) (29). Elias 239 (CAR) (7). Ernst s.n. (HBG) (3). Ewell 195 (MY) (23). Fendler 9 (MY) (27); 547 (MY) (31); 746 (МУ) (21); 2090 (MO) (26); 2091 (GH) (29); Ра (СН, МО) (31); 954 МО, Р) (13); 955 (СН, МО, (4); 956 (МО) (21); 958 (СН) (27); 959 (СН, МО, Му) (2); 961 (СН) (23); 962 (GH, NY) (22); 963 (GH) (31); 966 (GH, МО, NY) (12). Fernández 1266 (MY) (12); 2480 (MY) (4); 2671 (MY) (12); 3154 (MY) (20); 3888 (MY) m 4017 (MY) (12); 4018 (MY) (12). Fernández et al. MER) (2); 200 (CAR, MY, PORT) (2); 5239 (MY, PORT! (12); 84 (MER) (2). Ferrari 141 (MY) (27); 142 (MY) (27); 221 (MY) (27); 272 (MY) (12); 991 (MY) (27). Ferrari & Bunting 1672 (MY) (31). Field 282a (К, sk (15); 313 (K, MY) (15); 441 (K, MY) (15); 497 (MY) (16). Figuera (М Ү-68080) (27). y unc E & Schlim 429 (BM, P) (31); 627 (С) (31); 784 (G. P) García | 42 (VEN) (22); 153 Mes (4); 172 (VEN) (31). Gehriger 255 (F, Р, VEN) (31); 299 (F, MO, VEN) (21); 40 (F, MO, VEN) (5). Gentry 41165 (MO) (15). Gentry et al. 10439 (MO, VEN) (12). Gines & Rudd 1599 (CAR) (31). Gentry & Stein 47262 (MO, MY, VEN) (15). Gines 1763 (CAR, US) (5); 2003 (MY) (12); 4665 (US) (31). González 160 (MER) (2); 3 (VEN) (2). González & Ortega | MY, VEN) (15). Gragson & Gragson 48 (MY) (2). Guánchez & Mercado 1921 =“ — — ~ — (1). Grosourdy 13 (P) 348 Annals of the Missouri Botanical Garden (TFAV) (12). Guevara (MY-68083) (12). Gutiérrez 225 (VEN) (15). Meier & Silva 1565 M VEN) (21); 61 (MY, (TFAV) (12 VEN) (31); 768 (MY) 1 (MY, VEN) (13); 1205 Hernandez 1189 (MER) (27). Holst 3733 (MO) (9). Holst & Liesner 2724 (VEN) (15). Holst et al. 2024 (VEN (12). Holt 19 (VEN) (2). Horner 419 (MO, MYF) (12). Horner et al. 404 (MO, MYF) (12). Hoyos 2091 (CAR | & Delascio 4285 (CAR, VEN) (15); 4292 (CAR, VEN) (2). Hoyos & Foldats 3091 (CAR) (2). Huber 207 (VEN) (4); 6250 (MY, MYF) (23); 11033 (MY, MYF (32). Huber & Roth 1732 (VEN) (15). Humbert 26521 (P) (2). Hurtado (M Y-68082) (27). Ijjasz 305 (MY, VEN) (21). Jahn 767 (MO, US, VEN) (31); 811 (US, VEN) (5); 923 (US, VEN) (5); 1075 (US, VEN) (13). Jeffrey & Trujillo 2457 (MY) (5). Jimenez Saa 1320 (MER) (12). Johnston 385 (GH) (2). kis 37251 (Е, GH, US, VEN) (12). wski & Ramirez 70 (VEN) (12). Lasser a N) (2); “1091 (VEN) (13); 1114 (US, VEN) (22); 189 (US, VEN) (2); 2037 (VEN) (22); 2279 (VEN ) (26); а (VEN (2); 2334 (VEN) (15); 2339 (VEN) (23); 3469 (VEN) (18): 3533 (MY, VEN) (7); (VIA-3275) (26). Lasser & Ariste- guieta 3376 (F, VEN) (2). Lasser & Foldats 3150 (VEN (2). Lasser & Vareschi 6052 (VEN) (5). Lasser et al. 2904 (VEN) (23). Liesner 3611 (MO, MY, VEN) (1); 5373 (MO, VEN) (2); 6877 AVEN) (1); 7083 (MO, VEN) (1); 7134 7557 (VEN) (1); 8210 (MO, VEN) (15); 10024 (MO, VE N) (15); 11803 (MO) (15); 13384 (MO) > ), MY, VEN) (17); 16999 (VEN, MY) (15); 17789 (MO) (8); 18484 (MO) (8). Liesner & González 5822 (MO, VEN) (12); 9311 (MO, VEN) (1); 9322 (MO, VEN) (1); 9478 ier VEN) (15); 9844 (VEN) (4); 9930 — S -1 I c М7 == 11658 (MO, VEN E VEN) (26); 13559 (MO, MY) (21); 13578 (MY, VEN) (4). Liesner & Steyermark 12330 (MA, E VEN) (12); 2. (MO, VEN) (4). Liesner et al. 12617 (MY, VEN 12836 (MO, VEN) (15); 7755 (MO, V e (4). Little 1 з (MER, VEN) (23); 15522 (VEN) (13); 15745 (VEN) (1 Lgjnant & Molau 15835 (AAU, GB) (11). López & = doval 495 (CAR, MY, VEN) (12). López-Figueiras 8754 (МЕКЕ, MY) (21). 2. Palacios 86 (MER, MO) (7); 340 (MER, MO) (2); 1255 (МЕНЕ, MY) (12); 1373 (МЕКЕ) (5); 1505 (MO) an "1507 (MO) (31); 1886 (MO, VEN (15); 1988 (МЕКЕ, МО, MY, VEN) (12); 2145 (МЕКЕ, Му) (2); 2150 (MER, MO, VEN) (23); 2637 (MO) (21); 2732 (МЕКЕ) (4). López-Palacios & Bautista 3193 (MER (12). Lozada m 67062) ys 3). Luteyn 5250 (F, MO, NY, 5 25). Үч & Lebrón- ud 21). Luteyn et al. 5 (F, MO, 2. 13) € 6076 (F. YAN ТАС) 6171 (МҮ, PS ) 208 (MY, NY, VEN 3). — — => laguire & ue 35219 о (NY) 42500 (MO, US) (17). Manara 113381 (MO) (15 ан (2); s.n. (МО-2671613) (: n. (VEN-113158) (31); s.n. (УЕМ-113165) (2); s.n. IN. 115002 (21); s.n. (VEN-1805023) (21); s.n. (VEN- 115003) (2); s.n. e 172746) (2); s.n. (VEN-174788) 96) (15); s.n. (VEN-175090) (15); 2 Manara & Vera (MY-28669) (27); Marcano-Berti 1519 (MER) (12). Marcano-Berti & Torres- Lezama 207 (MER, VEN) (23). Marcano-Berti et al. 457- 979 (MER, MY) (23). Matos 120 (CAR) (12); 1122 (CAR, VEN) (13). Medina 520 (VEN) (4); 531 (VEN) (22); 533 1). 1. et al. ~ Yu -— м Ф 13); 90 А МҮ, VEN) (13); 1329 (MY, VEN) (13); 1723 (MY, VEN) (13); (MY, VEN) (31); 3174 (MY, VEN) (13); 3180 (МҮ, VEN) (13); 3284 (MY, VEN) (4); 3330 (MY, VEN) (21); 3603 (MY, VEN) (31). Meier & pe 1565 (MY, VEN) v Meier et al. 2642 (MY, VEN) (15). Mocqueris 1092 (MY Р, US, VEN) (31). Montes 61 (VEN) (12). Morales 215 (MY) (31); 249 (MY) (21). Moreno 23 (MY) (27). Mori et al. 14667 RN, (4). Morillo 1777 (MY, VEN) (2); 8346 у E 11105 (MERF) (5). Morillo & García ; 11475 (МЕКЕ, MY) (9); 11478 (МЕКЕ, MY) M TASA & Hasegawa 5026 (MY, VEN) (1). sil illo & Liesner 9130 (MO, VEN) (12). Morillo & Man 2135 (MER, MO, MY, VEN) (23); 667 (MY) (31). Morillo & 2. 8730 (V EN) (4). Morillo & Morillo 2966 (VEN) | LN 2 (MY, VEN) (12). N) T Morillo & Seres 8627 (VEN) (2 Hen 5856 (MY, ut Me ); et al; 9: 512 E (12 27); 2 (BM) (4) 309 (BM, G-DC, MO) (31); 348 (W) (4); 824 (BM, G-DC) (22); 1641 (BM) (26); 1931 (P) (12). 1, s.n. (BM) (21) Nee 17450 (F, VEN) (12); 17563 (F, VEN, WIS) (2); 30689 (MY, VEN) (8); 31259 (MY, VEN) (18). Nee & Mori 3961 (US, VEN) (27); 4107 (US) (14). Nee & Whalen 17147 (F, VEN) (18). Oberwinkler 13446 (MER, VEN) (13). Ortega 1671 MO, MY, PORT) (14); 549 (MY) (12); 631 (MY, PORT) (12). Ortiz 1196 2 VEN) (15); 1261 (MY, VEN) (15). Ortiz et al. 1145 (VEN) (12); 1157 (VEN) (12). Osorio (MY-84794) (27). Otto 704 (G, P) (27). Páez et al. 29 ^) (14). Pietrangeli (M Y-86956), (MY- 86957), (MY-86958) (31); (MY-86959) (5); 1226 (MY, VEN) (4); 1243 (MY, VEN) (13); 1269 (MY) (31); 2020 (МУ) (13); 2147 (МУ) (13); 2149 (МУ) (31); 2337 (МУ) (13). Pipoly et al. 6478 (MO) (5). Pittier 166 (VEN) (13); 5797 (F, NY, US, VEN) (12); 5925 (P, US, VEN) (27); 9200 (GH, NY. US, VEN) (27); 9245 (US, VEN) (4); 10045 (US, VEN) (21); 10393 (GH, US, VEN) (23); 11215 (VEN) (7); 12094 (NY, US, VEN) (12); 12332 (C, NY, US, Mew (12); 12639 (ЄН, NY, VEN) (31); 12858 (US, VEN) (31); V (US, VEN) (5); 12919 (US, VEN) (31); 13029 (С. NY, US, VEN) (18); 13210 (F, MO, US, VEN) (5); 13294 (F, Mo US, VEN) (31); 13323 (F, MO, US, VEN) (31); 13963 (US, VEN) (27); 14075 (VEN); Pittier & Nak- ic рт 15368 (VEN) (4); 15683 (VEN) (4). Plowman 7765 (F, МО) (13). Ponce & Trujillo 634 (MY) (5). Prance 28160 (MO) (15). Pulgar (MY-16420) (27). Pursell et al. 8432 (VEN) (12). Quintero 70 (MER, MY) (7); 134 (MER, MY) (23); 1 MER) (12); 526 (MER) (23); 1310 (MER) (31); ees MER) 15 2292 (МЕН) (13). Quintero & Carroz 1082 MER) 1. & Ricardi 1524 (MER) (14). Quin- ero et 1 p (M ). amia & Ortíz pus (VEN) (12). Ramirez 2148 (MY) (1); 234 (МҮ. VEN) (12). Ramirez & López 3162 (МҮ) (2). Reggio & de Scorza (VEN-118911) (18). Ricardi & Adamo 576 ue (31). Ricardi & Carroz 9 (MER) (12). Ricardi & Salcedo 5734 (MER) ч, ); 5744 (MER) (23); 5756 (MER) (23). Rivero 1678 (MO) (5). Rivero et al. 1888 (MO, MY, PORT) (14). Rodríguez 1 (MER) (4); 71 (MY) (12); 118 (MY) (12); 203 (MY) (12). Rodríguez & Cardozo 1729 (MY) (4). Rodríguez et al. 1357 (MY) (4). ~ Volume 85, Number 2 Benítez & D'Arcy Cestrum and Sessea in Venezuela 349 Romero 357 (MY) (12); 750 (MY) (31); 1035 (MY) (31). Ruiz 4183 (MY) (4); 4297 (MY) (4). Ruiz et al. 429 (VEN) (2); 525 (VEN) (2); 532 (VEN) (2); 579 (VEN) (2); 1507 VEN) (2); 1644 (VEN) (2); 2101 . (2); 4501 (МҮ) 12); 4718 nee (14); 4719 үт, 2 ыл Ruiz- кеч 1761 МҮ) Qu 13420 2 (5 n 13426 23 & Dae 12219 M (5 s ee M (13): 1 15985 UR MY) En 5) 1024 (MY) (5); 17: 5) 1998 (5); 8306 (MY) (5); 8791 (MERF) (13); 8898 (MY) (5); 949 (MY) (5); 13086 (MY) (5). Ruiz-Terán & López-Palacios 1651b (MY) (5); 7550 (МЕКЕ, MY) (21); 1552 pes (33); 7615 (MERF) (13); 9925 (MY) (12); 9975 ; 10223 (МЕКЕ, MY) (2). Ruiz-Terán & Mar- EL 1241 (MER) (23); 1325 (MER, MY) (12); 1390 (MY) (5); 1494 (MY) (5). Ruiz-Terán & Ruiz-Pérez 14956 (МЕКЕ, MY) (31); 15673 (MY) (13). Ruiz-Terán et al. 3818 (MERF, MY) (31); 3841 (MY) (5); 3859 (MY) (5): 3914 (MERF) (31); 3948 (MERF) (21); 3989 (MERF) (13): (MERF = = — ub — > е ( МЕКЕ) (13); 8226 (МЕКЕ, MY) (5); 10651 (МҮ) (12); 12357 (МЕКЕ, MY) (21); 14230 (МЕКЕ, MY) (21); 14266 (МЕКЕ, MY) (31); 14677 (МЕКЕ, ) (31); 15135 (МЕКЕ, MY) (21); 16114 (MY) (5); 16146 (МЕКЕ, MY) (21); 16154 (МҮ) (25); 16171 (МҮ) (25); 16303 (МЕКЕ, MY) (31). Rusby & Squires 327 (F, С, GH, MO, US) (12). Rutkis 373 (MY, VEN Saer 15 (US, VEN) (2); 17 (С, NY) (2); 162 (US. VEN) (27); 184 ey: US, VEN) (2); 445 (F. VEN) (31); 833 (NY, US, VEN) (12). Ree 4. (MY) (12). Schott 123 (F) 7). Schulz et al. 330 (VEN) (5). Smith 3507 (F, US) (13); V1555 (VEN) (31); ee (VEN) va V5242 (VEN) (4); V585 (VEN) (2); V7661 (VEN) (7); V8487 (VEN) (31). Mie 2974 ш. ‚ NY) (28). Stergios 821 (MY) (5); 1528 (MY) (31). Ster- 2108 & Aymard 7311 жш (28); 7358 (MO) (28); 7606 (PORT) (28) 7770 (MO, MY) (1). Stergios 5630 (MY. PORT) (12). Stergios & Delgado 12910 (MY, PORT) (12). Stergios & Utrera 2465 ( Stergi 3977 (MY, PORT) (Ра (МО, MY, PORT) (15 55008 (MY, VEN) (26); 55689 (MY, VEN) a Ae 55695 (Е VEN) (21); 55713 A MY, VEN) (5); 55972 (MY, VEN) (31); 56219 (F, VEN) (2); 56322 (К, MY, VEN) (14); 56453 (MY, VEN) (4); 56518 (F, MY, VEN) (5); 57006 ж VEN) ig 57031 (MY, VEN) (13); 57102 (VEN) ) EN) (12); 60925 (F. 2 (12); 23); 61816 (F, MY, VEN м2 Eu (F, VEN) (8); 75682 (F, VEN) (8); 86286 (NY, V EN) (4); 86845 (VEN) (12); 87218 (MO, VEN) (12); 88525 (US, VEN) (2); 88871 (F, US, VEN) (12); 89833 (NY, US, VEN) (4); 90560 (US, VEN) (12); 90566 (US, VEN) (12); 91594 (VEN) (13); 91611 (VEN) (13); 91659 (VEN) (26); 91672 (F. VEN) (13); 92106 (F. US, VEN) (26); 92147 (P, US, VEN) (11); 93274 (F, US, VEN) (8); 95075 (VEN) (20); 95083 (Е, Р, VEN) (15); 99040 (VEN) (4); 99361 (VEN) a 99368 (VEN) (4); 99894 (MO, US, VEN) (23); 101047 (US, VEN) (6); 104947 (VEN) (31); 105035 (P, VEN) (21); 105056 (P, VEN) (5); 106207 (US, VE ы (4); (VEN) (12); 107414 (MO, US, VEN) (12 VEN) (8) 120447 (VEN) (12); 122681 (VZU) (12); 125620 (VEN) (26); 127851 (MO) (13); 129677 (MO, MY, EN) (8). Steyermark & Agostini 91026 (US, VEN) (20). Steyermark & Braun 94598 (Р, VEN) (2). NM & Bunting 102318 (VEN) (12). Steyermark & Carr 108771 (VEN) (23); 111662 (F, US, VEN) (12). = айна mark & Liesner 118501 (VEN) (19); 118531 (VEN) (4); 118748 (МО, VEN) (4). Steyermark & Manara 110407 . VEN) (2); 125448 (MY, VEN) (5). eid US e & Јали 95906 (VEN) (4). Steyermark & Perki 2 (MY, VEN) (15). Steyermark & Rabe 96236 (VEN) (15). Steve & Steyermark 95366 (US, VEN) (15); 95463 ; 5, VEN) (29). Steyermark € Wessels-Boer 100397 (MO, VEN) (15); 98775 (MY, VEN, US) (5); 100219 (MO, US, VEN) (4); 100602 (F, US, VEN) (21): 100827 (F, MY, US, VEN) (9); 101047 (MO, VEN, US) : 101530 (MO, US, VEN) (1); 103403 (P, VEN) (4); 111522 (VEN) (15); 114590 (MO, VEN) (12); 119545 (MO, VEN) (1); 119881 (VEN) (4); 121278 (VEN) (12); 121518 (MO) (15); 121595 (MO, VEN) (20); 121863 (MO, is ТАС 122650 (МО, МҮК, VZU) (12); 123025 (МО, №. VZU) (2); 123352 (MO, VEN) (12); 124269 (MY, EN) (12); 124739 (MY, VEN) (15); 124887 (MY, VEN) (12): 127229 (MO, MY) (12); 131034 (CAR, MO) (15). Sugden 1192 y (15). Tamayo 116 (VEN) (13); 2077 (US, VEN) (12); 2500 (VEN) 2): : на (US, VEN) (12); 319 (VEN) (2); 4367 (M E mayo et al. 2504 (VEN) (13). Tate 24 (US) > (31): 885 (US) (8). Tengler 3967 (САВ) (18). dl >) (2). Thomas et al. 3392 (MY) (15). Tillett Y) (21); 746-454 (MY) (18); 747-1021 (AAU, MY, 390^ VEN) (30). Tillett € Kónig 737-276 (MY) Ч 738- 460 (MY) (21); 738-534 (MY) (21); 747-929 | Tillett & Sayago 843-35 (МУ, MYF) (18). Til- 15); Torres et al. 19 (MY) DN ). Trujillo 1306 (MY) (13); 1930 (MY) MY) (4); 2784 (MY) (31); 3993 dd (13); 9 (MY) (12); 4756 (MY) (2); 4816 (MY) (12); 5121 (MY) (31); 6147 my) em 6317 (му) (15); 4. (МҮ) (2); 6837 (MY) (2); 3 (MY) (2); 7635 (MY) (4); ( (1: 107001 МҮ) (21); 9124 B (4); 979 (MY) (31); 14541 (MY) 2). Trujillo & ШІ 30 (MY) (23); 856 (MY) (21); 16351 am (12 o & Ponce 18282 (MY) (13); 19702 (MY) (18). “Trujillo & Rodriguez 17987 (MY) (12). Valve um & Pefia 1061 (MER, MY) (12); 1069 (MER) (14); 1070 (MER, VEN) (14). van der Werff & Ortiz 5859 (MO, VE Ji (4). van der муы Wingfield 3107 (MY) (2). et al. 51 (WIS) (2); 74 (WIS) (2); 576 (CORO, MY) (4); 3209 (WIS) (15); 8771 (MO) (13). Va- reschi 5635 (V EN) (5); 7549 (VEN) (5). Vareschi & Lasser 374 (VEN) (5); 6052 (VEN) (5). Vogel 1238 (US) (12); A237 (BM) (26); 406 (BM, M, S, US, 27 Weitzman & Boom 95 (MYF) (16). 4. Bu 2210 (VEN) (5). Wessels Boer et al. 2421 (MER, VEN) (21). Williams 9937 (F, VEN) (13); 9948 (F, US, VEN) (13); 11476 (F, GH, US, VEN) (12); 11544 (F, US, VEN) (12); 11670 (VEN) (12); 13652 (F) (27). : (С око, MY) (7); 5214 (MO, WIS) (2); 6818 (WIS) (2). Ја € López-Figueiras 7640 (CORO, МУ) (4). Wood VEN) (22); 412 (VEN) (15); 429 (VEN) (22). Wood & Berry 88 (VEN) (30). Woronov 7079 (F) (13). Wurdack & Steyermark 1083 (VEN) (8); 1366 (VEN) (8). Xena 583 (MO, MY) (31); 1053 (MY) (15). = о N = N 350 Annals of the Missouri Botanical Garden Yeffrey 2457 (MY) (5 Zambrano & Alfonzo | 340 (HERZU, VEN) (12); 1404 (HERZU, VEN) (12). Zambrano & Gutierrez 1529 (HER- ZU, VEN) (15); 1790 (HERZU) (15). Zambrano et al. 1899 (HERZU) (12); 2217 (HERZU) (23). INDEX TO SCIENTIFIC NAMES itii жей сысы aad ante ke ae N 277 Cestreae С. Don ....................... 271 ( 5. ies dcs a i d Ыр aie. e ers е Шаа 271 Gestrum. L 4 e uw RR 271 acuminatissimum Dunal ................ 280 albopunctatum Dunal .................. 30( alternifolium (Jacq.) O. E. Schulz .......... 281 alternifolium var. mithanthum O. E. Schulz 281 alternifolium var. pendulinum (Jacq.) О. E. Sula 281 ambatense Francey ................... 337 amelanchier Папа|.................... 281 amplum Pittier ...................... 304 amplum var. grandifolium Francey ......... 304 aristeguiet Mirra 33; atrovirens Dunal ..................... 341 aurantiacum Lindl. ................... 285 aurantiacum var. chaculanum (Loes.) Francey 285 baenitzii Lingelsh. .................... 306 bigibbosum Pittier .................... 287 illbersianum Beurl. .................. 300 bogotense Roem. € Schult. .............. 304 bogotense var. edi Francey .......... 304 buxifolium Kunth .................... 289 caloneuru ad сыла uec be ee ees 316 calycinum Kunth ..................... 333 calycinum var. tum Francey ......... 334 calycosum Р\їпег..................... 316 cancellatum Dunal CRGO NEN E MS 334. сћасшапит Loes. .................... 285 chloranthum Dunal ................... 300 clausseni Юпапа|...................... 306 confertum Miller .................. 281, 283 costanensis Steyerm. ................... 304 cuneatum Francey .................... 289 cuneifolium Ётапсеу................ 290, 298 densiflorum Ктапсеу................... 337 densiflorum var. puberulum Francey ........ 337 depauperatum Dunal .................. 281 diasae Pittier ....................... 337 diurnum 1.......................... 290 diurnum var. fastigiatum (Jacq.) Stehlé ...... 292 diurnum var. odontospermum (Jacq.) O. E. Schulz 292 diurnum var. venenatum (Mill.) O. E. Schulz .. 292 dubium Рипег....................... 304 fasciculatum (Schltdl. Miers ............. 277 fastigiatum ]асд...................... 292 faucheri Dunal ...................... 306 floribundum Britton ................... 323 glabrescens (C. V. о em & Maguire 294 glabrum Klotzsch & Karsten ............. 304 grande AAA 320 hediundum Lam. ..................... 277 hirsutum Jacq. ...................... 337 1 Dents «une nS КГК es 300 hirtum SW. o каска a UR E Sem 300 humboldii Francey ................... 294 humboldtii var. calycinum Francey ......... 296 iumboldtii var. 4. Francey ........ 29 imbricatum Rusby ................. 290, 208 impressum Rusby ..................: 334, 335 telum Francey ...................... 287 lanatum M. Martens & Galeotti ........... 337 lanuginosum Ruiz & Pav. ............... latifolium Lam. ............... 277, 294, 300 latifolium var. genuinum Stehlé ........... latifolium var. tenuiflorum (Kunth) O. E. Schulz 300 laurifolium Fawc. .................... 292 laurifolium UHerit. ................ 292, 318 laxiflorum Dunal .................. H 329 lindenii Dunal ................... 5 longifolium Ruiz € Pavón ..... 277, 318, m 335 loretense Francey ..................... 280 lundianum Dunal .................... 334 mariquitense Kunth ................ 283, 304 mariquitense var. latifolium (Francey) Standl. & C. V. Morton «us mew ез 304 mathewsii Dunal ..................... 320 megalophyllum Dunal ........... 281, 306, 320 и аы са Dunal as 2 х meridanum Pittier .................... 337 2. France 12 kere a dea е даны ae Go 308 miersianum Pittier .................... 337 miersianum Wedd. .................... 337 moritzianum Klotzsch & Karsten .......... 318 ortzi Юипа1....................... 337 2. D Arcy & Benítez ............. 310 neomiersianum Benítez ................. 337 nocturnum а. 277, 312 о о Jaed. ОКИЛ СЕЛТ ЕГ 292 oliganthum Dunal .................... 300 oliganthum var. МИТРА Dunal .......... 300 22. Francey .................... 314 ovatum Roem. & Schult. ................ 300 panamense Standl. .................... 320 paniculatum Kunth ................... 327 pariense ®Чеуегт...................... 314 gui-EHer- soa vidas cic аи кои.» 277 parvifolium var. venezuelense Francey ....... 289 parvifolium Wild. .................... 289 paucinervium Ктапсеу.................. 285 pedunculare Пипа].................... 285 pendulinum Jacq. .................... 281 erilambanon Loes. ................... 329 petiolare Kunth ...................... 316 petiolaris (Kunth) Spreng. ............... 316 poeppigii Sendtn. ..................... 300 potalaefolium Пипа]................... 318 potali шойи ит Dunal ................... 318 prieurei Пипа... 300 pumilum Francey .................... 287 racemosum Ruiz & Pav. ................ 320 racemosum var. bolivianum Francey ........ 320 racemosum var. grande (Pittier) Francey ..... 320 racemosum var. panamense (Standl.) Francey .. 320 reflexum Sendtn. ..................... 323 reflexum var. densiflorum Francey .......... 323 rojasianum Hassl. .................... 339 rugosa Rusby ....................... 334 ruizteranianum Benítez & DAY ни 325 salicifolium Jacq. .................. 287, 325 salicifolium var. angustifolium Dunal ....... : scandens Уаћ].................... 287, 327 _ и var. terminale Dunal ............ 329 schulzianum Francey .................. 331 «інен тт Dammer ............... 308 sesseoides Francey .................... 337 silvaticum ин ey odo Gee ie uu a ee eae oe 310 Volume 85, Number 2 Benítez & D'Arcy 351 1998 Cestrum and Sessea in Venezuela standleyi Francey .................... 310 viridiflorum Hook. .................... 334 ние Ruiz & Pav. ................ 333 Chiococca strigilatum var. реге (Kunth) Kuntze .... 333 alternifolia E... E la id rs la var. laxiflorum Kuntze .......... 334 brothamnus Endl. ............. 273, 277, 285 strigilatum var. tenuiflorum Francey ........ 334 aurantiacum Steudel TOM онор suberosum Jacq. ..................... 312 tenuiflorum Kunth .................... 300 "әйелі Таб A з ERO ES ETT 281, 283 tenuiflorum var. glabrescens С. V. Morton ..... 2M Lomeria Шыр D ob eb кк деюн A 271 tenuissimum Francey .................. 310 Lomeria purpurea Raf. ................... 277 terminale (Dunal) Pittier ................ 329 Metternichia ари e EN. Cua ono dc 273 шепи Benítez & D'Arcy ............... 335: JMeyenia.Nees: У. Ла анаарар 277 tinctorium Jacq. ..................... 29 io SENE dar A ЕТІ tomentosum 1.Ғ.................. 296, 337 Meyenia бсеһіші!........................ 277 tovarense Francey .................... 3l Parqui Adans. ыи Boe es see. кишик з 271 tubulosum Sendtn. .................... 339 Рағаша Кас, 4: DAN as 271 umbrosum Ктапсеу.................... 287 Sessea Ruiz & Pav. ..................... 341 unibracteatum al 28 565464 sU toys 334 atrovirens (Dunal) B. D. Jacks. ............ 341 unibracteatum var. pm Dunal ...... 334 corymbiflora Rich. Taylor & R. Philips ...... 341 venenatum Burm. f. ................... 292 corymbosa Miers ..................... 341 venenatum Mill. ..................... 292 stipulata Ruiz & Pav. .................. 341 нр Francey .................. 287 Tsoala Bosser & D’Arey .................. 273 osum Roem. & Schult. ............... 316 Vestia Willd. .....2............ 273, 276, 3H verbascifolium. ado AA паса 337 diurnum var. tinctorium (Jacq.) M. Gómez ... vespertinum Griseb. ................... 300 fadea vespertinum Lucius x3 ТЕРТЕР 281 latifolia. [Do AA Mcr PR 271 vespertinum Lunan .................... 202 adea Каб ао еге rx EE 277 MATERIALS TOWARD A REVISION OF GRIMMIA (MUSCI: GRIMMIACEAE): NOMENCLATURE AND TAXONOMY OF GRIMMIA LONGIROSTRIS! 2 Jesús Muñoz? ABSTRACT Grimmia longirostris Hook. is described and 4. and its ecology апа phytogeography are considered. A world distribution map is presented. Fifty-five names, at s longirostris. с affinis Hornsch., predated b ecific and 2 "ific rank, are cons the name уе for G. longirostris Hook., a name hitherto used only in the A idered synonymous with G. nis taxon in the holarctic, is illegitimate and slightly Andean region. Cross-section costal morphology is € the best diagnostic character to distinguish G. longirostris from related species. Other useful features are e long, incrassate and nodulose basal paracostal cells of us lamina, the straight pines Grimmia longirostris grows on all continents exce these regions seems not ranges or latitudes. t Antarctica and Aus to be an artifact of зин сна The preferred bo ds acidic rocks in high mountain Grimmia is the largest genus in the Grimmi- aceae. The difficulty of its study is increased be- cause a vast number of taxa were described at the turn of the century without a critical analysis. In fact, more than half of the taxa in the genus were described in the 50-year period between 1875 and 1925. A critical revisionary study is required to understand the taxonomy and biogeography of the genus. Toward this aim, some years ago I started a nomenclatural database of the genus. The final goal is to present a checklist that could be the starting point of a complete taxonomic revision. One of my most striking conclusions is that at least 55 validly published names and 10 nomina nuda apply to a single species that has been most commonly known as Grimmia affinis Hornsch. Not less surprisingly, this has proven to be an illegiti- mate homonym predated by the legitimate Grimmia longirostris Hook. Тһе legitimate С. affinis Hornsch., described five months earlier, is a taxo- nomic synonym of G. fuscolutea (cf. Crum, 1994). MATERIALS AND METHODS All names in Grimmia found in Crosby et al. (1992), Crosby and Magill (1994), and Index Mus- corum (Wijk et al., 1962, 1969) and not transferred in these works to other genera have been included in the nomenclatural database. A few taxa excluded from Grimmia originally in Index Muscorum have een resurrected and included (G. brandegei Austin would be an example; cf. Ochyra & Bednarek- Ochyra, 1994). The database now contains 1300 names, which represent a possible maximum of 747 taxa; subsequent taxonomic judgments may reduce this number. The protologue of each name has been checked, involving the review of more than 500 different papers. I have so far been able to study ! This from the Spanish Minist Commission HCM Helsinki. The Friends of the быр and the The New York Botanical Gar d'Histoire Naturelle, Paris (PC ry of Education and Culture work was completed while the author was at the Missouri Botanical Garden, sponsored by a postdoctoral grant Travel and expenses at Helsinki ( E Contract n° ERBCHGECT940065 with the Division of nb s Biology of . Steere Fund supported my n (NY), 4... Гат indebted to these institutions and also to Muséum National pe especially to their staff, for help and support. H) were Por 4 ; the European e University of Harvard Univers sity (FH) and isits to — I also express my gratitude to Steven P. pes and Ryszard Ochyra for reading i uel ae the original. n I thank ALTA, BCB, BM, BP, CANM, PC. 4 ЕН, С, , H, JE, KUN, LE, MO, МІСН, PCD, TRH, and UPS for loan of specimens Special 1. to Francisco Pando (MA) for 2. 21. me to use aie program to generate the nomenc Таан list. William К. Buck, Alar ments on the manuscript. 1 T. Whittemore, and an anonymous reviewer are acknowledged for their constructive com- ? Instituto Asturiano de Taxonomía y Ecología Vegetal, Р.О. Box 8, E-33120 Pravia, Spain. Present address: Missouri Lo „А. Botanical Garden, P.O. Box 299, St. ANN. Missouni Bor. uis, Missouri 63166-0299, U.S GARD. 85: 352-363. 1998. Volume 85, Number 2 1998 Mufioz 353 Grimmia longirostris the type(s) or original material(s) of 328 taxa (ca 44% of the total Some 900 herbarium specimens from 22 herbar- ia were studied to determine the geographical dis- tribution and morphological variation of G. longi- rostris. Figure 2 shows data from 733 revised specimens for which geographical coordinates could be established. I have selected lectotypes whenever necessary. It has been pointed out (Blom, 1996: 11; Horton, 1982: 377) that the only extant specimen in the herbarium of the original author should not auto- matically be considered the holotype of a name de- scribed prior to the establishment of the type con- cept. I fully agree with this view, but I have followed the designations of previous authors, as did Blom (1996). Grimmia longirostris Hook., Musci Exot. 1: 62. 1818. TYPE: [Ecuador. Chimborazo: Mt. Chimborazo], Humboldt 76 (lectotype, desig- nated by Deguchi (1984), BM; isolectotypes, BM, PC). Grimmia obliqua Hornsch., Flora 2: 84. 1819 [February]. rimmia ovala var. Же (Hornsch.) Huebener, Muscol. Germ. 183. 1833. Dryptodon ovatus var. ob- liquus (Hornsch.) Hartm., Handb. Skand. Fl. ed. 3: 271. 1838. Grimmia ovalis var. obliqua (Hornsch.) I. Hagen, Kongel. Norske Vidensk. Selsk. Skr. (Trond- P 1909(5): 26. 1909. Grimmia ovalis f. obliqua (Hornsch.) Mónk., Laubm. Eur. 360. 1927. TYPE: EN Salaburg:] In alpibus Salisburgensibus, Hornschuch s.n. (lectotype, em designated, BM). Grimmia affinis Hornsch., Flora 2: 443. 1819 Padi nom. il leg. [non Hornsch. 1819, Flora 2: 85 [Febru e 2. Hartm., Handb. Skand. Grimmia mone var. affinis da & ded p. in Bruch, Schimp. & W. Gümbel, Bryol. Europ. (fasc. 25-28) 3: 123, Tab. 2558. 1845. Grim- mia ovalis var. e (Hornsch.) Broth., Acta Soc. Sci. Fenn. т 86. 1892. Grimmia ovalis f. affinis ónk., Laubm. Eur. 360. 1927. TYPE: ia] W mE ssmattrey Tauern, Hornschuch s.n. (lectotype, designated by Deguchi (1978), B). Grimmia affinis var. ramosissima Nees & Hornsch., Bryol. Germ. 2: 144, Tab. XXI fig. 13b. 1827. natn md var. elongata Nees & Hornsch. E Huebener, Germ. 709. 1833, nom. illeg. TYPE: [Aus- ie Heiligenblut b. po [Hornschuch s.n.] (lecto- e, here dE E schimperi Bruch & o ex Mill. Hal., Syn. Mus Schimper s.n. [Schim iter abyssinicum, Sectio II, n* 484. 1842] Пе. here designated, МУ; Ра РС), paria р шы Müll. Hal., Bot. Zeitung (Berlin) 1853. TYPE: India. Tam il Nadu: in montibus ы [Nilghiri Hills], Perrottet 1583 (lec- totype, here designated, PC; isolectotype, PC). ни jui atn De Not., Mem. Reale Accad. Sci. To- rino, ser. 2, 18: 447, fig. 7. 1859. TYPE: [Ecuador. Napo:] ad rivum Napo, Osculati s.n. (isotype, H- SOL). mun peruviana Sull., U. S. Expl. “Л 7 8, . 5А. 1860. TYPE: Peru. Andes, 15000 ft., U.S. Ex. Ex. Wilkes [Rick & Brackenridge] s.n. A here designated, BM; isolectotypes, BM, FH, NY) Grimmia 27. (Натре) А. 7. = Tatigk. St. Gallischen Naturwiss. Ges. 1872 s, 2800 m, July 1859, Lin- dig 201 1 л у Чейне, PC; isolecto- type, NY). Grimmia hausmanniana De Not., Cronac. Briol. Ital. 2(2): 66. Racomitrium fous Bryin. 2: 227. 1898. Puma oralis f. р Podp., p orbe Muscorum Europaeo- rum, 54, nom. illeg., non Warnst. 1904. TYPE: АЕ Tirol:] Rittner Horn, Hausmann s.n. (isotype, BM). Grimmia ovata var. praecox А. Kern., Flora exsiccata aus- tro- -hungarica n° 316, 1881. TYPE: [Austria. Toon] ad Trins in valle Gschnitz, 1200 m, Kern 316 (lec totype, here designated, BM). Grimmia integridens Müll. s бра 43: 460. Ls TYPE: Argentina. Tuc in der Cienaga, ca 11000', 1893, Eis yn here кылана ed, РС). ш: sii diui. Müll. Hal., Linnaea 43: 456. 1882. E: Argentina. Tucumán: Tafi, 1872, Lorentz s.n. Пе here designated, ЈЕ; isolectotypes, BM, ‚ NY). Grimmia ПРЕ Müll. Ex Linnaea 43: 459. 1882. TYPE: Argentina. Salta: Nevado de Castillo, 1873, Lorentz s.n. (lectotype, E sl JE). Grimmia d Müll. Ha., mg ie 43: 458. 1882. ntina. Salta rentz s.n. (lecto- type, bue деш, ЈЕ: . М). Grimmia cavifolia Lindb. € Arnell, Kongl. Svenska Ve- kapsakad. Handl. 23(10): 103. 1890. TYPE: [Russia. Krasnoyarsk Oblast:] Jenisei, Krasnojarsk, 11 June 1876, Arnell s.n. (lectotype, here designated, H-SOL; isolectoypes, JE, PC, UPS). xi dE Müll. Hal. ex E. Britton, Bull. Tor- Bot. Club 23: 477. 1896. TYPE: Bolivia. La Paz: Mapiri, 5000 ft, May 1886, Rusby s.n. (holotype, NY). Grimmia ortholoma Kindb., Rev. Bryol. 23: 19. 1896. T Canada. British шй Dear Park, 8 June 1890, Macoun s.n. (holotype, S; isotype, Grimmia breviexserta Müll. Hal., Bull. Herb. 2. T 200. 1897. TYPE: Guate mala. Quezaltenango: altenango, Bernoulli & Cario l5 (lectotype, жй designated, PC). 7. micro-ovata Müll. Hal., Nuovo Giorn. Bot. Ital., 128. 1897. TYPE: Bolivia. Cochabamba: prope 354 Annals of the Missouri Botanical Garden Choquecamata, June 1889, Germain 1142 (lectotype, here designated, JE; isolectotype, NY). Grimmia 1 Se nes ex ба а Hal., И Giorn. ot. Ital., n.s. 4: 128. 34 PE: | Joliv e- it 11. 4. rie а, hac le Peñas, 3700-4000 m, Apr. 1860, ). Mandon 1634 (lectotype, here designated, BM; ейин; BM 2 replicates, FH, С). Grimmia arctophila subsp. labradorica Kindb., Eur. N. . Bryin. 2: 221. 1898. TYPE: Canada. Labra- dor: 21 July 1896, Macoun s.n. (lectotype, here des- ignated, 5; isolectotype, CANM 198076). — 1 Müll. Hal., Bull. Herb. Boissier 6: 109, 18 ҮРЕ: Brazil. Minas Gerais: Serra do Itatiaia, set Negras, 2300 m, Mar. 1894, Ule 1830 (lec- ere designated, H-BR). ex Müll. Hal., Bull. Hen Grimmia itatiaien sis Broth. & Brotherus, Bryotheca brasiliensis, n type, here 2. H-BR; isolectotypes, СОЕТ, JE, NY, PC, UPS). me ea Kindb., Rev. Bryol. 32: 33. 1905. TYPE: dice. “Hunker creek, 25 July 1902, Ma- coun s.n. (lect . here designated, 5; isolectotypes, CANM 198084, n BR). Grimmia Ц Kindb., Rev. Bryol. 32: 35. 1905. TYPE: Canada. British Columbia: lake Louise, 6000 ft., 13 E 1904, Macoun s.n. (holotype, 5; isotype, CANM 198092). Grinunia praetermissa C ин, aa a 36: 105. 1909, TYPE: Mexico. México: In the crater of the Volcano of Toluca, 13,500 íi. i 20 иң 1892, Pringle s.n. [Pr а Plantae mexicanae n" 26a] 2. һеге designated, NY; isolectotypes, ЈЕ, PC 2 replicates). Grimmia herzogii Broth., in Herzog, Biblioth. Bot. 87: 55. 1916. TYPE: [Bolivia. Cochabamba:| An Felsen ei- nes Gipfel der Yanakakabastion, 4500 m, Juli 1911, Herzog 3826 (lectotype, designated by Deguchi (1987), JE; isolectotype, H-BR). Grimmia hg ee He т. к Bot. 87: 55, fig. 17. 1916. TYPE: Boliv i der Saittulaguna, 4300 m, не m 1, Herzog EON (lec totype, here designated, JE). me ке та е Biblioth. Bot. 87: . 1916. TYPE: Bolivia. Cochabamba: Torreni- Yan- akaka. July en l. pri s.n. (holotype, JE) Grimmia ия Herzog, Biblioth. Bot. 87: 55. 1916 T olivia. Cochabamba: Vanakskabsstion, 4500 m, June 1911, Herzog 3827 (lectotype, here designated, JE; isolectotype, JE). SYNTYPES zog 4871 (JE): Herzog 3148 (ЈЕ); Herzog 4811 (NY. PC). — ге Broth., Rev. Bryol. 47: 9. 1920. TYPE: or.] Azuay: In rupibus montis prope Cañar, 16 Nov. 1909, Allioni s.n. (lec oe here 5. Н-ВН; isolectotypes, H-BR, РС). Grimmia afro- ovata Broth. & Thér., Bull. Mus. Hist. Nat. (Paris) 30: 240. 1924. TYPE: Kenya. Kilimanjaro: Kilimanjaro, Alluaud s.n. (lectotype. here designat- ed, PC; isolectotype, H-BR). Grimmia catalinensis К. 3 2. БАДЫ 27:62, РІ. 10. 1924. TYPE: L . Arizona: Рипа Со., ravine near d of Mt. м Santa Catalir na Mountains, . 15 Jan. 1923, Bartram 387 (holotype, FH; suis FH, NY). Grimmia catalinensis var. mutica E. B. Bartram, Bryologist 217: 02. 1924. Grimmia ovalis f. mutica (E. B. Ва tram) G. N. Jones, in p oss Fl. 34. 1933. TYPE: isi Santa Cruz Co., 7 House iod pe a Rita Mountains, 5500 . 18 Feb. 1923, Bartram t peer inia FH). Grimmia trollii Herzog, Hedwigia 74: 102. TYPE: Bolivia. Oruro: Curahuara, Troll 58 a des- йшй. by Deguchi (1987), Grimmia cinerea Thér., Rev. Bryol. Lichénol. 9: 9, fig. 3. 936. T 1 ҮРЕ: Кс 2. Pichincha: rochers ди Соп- . 4 Nov. 1930, Benoist 3153 ч = % т N p trj dorguachana, 4150 r (holotype, PC) Grimmia stenopyxis Thér., Rev. Bryol. Lichénol. 9: 8. 1936. TYPE: Ecuador. Pichincha: Pichincha, 24 Oct. 1930, Benoist s.n. (lectotype, here designated, PC). Grimmia sumatrana Dixon, Ann. Bryol. 12: 50. 1939. TYPE: “С. Losir, 3250-3500 m, 4 Feb. 1937; van Steenis (10152); Hb. Bog (4035) [Buitenzorg Botan- ical Garden] (Syntype, fide Deguchi (1986), L not seen). Grimmia antillarum Thér., Rev. Bryol. Lichénol. 13: 13. 1944. TYPE: 2. 'an | Republic. Azua: Cordillera Central, Los Vallecitos de Yaque. 2500 m, 2 Oct. 1929, Ekman 13630 (lectotype, here designated, PC; isolectotype, NY). Grimmia ree 7 Takaki, Bot. Mag. (Tokyo) 64: 180, fig. 4. 1951. TYPE: [Japan.] South Alps, Sensui pass, UN m, 15 Aug. 1950, Takaki 10137 (holo- type, private herbarium of Takaki, not seen Grimmia maido Greven, Bryologist 99: 428, fig, 1. 1996. TYPE: Africa. La Réunion: Le Maido, at the end of RF 8, 21 50 m, 7 Oct. 1995, Greven & Khoe bal 4000/ I (isotype, MO — Autoicous. Plants forming dense cushions, yel- lowish green, olive-green to dark green, occasion- ally rusty, golden yellowish above, brownish to black below, occasionally hoary due to long hair- points of the leaves, only rarely muticous. Stems erect, to 3 cm high X 160-220 jum diam., central strand well developed. Leaves usually erect and ap- pressed, not flexuous when dry (larger plants can have more flexuous leaves), erect to erect-spread- ing, rigid when moist, 1.7-2.6 mm long (exclusive of hair-point) X 0.3-0.6 mm wide, lanceolate to ovate-lanceolate, acute, not or weakly keeled, not plicate; hyaline hair-points terete, from erect and rigid to moderately flexuous, to 2 mm long, smooth or slightly denticulate; margins entire, recurved on one side to 44 their length, elsewhere plane, sel- sides; dom recurved on both costa percurrent, scarcely prominent on dorsal surface on the upper Volume 85, Number 2 Mufioz 355 Grimmia longirostris half of leaves, in section semi-elliptical to reniform, curved and with the ventral sinus U-shaped, not al- ways clearly delimited from laminal cells, especial- y in the 2-stratose upper part of the leaf, at mid- leaf with 2-4 ventral guide cells, a medial layer composed of substereids, and a dorsal epidermis, the three layers not clearly differentiated from each oth- er, usually only the ventral layer composed of large cells, clearly differentiated from the other two, the latter more obscurely delimited from each other; lamina 2(3)-stratose in the upper У, more regularly 2-stratose with 3(4)-stratose areas toward the mar- gins; upper cells 4-11 um along their longest axis, isodiametric, rectangular and transversely rectan- gular intermingled, incrassate and somewhat sinu- ate, if long rectangular then more strongly nodu- lose; basal paracostal cells 25-60(90) um long X 7-13 pm wide, rectangular (3-7(11):1), with thick and nodulose walls; basal marginal cells 9-25 um long X 6-11 jum wide, usually rectangular (1-3:1), hyaline, with straight walls, the transverse walls м thicker than longitudinal ones. Perichaetial leaves convolute, 2.6–3.5 mm long X 0.6-0.7 mm wide, 3-5X larger than vegetative leaves, yellowish or hyaline at the base; hyaline hair-points as in vegetative leaves but longer, to 3 in inflated perigonial buds. Setae erect, straight, seldom weakly arcuate, 1—4(5) mm long (including the vaginula). Capsules immersed to long-exserted, ovoid to cylindric, 1.0— 2.5 mm long X 0.5—0.7 mm wide, smooth, with 7— 15 stomata at the base, yellowish when deopercu- late, chestnut when empty; exothecial cells 18—50 um long X 18-20 um wide, mostly rectangular (2: 1), although many isodiametric intermingled, with thin walls; annuli well developed, compound and revoluble, consisting of 3—4 rows of inflated, hya- line cells; peristome teeth 50-80 jum wide at the mouth, entire or fairly broken at the apex, the outer surface nearly smooth below, papillose above, the inner surface papillose throughout, orange to red- mm. Androecia terminal, dish, contrasting in color with the rest of the spo- rophyte (in recently deoperculate capsules that pre- serve intact peristome); opercula from mammillate to long-rostrate, beak straight, in some populations 0.75 mm long, red. Calyptrae mitrate, seldom cu- cullate, covering only the opercula. Spores 8-12 um, granulose. Illustrations. Figure 1; Afonina (1986: Ris. 1 figs. 1-8, sub С. affinis; Ris. 2 figs. 9-16, sub G. ovalis); Cao & Churchill (1995: plate 1); Cao & Vitt (1986: figs. 1, 2); Deguchi (1978: figs. 12, 13; 1984: fig. 6; 1987: figs. 6, 7); Ignatov & Cao (1994: fig. 7); Ireland (1982: pl. 134); Jóhannsson (1993: fig. 36); Maier & Geissler (1995: Abb. 1 Distribution (Fig. 2). Grimmia longirostris is known from the Americas, Eurasia, and Africa, and is not known from Australia and Antarctica. It rang- es altitudinally from 530 m at Lake Baikal in Ir- kutsk to 5300 m at Mururata in Bolivia. It grows on dry and exposed siliceous rocks, mostly above tree line, but also at lower elevations at high lati- tudes. е specimens examined. U.S.A. Alaska: Alaska district, Healy Quadrangle, along Denali daar T Sa avege River, Hermann 21555 (BM). Arizo- ruz Co., Baldy Trail, p or eae ви Bartram ms (NY). California: San Di ^ rom 6 Springs, Allen 648 ко. FS inn Roc dy Е Park, SW slope of Deer Mountain, Hermann 27789. pes Ha awaii: Maui, ари National iy narrow ravine 50 f s Grove, Hoe 346. NY). ligo cà gd ilegible] 10 Aug. 1902, 3h a d s.n. (FH). Montana: Madison Co., 12 mi. E of Ennis, Hermann 17937 С). New Mexico: Santa Fe Co., Santa Fe С anyon, 14 Oct. 1930, Arsene s.n. (FH). Oregon: Was- co Co., Dalles, Sep. 1933, Frye s.n. (MO). South Dakota: Pennington E. Black Hills, along Pine Creek, just W of Mount к, Churchill 8186 (IBA-7361). Texas: Jeff Dads Co., St. Davis, Orcutt 7073 (FH). coa mont: Newfane Mt., 27 le 1902, Grout s.n. (FH). W oming: Sheep Mt., Goodding 2101 (FH). CANADA. Brit- ish Columbia: head of Summit Lake above Fort Nelson (near mile 108), Correll 11990 (FH). Labrador: Churchill Falls, е - Camp area, Brassard 6511 (BP-156326). est Territories: Baffin Island, Pangnirtung, Po- lunin "m A20 (FH); Hunter Bay. Е of Sloa of McTavish Arm, Great Bear Lake, : s.n. (H). Ontario: Muskoka Dist. Mun., Foot's Bay, ca. 9 km S of Freeman Township along Moon River, Ireland 23944 (FH). Québec: Ungava Bay, valley slope and rol- ling upland N of Leaf River and 100 miles from І ,eaf Вау, Aylmer, 6 HILE. Andes (FH-Sull). ). Jacobshavn, Grónland, 1883, Hastrüp s.n. (ЕН); spi trt Narssarssuaq, Fjellgebiete zwischen Qoroq Fjord und Kiagtut ida pon 26 Aug. 1979, Frahm s.n. (IBA-5072) m Precor below El Salto, Sharp 1833 (FH). Méxi a desviación al E de Techuchulco, Castilla Joss e Hi idalgo: Dub- lan, Pringle 15076 (JE). Jalisco Colima, 27 July 1983, Delgadillo s n. (ALT. cán: Paracho, about 6 km N on way to Cherán, Frey 3072 FH). Puebla: ladera NW del Pico | de Orizaba, 22 Apr. 1980, Delgadillo s.n. (ALTA). i fornia Sur: summit of Sierra de La 5 mi. E of Todos los Santos, Delgadillo et ез 3093 (FH). Ve- = racruz: cima del Cofre de Perote, 7 Dec. 1979, Delga- dillo s.n. (ALTA). ER. tm Cerro la ufa, 9 June Cárdenas s.n. (ALTA). GUATEMA Quezalte- nango: Cerro La Pedrera, x of I TEA 2122 65530 (FH). Sacatepéquez: slope ЊЕ Santa María de Jesús, indio! "65262 (FID. HON. У mpira: Мотаћа Celaque, summit of Cerro e Тын “Allen 12260 (MO). COSTA RICA. Cartago: near 356 Annals of the Missouri Botanical Garden Figure l. Grimmia longirostris.—a, b. Leaves.—c. Basal marginal leaf cells.—d, f. Basal paracostal leaf cells.— e. Apical leaf cells.—g. Transverse section at basal part of leaf.—h. Transverse section at medial part of leaf.—i. Transverse section at apical part of leaf.—j. Basal exothecial cells and stoma.—k. Medial exothecial cells.—1. Annulus and peristome teeth (only contour shown, not papillosity). (b, f: isotype of Grimmia columbica, H-SOL; a, c—e, g-i: Redfearn 34610, МО; j-l: Liu 952096, МО). 357 Grimmia longirostris Mufioz Volume 85, Number 2 1998 'perpnis sueuiroeds uo poseq s1s0412u0] DUI) jo uornnquistp рром 72 әш 358 Annals of the Missouri Botanical Garden summit of [Inter American Highway at La Ascensión, Cros- y & Crosby 6137 (MO). DOMINICAN REPUBLIC. La Vega: 6.3 km S of Valle Nuevo, Steere 23052 (G). San á, Los Laches, ie 2011 (PC). Boyacá: near the Ritacuba glacier, a Nevada de Cocuy, Grubb & Guymer B.210 (BM). ^ ы. Páramo Frailejonale, near Vetas, Killip & Smith 17987 (BM). VENEZUELA. Mérida: 2. Lib- ertador, Sierra Nevada de Mérida, near Laguna de Los Anteojos just below the loma Redonda ш. Griffin et al. 398 (NY); Distr. Rivas Dávila, páramo La Negra, above ~ Napo: Cerro. Мн, Grubb 2500 (FH). PERU. eash: Huaraz Province, Monterrey bei Huaraz, Hegewald & Hegewald 7482 (MO). Apurimae: Andahuaylas Prov- ince, Pampa Runtojocha bei Chincheros, Hegewald & He- gewald 5761 (MO). de Wanda 1. Vargas 15863 (NY). Cuzco: Provin- cia Canas, El жи anso, Vargas 10039 (FH). Junín: Jauja Province, La . Laguna Mesapata bei Canchayllo, He- gewald & үлке? 5375 (MO). La Libertad: Otuzco Province, Huancamarca, Quebrada Hornillo, Hegewald & Hegewald 5156 (MO). Lima: Canta Province, 2 km vor Canta an der StraBe nach Lima, Hegewald & Hegewald 7425 (MO). Puno: Miajachi, Huancané, Aguilar 150 (FH). BRAZIL. Rio de Janeiro, Serra de Itatiaia, Mar. 1894, Ule s.n. (GOET). BOLIVIA. Cochabamba: 28 km NW of Cochabamba, NW slope of Mt. Tunari, near Liriuni Aguas Termales Hotel, Hermann 25149 (ALTA). Ama- suyos: vicinitis Achacache, Padchani, Mandon 1635 (BM). La Paz: Inquisivi Province, km 2 above Quime, Lewis — (MO); Mururata, Sep. 1924, Jaffuel s.n. (FH). Larecaja: vicinis Sorata, Ancouma, qudm Рећ- ав, "meia 1634 (BM). ARGENTINA. Tucumán: Ta cumbre del Potrerillo, Lamb 2 (ЕН). Tierra del Fue- go: lac Cami, base de la Chaloupe, Skottsberg 370 (PC). ICELAND. E-Iceland, Egilsstadir. Aptroot 4961A (NY). Northeast Iceland, Asbyrg, a glacial valley between Мў- vatn and Akureyri, 7 July 1964, Steere s.n. (NY). S-Ice- land, Kirkjubaeklaustur, Шш 4939 (NY). NORWAY. Hamar, Stift Foldaleu, Laughé, 20 July 1890, Conradi s.n. (TRH); Trondheim, Stift Dovre, Kongsvold, 24 July 1890, Conradi s.n. (TRH); in canyon of Skibotelva River about A km from Skibotn, Storfjord, Troms, Norris 69524 (MO). WEDEN. Gestriciae [Gástrickland], Hartman (Н); Hol- miam [Stockholm]. Hartman (H). RUSSIA. Tomsk Oblast: Bijskij u., Kusnezow 2066 (LE). Krasnoyark Kraj: Minusinsk, 88, Martinoff s.n. (H-BR). Yakutsk A.S.S.R.: middle inp of Indigirka river, near mouth of In'yali Creek, 23 June 1976, Afonina s.n. (LE) Magadan Oblast: Chukotka, ne Achchen, 30 a 1970, Afonina s.n. (LE). Gorno Altayskuya. Auto mous Oblast: Altai Mountains, ARAE kl Creek, Ignatov 8/48 (IBA- 7039). Irkutskaya Oblast: la- j: Klukhorskij ajon, Teberda. Tebe idis rve, 55, Abramova & Abramov s.n. (LE). UNITED KINGDOM. 2. а Seat, = inburgh, Aug. 1866, Hunt s.n. (NY). GERMANY. Falkenstein, 25 July 1898, Vocke s.n. Cano: Pohlberges b. Annaberg, Weicker (FH); Sylvae Nigrae (Hollinthal), 1824, Braun s.n. (BM); Westfalen, Rüthen, 13 Jan. 1862, Müller s.n. (BP-36103). POLAND. Probsthainer Spitzber- ges bei Bunzlau, 23 Apr. 1867, Limpricht s.n. (FH); Su- deten, Kaulfuss s.n. (BM). FRANCE. St. Etienne de Bai- . 9 May 1908, Fleischer s.n. (FH); Vogesen, Dép. ^ Наше-5абле, straße von Col 1898, Glowacki s.n. (JE). Tirol: ad margines rivi беш pr. pag. Oetztal, 30 July 1923, Vajda s.n. (BP-159024). CZECH Mu T он Арг. 1901, Schmidt s.n. (IBA 5361). Morava: Jesenik, Vrbno, FAS 1949, pecu s.n. з 159134Ь). SLOVAKIA. =. Mala Fatra, prope am Stary hrad, Pilous s.n. June 1935 (BP-112583). PORTUGAL. Azores, Fayal, May 1937, Persson s.n. (NY). SPAIN. Álava: Sur de Umandia 7 Sep. 1982, Heras s.n. (IBA-1018). Lérida: Coll de Can- tó, 9 Sep. 1981, Casas s.n. ium 3965). Santa ien de Tenerife: m (BM). ge e Sankt Michael, 25 June 1984, mhold s.n Vallvellena, 2. (RO). КОНТА. Distr. Chokha- tauri, vicinitas pagi Bakhmaro, mons Mzis-chasvlis-mta, 21 July 1980, Vasák s.n. (NY); in alpe Kasbek ad fl. Terek, Brotherus & Brotherus 121 (H-SOL). KAZAKSTAN. Alma Ata Oblast: Калу ЕЕ Allen 10849 (MO). Semipa- latinsk Oblast g. Narymskij, 22 July 1899, Lad- ж (МО). 2. Issyk-Kul’ Oblast: Thian Schan Mts., ~ fontes А. Narym, 27 July 1896, Brotherus 1. (FH). MONG vh ІА. Bulganskij aimak: Gurvan-bu- lak somon, S Khugne-Khan, NW ne of шыга МІ., - Nov. 1975, Tsegmed. s.n. (LE). ak: E cia -somon, Namasuren-ul, Schubert M183 (MO). Ubsu- : Turgen somon, N s un - Turgen Mt., Tegmei 248 (NY). Chubs a Aim y 1983, Huneck s.n. a. ae :hóchi-ul, : a westlich Chiarga 2. M295 (MO); about 7 km W of Ulan Bator, head of small valley, Jeffrey 1535 (BM). 1. ENISTAN. Prope Da-tzzjanj-en montes, 13 June 1893, Potanin s.n. (LE). PAKIS | Kashmir, Astor Val- ley, between 1 апа Harcho, 15 Aug. 1892, Duthie s.n. (H-BR). TIBET. Nielamu Co. Zhao 117 (ALTA); Ya- dong; vf an 373 (ALTA); Zhun Ba, Jie a Agi 1093 (KUN- ; Milin Co., ioni 658 (MO). CHINA. Anhui: Mt. . D. Li 2227 (ALTA). Guangxi: uM Co., Mt. Miao Er, Gao & Chang 1623 (ALTA). Heilongjiang: Mangui, Gao 12896 (ALTA). Jilin: Mt. Chang Bai, Aur 5091 1. Shaanxi: Taibai Co., Taibai Shan, Li 642A А n: Songpan Co., hwy. Zhangla to Huanglong, Redfearn 35234 (МО). НиЬе е fE 2% = Tian Shan Range, Zhao 225 (MO). Yunnan: Dali Co. E-slope of Diancang Shan > chi, Redfearn Jr. et al. 1091 gis TAIWAN. Ilan Co., Mt. Nan-hu-ta-shan, Peng Ching-1 83-31 (MO). NEPAL. Mo- raine droite du camp de base à Lobuje, 24 Apr. 1952, Zimmermann s.n. (G). JAPAN. Mt. Kitadake, de (ir mura, Nakakoma-gun, Yamaguchi Pr E 5 1974, Deguchi s.n. (TNS-50119); Simozuke, M e 13 Ju ly 1927, Sasaoka s.n. (BM). BHUTAN. B 197 (МО). INDIA. Kumáun, Kalámoni forest, 4 Aug. 1900, Iráyot s.n. (H-BR); Madras State, Madura District, Palni Hills, Kodaikanal and surrounding region, Kodaikanal, 1924, Volume 85, Number 2 1998 Mufioz 359 Grimmia longirostris Foreau s.n. (MO); Nilghiri Hills (SW India), Doddabetta, p -BR); Simla, рана imalaya S NKA. Central Province, Thwaites CM29 (BM). „оом Sabah: N Borneo, West Coast Res., Mt. Kinabalu, Meijer B10389 (MO). PHILIPPINES. Luzon Island, Mountain Prov., sum- mit of Mt. Pulog, Tan 82-195 (FH). PAPUA NEW GUIN- EA. Eastern Highlands, Bismark Ranges, Mount Wilhelm, Weber & McVean B-32146 (MO ETHIOPIA. Bale, Bale Mountains Natl. Park, Sanetti Plateau near the foot of Tullu Deemtu Peak, Petelin 37-7 (MO). CAMEROON. Cameroun Mountain, Meoli Ndiva Ridge, Victoria Div., Western Province, Brenan & Rich- ards 4269 (MO). UGANDA. Between head of Butandiga ridge and Jackson's Summit, Mt. Elgon, N. Bugishu Co., Mbale district, Eastern Province, Wood 1160 (NY). KE- NYA. Africa orient., Mt. Kenya, Troll 5868 (JE). ' ZANIA. Arusha: Arameru Dist., Mt. Meru, W side of cinder cone, just off trail from end of road from Forestry Training Institute, Gereau 1649 (MO). Kilimanjaro: Mt. Peters Hut, Hedberg 12384 (MO). SOUTH AFRICA. Dolme Hill, Kaffraria, Cape Pr., Sim 7218 (NY); Natal Drakensberg, Giants Castle Game Re- 1 1074 slopes above Mlanlane Forest Station, Rooy 2282 DE LESOTHO. gan Mountains, W tween Butha Buthe and Mokhotlong, i 3124 (MO): Sehlabathebe National Park, hills aro sandstone out- crops E of Lodge, Magill 4269 (MO). "MADAGASCAR. Fianarantsoa, Andringitra summit of Anjavidilava, 37 km S 2. Crosby & Crosby 6980 (MO); Mt. Tsara- 22 km S of E of St. Denis, Crosby & Crosby 9648 (MO). Grimmia. longirostris can be identified with con- fidence by the shape of the costa in transverse sec- tion (Fig. 1—1). It is reniform and curved, with both the ventral and dorsal surfaces concave. The ven- tral sinus is U-shaped. It is rather unfortunate that both Sayre (1952: 255) and Greven (1995: 24) ne- glected the importance of leaf cross sections as a source of taxonomic characters in Grimmia. From my studies I realized that information obtained from leaf cross sections is absolutely necessary to name some species correctly and extremely useful to name many others. In the case of G. longirostris, the leaf cross section alone serves to identify this taxon with certainty. The leaves of this species are usually described as keeled, but I reserve this term for laminae that are V-shaped in cross section, not U-shaped, as in G. longirostris. Characters that give additional support to the identification are long, in- crassate, and nodulose basal paracostal cells, the straight seta, and the usually cylindrical capsules with a compound, revoluble annulus of rectangular hyaline cells. Grimmia longirostris is widely distributed and varies accordingly. Most of this variation involves quantitative features that totally overlap in mea- surements. Length of setae, capsules, and opercula vary broadly, but there are no correlations among these or with other taxonomic features. Specimens distributed by Sullivant and Lesquereux in their usci boreali-americani ed. 2," n^ 214, for ex- ample, have both long and short setae in the same cushion. Consequently some capsules are relatively long-exserted, whereas others are included among the perichaetial leaves. Populations with immersed capsules from the southwestern United States were denoted as G. catalinensis E. B. Bartram, and from hina as G. subimmersa Broth. (nom. nud.). De- guchi (1987: 27) discussed the variation in oper- culum length and concluded that it is not correlated with other characters. Hair-point length is, as in many other Grimmi- aceae, quite variable. Some taxa were described on the basis of muticous morphs of G. longirostris (e.g., G. hausmanniana De Not. from the European Alps, and G. catalinensis var. mutica E. B. Bartram from the southwestern United States). Cao and Vitt (1986) have previously reported wide variability in all these quantitative characters. They discussed and illustrated seta, capsule, and operculum length variability in Chinese specimens of G. longirostris (as G. affinis Hornsch.). Grimmia longirostris also shows variability with regard to some qualitative characters. The calyptrae are usually mitrate, but occasionally also cucullate, even in the same cushion. Some populations, es- pecially from South America, show a rope-like dis- position of leaves around the stem, much like С. funalis (Schwügr.) Bruch & Schimp. This feature is especially pronounced in Bolivian plants that have been named С. speirophylla Herzog and Ecuadorian populations on which G. columbica De Not. was based, but there is no correlation between this fea- ture and other sporophytic or gametophytic char- acters. In North America G. longirostris is more abun- dant along the Alaskan Range, Sierra Nevada, and the Rocky Mountains in the west and the Adiron- dacks in the east, with scattered localities in Qué- bec, Newfoundland, the Northwest Territories, and Ontario in Canada. It is absent from the Interior Highlands, the Appalachian Mountains, and most of the interior of Canada Southward, Grimmia longirostris is à common plant of the high altitudes of the southern part of the Sierra Madre Occidental and transverse ranges between Veracruz and Jalisco states, Mexico. Col- lections from Guatemala, Honduras, and Costa Rica are from the highest mountains of these coun- tries. Grimmia longirostris also grows on several 360 Annals of the Missouri Botanical Garden Caribbean Islands, with most records from high al- titudes in Hispaniola. In South America it is common along the Andean range north of 28S, with a disjunct locality at Serra do Itatiaia, near Rio de Janeiro in Brazil, and an- other in South Patagonia. Other localities in the intervening Апдеап areas аге to be expected, where conditions are suitable for G. longirostris. African collections have been made mostly in the ranges along the east coast, with only one disjunct western locality on Mt. Cameroun. This distribution agrees with White's (1978, 1993) pattern of Afro- montane vegetation "islands," continuous corridor formed by the six main East African ranges (Ethiopian, Usambara, Kivu-Ru- wenzori, Uluguru-Mlanje, Chimaimani, and Drak- ensberg) and disjunct, outlier populations in high elevations of western Africa. The intervening low- land areas, dominated by savanna or tropical rain- forest, lack the conditions for the establishment of Afromontane vegetation. has not been recorded from the Chimanimani or Uluguru-Mlanje systems; however, its presence in Drakensberg and the northern Ethiopian and Usambara systems suggests that it may be found there. Malagasy collections fit the proposed distri- bution of Afomontane vegetation for Madagascar (White, 1978). Grimmia longirostris may yet be found in the high elevations of the Atlas Mountains in Morocco and Algeria, where conditions are ap- parently appropriate. n Europe, G. longirostris is rare toward the south, growing only in scattered localities in the Pyrenees and the Caucasus. In central Europe it is relatively common, mostly associated with the Alps and the Carpathian Mountains. It is also known at higher latitudes, in Scotland and Fennoscandia. In mainland Asia, Спттиа longirostris is com- with a more or less Grimmia longirostris mon along the main Siberian mountain systems and along the Himalayas, Tian Shan, and the Tibetan Plateau. In tropical Asia, G. longirostris is restricted to the highest peaks of the Nilgiri Hills in India, cen- tral Sri Lanka, Mt. Kinabalu in Borneo, northern Luzon in the Philippines, New Guinea, and Taiwan. Strikingly, I was not able to find Grimmia lon- girostris among the Australian and New Zealand collections studied. This is surprising since other common Grim species, viz., G. pulvinata (Hedw.) Sm. an с. trichophylla Grev., grow in both areas, where the environmental conditions should likewise support G. longirostris. In this study I have found the following hetero- typic names to be new synonyms of Grimmia lon- girostris: G. affinis, G. affinis var. ramosissima, G. afro-ovata, G. allionii, G. antillarum, G. arctophila subsp. labradorica, G. breviexserta, G. catalinensis var. mutica, G. columbica, G. elata, G. hausman- niana, G. herzogii, G. integridens, G. itatiaiae, G. leucophaeola, G. maido, G. nano-globosa, G. neilgherriensis, G. nigella, G. obliqua, G. ortholoma, G. ovata var. praecox, G. peruviana, G. praetermissa, G. raphidostega, G. schimperi, G. speirophylla, G. speirophylla f. humilis, G. stenopyxis, G. subovata, G. trollii, G. vernicosula. The following species had been previously considered synonymous with G. longirostris or G. affinis: G. bogotensis (Churchill, 1994), G. cinerea (Churchill, 1994), G. itatiaiensis (Cao & Churchill, 1995), G. micro-ovata (Deguchi, 1987), G. ortholoma (Allen, 1995), G. stenopyxis (Cao & Churchill, 1995), and G. sumatrana (De- guchi, 1986). The following names had been considered syn- onyms of Grimmia ovalis (Hedw.) Lindb., a species that can be separated from G. longirostris by its plane margins and its indistinct costa in the distal half of the leaf: G. akaisi-alpina (Takaki et al., 1970); G. catalinensis E. B. Bartram (Jones, 1933); G. cavifolia (Ignatov & Cao, 1994); G. ovataeformis Kindb. (Allen, 1995). The nomenclature around Grimmia affinis has а tortuous history. Three names must be kept in mind: Grimmia apiculata Hornsch., G. affinis Hornsch., and G. ovalis (Hedw.) Lindb. The prob- lems concerning the relationship and differentiation between the two latter species were resolved by Sayre (1951). There is one dioicous taxon, Grimmia ovalis (Hedw.) Lindb., and another autoicous taxon, for which Sayre (and most authors thereafter) used the name С. affinis Hornsch. But the latter name is an illegitimate homonym that cannot be employed, according to the ICBN (Greuter et al., 1994), be- cause it was applied to two different taxa in two separate issues of the same journal (Hornschuch, 1819a, 1819b). Hornschuch employed the epithet affinis in the first issue of the journal (Flora 19: 85, published 14 Feb. 1819) to describe the taxon currently known as G. apiculata, which has an arcuate seta. The same name was again employed by Horn- schuch in the second issue of the same journal (Flora 19: 443, published 28 July 1819) to describe the species with a straight seta, for which the epi- thet affinis has been widely used, and he changed the name of the first G. affinis to G. apiculata. The name change by Hornschuch was not the result of some printing error or mistake (Hornschuch, 1819b: 442). Thus, according to the ICBN (Greuter et al., 1994: art. 51), G. apiculata is an illegitimate name because when published it was a superfluous Volume 85, Number 2 1998 Mufioz Grimmia longirostris name for the previous valid and legitimate G. affinis (curved seta). The second G. affinis (straight seta) became an illegitimate homonym, and thus inap- plicable (Greuter et al., 1994: art. 53). All the above problems concerning the use of the e G. affinis were pointed out by Mártensson (1956: 117-118), who hesitated to employ this name for the autoicous plant with a straight seta here recognized as G. longirostris. Even without having seen the second Hornschuch paper, Маг- tensson was correct in considering the second Grimmia affinis (G. longirostris in the sense of this paper) as illegitimate. Later, Deguchi (1978: 161) justified his use of the illegitimate name by citing art. 64 of ICBN (Seattle Code, Stafleu et al., 1972), which corresponds with article 53 of the current Tokyo Code. Presumably, he based his argument on the last provision of this article (corresponding with the current art. 53.6, Tokyo Code), assuming that both names were published simultaneously. This is not, however, the case (cf. Sayre, 1959); actually, one was published five months earlier. NOMINA NUDA (ONLY NAMES ACTUALLY PUBLISHED EITHER IN LITERATURE OR IN EXSICCATAE) poss commutata f. brevipila Broth., in sched. Musci urkestanici, n* bis leucophaeoides Mull. in. Exot., Suppl. 1: Grimmia micromitria Schimp. ex x Müll. Hal., Nuovo Giorn. ot. Ital. n.s. 4: 165. 1897 Grimmia obliquata Host ex Brid., Bryol. Univ. 1: 180. Hal. ex Kindb., Enum. Grimmia ovata f. glacialis P. de la Varde, Ark. Bot., n.s 3: 55. Grimmia preussii Broth. ex Paris, Index Bryol. Suppl.: 175. Grimmia semipilosa Hampe ex Müll. Hal., Linnaea 43: 4 Grimmia paeem Schimp. ex Müll. Hal., Linnaea 43: 461. 1882. Grimmia subovata var. laevipila Schimp. ex Paris, Index Bryol. 2: 538. 1895. Grimmia tenuicaulis Hampe allischen Naturwiss. 354). 1874 e ex A. Jaeger, Ber. Tatigk. ч, Сез. 1872—73: 72 (АЧ. 1 Literature Cited Afonina, O. M. 1986. Additamenta ad bryofloram pen- 2” Czukotka. 4. Novosti Sist. Nizsh. Rast. 23: 222— | эң в. 1995. Eight neglected species of Grimmiaceae (Musci) from North America. Fragm. Florist. Geobot. 40: 159- Blom, H. H. 1996. A revision of the Schistidium apocar- m complex in Norway and Sweden. Bryophyt. Bib- lioth. 49: 1-333. Cao, T. & S. P. Churchill. 1995. New synonyms in Grim- mia and Schistidium (Bryopsida: Grimmiaceae). Nova Hedwigia 60: 505—513. D. Н. Уш. 1986. А taxonomic revision and phylogenetic analysis of Grimmia and Schistidium (Bryopsida; Grimmiaceae) in China. J. Hattori Bot. Lab. 61: 123-247. Churchill, S. Р. 1994. New synonyms and deletions for the moss floras of Colombia and Ecuador. Trop. Bryol. 9: 1—4. Crosby, M. R. & R. E. Magill. 1994. Index of ен 1990- ps Monogr. Syst. Bot. Missouri Bot. Gard. 5 [ivi], 1-87. . Bauer. 1992. Index of Mosses 1963- 1989. Monogr. Sl Bot. Missouri Bot. Gard. 42: [i-vi], 1-646 Crum, H. 1994. Grimmiales. /n A. J. Sharp, H. Crum & P. M. Eckel (editors), The Moss Flora of Mexico. Part One. Sphagnales to Bryales. Mem. New York Bot. Gard. 69: 386-415. Deguchi, H. 1978[1979]. A revision of the genera Grim- Schistidium and Coscinodon (Musci) of Japan. J. Sci. Hiroshima Univ., Ser. B, Di ы 9 Grimmiaceae (Musci, oue (editor), inia on Cryptogams in Southern Chile. Kensei- бе То 198 News on some Asian species of Grimmia. Hikobia Ee 3071829 ——— . Studies on some Peruvian species of the 2. (Musci, Bryophyta). Pp. 19-74 in H. In- oue (e ditor), Studies on роле: in Southern Peru. niv. Press, T: Barrie, н. M. Burdet, W. С. Chaloner, V. Demoulin, D. L. Hawksworth, P. M. Jørgensen, D. H. Nicolson, P. C. Silva, P. Trehane & J. McNeill (Editors). 1994. International Code of Botanical Nomenclature (Tokyo E Less Veg. 131: d; Т жүй, 1-389. Greven, H. C. Grimmia Hed ci) in eds ры кн Leiden Hornschuch, C. F. 1819a. Neue Laubmoose. Flora 2: 81- 112. Mus- -----. 1819b. py botan. Notizen aus England und E Flora 2: 440-444 Horton, D. H. 1982. A revision of the Encalyptaceae (Musci), with Caprio reference to the North Ame can taxa. Part 1. J. Hattori Bot. Lab. 53: 365-418. Ignatov, M. S. & T. Cao. 1994. Bryophytes of the Altai Mountains. IV. The family Grimmiaceae. Arctoa 3: 67— 1 Ireland, R. R. 1982. Moss Flora of the Maritime Prov- inces. Publ. Bot. (Ottawa) 13: 1-738 Jóhannsson, B. 1993. Islenskir mosar. Skeggmosaztt [Grimmiaceae]. Fjölrit ДР ам ын Has 24: Jones, G. N. Grimmiaceae. /n A. J. Grout (editor), Moss al o hs America 2: 1—66, pl. 1-25. New- fane, Verm Maier, E. & P "Geissler. 1995. Grimmia in Mitteleuropa: i 0. gei Musci, Grimmiaceae. Fragm. Florist. Geobot. 392: 667-670. Sayre, G. 1951. The identity of Grimmia ovalis and G. commutata. Bryologist 54: 91—94. 1952. Key to the кон of Grimmia in North diuites: Bryologist 55: 251-25 Annals of the Missouri Botanical Garden ----- 1959. Dates of dw ations describing Musci, 1801— 1821. 4. Sage College. Troy, New York. Stafleu, F. A., ;. B. Bounss R. MeVaugh, R. D. Meikle, R. C. Rollins, 3 ies J. M. Schopf, C. M. Schulze, R. и Vilmorin & E. G. Voss (editors). 1972. International ode of Botanic | мане: lature. focum Veg. 82: 426. Takaki, Т. Amakawa, T. Osada & E. Sakuma. 1970. Bryo Hed flora of Mt. Kaikoma, Mt. Senjo and Mt. Ki- tadake (Southern Japan Alps). J. Hattori Bot. Lab. 33: 171-202. White, К. 1978. The Afromontane Region. Pp. 463-513 ] Werger (editor), Biogeography and Ecology of к Africa. Dr. W. Junk, The Hague 1993. Refuge theory, ice- age aridity aud the his- tory of 52 biotas: An essay in plant geography. Fragm. Florist. pq. c Wijk, R. van der, . Florschütz. Es Index muscorum 2. D-Hypno. Rus Veg. 26: l- 1969. Index muscorum 5. T-Z, ‘Appendix: | eee Vex. 65: [I]-xii, 1-922. INDEX TO EXSICCATAE Aguilar, P. G. 150 (FH), 18454 (FH), 18493 (FH). 18545 (FH), 18709 (FH), 18775 (FH); Allen, B. Н. 648 (MO), 9511 (С), 10849 (MO), 12260 (MO); Allioni, M. 8095 (H-BR); Allred, К. W. 6363 (MO), 6376 (MO); Al- E P. 228 (FH); Amatt, T. 1401 (FH); André, E. 591 (NY); iv А. 4939 (NY), 4955A (NY), 4961A (NY); Arnell, 2368 (H-SOL); Asplund, E. I: = D GE); 78 nL JE, UPS); Aur, C.-W. 5091 t aldwin, D. D. 224 (H-BR); Barclay, Н. С. > Jua- iio 1950 (IBA); Barnes & Land 368 (JE); рис. В al. 990 (ЕН); Bartram, E. B. 268 (ЕН), 283 (FH), 3: 56 (Е Н), 383 (ЕН), 446 (ЕН), 621 (ЕН), 683 TN 083B (FH), 788 (ЕН), 808 (FH), 819 (FH), 829B (ЕН), un 855 a NY), 857 (FH), 862 (FH). 1102 (FH). 1197 (FH), 1198 (FH), 1200 (FH), 1250 (FH), 1276 (FH, NY), 1310 (FH), 1695 (FH); Bell, B. С. 154 (BM, С, IBA), 155 (BM, FH, G); Benoist, R. 2276 (NY), 3153 (PC), 3386 (PC); Brass, L. J. 30026 (ЕН); Heres G. R. 6511 (BP), 15284 (FH); sud A. H. © Brotherus 121 (H- SOL); Brotherus, V. F. 244 (H- ER p ys G. & O. Buchtien 1750 n: BR): Buck, W. 7 (NY), 8396 (NY), 13341 (NY), 14040 (NY), уе A 14277 (NY), 14289 (NY). Cao, T. & B. Yan 199 (ALTA), 266 (AL TA): Castilla, M. 2030 (MO); € Chang, K.-C. & J.-Y. К 4744 (ALTA), 5112 (ALTA), 5112 (KUN ; C he зла Sg I. V. 8 (LE), 35 (LE), 53 (LE); Churchill, S. P. 8186 1B A); Correll, D. S. 11990 (FH); Cremers, G. 1705 y, M. R. & С. A. Crosby 6980 (МО), 8438 (МО), 8824 (MO), 9648 (мо, ВАЗОА (МО), 8941 (МО), 6137 (МО); Crum, H. А. & L. E. Anderson 7797 (С). De Sloover, J. 17934 (MO); Deall & Killik 133 (MO); Delaney, K. S. 1646 (BM); Delgadillo, C. et al. 3093 (FH); Doulea, J. 3110 (Н-ВЕ); Dusén, P. 605 (H-BR), 924 (H- T p mid а] 2519 (ВМ), oc T. S. et Ki B-65190 (NY). Fam. J. -Р. E (IBA), 935069 (IBA); & E rey "H). . H. ap (ALTA), 2520 (BM), 2548 (BM), 257: 2. Sad E. L. 13630 (NY, PC); Elias, Frey, T. C. 974 (PC); Gao, С. 1194 (ALTA), 1197 1383 (AL 1. 12881 (ALTA), 12895 (ALTA). 12896 (ALTA), 13092 (ALTA), 13119 (ALTA), 13227 (ALTA), 22186 (MO), 22252 (MO), 22445 (ALTA); Gao, C. et al. n (ALTA); Gao, C. & E.-Z. Bai 34124 (MO); Gao, С. & K.-C. Chang 1623 (ALTA), 9132 i TA), 9202 `А); саме ans 4812 (ЈЕ); a К. 1649 (МО); Ger- main, P. 1142 (ЈЕ); Соод па, L. N. 433 (ЕН), 2101 (ЕН); Griffin, D. et al. 398 (NY), 2047 (МУ); eer 127 (MO); Grubb, F. J. 2500 (FH); Grubb, F. J. & D. A. Guymer B.8 (BM), B.216-A (BM), B.210 (BM), T (BM), B8 (IBA): Guo, H.-G. 17 (ALTA). Handel-Mazzetti, H. > © 240 (ВМ), 748 (BM), 966 (JE). 2058 (H-BR, JE), 3183 (JE), 4670 (JE); He, S. 31243 (МО), 31245 (MO). 31248 (MO), 31304 (MO), 31792 (MO), 31800A (MO), 31834 (MO), 951051 (MO), 951156 . 5076 (MO). 5 513: ; "5274 (MO), 5375 (MO), 5761 (MO), . 6029 en 6151 oed 6220 (IBA, (IBA. МО), 654: т 7 . 7495 (MO), 7482 (мо) (IBA, укн 7638 ee . 7605 .T 24 v 2619 2. . 3579 (Н- 4871 (JE). 4912 (BM). 4938 (JE); Hoe, W. J. fe 4. 3463.0 (NY); Hu 950144 (MO); Humboldt, 6 (BM); Hyvónen, J. & S. Stenroos 4032 (IBA), 3924 E 3946 (IBA). Ignatov, M. 0/918 (IBA), 0/475 (MO), 3/103 (IBA), 3/ 67 (IBA), 3/102 (IBA), 3/273 (IBA), 4/7 (IBA), 5/14 (IBA), 6/36 (IBA), 8/48 (IBA), 8/307 (IBA), 8/147 (IBA), 9/37 (IBA), 17/54 (MO), 17/43 (IBA), 34/161 (IBA), 34/218 (IBA), 36/103 (IBA); Ireland, R. R. 23944 (FH); Ireland, R. R. & G. ВеПоћо-Лтиссо 18557 (NY). Jeffrey, C. 1535 (BM); Jin, Z.-Z. 87A (ALTA). Kabir Khan 1835 (H-BR); к ^ner, Ae 18517 5. Kil'dyushevski 54/7 (LE); Killip, | А. L. Smith 17434. (BM), 17985 (BM), 17987 2" 17988 (BM); Ko- lenati 2568 n E); Koponen, T. 18252 (ALTA); Krivoshap- kin, К. К. 010404. (LE), 010602 (LE), 060107 (LE), 060108 Iv 060110 (LE); Kuchar, P. B8980 (MO); Kun- а С. 186 (МУ). Lai, E a 80- 171 (AL TA), 80- 184 ( i Du 6886 (MO); Lamb, 1. M. A (ALTA), 321 (ALTA), “ош (MO), (МО), 1102 (АТТА), (АТТА); 4. Н. pua с Lewis, M. 87-947 87-958 (МО); Li, X.-J. 28 (MO), 80-16 (KUN), (KUN), 642 MO 642A Mo 799 (MO), 1 D.-K.. 222 2. 2233 EIN Lin, 7. ы T а. 1102 (МО). agill, R. E. 4186 (МО), 4269 (МО), 6607 (МО), 6674 (МО), 6729 (ІВА); Mandon, С. 1631 (BM), 1632 (РС), 34. (BM, FH-Sull, С), 1635 (BM): Matthews, К. 1. 1125 FH); Meijer, W. B10374 (MO), B10389 (MO), B11947 (МО), B11952 (МО); Meyer, F. С. 5435 (ЈЕ), 5518 (JE). Norris, D. H. 69524 (MO). Oba, 1 Т. 103 (МСН); Огеш, С. 4. 5353 (ЕН), 7073 Peng Ching-I 83-31 (MO), 83-34 (MO), Pennell, F. W. 83-35 (MO); 18400D (FH); Petelin, D. A. 37-7 (МО); Volume 85, Number 2 1998 Mufioz 363 Grimmia longirostris Phillips, ХУ. 5. 2902 (FH); Pittier, Н. 13198 (МҮ); Polunin, №. 2293A-5 (ЕН), 2611A20 (FH); Poplavskaya, С. 141 (LE); Preuss 859 (H-BR); Pringle, C. G. 26A (JE), 15076 (JE) Redfearn Jr.. P. L. 34610 (MO), 34945 (MO), 34976 MO), 35162D (MO), 35213 (MO), 35234 (MO); Redfearn Jr.. P. L. et al. 599 (ЕН, МО), 1079 (MO), 1091 (FH, MO), Я 5 d 1659 (MO), 1814 (FH, MO); D. M. Eggers 16340 (MO); Richards, P. W. 4269 Tm Hibbing. | к 706 (FH); Rojkowski, 55 Ruinard, C. & ; . M. Jayasuriya 17/40 РЦ І (H-BR), 773 (ЕН, Н-ВК); Schubert. W. M183 (MO). М295 (MO); Selivanova-Gorodkova, E. А. 792 (LE). 5259 (LE); Sharp, A. J. 1833 (FH), 22904 (FH); . 5610 (NY); Shushan, S. 1038 (С); Sim, T. R. NY); Skottsberg, С. 370 (PC), 1600 (H-BR), 1603 (H-BR), 1607 (Н-ВЕ), 1618 (H-BR), 1628 (Н-ВН); Stan- q P. C. 65262 (FH), 65526A (FH), 227 Stark. L. В. & R. Castetter 2055 (MO); Steere, W. C. 7 - de 10129 (NY), 12436 (NY), 23052 (G), pie (NY); Stenroos, 5. 3050 (IBA). Takeda, ‚ J. Wang 860420-11 (NICH), 860406-7 (МЄН), 860406- BA (NICH), 860412-1A (NICH); Tan, B. C. 82-195 (FH), 89-521 (NY); The Expedition to the Tibet 7446 (ALTA), 7608 (ALTA), 7884 (ALTA); Thwaites, G. ~ | К. CM29 (BM); Townsend, С. C. 85/860 (ALTA); Troll, C. 64 (JE). 5570 (JE), 5860 (JE), 5868 (JE). 5868 (FH), 5868 (JE); Tsegmed, T. 228 (LE), 248 (NY), 747 (LE), 2442 (LE); Tuomikoski, R. & J. Kucyniak T1472 (JE). Ule, E. 1830 (H-BR), 1913 (H-BR). Valcarcel, L. 40 (FH); van Rooy, J. 2294 (NY), 3124 (MO). Su + (ЧҮ), 1104 qu 2282 (NY); Vareschi, L. 1274 argas, J. C. 5039 (FH), 7099 (FH), 7676 (FH). E | (FH), 8709 (E. 8780 (FH), 10039 (FH), 15863 (NY). Wang, M.-Z. 12503 (ALTA). Me (AU ңы! 'eber, W. A. & D. ҮТ. е (МО), В-32146 (М 0); Wed- dell, M. H. Weiss, J. zd y PC): ch, H. 16694 ы A Piesi R. S. M); Wood 1160 (NY); Wooton, E. O. 1501 (FH); m 8. Т. 1076 (ALTA), 1076A,B (KUN). 4. Biological Institute 47085 (МО); Xizang Team 7701 (MO); Xu, Ey "v (ALTA). ; E Zang, M. 21 REEN 93 ( (ALTA), 373 (ALTA) 428 (ALTA). 682 KUN), 890 (ALTA, KUN), 1093 (KUN). 1119 (KUN), 1358 (ALTA), m (AL ТА), 1865 (ALTA), 1996 ү, 5217 (ALTA), PON (ALTA), 5298 (ALTA), 5317 (ALTA); Zhao, K.-Y. A (ALTA), 117 (ALTA), 953299 (MO): Zhu, W.-M. ли ТА), 7716 (ALTA), 64020 (ALTA), 64627 (ALTA); Zhu, W.-M. & J.-L. Wu 64067 (ALTA). Volume 85, Number 2, pp. 215—366 of the ANNALS OF n MISSOURI BOTANICAL GARDEN was published on August 31, 1998. Flora of the Venezuelan Guayana Located in the southeastern half of Venezuela, the Venezuelan Guayana is the core area of what has been called *The Lost World." The area is dominated by massive table mountains e as tepuis and includes many endemic species and genera, with much of the area still n pristine condition. There are nearly 10,000 species in the flora area, and over half will be isse by line drawings. Volumes 3 and 4 of the Flora of the Venezuelan Guayana are now available from Missouri Botanical Garden Press: Berry, P. E., B. K. Holst, and K. Yatskievych, editors. Flora of the Venezuelan Guayana. Volume 3, сараас 1997. ISBN 0-915279-46-0. 774 рр. 1113 species treated. 628 line drawings. $67 Volume 4, и 1998. ISBN 0-915279-52-5. 799 pp. 1329 species treat- ed. 621 line drawings. $67.95. Also still available: Volume 1, Introduction (includes Vegetation Map and Topographical Map). 1995. ISBN 0- 88192-313-3. 320 рр. of text, plus 44 pp. of color plates, 10 b/w photos, 51 line drawings. $52.95. Volume 2, Pteridophytes, Spermatophytes (Кыша res Апен. 1995. ISBN 0-88192-326- 9. 706 pp. 1285 species treated. 618 line drawings. .95. Vegetation Map and Topographical Map, 2-map set: rolled and крвна: in tube $17.00; or professionally folded $15.00. Price. Ко. Total- 8 Volume 1 (& 2-map set) $5295. —— Make checks payable to: Volume 2 66795. == Missouri Botanical Garden Press Volume 3 $6795 ~ ш 4344 Shaw Blvd. Volume 4 $6795 "= St. Louis, MO 63110-2991 2-map set, rol 17. tes А Mess set, f deer Es SERA Or order using Visa or MasterCard: Add for shipping and handling: - Check one: Visa ——— MasterCard —__—_ Within the U.S. Card №. Exp. date for the first book $4.00 Signature --------- — | 1 ЕШ ІШ for each additional book — $1.00 phone, 314-571-9534 Canada and Mexico fax, 314-577-9591 for the first book : $6.00 e-mail, шу org for each additional book $150 .— web site, www. mobot.o Internation for the first book $8.00 Ship to: ierit i ааа for each additional Шок: МИР» Street Air rates available on request Total enclosed 22% CONTENTS Pollination of Petaloid — by Monkey Beetles (Scarabaeidae: Rutelinae: Hopliini) in Southern Africa ____---- Peter Goldblatt, Peter Bernhardt & John C. Manning 215 Pollination Ecology and Maintenance of Species Integrity in Co-occurring Disa racemosa L.f. and Disa venosa Sw. (Orchidaceae) in South Africa ______ ____ S. D. Johnson, K. E. Steiner, V. B. Whitehead & L. рде 231 Tribal ‘Phylogeny of the Asteraceae Based on Two Non-coding Chloroplast Sequences, the trnL Intron and ete oe Spacer __ ndall J. Bayer & Julia R. Starr 242 A Synopsis of the Genus Echinopepon 252221 Sicyeae), Including Three . Alex К. Monro & Peter J. Stafford 257 The Genera Cestrum and Sessea (Solanaceae: Cestreae) in Venezuela _ Carmen Benítez de Rojas & William С. D'Arcy 273 Materials Toward a Revision of Grimmia (Musci: Grimmiaceae): Nomenclature | and Taxonomy of Grimmia longirostris ess Muñoz 352 er illustration. Drawing by John Myers reproduced from cover of booklet for I à иаи Symposium, Missouri Botanical Garden. Annals © of the Missourl - Botanical Carden, T olume 82 Volume 85, Number 3 Summer 1998 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating out- side the Garden will also be accepted. All manuscripts are reviewed by qualified, independent reviewers. Authors should write the Managing Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Committee Michael H. Grayum Editor, Missouri Botanical Garden Amy Scheuler McPherson Managing Editor, Missouri Botanical Garden Diana Gunter · Editorial Assistant, Missouri Botanical Garden Vicki Couture Secretary Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical Garden Roy E. Gereau Missouri Botanical Garden Peter Goldblatt Missouri Botanical Garden Gordon McPherson Missouri Botanical Garden P. Mick Richardson Missouri Botanical Garden Henk van der Werff Missouri Botanical Garden For subscription information contact ANNALS OF THE MISSOURI GARDEN, % Allen Marketing & Management, Р.О. Box 1897, Lawrence, KS 66044-8897. Subscription price is $125 per volume U.S., $135 Canada & Mexico, $160 all other countries. Four issues per volume. The journal Novon is included in the subscription price of the ANNALS. 4€ mobot.org oe que- hip: www.mobot. org © Missouri Botanical Garden 1998 The mission of the Missouri Botanical Garden is to discover and share knowledge about plants and v ant Ger TU HU The ANNALS OF THE MissoURI BOTANICAL GARDEN (ISSN 0026-6493) is published quar- terly by the Missouri Botanical Garden. 2345 - Tower Grove Avenue, St. Louis, MO 63110. Pe- riodicals postage paid at St. Louis, MO and ad- ditional mailing offices. PosrMASTER: Send ad- | dress changes to ANNALS OF THE MISSOURI | BOTANICAL GARDEN, % Allen Marketing & Management, P.O. Box 1897, Lawrence. 66044-8897. . their environment, in order to preserve and enrich life. e Thi, "M P В sam | Аш. р ИЕ | ts of ANSI/NISO 239.48-1992 (Permanence of Pape. — Volume 85 Annals Number 3 of the NZ 1998 Missouri Botanical Garden A TAXONOMIC REVISION OF Jesús Mufioz GRIMMIA SUBGENUS ORTHOGRIMMIA (MUSCI, GRIMMIACEAE)! ABSTRACT Grimmia subg. Orthogrimmia comprises nine species distributed on mountain chains or high latitudes mainly in the Northern Hemisphere. This subgenus is defined by the following combination of characters: leaves V-shaped in trans- verse section, margins flat (seldom recurved in G. reflexidens), proximal жаныш, cells with the transverse rall thicker than the longitudinal walls, setae erect and straight (curved on y in G. are aria), and capsules symmetric and smooth. I recognize two sections: sect. Donnianae, including G. arenaria Ham “С. donniana Sm., апа 6. triformis Carestia & De Not.; and sect. Montanae, comprising G. alpestris (F. Weber & D. Mohr) Schleich., G. gy matan (Brid.) Јиг., С. montana Bruch chimp., G. nivalis Kindb., G. reflexidens Müll. Hal., and G. ungeri Jur. Grimmia brachyphylla Cardot, considered by other authors to be synonymous with G. montana, is shown to be neo 2 Coscinodon 2” and is here Гетвик бод. Grimmia 2. is considered to be synonymous with Coscinodon cribrosus, not with G. caespiticia, as previously believed. All taxa are keyed and described, and Kiei ide indicated. | secs and distribution maps are provided for each species. IL Grimmia Hedw. is the largest genus in the Grim- other subgenera of Grimmia by the following com- miaceae (Musci). It is distributed throughout the — bination of characters: leaves keeled, V-shaped in world, mostly in mountainous areas. Several infra- transverse section; margins flat (rarely recurved in generic taxa have been proposed within Grimmia, . reflexidens); proximal marginal cells with the always based on the European species. transverse walls thicker than the longitudinal walls; Grimmia subg. Orthogrimmia differs from the setae erect and straight (curved only іп G. arenaria); ! [ thank the curators of the herbaria cited in the text for the loan of specimens, and the individuals that have helped my efforts to complete this work, including C. Aedo, B. Allen, J. J. Aldasoro, S. he ran M. Mr #4 Rosa García, and Culture, and the visit to Helsinki (H) was paid for by the European Commission HCM Contract no. ERBCH- СЕСТ940065 with the Division of Systematic Biology of the University of Helsinki. Elena García and César González drew i plant figures. Finally, the help from my family was, and is, beyond measure. to Asturiano de Taxonomía y Ecología Vegetal, P.O. Box 8, E- 33120 Pravia, Spain. Present E. Missouri ды Garden, Р.О. Box 299, St. Louis, Missouri 63166-0299, U.S.A. E-mail: jmunoz@admin.mobot.o ANN. MISSOURI Bor. GARD. 85: 367-403. 1998. 368 Annals of the Missouri Botanical Garden and capsules ovoid to fusiform, symmetric, and smooth. I have recognized two sections: sect. Don- nianae, with three species, and sect. Montanae, with six. The species of this subgenus are distrib- uted along mountain chains mainly in the Northern Hemisphere. Only G. reflexidens is also known from the Southern Hemisphere. The taxonomic history of Grimmia has been re- viewed in an excellent manner by Deguchi (1978: 123-126). Table 1 summarizes several major tax- onomic treatments of the genus. Recently, revisions of Grimmia have been pub- lished for the Altai Mountains (Ignatov & Cao, 1994), China (Cao & Vitt, 1986), Europe (Greven, 1995; Maier & Geissler, 1995), Japan (Deguchi, 1978), and South America (Deguchi, 1984, 1987). However, much more work must be done to resolve the taxonomic and nomenclatural difficulties of this enus. Traditionally, the species dealt with in the pre- sent work have been treated (Hagen, 1909; Lim- pricht, 1890; Loeske, 1913) in Grimmia subg. Guembelia (Hampe) Schimp. (Schimper, 1856). However, the lectotype of Guembelia, designated by Pfeiffer (1874: 1511), is Grimmia elliptica Funck (nom. illeg.= G. ovalis (Hedw.) Lindb.), which be- longs to a different subgenus, usually known by the later and arguably illegitimate name Grimmia subg. Litoneuron I. Hagen (Hagen, 1909). Therefore, Grimmia subg. Guembelia is the oldest name for the group usually known as subgenus Litoneuron, and the group usually known as subgenus Guem- belia must be called Grimmia subg. Orthogrimmia Schimp., here lectotypified by G. donniana Sm. The species in the G. alpestris-G. donniana com- plex of subgenus Orthogrimmia have been treated at various taxonomic ranks in the literature. Where- as G. alpestris, G. caespiticia, G. donniana, G. mon- tana, and G. reflexidens (as G. sessitana) have usu- ally been considered worthy of specific rank, G. arenaria, G. nivalis, G. triformis, and G. ungeri have rarely merited more than subspecific recognition. Nevertheless, all are discrete entities, without mor- phological intergradations. These taxa vary across their geographical ranges, but always in secondary features, mostly quantitative and presumably hab- itat-induced. These quantitative features have been much employed in the literature, yielding disparate taxonomies. On the other hand, major structural characters, such as presence or absence of stomata, have usually been neglected. From Index Muscorum (Wijk et al., 1962, 1969) onward, all the species treated in this study have been commonly lumped with Grimmia donniana following Habeeb (1950: 75). His knowledge of the group, however, was in- adequate, as noted by Crum and Anderson (1981: 429). MATERIALS AND METHODS This revision is based on 1939 specimens from BCB, BCC, BM, BP, CANM, COLO, FH, G, GLM, H, JE, KRAM, KUN, LE, LISU, M, MA, MO, MUB, NY, ОР, РАУ, PC, S, ТСО, TNS, ТЕН, VIT, W, WRSL, Z, as well as from the private herbaria of J.-P. Frahm (Bonn), E. Fuertes (Madrid), and R. B. Pierrot (Dolus). Some type specimens requested from DR, GLM, LZ, MPU, and ROST were not found in these herbaria. Specimens were moistened in water with domes- tic detergent and then transferred to clean water. At the beginning of this study and, later, for types and specimens not immediately referable to any taxa, 10 to 15 mature but green leaves from the apical part of the stem (excluding the uppermost ones), transverse sections, the two or three inner- most perichaetial leaves, and a capsule sectioned in eight parts were mounted in lactophenol gel (Zander, 1983) or Hoyer's (Anderson, 1954) medi- um. In this treatment, seta length includes the va- ginula. Lamina length excludes hair-point, lamina width has been measured at the broadest part of the leaf, and fractions (e.g., “margin recurved in the proximal %”) always refer to lamina length. Cell measurements include the wall. Proximal paracos- tal and marginal cells refer to the two or three rows of cells closest to the costa and margin, respective- I have distinguished species on the basis of mor- phological characters. They are “taxonomic spe- cies" according to Grant (1981: 78—80). The char- acters employed are important throughout the genus, and even the family. For mapping distributions I have used ArcView GIS, which almost automatically loads geographical coordinates from database files with the “.dbf” for- mat. Distribution maps are based solely on exam- ined herbarium specimens. In “Selected specimens examined" I have cited only one specimen per geo- graphical unit, and the total number of specimens parenthetically at the head of the section. А com- plete list of specimens studied is available upon request. I have designated lectotypes for all names except those for which the author specifically selected in the protologue a single collection as holotype (Greuter et al., 1994, ICBN 9.1). The common prac- tice of accepting as holotype a specimen kept in the personal herbarium of the original author can be a source of error. Good examples of problematic 369 ufioz Revision of G Volume 85, Number 3 1998 ја Subgenus Orthogrimmia rimmia вашлојора грооду грорлођаца ја штрпи“ {әс штри« у "qns штрпу“ об umpisnpog штрп у "qns UMpusiyos DIUWUNLIOLP Y отшшпаволрхн "qns иороја ла р2]020]82.116 "qns uopojdKa(] әттолуү appui) „20194 umpiisnpog-opnosq sappun y ЕС ШТ) anuid]y “sqns ¿2DUDJUO y] 71296 4 SAULLOfi10]091U/) x $24152d] y “‘Sqns y DIaquiansy "qus x 52uLIOf9DIDA() 4 DYjaquany "qns 49D]DImumuor) uo4nauo]r] "9416 uo4nauojr] "qns „арорлојјАудоцлај + 21шишп180111() "9406 „раруаото] 1928 әрп”) apjasij2ay пц], грхојпара штапшозрлорпоа) ¿D19QUIINL) "iqus грорлогралриу жзәшаоЙ зә йү anjpnbao[ "п „220 шајпа ж01шштл) "qns aDIDUIIU/) әртоишат qns „зриптјал punu) "qns 8910110] отшипі8ордоцу :3qns piumunidopqpuy “sqns sauLiofi]opy piuuna4dopqpi[y 79406 apDj]&uydououg puwang дортлата pruunago43]sp«) 79406 отишла олојгр«) "24016 штујАцаоја2.16 DIWUNIZOLIISDY) "qns ртишлаболојр у "dqns anui) 71295 әрјәғалт”) пац], prunas) отишла“) отишло отишла) ю1шш1лг) piura) (0261) 24890] (6061) човен (peieorpur jou squex) (0681) 1joudur] (9681) 19duiuos (crgD (8681) злодритм 1oduiqos Y yong ром siy} ur Ҷим ЦРәр saroads opn[our yey] exej sojeorpur (y) xsuojse uy отшил) JO SUOISIAIDp оџоповелји SNOLIBA Ч 3Id*eL 370 Annals of the Missouri Botanical Garden typification are the many epithets associated with Kindberg or Müller (also cf. Ortiz, 1989). TAXONOMIC CHARACTERS This character overview refers to members of Grimmia subg. Orthogrimmia, except where oth- erwise indicated. GAMETOPHYTE Нађи. Тһе plants grow in dense compact cush- ions on rocks, mainly siliceous. These cushions are usually round and semi-spherical, although when growing in rock crevices they adopt the form of the crevice. The color of the plants varies among пи- ances of olive-green above, and usually blackish below. Stem. Stems of G. triformis may reach 2 cm high, but most species rarely exceed 1.5 cm. The stem diameter varies between 100 and 210 шп. Stem internal structure in Grimmia was studied in depth by Kawai (1965: 113-117) and Deguchi (1978: 128-131). Kawai distinguished three parts in transverse section: an epidermal layer, a cortical layer, and a central strand. According to the degree of differentiation among the three layers, Kawai de- fined four stem types: “а” type, with no differenti- ation among the three bu. because of their near- ly uniform cells; *b" type, where the epidermal and cortical layers are differentiated but the central strand is not; “с” type, where all three layers are differentiated md the central strand has less than 15 cells; and finally “4” type, as in the previous type but with a more developed central strand, i.e., of more than 15 cells. Deguchi's system varies from that of Kawai. He joined Kawai's types “с” and “d” into his “Type IL" whereas his Types “I” and “III” are identical to Kawai's types “a” and “b,” respectively. In this study һе internal structure of the stem in Grimmia subg. Orthogrimmia was found to be characterized by a weak differentiation between the epidermal and cortical layers (which together form a more or less homogenous layer) on the one hand, and the central strand (which is well developed in nearly all species) on the other. This stem structure is intermediate between Kawai's types “а” and “с” or Deguchi's “Т” and “II.” Branching. Branching systems in Grimmia have been studied in depth by Deguchi (1978: 128, figs. 1, 2). Branching pattern is usually sympodial, and this is the only type observed in Grimmia subg. Orthogrimmia. Innovations arise at the base of the perigonia and perichaetia. When these sexual structures are. abundant, the stems appear like stairs (Deguchi, 1978: figs. 1, 2). Occasionally, the plants appear monopodial because damaged sexual structures have fallen off. This is especially true for old plants. Rarely, the perigonia can be monopo- dially attached in cladoautoicous species. All species of subgenus Orthogrimmia except Grimmia caespiticia show intricately branched stems resulting in a cohesive net and compact cushions. Grimmia caespiticia has less extensively branched stems, and its cushions are easly decom- posed. Some young branches have rhizoids at their base and are easily detached from the stems. This could reflect their role as diaspores in nature (Correns, 1899: 102; Deguchi, 1978: 128). Rhizoids. Usually, rhizoids are limited to the base of stems. Plants subject to periodic inunda- tion, however, have rhizoids throughout the stem — ength. Axillary hairs. Axillary hairs in subgenus Or- thogrimmia consist of 3-8 hyaline, uniseriate cells, of which the 1-2 most proximal ones are shorter. Length may vary between 50 and 175 um. Leaf orientation. When dry, all taxa have erect and appressed leaf bases, whereas the apex can be incurved and appressed (e.g., G. caespiticia, G. montana, and G. ungeri), variously flexuous (e.g., G. arenaria, G. donniana, and G. triformis) or ap- pressed (e.g., G. alpestris, G. nivalis, and G. reflex- idens). When moist, the leaves vary from erect to spreading, and G. montana exhibits sigmoid leaves in lateral view. Leaf size and shape. Leaf lengths fall into two groups: shorter than 1.5 mm (e.g., G. alpestris, G. caespiticia, G. montana, G. nivalis, G. reflexidens, and С. ungeri); and longer than 1.5 mm (e.g., arenaria, G. donniana, and G. triformis). Mean leaf width is more uniform than length, always around 0.35 mm. Grimmia alpestris, G. caespiticia, С. nivalis, G. reflexidens, and G. ungeri (Figs. 8a, b, 10a, 14b, lóa, 18a) have ovate leaves, with a length/width ratio of 2-3: 1. Grimmia arenaria, G. donniana, С. triformis, and G. montana (Figs. 2a, 4a, 6a, 12a, b) have narrowly ovate leaves, with a length/width ra- tio of 3-6:1. Leaf margin. Тһе leaf margins are entire in all species studied, and plane at the base on both sides in most species. Ап exception to this rule are some populations of G. reflexidens, with recurved margins in the proximal half on one side and at the very base on the other side, or occasionally only briefly and narrowly recurved at the base on one side. Un- fortunately, in this taxon it is an inconsistent char- acter used profusely in the literature. The distal Volume 85, Number 3 1998 Mufioz 37 Revision of Grimmia Subgenus Orthogrimmia leaf margins are plane or incurved. Exceptions are the muticous leaves of G. caespiticia, G. montana, and G. nivalis, which have a more or less cucullate apex. Leaves in transverse section. leaves can be distinguished in Grimmia according to the shape of their transverse sections: concave, U-shaped (= canaliculate) and V-shaped (= keeled, carinate). Species in subgenus Orthogrim- mia. have strongly keeled leaves, although G. mon- tana and more often G. ungeri can have only slight- ly keeled leaves in some populations Grimmia alpestris, С. nivalis, and especially б. caespiticia have leaves with a longitudinal plication on each side of the costa. The cells of these plicae are usually longer and narrower and have thicker walls, although they can be жашашы; main- ly in the two former spec Species of subgenus Orthogrimmia have a semi- terete costa mostly projecting on the dorsal surface and clearly delimited from the lamina. Costal cross sections consist of three cell layers: ventral epider- mis, internal band of stereids or substereids, and dorsal epidermis. Based on the morphology and dif- ferentiation of these three layers, Kawai (1965: 111, 1968: 128) recognized four types of costal structure in Grimmia. Species in subgenus Orthogrimmia be- long to Kawai's “C” type, characterized by a more or less clear differentiation of the three layers, and a ventral epidermis two cells wide. Whereas costal structure and morphology are of great taxonomic value in some taxa of Grimmiaceae (e.g., Racomi- trium, Dryptodon, other Grimmia subgenera), they have a limited utility in Grimmia subg. Orthogrim- The leaf lamina is unistratose in the proximal part and 2(3—4)-stratose in the distal 4424, mainly at the margins. Laminar cells. The distal cells vary greatly in length, width, wall thickness, and sinuosity, and cannot be used to distinguish species. Their shape varies from isodiametric to rectangular or trans- versely rectangular in the same leaf, without any definite pattern. Length of major diameter ranges between 4 and 8 шп in С. montana and 8 and 13 pm in б. alpestris. Cell cross-sectional shape, on the other hand, is of great importance in defining taxa: Grimmia alpestris, G. caespiticia, G. nivalis, and some populations of G. reflexidens have round- ed cells bulging on both surfaces, whereas G. ar- enaria, G. donniana, G. triformis, G. montana, and G. ungeri have isodiametric or rectangular cells plane on the lamina surface (seldom G. montana has cells slightly bulging on the dorsal surface). rimmia nivalis and many populations of G. caes- Three types of piticia have papillose distal laminar cells. The pa- pillae are usually better developed on the dorsal than on the ventral surface. The proximal laminar cells in Grimmia arenaria, G. donniana, and G. triformis are alike: hyaline, and long and narrow (length/width ratio 3—10: 1) Their walls are thin and even, straight and nearly indistinct, although the paracostal cells can have somewhat thickened and sinuous walls (Figs. 2e, 4c, 6c). Grimmia alpestris, G. caespiticia, G. mon- tana, G. nivalis, G. reflexidens, and G. ungeri have proximal cells that range from isodiametric to rect- angular (to 6:1 in paracostal cells), with the trans- verse walls always thicker than the longitudinal ones, which usually are thicker than in the other species of the subgenus (Figs. 8d, 10c, 12d, 14e, 16e, 18d) Hair-points. Ав in almost any other species іп the genus, the length of the hyaline hair-points varies significantly in subgenus Orthogrimmia, and can often be correlated with ecological conditions. Populations from exposed habitats usually have longer hair-points. Grimmia arenaria has the lon- gest hair-points in the subgenus despite habitat, and they are always strongly flexuous. Grimmia caespiticia, a species of exposed sunny and dry habitats, has the shortest hair-points, at times re- duced to as few as one hyaline cell. Male plants of dioicous species always have shorter hair-points than their female counterparts, and when cushions of different sexes grow intermingled, they can ap- pear quite distinct. The hair-points are usually erect, and their de- gree of flexuosity depends on their length. Short hair-points are straight, whereas longer ones are of- ten flexuous. Grimmia arenaria always has strongly flexuous and homomallous hair-points. In Grimmia alpestris, G. caespiticia, G. montana, G. nivalis, G. reflexidens, and G. ungeri the hair- points are always terete, whereas in G. arenaria and G. triformis they are always flat. The structure of the hair-points in G. donniana depends on their length: longer hair-points are flat, whereas shorter ones are more or less terete. Perichaetial leaves. leaves of most species of subgenus Orthogrimmia are well differentiated from the apical vegetative ones, i.e., they are 2—3 times larger and convolute. Moreover, the cells in the proximal, sheathing half are enlarged and hyaline-yellowish, with very thin walls, at least along the margins. Exceptions are G. arenaria, with undifferentiated perichaetial leaves, and G. caespiticia, with convolute but only slightly larger perichaetial leaves. The inner perichaetial 372 Annals of the Missouri Botanical Garden The hyaline hair-points are always longer in per- ichaetial than in vegetative leaves. Reproductive organs. Таха in subgenus Ortho- grimmia are dioicous, gonioautoicous (androecium bud-like and axillary on the same stem as the ter- minal gynoecium) or cladautoicous (androecium on a separate stem). In the last-mentioned case, ram- ifications and the growth of branches at times pro- duce a notable separation of perichaetium and per- igonium-bearing branches. In this case, the plants could seem dioicous. Androecia can be axillary or terminal. The latter are easy to observe, since they are inflated and bul- d giving the stem a clavate appearance. Ax- ry buds are smaller and very difficult to find, especially | in autoicous species. The antheridia are surrounded in both types by strongly differentiated, cochleariform perigonial leaves with acute apexes. Perichaetia are always terminal, but soon appear axillary due to elongation of the subfloral branch (Deguchi, 1978: figs. 1, 2). SPOROPHYTE Autoicous species, i.e., G. arenaria, G. donni- ana, G. reflexidens, G. triformis, and G. ungeri, had sporophytes in 70-100% of the studied collections. Dioicous species show important differences in the percentage of fertile collections: Grimmia nivalis had sporophytes in all examined samples except one, Grimmia alpestris showed a high percentage, near 90%, whereas G. caespiticia and G. montana exhibited lower percentages, ca. 7096. etae. The setae are straight except in С. ar- enaria, which has curved setae. Other species can occasionally have slightly curved setae, particularly G. reflexidens. The setae are longer than capsules in all species except G. triformis. They range in length from 1 mm in G. triformis to 4 mm in alpestris and G. montana. In all species the seta is twisted counterclockwise when dry. Capsules provide important taxonom- ic characters in subgenus Orthogrimmia. Recently dehisced or, better, non-dehisced capsules must be used. Older capsules have greater deposits of wax over their exothecial cells, and most are infected by fungi, making it difficult to recognize important features, like the presence of stomata or the thick- ness of the exothecial cell walls. All species of subgenus Orthogrimmia have smooth capsules. They are exserted except in Grim- mia arenaria, which has emergent capsules, and С. triformis, which has immersed capsules. apsules are mostly ovoid, with a wide base abruptly connected to the seta. The only exception is G. alpestris, with ellipsoid-fusiform capsules at- tenuated at the base and not abruptly connected with the setae (Fig. 8c). The base is usually sym- metrical, but in some populations of G. alpestris, G. montana, and G. reflexidens the base is slightly asymmetrical and the capsule is weakly inclined. Capsules of Grimmia arenaria, G. donniana, G. reflexidens, G. triformis, and G. ungeri are usually stramineous, although in the last two species the color can turn to brownish in older capsules. The capsule color of G. alpestris, G. caespiticia, G. mon- tana, and G. nivalis is castaneous. At the capsule mouth there are several rows of small, transversely rectangular, usually reddish or brownish, more intensely colored cells with thick cells of the rest of the urn are isodiametric, rectangular, or transversely rectan- gular; usually all types are present, with one of them dominating. The exothecial cell walls are thin (less than З шт) іп С. arenaria, G. caespiticia, G. donniana, G. montana, G. reflexidens, and G. un- geri, and thick (more than З ит) in C. alpestris, G. nivalis, and G. triformis. Stomata can be found in the neck region of the urn in all species of subgenus Orthogrimmia except G. alpestris, G. montana, and G. ungeri. The best procedure for observing the stomata is to cut a cap- sule in half and then cut the proximal half again (for a total of four parts) and search for stomata (in uncut old capsules it can be hard to see the stomata because of secondary deposits of wax and other substances). The guard cells are reniform, and un- der the compound microscope appear dotted or col- ored, with a cellular content different from that of the adjoining exothecial cells. The cells surround- ing the guard cells are usually undifferentiated from the other exothecial cells, thus the stomata can be a as anomocytic. Sometimes, how- ever, the row of subsidiary cells surrounding the guard cells i is slightly differentiated and smaller ки the other exothecial cells, and such stomata could be considered stephanocytic (Baranova, The absence of stomata is a good character to separate Grimmia alpestris from the closely related G. nivalis and G. reflexidens. Two annular types can be recognized in subge- nus Orthogrimmia (Deguchi, 1978: 143-144). An- nulus cells in the Schistidium-type are undifferen- tiated from the exothecial rim cells апа non-revoluble. All species in section Montana have this type of annulus (simple and persistent). On the other hand, the elongata-type is characterized by ecial having two or three rows of isodiametric, inflated Volume 85, Number 3 1998 Mufioz 3 Revision of Grimmia Subgenus Orthogrimmia and hyaline, revoluble cells. All species in section Donnianae have this type of annulus (compound and revoluble). The peristome of Grimmia subg. Orthogrimmia consists of 16 more or less entire, irregularly divid- ed or cribrose teeth. Their width, measured at the outer surface of the peristome base, varies from 35— 50 um in С. caespiticia and G. ungeri (Figs. 10d, 18f) to 70-100 ¡um in G. triformis (Fig. 6f). The teeth are densely papillose on the inner surface, whereas on the outer surface they are papillose dis- tally and smooth or slightly papillose proximally. Tooth color (castaneous-brown or orange) is con- stant in each species. The operculum is conic with a short mammilla in all species except G. montana, in which it is rostrate. Calyptrae. Calyptra morphology is much em- ployed in the taxonomy of Grimmiaceae. All spe- cies in Grimmia subg. Orthogrimmia have smooth calyptrae, which may be mitrate (sect. Donnianae) or cucullate (sect. Montanae). Only one collection of Grimmia donniana out of the 555 fertile speci- mens of section Donnianae studied had some cu- cullate calyptrae, and I considered this an abnor- mality. Spores. Orthogrimmia are spherical and homogenous in size (isosporic, Mogensen, 1983: 334—336). Under the compound microscope they appear smooth, but are minutely granulose (Cao & Vitt, 1986: fig. 16c, f). Hirohama (1978: 37, figs. 57, 58) described the spores as smooth, but his figures show a minutely granulose surface similar to Cao and Vitt's figure The spores of all species of subgenus Once again, it is possible to separate the group consisting of G. arenaria, G. donniana, and G. tri- formis (i.e., sect. Donnianae), with smaller spores (6.5—11.0 um), from that comprising G. alpestris, G. caespiticia, G. montana, G. nivalis, G. reflexidens, and G. ungeri (i.e., sect. Montanae), with larger spores (9-14 шп). CHROMOSOME NUMBERS It has been pointed out that the е chromo- me number in Grimmiaceae is x — 7 (Smith, 1978; Vitt, 1984). From this number have been de- rived the remaining ones in the family: 10, 12, 124m, 13, 134m, 14, 144 1-4acc., 22, 26, and 26-4 m (Fritsch, 1991; Vaarama, 1949). The most common haploid complement in Grim- mia is n = 13, with n = 10, 124m, 134m, 14, and 14+ 1-4acc also common. These numbers were probably derived through doubling of the basic x = 7 complement of the Grimmiaceae (primary dip- loids). Other species have a haploid complement of = 26 or n = 26+m (Fritsch, 1991), probably resulting from secondary doubling of the n — 13 chromosome complement (secondary diploids). It is generally acknowledged that the duplication of the chromosome number implies a directional transformation from a dioicous to an autoicous sex- ual condition [but see Wyatt & Anderson (1984) for a thorough discussion on this topic]. Ramsay (1983: fig. 147, see also discussion on pp. 202-206) dia- grammed the various ways of maintaining the di- oicous condition by aneuploid reduction. Some of the primary diploids in Grimmia may have reverted to the dioicous sexual condition (e.g., 6. alpestris), whereas others have become autoicous (e.g., 6. ar- enaria and G. donniana (Fig. 1)). Chromosome counts for the species studied here are scarce (Table 2). Known numbers are n — or 134m for G. alpestris, n = 12+m and n = 13 for С. donniana, and n = 13 for С. arenaria and С. montana. 1 was only able to verify the identity of the voucher for the count of G. donniana (Khan- na, 1964: 348, fig. 5). TAXONOMIC TREATMENT Grimmia Hedw., Sp. Musc. Frond. 75. 1801. TYPE: Grimmia plagiopodia Hedw. (lectotype, designated by Mártensson (1956: 106-107). Autoicous or dioicous. Plants in dense cushions or compact to loose tufts, glaucous, green, greenish yellow, or dark green. Stems erect or ascending, with or without central strand. Leaves erect, ap- pressed or flexuous, occasionally with homomallous tips when dry, erect to spreading when moist, lin- ear, ovate, lanceolate, ligulate or oblong, obtuse to acuminate, concave, canaliculate or keeled, plane or plicate; margins entire, plane, recurved or in- curved; costa single, percurrent, terete, semi-terete, semi-elliptic, or almost indistinct in cross section; lamina 1- to 4-stratose in the distal half, smooth or pseudopapillose; distal cells isodiametric to rect- angular or transversely rectangular, with straight or sinuous walls, plane or bulging, smooth or papil- lose; proximal cells isodiametric to rectangular or transversely rectangular, the walls straight or sin- uous, uniformly thickened or with the transverse walls thicker than the longitudinal walls; with or without hyaline hair-points. Perichaetial leaves con- volute and larger, or similar in shape but slightly larger than vegetative leaves; hyaline hair-points entire to dentate, or lacking. Androecia axillary or terminal. Setae straight, curved, or coiled, longer or shorter than capsules. Capsules immersed, emer- Annals of the Missouri Botanical Garden duplication haploid (n = 7) DIOICOUS —————> AUTOICOU | CHROMOSOME NUMBER primary diploid (n = 13) S reversion by aneuploidy DIOICOUS duplication secondary diploid (n = 26) b AUTOICOUS Figure 1. gent or exserted, subglobose, ovoid, ellipsoid, or fusiform, symmetric or asymmetric and ventricose at the base, smooth or furrowed, stramineous or castaneous; with stomata at the urn base or lacking stomata; exothecial cells isodiametric to rectangular, thin- or thick-walled; annulus simple and persistent or compound and revoluble, of 1—3 rows of isodi- ametric to rectangular cells; peristome teeth 16, tri- angular, entire, perforate in the distal half or cri- brose throughout their length and irregularly cleft in the distal %-3, orange or brown; opercula сопіс to long-rostrate; calyptrae mitrate or cucullate, smooth, covering the operculum; columella persis- tent; spores usually 8-16 шт, occasionally up to 20 шт, smooth to granulose. Correlation between chromosome number and sexual condition in Grimmia. The supraspecific taxonomy of Grimmia has yet to be fully resolved. Whereas some recognized su- praspecific taxa (cf. Table 1) are natural and well established (e.g., subgenera Grimmia and Ortho- grimmia), others are at present tentative. The fol- lowing key is intended to separate subgenus Ortho- grimmia from all other members of Grimmia, not to present a taxonomy of the genus. No formal sta- tus or circumscription is given to either “Ећађдо- grimmia" or “Alpinae” until further work on these groups is carried out. Кеу TO MAJOR SUPRASPECIFIC ТАХА OF GRIMMIA la. Capsules strongly asymmetric at base, ventricose Grimmia subg. Grimmia Table 2. Chromosome numbers reported for Grimmia subg. Orthogrimmia. Species n = Origin Author Grimmia alpestris 13 California Steere et al. (1954) 134m Georgia (Republic) Lazarenko et al. (1971) 13 Colorado Khanna (1971) Grimmia arenaria 13 Great Britain Smith & Newton (1968) Grimmia donniana 124m New York ае. (1964) 13 Great Britain mith & Newton (1966) 13 Great Britain 5 (1969) 13 Kazakhstan Vysotskaya & Lesnyak (1984) 13 Poland Kuta et al. (1984) Grimmia montana 13 Canada Anderson & Crum (1958) Volume 85, Number 3 1998 Mufioz 375 Revision of Grimmia Subgenus Orthogrimmia lb. Capsules үш or very slightly asymmetric, not ventricose at base. 2a. jm variously curved; capsules mostly ribbed, or if smooth, leaf margins recurved to some degree or leaves variously crisped *Rhabdogrimmia" . Setae straight or, if hirt capsules smooth, leaf margins ES and leaves straight to Hapus not cri . Leaves concave; uh semi-elliptic, in- isti ;rimmia subg. Guembelia . Lasós canaliculate to keeled; costa semi-terete, clearly delimited from lam- ina. 4a. Leaf margins plane or incurved Grimmia subg. Orthogrimmia 4b. Leaf margins recurved. nnulus compound and revo- uble |. “Alpinae” 5b. Annulus simple and persistent Grimmia subg. Orthogrimmia (G. reflexidens) N c w > Grimmia subg. Orthogrimmia Schimp., Coroll. Bryol. Eur. 48 1856. TYPE: Grimmia donni- ana Sm. (lectotype, here designated). Autoicous or dioicous. Plants in dense, compact tufts, glaucous, green, greenish yellow, or dark green. Stems erect, to 2 cm tall X 110-210 um diam., with central strand well developed, some- times with rhizoids present nearly throughout; ax- illary hairs 3—8-celled, hyaline, 50—175 шт long. Leaves erect, appressed to flexuous (occasionally with homomallous tips) when dry, erect-spreading to spreading (occasionally sigmoid in lateral view), id when moist, 0.8-2.2 x 0.25-0.65 M ovate to ovate, acute to acuminate, keeled, although sometimes only weakly, plane or plicate; margins plane, seldom recurved; costa semi-terete in cross section, usually prominent on the dorsal surface, slightly to clearly delimited, with 2 cells in the ventral epidermis, an internal band of stereids or substereids, and a dorsal epi- ermis, the three layers — differentiated; lamina 2—3(4)- оин in the distal half, mainly along the flaccid or rig KEv TO SECTIONS OF GRIMMIA SUBG. ORTHOGRIMMIA la. @ ¡EE pP Setae straight or curved, longer than 1 m Setae strai ight, to 1 mm; capsules wor i exothecial се thick-walled margins, smooth or pseudopapillose; distal cells 4— 20 x jum, isodiametric to short or transverse- ly la with straight or sinuous walls, plane or bulging, smooth; proximal paracostal cells 10-5 X 7-20 um, green, isodiametric to rectangular (1— 6:1), the walls straight or sinuous, uniformly thick- ened or the transverse walls thicker than longitu- dinal walls; proximal marginal cells 9-50 X 6-20 шт, isodiametric to rectangular (1-5:1), with transverse walls thicker than longitudinal cell walls; or proximal and paracostal ce 00 x 5.5-25 um, alike, hyaline, long and narrow (3-10: 1), with even, very thin, straight, scarcely discernible walls; hyaline hair-points flat or terete at the base, somewhat to strongly flexuous, at times homomallous and twisted, to 2 mm long, denticu- late to dentate. Perichaetial leaves 1.5-3.1(4 0.4-0.9 mm, convolute and larger (2-3 X) or similar in shape but slightly larger than vegetative leaves; hyaline hair-points plane or terete, flexuous, to 2.8 mm, entire, denticulate or dentate. Androecia axil- lary or terminal. Setae straight or curved, to 1.0— 4.5 mm long. Capsules immersed, emergent or ex- serted, ovoid, ellipsoid or fusiform, symmetric (sel- dom slightly asymmetric at base), smooth, strami- neous or castaneous; with stomata at the urn base, or lacking stomata; exothecial cells 16-70 X 10-55 um, isodiametric to rectangular (1—4.5 : 1), thin- or thick-walled; annulus simple and persistent or compound and revoluble of 2-3 rows of mostly iso- diametric cells 6-10 ¡um high; peristome teeth 35— um wide at the base, entire, perforate in the distal half or cribrose throughout their length and irregularly cleft in the distal 25—4, orange or brown, concolorous or contrasting in color with the urn, outer surface papillose distally and smooth or slightly papillose proximally, inner surface papil- lose throughout; opercula conic to rostrate; calyp- trae mitrate or cucullate. Spores 6.5-14 шп, mi- nutely granulose Distribution. АП continents. Proximal marginal cells of leaf + hyaline and inflated, length/width ratio 3-10:1, walls thin, scarcely 2... the transverse walls similar to the longitudinal walls; с voluble 'alyptrae mitrate; annulus compound and rimmia subg. Orthogrimmia a pec а G. triformis ; capsules emergent or exserted; exothecial cells Ta walled. a. eem curved; hyaline hair-points to y mm, those of the perichaetial leaves strongly flexuous and 1 С. arenaria 3b. Selas straight; hyaline hair-points to 1 mm, those of the perichaetial leaves weakly w^ and not twiste G. donniana ғ roximal S cells of leaf neither hyaline nor inflated, length/width ratio 1—4.5(6) : 1, walls "hb TA iin ii transverse walls thicker than the longitudinal d alyptrae cucullate; annulus simple and Тыш cells not bulging (Figs. 12e, 164, 18c). . Grimmia subg. Orthogrimmia sect. Montanae 376 Annals of the Missouri Botanical Garden за. 5b. Stomata present at the urn base Stomata lackin Opercula РР rostrate; setae 2-4 mm long; peristome teeth 50-90 ит wide at mouth, irregularly splitting above and = cribrose; proximal paracostal leaf cells mostly long-rect- angular, to 4.5:1; dioicous _ 7. montana Opercula ue to mammillate; setae to 2 mm; peristome teeth 40—50 шт wide at mouth, ROM a RT TRE . 8. G. reflexidens 6b. entire or slightly cribrose at apex; proximal paracostal leaf cells isodiametric to ine 9, G. = Laminar cell bulging (Figs. 8e, 16c). Та. Laminar cells papillos 8a. Leaves strongly du on eh sides of costa 8b. Leaves plane or weakly plic Laminar cells not papillose 7b. 9b. Leaves strongly «е on both sides of costa . A ungeri o 5. б. caespiticia REED 7. С. nivalis dl 5. G. caespiticia Leaves plane or weakly plicate. 10a. Stomata lacking at the urn base; capsules usually fusiform, castaneous, — with peristome em exothecial cells isodiametric, thick-walled . 4. G. alpestris I. Grimmia (subg. Orthogrimmia) sect. Don- nianae (Loeske) J. Мипог, comb. et stat. nov. Grimmia [unranked] Donnianae Loeske, Bib- lioth. Bot. 101: 110. 1930. TYPE: Grimmia donniana Sm. Autoicous. Plants in compact tufts, green, green- ish yellow, or dark green. Stems to 2 cm tall, with central strand well developed, sometimes rhizoids present nearly throughout; axillary hairs 3-Т- celled, 70-160 ¡um long. Leaves егесі and ap- pressed or flexuous (occasionally with homomallous tips) when dry, patent to spreading, and flaccid or rigid when moist, 1.3-2.2 X 0.25-0.65 mm, паг- rowly ovate, acute to acuminate, keeled, plane; margins plane; costa semi-terete, prominent on the dorsal surface, clearly delimited; lamina 2-stratose at margins and in streaks in the distal half, smooth or pseudopapillose; distal cells 7-11 X 7-14 um, isodiametric to rectangular or transversely rectan- gular, plane, smooth; proximal paracostal and mar- ginal cells 35-100 х 5.5-25 um, alike, hyaline, narrowly rectangular (3-10:1), with even, very thin, straight, scarcely discernible walls, or the par- acostal cells with thickened and sinuous walls; hy- aline hair-points flat or terete at the base, somewhat to strongly flexuous, at times homomallous and twisted, to 2 mm long, denticulate to dentate. Per- ichaetial leaves 1.8—3.1 X 0.4—0.8 mm, convolute and larger (2-3X) or similar in shape but slightly mm, entire, denticulate or dentate. Androecia axillary or terminal. Setae straight or curved, to 3.5 mm long. Capsules im- mersed, emergent or exserted, ovoid or ellipsoid, symmetric, smooth, stramineous, with stomata at the urn base; exothecial cells 30—70 х 10—46 шп, . Stomata present at the urn base; capsules the orange peristome teeth; exothecial cells irregularly кшш thin-walled ovoid, stramineous, different i in n color from REEE EEA AE 8. G. reflexidens isodiametric to rectangular (1—4.5:1), thin- or thick-walled; annulus compound and revoluble of 2-3 rows of isodiametric cells 6-12 um high; peri- stome teeth 50-100 шт wide at the base, entire, perforate in the distal half or cribrose throughout, irregularly cleft in the distal 244, orange, contrast- ing in color with the urn; opercula conic or with a short mammilla; calyptrae mitrate. Spores 6.5-11 Distribution. temperate Asia. Northern America, Europe, and Grimmia subg. Orthogrimmia sect. Donnianae is characterized by thin-walled proximal marginal cells, mitrate calyptrae, and a compound and re- voluble annulus. l. Grimmia arenaria Hampe, Linnaea 10: 405. Schimp., in Bruch, Schimp. & W. Gümbel, Bryol. Europ. 3: 113. 1845, nom. inval. pro syn. Grimmia donniana var. curvula Spruce, Musci pyrenaici n° 281, 1847. Grimmia don- niana subsp. arenaria (Hampe) Dixon, Stud. Handb. Brit. Mosses Ed. 2: 155. rim- mia donniana var. arenaria (Hampe) Loeske, Laubm. Eur. Part I: 93, figs. 1c, 17a, 26b, 28. 1913. TYPE: [Germany. Magdeburg:] Regen- stein Hercyn[iae], June, Hampe s.n. (lectotype, here designated, BM; isolectotype, FH). Illustrations. Figure 2; Bruch et al. (1845: tab. 238, sub G. curvula) Autoicous. Plants in hoary tufts, dark-green to blackish. Stems to 1.5 cm tall, with central strand Volume 85, Number 3 1998 Mufioz Revision of Grimmia Subgenus Orthogrimmia Figure 2. Grimmia arenaria. —a. Lea verse sections of leaf. —e. Proximal leaf cells. —f. | —h. Annulus and peristome teeth. (а, d-h, Garovaglio s.n. (С); b, Spruce, Musci Pirenaici n° 281, (TCD).] well developed; axillary hairs 5—7-celled, 140—160 um long. Leaves erect (occasionally with the tips homomallous) when dry, patent to spreading, flaccid when moist, 1.3-2.2 X 0.25-0.65 mm, narrowly ovate, acute to acuminate, keeled, not plicate; mar- gins plane; costa semi-terete, prominent on the dor- sal surface, clearly delimited; lamina 2-stratose at margins and in streaks in the distal half, occasion- ally pseudopapillose; distal cells 7-10 шп, trans- ves. —b. Perichaetial leaf. —c. Sporophyte and perichaetial leaf. —d. Trans- Proximal exothecial cells and stoma. — Medial exothecial cells. versely rectangular to isodiametric or rectangular, plane, smooth; proximal paracostal and marginal cells 35-95 х 8-25 um, alike, hyaline, narrowly rectangular (3-10: 1), with even, very thin, straight, scarcely discernible walls, or the paracostal cells with thickened and sinuous walls; hyaline hair- points flat, strongly flexuous, usually homomallous and twisted, to 2 mm long, rarely shorter than 1 mm, denticulate to dentate. Perichaetial leaves 1.8— 378 Annals of the Missouri Botanical Garden Figure 3. Distribution of Grimmia arenaria. 2.6 X 0.4—0.7 mm, similar to vegetative leaves but slightly larger; hyaline hair-point similar to those of vegetative leaves but longer, to 2.8 mm. Androe- cia terminal. Setae curved, ca. 2 mm long. Capsules exserted or, more commonly, emergent among the perichaetial leaves, ovoid, symmetric, stramineous, with stomata at the urn base; exothecial cells 30— 5 pm, rectangular (1.5—4.5:1), thin- walled; annulus compound and revoluble of 2—3 rows of isodiametric cells 10 um high; peristome teeth 50—90 um wide at the base, perforate in the distal part, eroded and irregularly divided in 2—3 branches, orange, contrasting in color with the urn; opercula conic or mammillate; calyptra mitrate. Spores 6.5—10.5 Diagnostic characters. (1) Proximal laminar cells hyaline, thin-walled. (2) Hyaline hair-points very long, to 2 mm, twisted and usually homomal- lous. (3) Capsules emergent and facing down among the perichaetial leaves. (4) Setae short and curved. (5) Annulus compound and revoluble. (6) Peristome teeth 50-90 um wide at the base, perforate in the distal part, eroded and irregularly divided in 2-3 branches. Distribution (Fig. 3). Central and western Eu- rope, Finland, and Great Britain; open areas from o 1800 m elevation, on dry sandstone and slate. Mature sporophytes were present in 100% of the studied specimens. This species is distinguished at first sight by the curved setae and small ovoid capsules emerging among the long and twisted hair-points. With more than half (ca. 68%) of the samples studied from the vicinity of Regenstein (Germany), the type locality, low variability may be expected. Indeed, this taxon varies only in the length of the hyaline hair-points, which only rarely is less than 2 mm. Even the shorter points are strongly curled and flexuous, a feature not seen in other species of the subgenus. Grimmia arenaria has been treated mostly as an infraspecific taxon of G. donniana, both as a sub- species (e.g., Dixon, 1904: 155) and a variety (e.g., Podpéra, 1954: 280). Only recently have Greven (1994, 1995: 44—47) and Touw and Rubers (1989: 212-213) agreed with Limpricht's (1890: 735-736) view that this taxon should be treated at the specific level. ‚ Selected specimens агии (155). AUSTRIA. Са- nthia: im Kressbrünngraben bei Raibl, (ВР. 36050). Tirol: issu уйи Gander s.n. (BP- 36053). FINLAND. Turku ja Pori: Lojo, т. 5 Aug. 1880, 2” (б). FRANCE. Ікег de Isans, Aug. 1860, Ravaud s.n. (TRH). клы сы Angers, Schimper s.n. (S). Pyrénées Centrales: in fauce dict. la Gorge de Labassere, Spruce s.n. (NY). Vosges: Le Hohneck, Boulay s.n. (FH). GERMANY. Baden-Würt- temberg: Südl. Schwarzwald, nahe südwarde du e des Schauinsland, Aug. 1925, Schmidt s.n. (JE). Magde- burg: Blankenburg. Regenstein, 23 Mar. 1902, [rn s.n. (JE). ITALY. Biellese ne'monti dell'Oropa e di S. Gio- vanni d'Andorno, Aug. 1861, Cesati s.n. (G). ROMANIA. .. Alpe 2. Péterfi s.n. (BP- б SP. Lérida: Vallferrera, Casas s. | 2. 952). SWITZERLAND. Tessin: (im July 1 ин s.n. (S). UNITED KINGDOM. Wales: . Talsa nan, Harlech, Sep. 1911, Rhodes s.n. (S). 2. Grimmia donniana Sm., Engl. Bot. 18: pl. 1259. 1804. бош sudetica Schwügr., Sp. Musc. Frond. Suppl. 1(1): 87, tab. 24. 1811, nom. illeg. incl. sp. prior. Grimmia donnü Gray, Nat. Arr. Brit. Pl. 1: 728. 1821, nom. Шер. Dryptodon donnianus (Sm.) Hartm., Handb. Skand. Fl. Ed. 3: 270. 1838. Grimmia obtusa var. donniana (Sm.) Hartm., Handb. Skand. Fl. Ed. 5: 377. 1849 (1850?). Grimmia donii Sm. ex Lindb., Musci Scand. 30. 1879 nom. illeg. Grimmia donniana var. eudonniana Loeske, Laubm. Eur. Part I: 91. 1913, nom. inval. Guembelia donniana (Sm.) Loeske, Laubm. Eur. I: 90. 1913, nom. inval. Grimmia donniana subsp. eudonniana Giacom., Atti Ist. Bot. Lab. Crittog. Univ. Pavia, ser. 5, 4: 221. 1947, nom. inval. TYPE: [United Kingdom. Caernarvon:] North Wales, Beddgelart, July 1802, Turner s.n. (lectotype, here designated, BM) Grimmia obtusa Schwügr., Sp. Musc. Frond. Suppl. 1(1): 88, tab. 25. 1811, nom. Шег., non Brid., 1801. Grimmia donniana var. obtusa (Schwiigr.) Steud., No- mencl. Bot. 2: 189. 1824. Dryptodon erostris Hartm., Handb. e Fl. Ed. 4: 374. 1843, nom. illeg. incl. TYPE: 22. Glockner, Kaulfuss s.n. (hok 2 G; isotype, ( М е зидепса Spreng. ex S TE Deutschl. Krypt. у ећ 2: 48, Тађ. 2 . G. donniana f. зи- ps (Spreng. ex 5. lo . Biblioth. Bot. 101: 113. 1930, nom. Шег., non Chal. 1882. TYPE: T Volume 85, Number 3 1998 Mufioz 3 Revision of Grimmia Subgenus Orthogrimmia [Poland. Sudety. Ludwig s.n.| “Kryptogamische Gewüchse der Riesengebirgen. Laubmoose" [hand- written label, not the original of Ludwig's exsiccata] (lectotype, here designated, M). Grimmia а var. bohemica Schkuhr ex Brid., Bryol. Univ. 1: 176. 1826. Grimmia bohemica Sc hkuhr ex - Nomencl. Bot. 2: 188. 1824 syn. TY Е: [Poland.] 5с hneekoppe "бейи, 1814, ыы s.n. (lectotype, here designated, B). Illustrations. Figure 4; Bruch «€ Schimper (1845: tab. 249, sub G. obtusa); Cao and Vitt (1986: figs. 15, 1ба-с); Chałubiński (1882: tab. 7 fig. 11); Deguchi (1978: fig. 27); Ignatov and Cao (1994: fig. 10, but not figs. 8, ^ Jóhannsson (1993: fig. 32); Limpricht (1890: fig. 198); Noguchi (1988: fig 140B); Nyholm ed fig. 69F Autoicous. Plants in tufts, greenish yellow above, dark green to blackish below. Stems to 1.5 cm tall, with central strand well developed; axillary hairs 4—5-celled, 90-125 ¡um long. Leaves erect and ap- pressed (occasionally with the tips somewhat flex- uous) when dry, patent and rigid when moist, 1.3— 5—0.60 mm, narrowly ovate, acute to acu- minate, keeled, plane; margins plane; costa semi- terete, prominent on the dorsal surface, clearly de- limited; lamina 2-stratose at margins and in streaks in the distal half, occasionally pseudopapillose; dis- tal cells 7-11 X 7—9 um, isodiametric to rectan- gular (1—1.5: 1), plane, smooth; proximal paracostal and marginal cells 38-80 X hyaline, narrowly rectangular (4—9:1), with even, very thin, straight, scarcely discernible walls, or the paracostal cells with thickened and sinuous walls; hyaline hair-points terete or flat when longer, slight- ly flexuous, usually to 1 mm long (seldom to 2 mm and then strongly flexuous), smooth to weakly den- ticulate. Perichaetial leaves 2.0-2.5 X ca. 0.8 mm, convolute and larger than vegetative leaves (2.0— 2.5X); hyaline hair-points slightly flexuous when short, strongly so when longer, to 2.2 mm, smooth pm, alike, or denticulate. Androecia axillary or terminal. Setae erect and straight, to 3.5 mm long. Capsules ex- serted, ovoid, symmetric, smooth, stramineous, with stomata at the urn base; exothecial cells 35—55 X 24-46 шт, isodiametric to rectangular (1-2: 1), thin-walled; annulus compound and revoluble of 2— 3 rows of isodiametric cells 10-12 ит high; peri- stome teeth 50—70 jum wide at the base, entire or weakly broken at the tips, orange, contrasting in color with the rest of the sporophyte; opercula conic or mammillate; calyptrae mitrate. Spores 7-11 ¡um Diagnostic characters. (1) Proximal laminar cells hyaline, thin-walled. (2) Hyaline hair-points to 1 mm, slightly flexuous, seldom longer (to 2 mm) and then more flexuous. (3) Capsules exserted. (4) Setae straight, to 3.5 mm. (5) Annulus compound and revoluble. (6) Peristome teeth 50-70 шп wide at the base, entire or only slightly cleft at the tip. Distribution (Fig. 5). Common in Europe, it is scattered through Siberia, Nepal, Japan, North merica, and Greenland; open areas and forests from 80 to 3800 m elevation on all types of non- calcareous rocks. Mature sporophytes were present in 96.596 of the specimens studied. Grimmia donniana is rather stenotypic and easy to recognize because of its hyaline proximal cells, straight setae, and exserted capsules. The most var- iable features are the length and twisting of the hair-points, but they are rarely as long and flexuous as in G. arenaria. However, I have studied two puz- zling specimens in which the variation observed is difficult to interpret. Chernyadjeva 37 (Kamchatka, Kosheleva volcano, LE) exhibits some cucullate ca- lyptrae, but this is the only deviant feature ob- served and is here considered an abnormality. An- other anomalous specimen is Sharp 4761 p.p. (Mexico, Popocatépetl, TENN). It has the shortest setae seen in С. donniana (i.e., 1 mm) and the cap- sules are mostly immersed, approaching G. trifor- mis. Nevertheless, it matches typical G. donniana n all other respects, especially in the thin-walled Іі! cells and the entire and narrow регі- stome teeth. See comments under Grimmia arenaria and G. triformis for differences between these species and G. donniana. Most collections from eastern North America identified as G. donniana are specimens of G. in- curva Schwagr. with nearly straight seta and auto- icous inflorescence. The latter species can be dis- tinguished from С. donniana by its longer, narrower, and more acuminate leaves, which are crisped and contorted when Grimmia sudetica lr: ч was considered the legitimate name for G. alpestris by Geissler and Maier (1995: 503). However, in the original pub- lication, Schwagrichen (1811: 87) cited G. donni- ana as a synonym of his new species, which there- by became illegitimate (Muñoz, 1998). rimmia donniana has a non-continuous, cir- cumboreal distribution. It is quite common in mountainous areas in central and northern Europe, but becomes rare toward the south and the east. In North America, it is known from scattered localities in the United States and Mexico. In Asia Grimmia donniana is rare, growing on Honshu (as cited hereunder) and Hokkaido (Deguchi, 1978: fig. 28), Japan, and from the Altai Mountains and Tibet in 380 Annals of the Missouri Botanical Garden Grimmia donniana. —a. Leaves. —b. Sporophyte and perichaetial leaves. —c. Proximal leaf cells. —d. af. igure 4. verse sections of leaf. —e. ral е ial cells and stoma. —f. Medial ш ial cells. —g. Annulus and rans peristome teeth. [a-g, Turner s.n. (BM). continental Asia. Two of the three Tibetan reports (Cao & Vitt, 1986: 161) actually pertain to G. elon- gata Kaulf. (Lang Kaiyong 598, ALTA) and G. lon- girostris Hook. (Lang Kaiyong 5302, ALTA). Grimmia donniana has also been reported from Africa (Ochyra & Sharp, 1988: 344) and Antarctica (Bartram, 1957: 141; Kuc, 1969; Savicz-Lyubit- skaya & Smirnova, 1969). However, the collections from these areas studied by me represent other taxa (e.g., 6. kidderi, 6. lawiana, or 6. reflexidens). Volume 85, Number 3 1998 Muñoz 3 Revision of Grimmia Subgenus Orthogrimmia Distribution of Grimmia donniana. Figure 5. Selected specimens examined (380). AUSTRIA. Ca- : im Kremsthale, Breidler s.n. one 36075). Salz- 887, Schliephacke wülz, Breidler (BP-36077). 1892, Gander s.n. (TRH). CZECH REPUBLIC. Шешә, Montes Krkonoše, vallis Obří důl, Apr. 1949, Pilous s.n. (Н); Riesengebirge, Felstriimmer der Schneekoppe, July 1887, Eampricht s.n. (RO); Tatra 74869). FRANCE. Vosges: prope Lacum viridem, Mou- geot & Nestler s.n. (FH). Haute-Loire: Le Mégal, 2. s.n. (BP-112577). Puy-de-Dóme: sous le met du Puy-de Dóme, Pierrot s.n. (BP-112575). Pyrénées n au sommet du port d'Oo, 5 Sep. 1856, Zetter- TRH). Savoie: бы йш, cirque de l'Arcelin, Parriat s s.n. (BP-112578). GEORGIA. South Ossetia Au- ; d rajon, Егтат, 9 Sep. Abramov s.n. (LE). GERMANY. Bavaria: Bayer- bbs: Wald, Arberkuppe, Progel s.n. (BP-36101). Bay- ern: auf dem Ochsenkopf, Funck s.n. (RO); bei Annaberg, Weicker s.n. (ВР-36102). Karl-Marx Stadt: Birenstein, Kopsch s.n. (BP-112572). Magdeburg: Regenstein bei Blankenburg, /tzigsohn s.n. 1843 (S). Niedersachsen: Oberharz, Luisenklippen auf dem Quitschenberg zwischen Torfhans und Oderbrücke, 11 July 1988, Heimhold s.n (СОЕТ). Rhón: Gipfel der Milseburg, Anonymous s.n. (BP-36108). Saxonia: prope Altenberg, Rabenhorst s.n. (BP-80338). Thüringia: Schmidt s.n. (GLM-2850). HUN- Montes Szeben, prope Paltanis, Vajda s.n. (BP-73974). ICELAND. North Iceland, 2. агеа, above farm Nipá, NE of Akureyri, 1 July 4, Steere s.n. (NY); South Iceland, Skaftafell, uot 271 Y); hill near Grótta by Rejkjavik, Andrews 1/2 (NY). IRELAND. Down, Mourne Mts., Slieve Donard, Ballagh Park, 15 Oct. 1884. Lett s.n. (TCD). ITALY. Mte. S. 52 July 1853, ЖИЛ ; Rima, precipizii del Mte. Tagliaferro, 19 Aug. 1863, Carestia s.n. (RO). JAPAN. Toyama Pre- fecture, Mt. Tateyama, 11 Aug. 1955, /watsuki s.n. (G); prov. Iwashiro, mt. lida, /shiba 1331 (H-BR). NORWAY. Bergen: Arstad, mellem Haulelandsvadtnet og Kronstad, 9 Sep. 1871, Wulfsberg s.n. (TRH). Kristiams: Lom, Вау- erdalen, Róshejm, 19 Aug. 1887, Hagen s.n. (FH). Son- dre d ms Stóren, уор, 16 A 1884, пена sub cacumi- r 6071). ROMANIA. Beszterce-Naszód: Mt. R Posi cacuminis montis Опбкб, Péterfi s.n. (BP-112955). asov: montes Fogarasi лаб. vallis Arpasul, Vajda п. (BP-70413). Maramures : sept. jugi Lunca ciasa sub nemi Toroiaga prope pagum Borsabánya, Boros s.n. (BP- Szeben: Cibin, ad "Wasserleitungs- Weg" supra Hohe Rinne Boros s.n. (BP- me RUSSIA. Gorno Al- las "e Mountains, Trekh- ВА )25). Kamchatka: slopes of Как эсин, г. желдік скі 37 (LE). Ma- adan: otka peninsula, Lake loni, 5 July 1977, 5 Gömör: Ж montis ay in hjo 4128). $ der P jaian aiii т, July 18: (RO). Tessin: Fusio, July (om s.n. (TRH). Valais: Vallon d' odi 30 a 889, Bernet s.n. (FH). TIBET. Yatong County, Zang 669 (ALTA). а i y i hornogora range, Turku Partyka s.n. (IBA-6993). UNITED KINGDOM. England; Cumberland, Grange, Borrowdale s.n. (FH). Wales: Carnarvonshire, Capel Curig, Feb. 1880, "Bleckel s.n. (FH). MEXICO. amps агро Sharp 4761 p.p. (TENN). Veracruz: e Orizaba, Galeotti s.n. (PC). ' 1892, GREEN) AND. dd Field (the summit), 12 May U : Circle Quadrangle: vicin- 1939, : E ѕѕех | New Hampshire Mt. Washington, Harring, Wickes & Grout s.n. (FH). County trail from Adirondack Loj to 1. Peak of the MacIntyre Mountains, Redfearn 13325 (МО). 3. Grimmia triformis Carestia & De Not., Com- ment. Soc. Crittog. Ital. 2: 102. 1866. Grimmia donniana var. triformis (Carestia & De Not.) Loeske, Laubm. Eur. Part I: 96, figs. 26d, 30. 1913. TYPE: [Italy. Vercelli:] In Tagliaferro, a quelche metro sotto la veta sul versante di Rima, 19 Aug. 1963, Carestia s.n. (lectotype, here designated, BM; isolectotypes, BM [2 rep- licates], BP-37541, JE). Grimmia ume RS Jahresber. Schles. Ges. Vaterl. Cult. 215. 1884. Grimmia 25. var. brevi- seta Ad y ex bou Laubm TYPE: M pon im menie, “Kalchstein” 4400 ft., 15 Mar. er s.n. (lectotype, here designated, BP- xu isolectotypes, JE [2 repli- cates], Н). 382 Annals of the Missouri Botanical Garden Figure 6. Grimmia triformis. —a. Leaves. Transverse sections of leaf. —f. Medial exothecial cells. —g. Annulus and peristome teeth. [a, c, d, f, Gander s.n. (BP- 37547); b, e, Gander s.n. (BP-37549).] Illustrations. Figure 6; Ignatov and Cao (1994: fig. 5, sub Coscinodon cribrosus); Limpricht (1890: figs. 196, 197). Autoicous. Plants in tufts, green. Stems to 2 cm tall, with central strand weakly developed; rhizoids abundant to middle of stems; axillary hairs 3- ves. —b. Perichaetial leaf. —c. а 1 mm b 1 mm f ——— 50 um с-е — 20 um Sporophyte. —d. Proximal leaf cells. —e. celled, ca. 70 шт long. Leaves erect and flexuous when dry, patent and flaccid when moist, 1.5-2.2 X 0.3-0.5 mm, narrowly ovate, acute, keeled, plane; margins plane; costa semi-terete, prominent on the dorsal surface, clearly delimited; lamina 2- stratose at margins and in streaks in the distal half, Volume 85, Number 3 Mufioz 383 Revision of Grimmia Subgenus Orthogrimmia 1998 wo е? 9 30° 0° 30 60° Figure 7. Distribution of Grimmia triformis. occasionally pseudopapillose; distal cells 7-9 X 7— jm, isodiametric to rectangular (1-1.5:1), plane, smooth; proximal paracostal and marginal cells 40-100 х 7-18 um, alike, hyaline, narrowly rectangular (3-10: 1), with even, very thin, straight, scarcely discernible walls, or the paracostal cells with thickened and sinuous walls; hyaline hair- points flat, flexuous, to 1.3 mm long, denticulate. Perichaetial leaves 2.5—3.1 X 0.8 mm, convolute and larger than vegetative leaves, yellowish at the base; hyaline hair-points flat, strongly flexuous, to 2 mm, denticulate. Androecia terminal. Setae erect and straight, to 1 mm long. Capsules immersed, ovoid, symmetric, smooth, stramineous, with sto- mata at the urn base; exothecial cells 46-75 X 10— 21 jum, irregularly rectangular (2—4:1), thick- walled; annulus compound and revoluble of 2 rows of isodiametric cells 6-10 ¡um high; peristome teeth T jm wide at the base, cribrose throughout and irregularly cleft in the distal 44, orange, con- trasting in color with the urn; opercula conic or mammillate; calyptrae mitrate. Spores 8-11 Lum. Diagnostic characters. (1) Proximal laminar cells hyaline, thin-walled. (2) Capsules immersed. (3) Setae erect and straight and very short (to 1 mm). (4) Annulus compound and revoluble. (5 Peristome teeth 70—100 дт wide at the base, cri- brose throughout and irregularly cleft in the apical 24—34. — Distribution (Fig. 7). Very rare, known only from the European Alps and Pyrénées, and from an outlying locality in the Altai Mountains of Siberia; in open areas above the tree-line between 1385 and 4500 m elevation on non-calcareous rocks. Mature sporophytes were present in 100% of the specimens studied. This seldom-collected taxon hardly varies, and is well characterized by sporophyte features. Grimmia triformis can only be confused with С. donniana, from which it is nearly indistinguishable gameto- phytically. The most conspicuous difference is the included sporophytes of G. triformis; those of G. donniana are exserted. Other characters separating these species are the wider (70—100 шт), cribrose peristome teeth and thick-walled exothecial cells of С. triformis versus the narrower (50—70 шт), entire peristome teeth and thin-walled exothecial cells of G. donniana. Some collections of G. triformis have been identified as Coscinodon cribrosus (Hedw.) Spruce (Ignatov & Cao, 1994: 76), which also has included capsules, but the shape of the proximal laminar cells precludes any confusion. Selected. specimens examined (33). AUSTRIA. Salz- burg: Rauriser Goldberg, Mielichhofer s.n. (BP-36855, Tirol: Innervillgraten, “Kalchstein,” bius s.n. (BP- 37549). FRANCE. Haute-Garonne: Port de Ben- asque, Pierrot s.n. (BCB- -2578). ITALY. Vercelli: рм Penninae Pedemontii, in monte T prope cacu- men, 19 Aug. 1863, Carestia s.n. (JE). RUSSIA. dise Altayskaya у збы? т Oblast: Altai Mountains, Ко- biguayuk Creek, /gnatov 0/474 (IBA-6997). SWITZER- LAND. Bern: Fungfraujoch am den “Hotelfelsen,” Kol & Chorus s.n. (ВР-80466, p.p.). II. Grimmia (subg. Orthogrimmia) sect. Mon- tanae I. Hagen, Kongel. Norske Vidensk Selsk. Skr. (Trondheim) 1909(5): 16. 1909. TYPE: Grimmia montana Bruch & Schimp. Autoicous or dioicous. Plants in compact tufts, seldom fragile, glaucous, green, olive-green, or brownish green above, blackish below. Stems to 1.5 cm tall, with central strand well developed, seldom with rhizoids the entire stem length; axillary hairs 3-8-celled, 50-175 ¡um long. Leaves erect and ap- pressed or flexuous when dry, erect to spreading, sigmoid or straight, and flaccid or rigid when moist, 0.8-2.0 X 0.26—0.65 mm, narrowly ovate to ovate, acute to acuminate, keeled (occasionally weakly), plane or plicate; margins plane or partially re- curved at the base; costa semi-terete, usually prom- inent on the dorsal surface, slightly to distinctly delimited; lamina 2—3(4)-stratose in the distal half, smooth; distal cells 4-20 um, isodiametric, plane or bulging, smooth or papillose; proximal paracostal cells 10-55 X 7-20 um, not hyaline, isodiametric to rectangular (1—6: 1), the walls straight, the trans- verse walls thicker than the longitudinal walls; proximal marginal cells 9-50 X 6-20 шп, isodia- metric to rectangular (1—5:1), with straight walls uniformly thickened or the transverse walls thicker than longitudinal walls; hyaline hair-points terete, straight, to 1.5 mm long, smooth or denticulate (ser- rate, with many acute teeth in some populations of Perichaetial leaves 1.5-2.8(-4) X .9 mm, convolute and larger (2-3X) than veg- etative leaves; hyaline hair-points terete, straight, o 1.5 mm, nearly smooth. Androecia axillary or ter- minal. Setae erect and straight or somewhat curved, 1—4.5 mm long. Capsules exserted, ovoid, ellipsoid or fusiform, symmetric (seldom slightly asymmetric G. montana). > Annals of the Missouri Botanical Garden at base), smooth, stramineous to castaneous, with stomata at the urn base or lacking stomata; exothe- cial cells 16-70 X 10—55 шп, isodiametric to rect- angular (14:1), thin- or thick-walled; annulus simple and persistent; peristome teeth 35-90 um wide at the base, entire, split and cribrose or irreg- ularly 2-3 cleft in the distal half, castaneous or orange, concolorous or contrasting in color with the urn; opercula conic, mammillate or rostrate; calyp- trae cucullate. Spores 9-14 jum. Distribution. АП continents. Grimmia subg. Orthogrimmia sect. Montanae is characterized by proximal marginal cells with the transverse walls thicker than the longitudinal walls, cucullate cayptrae, and a simple and persistent an- nulus. 4. Grimmia alpestris (Schleich. ex F. Weber & D. Mohr) Schleich., Cat. Pl. Helv. Ed. 2: 29. 1807 [1808]. Grimmia alpestris Schleich., Neues J. Bot. 1: 196. 1806 [1805], nom. inval. Trichostomum pulvinatum var. alpestre Schleich. ex F. Weber & D. Mohr, Bot. Tas- chenbuch 110. 1807. Campylopus pulvinatus var. alpestris (Schleich. ex F. Weber & D. Mohr) Brid., Muscol. Recent. Suppl. 4: 75. 1819. Dryptodon pulvinatus var. alpestris (Schleich. ex F. Weber & D. Mohr) Brid., Bryol. Univ. 1: 198. 1826. Grimmia donniana var. sudetica Huebener, Muscol. Germ. 175. 1833, nom. Шер. incl. var. prior. Grimmia don- niana var. alpestris (Schleich. ex F. Weber & D. Mohr) Hampe, Flora 20: 281. 1837. Guem- беба alpestris (Schleich. ex F. Weber & D. Mohr) Hampe, Bot. Zeitung (Berlin) 4: 125. 1846. Grimmia alpestris var. eualpestris Loes- ke, Laubm. Eur. Part I: 101, figs. 25b, 27а-с. 1913, nom. inval. TYPE: [Switzerland. Valais: ] In M[onte]. Sylvio, Schleicher s.n. [Schleicher, Plantae cryptogamae helvetiae, n^ 13] (lecto- type, designated by Мипог (1998), BM). = 2. Cardot & Тћег., іп Holz., Bot. Gaz. , tab. 11 fig. 2. 1900. Grimmia alpestris var. holsingeri (Cardo & Thér.) G. N. Jones, in Grout, Amer. 2: 31, 1933. Grimmia don- niana var. ү = (Cardot & Thér.) Wijk & Mar- gad., Taxon 9: 190. 1960. TYPE: [U.S.A.] Montana: Flathead Co. ы of Lake 1. А mi. No Belton, a station 000-7000 g here designated, PC; isolectotypes, NY, РС). Illustrations. Figure 8; Abramov and Abramova (1983: figs. 27, 9-12); Maier and Geissler (1995: fig. 24, sub G. sudetica). Dioicous. Plants in compact tufts, glaucous or green above and blackish below. Stems to 1.5 cm tall, with central strand well developed; axillary hairs 5—6-celled, 100-145 шт long. Leaves erect, appressed, and straight when dry, patent to spread- ing and somewhat flaccid when moist, 1-1.6 X 0.3-0.5 mm, ovate, acute, keeled, plicate, some- times weakly so, plicae cells undifferentiated or more commonly longer and narrower than the other laminar cells; margins plane proximally and in- curved at the apex; costa semi-terete, prominent on the dorsal surface, clearly delimited; lamina 2(-3)- stratose in the distal 25, smooth; distal cells 8—13 шп, isodiametric, bulging, smooth; proximal para- costal cells 10-35 X 8—20 um, isodiametric to rect- angular (1-3.5:1), the walls straight, uniformly thickened or the transverse walls thicker than the longitudinal walls; proximal marginal cells always with the transverse walls clearly thicker than the longitudinal walls, otherwise similar to the proximal paracostal cells; hyaline hair-points terete, straight, to 1 mm long, nearly smooth. Perichaetial leaves 2— 2.5 X 0.75–0.9 mm, convolute and larger than veg- etative leaves (2-3X); hyaline hair-points straight, mm, nearly smooth. Androecia terminal. Se- tae erect and straight, 2-4 mm long. Capsules ex- serted, mostly fusiform, seldom ovoid and then with a narrowed mouth, symmetric, smooth, castaneous, lacking stomata; exothecial cells 16-35 X 1 шт, isodiametric (1[2]:1), thick-walled; annulus simple and persistent; peristome teeth 50-70 шт wide at the base, entire or irregularly cleft in the apical part, brownish, concolorous with the urn; opercula conic or with a short and obtuse mammil- la; calyptrae cucullate. Spores (9)10-13(-14) um. Diagnostic characters. (1) Lamina 2(-3)-strato- se and plicate. (2) Cells always bulging. (3) Cap- sules mostly fusiform, castaneous, lacking stomata. (4) Exothecial cells + isodiametric with thick walls. Distribution (Fig. 9. Common in mountain chains in Europe and western North America, but known only from scattered localities in the inter- vening Asian areas. Mostly in open areas above the tree-line, between 1500 and 3400 m elevation on dry, exposed siliceous rocks. Mature sporophytes were present in 89.6% of the specimens studied. Grimmia alpestris is somewhat variable. The amount of cuticular wax and consequent glaucous color vary according to habitat conditions. A greater amount of cuticular wax develops in dry and sunny habitats. The capsules are usually fusiform, but ovoid capsules are occasionally found. Grimmia alpestris is easily distinguished from С. reflexidens when sporophytes are present. Diagnos- tic characters are the color and shape of the cap- Volume 85, Number 3 1998 Mufioz 385 Revision of Grimmia Subgenus Orthogrimmia 1 mm a,b С 1 mm 9g —— 50 um d-f — 20 um Figure 8. Grimmia alpestris. —a. Non-plicate leaf. —b. Plicate leaves. —c. Capsule. —d. Proximal leaf cells. — e. Transverse sections of d, e, Schleicher s.n. (JE).] sules: brown and fusiform with the base attenuated into the seta in G. alpestris (Fig. 8c); stramineous and ovoid and abruptly connected with the seta in С. reflexidens (Fig. 166). Dissecting capsules to study the shape of the exothecial cells and search for stomata may sometimes be necessary: G. alpes- tris has isodiametric, thick-walled exothecial cells leaf. —f. Medial exothecial cells. —g. Peristome teeth. [a, c, |, g, Vajda s.n. (BP-64281); b, and lacks stomata, whereas the exothecial cells in G. reflexidens are rectangular and thin-walled, and 4—8 stomata can be found at the urn base. On the other hand, sterile specimens can be dif- ficult to name. When present, a diagnostic feature is the recurved margin of G. reflexidens (always plane in G. alpestris). The axillary hair-points are 386 Annals of the Missouri Botanical Garden Ai ас Figure 9. Distribution of Grimmia alpestris. also useful: in Grimmia alpestris, they are 5—6- celled and 100-145 ¡um long, whereas in G. reflex- idens they are 3—4(—5)-celled and 50-85 шп long. Geissler and Maier (1995) and Maier and Geis- sler (1995) have employed the name G. sudetica Schwügr. for this taxon. This is an illegitimate name, however, as noted by Muñoz (1 rimmia alpestris has a discontinuous circum- boreal distribution pattern. It has a western Eu- rope—western North America disjunct distribution, and is not known from eastern North America or most of Asia. While this species is very likely ab- sent from North America east of the Rocky Moun- tains, the lack of records from temperate Asian re- gions probably has more to do with undercollection. This species is typically associated with mountain chains: the Swiss Alps, the Fennoscandian Alps, the Balkans, the Carpathians s.l., and the Pyrenees in Europe, the Rocky Mountains in western North America, and the Altai, Caucasus, and Northwest- Himalayan chains in Asia. Bardunov (1974: 40) re- ported this species from the western Sayan Moun- tains (Russia, Krasnoyarsk Kray), but I was unable to obtain these specimens on loan. Selected specimens examined (234). AFGHANISTAN. Panjskir 2. Ruka, Kóie s.n. T 1. ANDORRA. ei Turrach, 16 Aug. FH). Tirol: Sulden (am Ortler), lower part of the 7 92/522 (IBA-6279). BULGARIA. a, 2 Aug. 1908, Роарета s.n. (ТАН). CZECH RE PUBLIC. de Schlesien, Gesenke, Kessel, Limpricht s.n. (IBA-5468). FRANCE. Alpes Maritimes: S. Martin-Vésubie au о 28 July 1910, Durand s.n. (С). Ariège: pres de l'étang d'Areon, Pozo d'Aula, 5 Sep. a Y TEES 5 Ба E ue 1923, ari s.n. (Z). Corsica: montis Rotondo, supra “il Tim 7 July 1880, Levier s. autes Al- pes: mon ане d Galibier du ail a Lantiret, 26 Aug. 1926, Culmann s.n. (Z). Isére: La Lauvitel, 17 Aug. 1894, Thériot s.n. (TRH). Normandie: Montagne s.n. (RO). Pyrénées Centrales: marginem lacus Lac de Gaube dic- ti, non longe a Cauterets, Spruce s.n. (NY, TCD). Savoie: Col de la Croix de Fer, Cuynet 22 (herb. Pierrot). GEOR- GIA. Gurschevi ad fontes fl. Dschandschachi Tschali, July 1877, Brotherus s.n. (Н). Svanetia libera or. supra Chal- dechi et Kala, montis Djanga-tau, 8 Aug. 1890, Sommier ¿RMANY, Kleinhans. Anonymous s.n. (TRH). ITALY. Mte. Lineone, Balsamo: crie s.n. (RO); itte, : а, Kern s.n. Dan shove Айша jh E 10798 (MO). i sk. distr. Atbassar, montes Dshaksy Gordjagin 198 (H-BR). NEPAL. NW Himalaya, Troll 67 (JE). NORWAY. Christiams: Jotumheimen, Bukkelaegret, 31 July 1879, Bryhn “| (ТЕН). Sondre Trondhjems: Opdal, Sliper, 19 May 1882, Kaurin s.n. (TRH); Valders, UN dn 8 iu. 1889, Bryhn s.n. (TRH). PAKI- STAN. Hindukusch: Tschitral, Bumboret-Tal, 1935, Ker- Jed s.n. (JE). К 2. Kalapani, Kamri Nala, Astor Val- ey, 10 July 1901. yal-Khan s.n. (H-BR). 2. o Tatry Mts. + la W of Czarny Staw tarn, 24 ; 1987, Wójcicki s.n. (IBA-7417, KRAM) ROMANIA. Fo. garas: ad lacum Bulea, Boros s.n. (BP-112946). Beszter- ce-Naszód: Мі. Radnai havasok, montis Unókó, ad Опб- Кбі menedékház, Felfoldy s.n. (BP-112943). RUSSIA. Altayskaya Autonomous Oblast: Altai, Stonovajatal in кашк 1 Aug. 1915 Gand s.n. (H-BR). Kam- chatka: slope of Kosheleya er cano, 1 16 (LE). North Ossetia: 2 mare glaciale Zei, Aug. 1881, Brotherus s.n. (PC). SLOVAKIA. Siroka (Jaworiner-). Chałubiński s.n. (BP- 36661); versus bo od vallis Felkai-vólgy, Velická dolina, Boros s.n. (BP-112912). SPAIN. Gerona: Cire de Concrós, Lloret s.n. А 23455). Huesca: Benasc, La Renclusa, а l'ermita, Casas s.n. (ВСВ-27596). Lérida: Вог, 2. de Cavallers, Са- sas s.n. (BCB-39886). Cantabria: Peña Prieta. u 1987, Мипог s.n. (IBA-606). SW Г IZERL Abhange des Piz Pischierva, 21 July 1 WRSL). Rhaetia: Davos, 20 Apr. 1890, (TRH); Grand San Bernard, au Val Soren, : 1 s.n. : e 888, Adlerz s.n. (TRH). Wallis: Südabhang des Santsch- passes, 11 Aug. 1912, Culmann s.n. (FH). U din: Zernez, Mu М Baseglia, (WRSL). TADZHIKISTAN. N slope of Darvazhski moun- lain range p. Zhingou, 19 July 1964, Mamatkulov s.n. (LE). TURKEY. Art i cam Deglari), : уа) y (IBA-5098). Kayseri: Erciyas-dagh, Zederbauer s.n. (W- => — s.n. (BP-36623). Vandam-czai, 23 Aug. 1900, Alexeenko s.n. (LE). KINGDOM. Wales: Carmarthenshire, near Marros, 29 : Revelstoke, 7 May 1890, Macoun s.n. (S). U.S.A. Arizona: Kaibab, Mead 990 (FH). California: Alpine Co., Monitor, 14. s.n. (FH). Colorado: Larimer Co., Dream Lake, Rocky Moun- tain National Park, 1 Aug. 1929, Braun s.n. (MO). Idaho: Elmore Co., Atlanta, Boise National Forest, Anonymous s.n. (MO). Montana: Flathead Co.. pr. Lake McDonald, 4 Volume 85, Number 3 1998 Mufioz 3 Revision of Grimmia Subgenus Orthogrimmia mi. N of Belton, a station 30 mi. E of Kalispell, 25 e 1898, Holzinger & Blake s.n. 2. ). Utah: San Juan Elk Ridge, at Kigalia Ranger Sta., Flowers 3633 M Wyoming: Evanston, Degener & Peiler 16928 (FH). 5. Grimmia caespiticia (Brid.) Jur., Laubm. “Fl. Oesterr.-Ung. 172. ticius Brid., Muscol. Recent., yd 1819 [1818]. Grimmia pen var. с аи (Brid.) Hampe, Flora 282. 1837. Gue belia caespiticia rg x Hal., Syn. Musc. Frond. 1: 773. . Grimmia alpestris var. caespiticia 1. pu N. Jones, in Grout, Moss Fl. N. Amer. 2: 30. 1933. TYPE: [Switzerland. Bern:] St. Bernard, Bridel s.n. (holotype, B). Grimmia sulcata Saut., Flora 24: 39. 1841. Guembelia sul- a es Hampe, Bot. Zeitung (Berlin) 4: 125. 1846. TYPE: [Austria. Salzburg:] Pinzgauer 1. ке: s.n. (lectotype, here designated, H-SOL; lectotype, Z). Grimmia jacquinii Garov. var. subimberbis Lindb., Öfvers. “бгһ. Kongl. Svenska Vetensk.-Akad. 23: 552. 1866 [1867]. Grimmia 2. var. subimberbis (Lindb. Berggr.. Kongl. ska Vetenskapsakad. Handl. 13(7): 49. 1875. Cumis caespiticia var. subimberbis (Lindb.) Limpr., Laubm. Deutschl. 1: 780. 1889. Grimmia sulcata var. subimberbis (Lindb.) G. Roth, Eur. Laubm. 1: 431. 1903 [1904]. Grimmia caespi- ticia f. 4. (Lindb.) Родр., Consp. Musc. Eur. 283. 1954. TYPE: [Norway.] Spitsbergen: Am sterdam Island, bol. Holmgren s.n. (lectotype, et A Grimmia manniae Miill. Hal., Flora 70: 223. 1887. Grim- mia alpestris var. manntae ba Hal.) G. N. Jones, in Grout, Moss Fl. N. Amer. 2: 31. 1933. Grimmia donniana var. oa өші, Hal.) Wijk € Margad.. Taxon 9: 50. 1960. Т c S.A.] California: Napa Soda Springs, 3 Ha . Mann s.n. (lectotype, here Дейш, NY; 5. РС) auf eisenhalbsigen Felsen de ort de charo spanische Seite, 13 July 1914, Kern s.n. (lectotype designated by Bednarek-Ochyra et al. (1992), /RSL: 2 BP-80423, KRAM, MA-8158, MA-10927). Illustrations. Figure 10; Bednarek-Ochyra et al. (1992: fig. 1); Bruch et al. (1845: tab. 252, sub G. sulcata); Chałubiński (1882: tab. 9 fig. 16); Lim- pricht (1890: fig. 203); ps and Geissler (1995: fig. 6); Nyholm (1956: fig. 69C). Dioicous. Plants in fragile tufts, glaucous or ol- ive-green to blackish. Stems to 1 cm tall, with cen- tral strand well developed; axillary hairs 4-celled, 75 um long. Leaves erect and appressed, and with incurved apices when dry, erect and rigid when moist, 0.8—1.8 X 0.3—0.6 mm, ovate, acute, keeled, strongly plicate, the plicae of 2—5 rows of longer and more narrow cells with thicker walls; margins plane in the proximal half, incurved in the distal half and cucullate at tip; costa semi-terete, prominent on the dorsal surface, clearly delimited; lamina 2(—3)-stratose in the distal half, smooth; dis- tal cells 7-9 um, isodiametric, bulging, usually pa- pillose; proximal paracostal cel 35 X um, isodiametric to rectangular (1—3: 1), the walls straight, uniformly thickened or the transverse walls thicker than the longitudinal walls; proximal marginal cells 12-32 X 10-14 um, isodiametric to rectangular (1-3:1), with the transverse walls thicker than the longitudinal walls; hyaline hair- points terete, straight, to 0.4 mm long, smooth. Per- ichaetial leaves ca. 1.5 X 0.4—0.5 mm, convolute and slightly larger than vegetative leaves; hyaline hair-points straight, to 0.5 mm, smooth. Androecia terminal. Setae erect and straight, 2.5—3.5 mm long. Capsules exserted, ovoid, symmetric, smooth, cas- taneous, with stomata at the urn base; exothecial cells 24-70 X 10-28 um, very irregularly isodia- metric to rectangular (1—3: 1), thin-walled; annulus simple and persistent; peristome teeth 50 um wide at the base, = entire, brownish, concolorous with the urn; opercula mammillate or rostellate, and then with an oblique beak; calyptrae cucullate. Spores 10—14 um. Diagnostic characters. (1) Leaves with а strong- ly marked longitudinal plication on each side of costa. (2) Margin involute in the distal half and becoming cucullate at tip. (3) Hyaline hair-point very short, to 0.5 mm on the perichaetial leaves, but usually much shorter. (4) Laminar cells bulging. (5) Distal leaf cells usually papillose. (6) Capsules castaneous, with stomata at the base. Distribution (Fig. 11). Europe, the Caucasus, Svalbard, and the west coast of North America; in open areas above the tree-line from 1000 and 2797 m elevation on dry, siliceous rocks. Mature sporo- phytes were present in 7796 of the specimens stud- ied. Grimmia caespiticia is rather invariable in most respects, except glaucousness, which depends on the amount of wax deposited on the leaves. As in G. alpestris, this variation correlates with habitat conditions. The degree to which the cells bulge also varies. Usually they bulge prominently, but may be nearly smooth in some populations. In fertile condition, the combination of plicate leaves, very short hair-points, and exserted cap- sules clearly distinguishes G. caespiticia from any other species in Grimmia or Coscinodon. Another useful character for separating this taxon from G. alpestris is the presence of stomata at the base of the capsule, absent in the latter species. In sterile condition, it is difficult (or at times impossible) to 388 Annals of the Missouri Botanical Garden 1 mm e b —— 0.1 mm d ——— 50 um c,f-h — 20 um Figure 10. Grimmia caespiticia. —a. Leaves. —b. Dorsal view of papillose leaf apex. —c. Proximal leaf cells. — d. Peristome teeth. —e Capsule. —f. Transverse sections of leaf with non-papillose cells. —g. Medial exothecial cells. —h. Proximal exothecial cells and stoma. |а, b, с, f, Pierrot 201/100 (herb. Pierrot); d, e, g, h, Rupidera 3 (SALA).] separate this taxon from Coscinodon cribrosus, а species also difficult to distinguish from G. alpestris and С. reflexidens when sterile. The differences al- leged by previous authors (Cao & Vitt, 1986: 168; Nyholm, 1956: 151) are not reliable; the length of the proximal cells, the thickness of the cell walls, and the shape of the leaf apex vary inordinately in both taxa, as can be verified from fertile collections. The only useful character for separating sterile ma- terial of G. caespiticia and C. cribrosus is the usually papillose laminar cells in the former, as opposed to the always smooth laminar cells in the latter. This feature, though noted by Loeske (1913: fig. 32), has been neglected by later authors. Another useful character is the length and shape of the hair-points. Grimmia caespiticia has very short hair-points, nev- er longer than 0.5 mm, which are always terete, even on the longest hairs. Hair-points in Coscino- don cribrosus, however, are usually longer than 0.5 mm and flat below. From these differences, and af- Volume 85, Number 3 1998 Mufioz Revision of Grimmia Subgenus Orthogrimmia Distribution of Grimmia caespiticia. Figure 11. ter studying the lectotype (H) and two isolectotypes (BM, of G. sinensianodon Müll. Hal. (Müller, 1898: 188), I conclude that it is synonymous with C. cribrosus (Hedw.) Spruce and not with G. caes- piticia as proposed by Cao and Vitt (1986: 167). Those authors reported the type locality of G. si- nensianodon to be the only known Chinese station for G. caespiticia. Because I have not found any specimens referable to G. caespiticia among the Chinese collections studied, this species must be excluded from the Chinese bryoflora. The protologue of Grimmia manniae states “2 Majo 1886" as the collection date. The only two specimens I can find collected by Martha R. Mann in Soda Springs, the type locality, have “3 Majo 1886" as the collection date (NY, PC). Considering that they agree in every other respect with the pro- tologue and that there are more typographical mis- takes in the same paper (Müller, 1887), I have as- sumed that a mistake could have been made in transcribing the label in the original publication. Selected specimens examined (163). ANDORRA. Tris- taina, Casas s.n. (BCB-10537). ARMENIA. Sisianskij ra- jon, s. Arevis, 5 July 1966, Manakyan s.n. (LE TRIA. Carinthia: Monte Caglians, auf dem Piz Ciadin, 8 July 1908, Kern s.n. (WRSL). Steiermark: Würslingen Hóhe bei Stadl, 12 July 1878, Breidler s.n. (GOET). Tirol: Alpe Rosstall bei Innervillgraten, Gander s.n. (BP-36291). BELGIUM. Promenade Annette Lübin à Spa, Feb. 1906, Cornet s.n. (NY). BULGARIA. Mt. Pirin, decl. borealium mt. Mangar Tepe, Simon s.n. (BP-68488); Wichryn, ad viam Zwusticam, Kuc s.n. (BP-70237). FRANCE. Alpes- Maritimes: Saint-Dalmas-de-Tende, les Ciappe de Fon- tanalba, Parriat s.n. (BP-112934). Cantal: Plomb-du- Cantal, Cuynet s.n. (BP-112904). Hautes Alpes: plane nord du Combegnot, 21 July 1926, P. Culmann s.n. (Z). Isére: les Grandes Rousses, environs du lac Blanc, 20 Aug. 1894, Thériot s.n. (7). Puy de Dóme: vallée du Chandeferer, 19 Aug. 1919, Culmann s.n. (Z). Pyrénées Centrales: Port de Bénasque, 12 Sep. 1845, Spruce s.n. (NY). GEORCIA. In jugo alpino inter flumina Neusksa et Sckun pr. Svaniae occid. confine, Sommier & Levier 364 (H). South Ossetia Autonomous Oblast: in alpe Zo- July 1895, Kern s.n. (WRSL); Rizlembrai juxta Cól di Stel- vio, Lorentz s.n. (BP-36283). NORWAY. Svalbard: Am- Kozi-Wirch, den Gipfel, 3 Sep. 1876, Chatubirisky s.n. (BP-36297). PORTUGAL. Beira Alta: Serra da Estrela, Cántaro Raso, pr. da vista para o Covao da Ametade, Sér- gio & Séneca 8164 (IBA-5050). ROMANIA. Brasov: val lis ragu, montes Fogarasi havasok, Vajda s.n. (BP 72021). RUSSIA. Irkutsk: Western Sayan, the river Ona in the upper part, Bardunov s.n. (NICH 306126). Stav- ropols'kij Kraj: Teberdinsky Reserve, 14 Sep. 1954, Patrabolova s.n. (LE). SLOVAKIA. Tatra Magna, convallis “Sirkert” sub monte Lomnicki-csucs, Boros s.n. (ВР- (IBA-6723). Cantabria: Peña Prieta, 14 Aug. 1987, Ми- fioz s.n. (BCB-25772). Gerona: Coma de l'Orri, Lloret s.n. (BCB-28890). Huesca: Benasc, La Renclusa, Ibón de Paderna próximes al refugi, Casas s.n. (BCB-33173). Palencia: ico Curavacas, 12 July 1988, Muñoz s.n. (IBA-68 SWITZERLAND. Bern: Urbachtal bei der Gaulihütte, Culmann s.n. (MA 5870). Graubünden: Pischahorn im Fluelathal, Amann BH-12 (Z). Valais: Distebalp, Saastal, Amann BH-24 (Z). TURKEY. Bursa: Nordabhang des Ulu Dag, Walther 3318 (NY). S Lake, Blue Mts., 17.5 mi. W 28821 (G). Washington: Pierce County, Mt. Rainier Park, about 0.5 mi. from ranger station, Lawton 4792 (MO). 6. Grimmia montana Bruch & Schimp., in Bruch, Schimp. & W. Gümbel, Bryol. Europ. 3: 128, tab. 250. 1845. Guembelia montana 1858, nom. inval. pro syn. TYPE: [Germany. Rhineland-Palatinate:] Donnersberg, Арг. 1843, Gümbel s.n. (lectotype, designated by Cao & Vitt (1986), BM). Grimmia laxa Müll. Hal., Bot. Zeitung (Berlin) 5: 801. 1847. TYPE: Mexico. In monte Orizabae, Deppe & Grimmia fragilis Schimp., Syn. $251. 76, nom. illeg., non F. Weber, 1804. Grimmia montana. var. fragilis (Schimp.) Loeske, Laubm. Eur. Part I: 99. 1913. TYPE: [Portugal.] Ad rupes granit. in alpestribus Prov. Beira, Aug. 1848, Welwitsch s.n. (lectotype, here designated, BM). 390 Annals of the Missouri Botanical Garden Guembelia tenella Müll. Hal., Bot. Centralbl. 44: 388. ка Grimmia tenella (Müll. Hal.) Kindb., Enum. . Exot., Suppl. 2: 107. 1893. TYPE: [U.S.A.] Idaho: Coeur d'Alene, 6 Aug. 1888, Roll s.n. (lec- totype, here designated, G; isolectotypes, H-BR, JE, 5). Grimmia montana var. longifolia Cardot, in Gasilien, Rev. Bryol. 21: 24. 1893 [1894]. Grimmia montana f. lon- ма Е Podp., Consp. Musc. Eur. 281. 1954. E: [France. Puy de Dóne | Auvergne, Pierre sur un 1884. Gasilien s.n. (lectotype, here designat- ed, ). Grimmia montana var. abnoba H. Schmidt, Mitt. Bad. esvereins Naturk. Naturschutz Freiburg 2: 121, 1927. ТҮРЕ: [Germany. Wiirttemberg:] Süd. Schwarzwald, südweste dem Schauinslandgipfel, Schmidt s.n. (lectotype, here designated, JE). m montana f. submutica J. E. Zetterst. ex H. Möller. Bot. 26A(2): 31. 1933 [1934]. TYPE: [Sweden tae Husbyborg prope Upsaliam, 15 May 1855, Zet- terstedt s.n. (Zetterstedt, Grimmiae et Andreaeae exsiccatae, n° 21c] (lectotype, here designated, H- BR). Illustrations. Figure 12; Bruch et al. (1845: tab. 250); Chałubiński (1882: tab. 8 fig. 14); Jóhannsson (1993: fig. 26); aid and Geissler (1995: fig. 17); Nyholm (1956: fig. 6 Dioicous. Plants in s or bulging cushions or tufts, olive-green at the tips, dark green or blackish below. Stems to 1 cm tall, with central strand well developed; axillary hairs 5-8-celled, 95-175 шп long. Leaves erect, loosely appressed and flexuous when dry, with patent proximal part and incurved apex, sigmoid in lateral view, rigid when moist, 1- 2 X 0.3-0.6 mm, abruptly acuminate from an ovate base, apex from somewhat to distinctly keeled, plane or with very weak plicae; margins plane prox- imally and incurved distally, forming a canaliculate apex; costa semi-terete, prominent on the dorsal surface, slightly to distinctly delimited; lamina 2(3—4)-stratose in the distal half, smooth; distal cells 4-8 um, isodiametric, plane or slightly bulg- ing on the dorsal surface, smooth; proximal para- costal cells 20-50 X 8-15 jum, rectangular (24.5: 1), the walls straight, uniformly thickened or the transverse walls thicker than the longitudinal walls; proximal marginal cells 20-50 X 8-15 jum, rect- angular (24.5: 1), with the transverse walls thicker than the longitudinal walls; hyaline hair-points te- rete, straight, to 1.5 mm long, obtusely denticulate (rarely serrate). Perichaetial leaves 1.7-2.4(—4) X mm, convolute and larger than vegetative leaves (2X), yellowish at the base; hyaline hair- points similar to those of vegetative leaves. Androe- cia terminal. Setae erect and straight, 2-4 mm taneous, lacking stomata; exothecial cells 30—70 х 10-25 um, rectangular (2—4: 1), thin-walled; an- nulus simple and persistent; peristome teeth 50—90 шт wide at the base, irregularly split and perforate, = cribrose, castaneous, concolorous with the urn; opercula rostrate, the beak oblique; calyptrae cu- cullate. Spores 10—14 шт. Diagnostic characters. (1) Leaves flexuous, sig- moid in lateral view when moist, ending in a long acuminate apex. (2) Laminar cells not bulging. (3) Plants dioicous. (4) Capsules lacking stomata. (5) Opercula rostrate with oblique beak. Distribution (Fig. 13). | Western Europe, the Ca- nary Islands, Greenland, and North America; co- niferous and broad-leaved formations, and also open areas, mostly below the tree-line, between 300 and 2000 m elevation on dry, siliceous rocks. Ma- ture sporophytes were present in 69% of the spec- imens studied. Grimmia montana is a rather stenotypic taxon throughout its entire distribution range. Some minor gametophytic differences between North American and Eurasian specimens may be observed, but the sporophytes are identical worldwide. Eurasian pop- ulations from lower latitudes and/or elevations have longer and more acuminate leaves than those from higher latitudes and/or elevations. North American populations resemble northern Eurasian popula- tions in having leaves with a short leaf acumen. In very exposed places at high altitudes, cells bulge slightly at the dorsal laminar surface, and the ac- umina are shorter. Male plants are shorter and have leaves with shorter hair-points, sometimes muti- cous. They are easy to recognize because their stems end in globose perigonia. Circumboreal but scattered, Grimmia montana is the only species of subgenus Orthogrimmia known from Greenland and Baffin Island. Oddly, there are no collections from continental eastern North America; all material from that region so identified actually represents G. incurva with erect ап straight setae. Brodo and Alstrup (1981: 231-233) documented the same distribution pattern for two lichens, Bryoria subdivergens (Dahl) Brodo & D. Hawksw. and Rhizocarpon bolanderi (Tuck.) Herre, and noted that such a pattern is unknown in other plant groups. Subsequently, Blom (1996: fig. 46) illustrated the same pattern for Schistidium umbro- sum (J. E. Zetterst.) H. H. Blom (Musci, Grimmi- aceae). Grimmia montana was recorded from China and Tibet by Cao and Vitt (1986: 161—164, fig. 19), but all specimens listed by those authors represent oth- er species, mainly G. longirostris; hence tana can be deleted from the Chinese bryoflora. . mon- Volume 85, Number 3 1998 391 Revision of Grimmia Subgenus Orthogrimmia алта ыы а ех JO% , Figure 12. Grimmia montana. —a leaf. —f. Medial exothecial cells. —g. s.n. (TRH).] eaves. —c. Cao and Vitt (1986: 161) also synonymized 6. bra- chyphylla Cardot with G. montana, based on sev- eral character states shared by these two species. Nevertheless, the features listed by these authors are common to many other species in the genus, and thus inconclusive. Based on a study of the .b Capsule. Peristome teeth. [a, Gasilier — d. Proximal leaf cells. —e. Transverse sections of 1 s.n. (ТЕН); b. d, е, Gümbel s.n. (ВМ); с, f, g, Kaurin types, I conclude that G. brachyphylla? is synony- mous with Coscinodon humilis Milde. 3 TYPE: South Korea. Pomasa, 800 m, 1906, Faurie 218 (lectotype, here designated, PC; isolectotypes H, H- BR, NY). Annals of the Missouri Botanical Garden (> у Figure 13. Distribution of Grimmia montana. Grimmia montana has been also reported from the Russian Far East (Afonina, 1986: 223, ris. 2 figs. 1—8; Ignatov & Afonina, 1992: 42); however, all the specimens supporting these reports are G. longirostris or G. reflexidens. Selected specimens examined (294). ANDORRA. Tris- taina, Casas s.n. (BCB- 2. AUSTRIA. Niederoster- gun-russes bei Spa, Cornet s.n. (BP-36286). FRANCE. Maine-et-Loire: Angers, 10 Apr. 1857, Perraudiére s.n. (G). Calvados: Condé-sur-Noireau, Husnot s.n. (FH : circa dei sero iy n. (FH). North Rhine- Wes tphalia: Bertram s.n. (IBA-4125). ITALY. Adamellogruppe, Val di Genova, 28 July 1895, Kern s.n. (WRSL). NO jugis alpinis editionibus, 31 July Beira Baixa: Sabugal, a 2 km de Quadri e = río Cóa, Sérgio s.n. (IBA-4506). Douro Litoral: n do Serra do Marao, Ervideira s.n. (LISU- -P53901). Minho: Serra de Peneda-Geréz, Soajo, Branda da Bouga dos Ho- mens, Sérgio & Sim-Sim s.n. (LISU-154131). Trás-os- Montes: Pitoés das Junias, Montalegre, Sérgio = Schumacker s.n. (LISU-154132). SPAIN. Almeria: Sierr de los Filabres, barranco de la Verruga, 26 May 1990, García-Zamora & Ros s.n. (MUB). Asturias: Puerto de Leitariegos, 15 July 1835, Durieu s.n. (PC). Ávila: bajada a las lagunas de El 2. 9 Sep. 1 (VIT 1050/84). Cantabria: Puertos de Riofrío, 12 July (IBA-618). Cáceres: Acebo, Cros & 88, Muñoz s.n. Urdiceto, Casas s.n. (IBA-3984). Lérida: Alta dure marge esquerre de la Noguera Ribagorcana, Canalís s (BCC-1278). León: Ancares, subiendo al Сшћа, Cros & Lloret s.n. (BCB-20118). La Rioja: Posadas, Sierra de la Demanda, Casas s.n. (BCB-2472). Madrid: Puerto de Na- vacerrada, Casares Gil s.n. (MA 8156) Mendaur, Arraiza s.n. (NAU-3284). Fonte da Cova, Aedo s.n. (IBA-3177). cebollas, Aedo s.n. (IBA-3933). Salamanca: Alto de Los Lobos, subiendo a la Pefia de Francia, 26 Sep. 199 йог s.n. (IBA-3595). 1. Cañadas del Teide, Apr. 1906, Pitard s.n. (MC amora: San Pedro de las Her- rerías, Rupidera i er SWEDEN. Mt. Huddinges, 2 June 1902, Arvén s.n. (FH); Norrkáping, 1878, Olsson s.n. (TRH); E ‘Lidingo. 30 Apr. 1859, Zetterstedt s. n. ymous s.n. (BM). UNITED KINGDO OM. England erset, near Highbridge, 24 Mar. 1913, Nicholson s.n. (FH). CAN i "Yers peni AND. Gro enlandia boreal, Claushavn, 1870, Berggren s.n. (TRH); Greenland, Narssaq, on Narssaqsund at Gronlands Geologiske Undersogelser camp and E vicinity, Steere 62-982 (NY). MEXICO. Baja Californi National Park, San Pedro Mártir Mts., Hammond 10837 ur ая Island, summit, Morin 5664 (FH). Méx- atépetl, Sharp 4749 (TENN). Puebla: S slope of accua, Vitt 17520 (TENN). Veracruz: Monte Ori- zaba, л 4277 (PC). U.S.A. California: San Bernar- dino Тұ face of Cucamonga Mt., Sweet Jr. 189 (FH). med Eagle Co., 1 mi. W of Wolcott, Weber B-15354 (MO). Idaho: Elmore Co., Hot Springs, Atlanta, Berse National Forest, MacFadden 18766 (FH). Montana: Min- eral Co., shore of Clark Fork, Rest Area on US 10, 3 mi. У of Alberton, Hermann 22562 (FH). Nevada: Kings n o 11541 (MO). Washington: Spokane, 11 May Bonser s.n. (FH). Wyoming: Yellowstone, 2 Sep. үө Roll s.n. (JE). 7. Grimmia nivalis Kindb., in Macoun, Bull. Tor- rey Bot. Club 17: 271. 1890. TYPE: [Canada.] British Columbia: Gold Range, 7000 ft., 10 Aug 1889, Macoun s.n. (lectotype, here des- ignated, S; isolectotypes, CANM-198090, H- BR) Grimmia 2. Kindb., Rev. Bryol. 34: 89. 1907. T Canada. British Columbia: Skagit summit, 18 July 1905, Macoun s.n. (holotype, S). Illustrations. Figure 14; Allen (1995: figs. 3- 5 Dioicous. Plants in tufts, olive-green. Stems to 1.5 cm tall, with central strand weakly developed; axillary hairs 5—7-celled, ca. 130 ¡um long. Leaves erect and appressed when dry, patent, rigid when moist, 1.1—1.8 X 0.3-0.5 mm, ovate, acute, strong- Volume 85, Number 3 1998 393 Revision of Grimmia Subgenus Orthogrimmia — = DI + rm mE. E rr в A Yam», өз: ES 72 um => AS e EN ee: lev t on = ж Ons PO Figure 14. —b. Leave Grimmia nivalis. —a EA \ M | Ec . —c. Perichaetial leaves. —d. Transverse sections of leaf. — g Habit. e. Proximal leaf cells. —f. Medial rte ial cells. —g. Modi teeth. [Howell s.n. (FH).] ly keeled, plane or weakly plicate; margins plane proximally and incurved distally; costa semi-terete, prominent on the dorsal surface, clearly delimited; lamina 2-stratose in the distal half, smooth; distal cells 6-9(-11) um, isodiametric, bulging, papillose; proximal paracostal cells 17—3 15 шп, isodia- 1), the walls straight, uniformly thickened or the transverse walls thicker metric to rectangular (1-2: than the longitudinal walls; proximal marginal cells 9—14 X 9-11 pm, isodiametric to rectangular (1— 1.5:1), with the transverse walls thicker than the longitudinal walls; hyaline hair-points terete, straight, to 0.5 mm long, smooth. Perichaetial leaves 1.7-2.8 X 0.6—0.9 mm, convolute and larger than vegetative leaves (2X); hyaline hair-points straight, to 0.75 mm, smooth. Androecia not seen. Setae erect Annals of the Missouri Botanical Garden 394 VAI o YO \ © ax 9% 4) SN a 0 30? сь Е 120° 90° Figure 15. Distribution of Grimmia nivalis. and straight, ca. 3 mm long. Capsules exserted, ovoid, symmetric, smooth, castaneous, with stomata at the urn base; exothecial cells 28-55 x 10-35 im, rectangular or (seldom) isodiametric (1.4—4: 1), thick-walled; annulus simple and persistent; peristome teeth 55—80 um wide at the base, split in 2-3 branches in the distal 4—%, castaneous, con- colorous with the urn; opercula conic, mammillate or short-rostellate, and then the beak oblique; ca- lyptrae cucullate. Spores 9-14 um. Diagnostic characters. (1) Lamina plane or weakly plicate. (2) Cells bulging and papillose. (3) Plants dioicous. (4) Capsules castaneous with sto- mata at the base. Distribution (Fig. 15). Known only from the northwestern United States (Washington) and west- ern Canada (British Columbia); open areas between 620 and 2100 m elevation on siliceous, seemingly dry rocks. Mature sporophytes were present in 94% of the studied specimens. Both Grimmia nivalis and G. alpestris have ovate, somewhat glaucous leaves, long hair-points and bulging laminar cells, and their overall macroscop- ic aspect is quite similar. Nevertheless, microscopic examination distinguishes both species at once: transverse leaf sections of G. nivalis reveal scat- tered thick papillae on the lamina, and stomata on the capsules. Leaf cells of G. alpestris are bulging but not papillose, and capsules of this species lack stomata. Moreover, capsules of G. alpestris are mostly fusiform in shape, whereas those of G. ni- valis are ovoid. Grimmia caespiticia, the only other species in subgenus Orthogrimmia with papillose laminar cells and stomata on the capsule, usually has very short hair-points and strong plicae along both sides of the costa. Allen (1995: 162-164) considered Grimmia ni- valis synonymous with G. tenerrima Renauld & Cardot (G. reflexidens in the sense of this paper). Both species are macroscopically similar, but they can be separated because б. nivalis is a dioicous species with papillose cells and castaneous cap- sules, whereas G. reflexidens is autoicous, never has papillose cells, and the capsules are stramineous. CANADA. British Colum- ud Macoun s.n. (CANM- ‚ NY [4 oie E A Macoun s.n. 18 July 1905, Macoun s.n. pa 1889, Macoun s.n. (PC). U.S.A. 15 July 1931, Howell s.n. (FH); 23 July 1931, Howell s.n. (FH). Specimens examined (13). bia: Gold Range, 10 e 198089, CANM-198090, I Summit Lake, 18 July 135430); Skagit ui Spence's Bridge, 28 May Baker, Washington: Mt. Clallam Co., Mt. Angeles, 8. Grimmia reflexidens Müll. Hal., Syn. Musc. Frond. 1: 795. 1849. TYPE: Chile. Póppig s.n. (lectotype, here designated, BM; isolectotypes, JE, NY, PC) Grimmia subsulcata Limpr., Laubm. Deutschl. 1: 757. „птита sessitana var. subsulcata (Limpr.) - Nene Naturwiss. Vereines Steiermark 28: 88. 1892 [1891]. ЕЕ alpestris subsp. subsulcata (Limp) Kindb., Eur. N. Amer. Bryin. 2: 221. 1898 а d n subsulcata (Limpr.) Loeske, Laubm. Eur. Part 1: 104. 1913. (ти Као var. subsuleata impr Broth., in Engl., Nat. Pflan- zenfam. Ed. 2, 10: : 1924. Grimmia sessitana f. a | impi) ee Biblioth. Bot. 101: 117 1930. TYPE: [Austria.] Steiermark: auf Glimmer- schiefer . bei Schoder,” 9 Aug. 1888 er s.n. (holotype, BP). Guembelia lamellosa Müll. Hal., Bot. ae (Berlin) 12: 318. 1854. Grimmia lamellosa (Miill. Hal.) A. Jae- ger, Ber. Thütigk. St. Gallischen Naturwiss. Ges. 1872/73: 72. 1874. TYPE: | Егапсе. Наше Garonne: | in Pyrenaeis centralibus, ad lacum Espingo (lec- ж designated by Deguchi (1978), PC; isolecto- type, H-SOL). Ыш. нй De Not., Atti Reale Univ. Genova 1: 869. e exannulata Lindb. ex Broth., Acta Soc. Sci. Fer 1892, nom. inval. eii alpestris var. stomata Loske, Laubm. Eur. Pa 1913, nom. illeg. Grimmia alpestris var. sessitana De Not.) I. Ha- gen, Kongel. > Vidensk. Selsk. heim) 1909(5 E tana (De Not А 1913. ТҮРЕ: ‘tay Vercelli:| Frane alle өсештігіні del Vogna, sotto при della Valdobbia in Val Ses- sia, Carestia s.n. (lectotype, designated by Cao & Vitt (19 d RO: isolectotypes, BM [2 replicates]. FH. VRSL 2 igs. 26e, f, g, 27d, e. i Grimmia anceps Boulay, Musc. France 1: 371. 1884. TYPE = rance. Haute-Savoie: Col de Berard a d'Anchane-Mt. Blar ayot s.n. (lectotype, here denia PC; Des totype, BP-36692). Volume 85, Number 3 1998 Mufioz 395 Revision of Grimmia Subgenus Orthogrimmia Grimmia tenerrima Renauld & Cardot, Bot. Gaz. 15: 40, l. 6 90. TYPE: [U.S.A.] Oregon: Mt. Hood, Henderson 1239 (lectotype, here designated, PC; is- olec toy pe, NY). Grimmia grisea Cardot, Bull. Herb. gom sér. 2, 6 1906. TYPE: South Georgia. mberland Bay. im 307 (holotype, PC? not Phe isotypes, H- d pt Kindb., Rev. Bryol. 36: 98. 1909. TYPE: [Canada.] British Columbia: Ambean Valley, 10 July 1908, 7 s.n. (lectotype, here designated, CANM-19810 Grimmia subcaespiticia Sc m. An aturhist. Hofmus. 27: 490. 1913. TYPE: Fieles: pr Kurdistania occidentalis, Taurus Cataonicus, in con- Bekikara inter ur- ed, FH; isolect Grimmia ppc ha ШТ & Sainsbury, . & Ps Roy. Soc ealand 75: 173. 0 TYPE: —€— fans poer Tasman Glacier, near D la Beche Hut, Sainsbury 756 (lectotype, here n= ed, BN: syntypes, BM [Scinabury 753, 157, 1910, Handel- Vni s.n. (lectotype, here designat- -BR). 772, 799). Grimmia ds f. d am Pamietn. Fizyogr. 2 68. ;rimmia alpestris var. hybrida (Chal.) Chal Eae Мике. уља Tatr 56. 1886. M in )] . Biblioth. 116. 1930, nom. illeg. incl. f. prior. а [Slovak- ia.| Tatra, Polnischer-Kaum, v.d. Felkaér-Thal, 27 Aug 1879, ai s.n. жүбы here desig- ted, BP-367( Gia бов а antarctica Кис, Rev. Bryol. Lich- ( 5‹ ar 1, 2, 3d, e, 4, 5. 1969. TYPE: Ant ca. Queen Mary Lan i 5. of the Polish Base, 25 June 1959, Rózycki s.n. (holotype, ККАМ; isotype, ТВА-7386). Illustrations. Figure 16; Cao and Vitt (1986: fig. 20a, c, e-l, n, р, д, s, и; Deguchi (1978: fig. 41, sub С. subsulcata); Maier and Geissler (1995: fig. 23, sub G. sessitana). Autoicous. Plants in tufts, yellowish green at the tips, brownish green, brown, or blackish below. tems ca. , occasionally with rhizoids throughout, же central strand well developed; ax- illary hairs 3A(5)-celled, 50-85(-100) um long. Leaves erect and appressed when dry, patent to spreading and rigid to somewhat flaccid when moist, 1-1.8 X 0.2 mm, ovate to narrowly ovate, acute, keeled, plane or very weakly plicate; margins plane or recurved to % the leaf length on one side and proximally on the other side, occa- sionally и briefly and narrowly recurved proxi- mally on one side; costa the О, ris clearly delimited; lamina 2-stra- tose in the distal half, smooth; distal cells 8-12 um, isodiametric, plane or bulging, smooth; proximal paracostal cells 20-55 X 8—10 um, rectangular (2- 6:1), the walls straight, uniformly thickened or the transverse walls thicker than the longitudinal walls; a semi-terete, prominent on proximal marginal cells 20-50 х 8-13 шит, rect- angular (2-5: 1), with the transverse walls thicker than the longitudinal walls; hyaline hair-points te- rete, straight, to 0.5 mm long, smooth or very weak- ly denticulate. Perichaetial leaves 1.8-2.7 X 0.5- mm, convolute and larger than vegetative leaves (2X); hyaline hair-points terete or somewhat flattened proximally in longer hair-points, straight, to 1.5 mm, nearly smooth. Androecia axillary or ter- minal. Setae erect and straight, 1.5-3 mm long. Capsules exserted, ovoid or ellipsoid, symmetric (rarely slightly asymmetric at the base), smooth, stramineous, with stomata at the urn base; exothe- cial cells 35-70 X 10—35 jum, irregularly rectan- gular (1.5-3:1), thin-walled; annulus simple and persistent; peristome teeth 45-70 um wide at the base, entire or split, orange, contrasting in color with the urn; opercula conic, obtuse or mammillate; ш = cucullate. Spores 10-14 ш Diagnostic Я. (1) Axillary hairs 3— 4(5 )-celled, 5 00) um long. (2) Proximal cells mainly PE 2—5:1. (3) Capsules stra- mineous, with stomata at the base. (4) Exothecial cells rectangular, with thin walls. Distribution (Fig. 17). Known from all conti- nents; woody formations and open areas from 850 to 3500 m, except in Antarctica and surrounding islands, where it grows at sea level. It prefers damp or moistened habitats close to rivulets and snow- beds, and is not uncommon in shaded places. With- in section Montanae, this species is the least xe- ophilous member, even growing on rocks that are periodically flooded, mainly from spring thaw. Ma- ture sporophytes were present in 82% of the stud- ied specimens. Grimmia reflexidens 1s a widely distributed and gametophytically variable species, and identifica- tion of sterile specimens can be impossible. Plants growing in drier and more sunny habitats were de- scribed in Europe as G. subsulcata. They have shorter and more ovate leaves with (usually) strong- ly bulging cells, and can also have weak longitu- dinal plicae. On the other hand, plants from ex- ceedingly moist places have more flaccid leaves and a high number of rhizoids, and are closer to the traditional concept of G. sessitana. This dis- tinction of these two taxa is untenable, as the cor- relation between the observed characters is not ab- solute. Plants with very long proximal paracostal cells and recurved margins, typical of the G. ses- sitana concept, can exhibit strongly bulging cells, and do not differ from other material in sporophytic features. The leaf margins of G. reflexidens are usually Annals of the Missouri Botanical Garden b Figure 16. Grimmia reflexidens. —a. Transverse sections of leaf with smooth cells. eaves. —b. Capsu —e. Proximal leaf cel h —— 50 рт c-g — 20 um le. —c. Transverse section of leaf with bulging cells. — Is. —f. Medial exothecial cells. —g. Proximal exothecial cells and stoma. —h. Peristome teeth. [a, b. e-h, Carestia s.n. (PAV); c, Brinkman s.n. (CANM-198100); d, Breidler s.n. (ВР-26598).] plane in the proximal half. However, some popu- lations have the margins recurved for part of their length, a feature that has been considered stable and of prime taxonomic value, causing many mis- identifications. Grimmia reflexidens was originally described as dioicous, because the perigonia and perichaetia usually occur at the ends of separate branches and so the autoicous condition can be extremely diffi- cult to demonstrate. The sporophyte is rather sten- otypic, and when fresh is easy to recognize with a 16-20X hand lens. The combination of straight se- tae and usually stramineous urns crowned with a distinct orange peristome is diagnostic and makes it unnecessary to dissect capsules to observe sto- mata. Populations with old capsules, already brown and damaged, are more problematic. In such cases, capsule color can be the same as in G. alpestris. The only sporophytic characters that vary to some degree are the orientation of the setae and the Volume 85, Number 3 1998 Mufioz 397 Revision of Grimmia Subgenus Orthogrimmia Figure 17. Distribution of Grimmia reflexidens. shape of the opercula. The setae can be straight or weakly inclined, even in the same population. The operculum can be conic and obtuse, or mammillate. Both characters vary randomly, and so are taxonom- ically uninformative. Grimmia reflexidens is a bipolar species wide- spread in the Northern Hemisphere, although un- common in temperate Asia. In eastern North Amer- ica it is rare, known only from a few localities in New Hampshire, New York, and Québec. It is the most common and widespread species of Grimmia in Antarctica, and it is relatively common in south- ern South America. Outside these two continents, it is rare in the Southern Hemisphere, known only from scattered localities in Australia, New Zealand, and Uganda. | Grimmia reflexidens (аз G. sessitana) has been recorded from China by Cao and Vitt (1986: 166, fig. 21). I was able to study the materials on which their reports are based. Whereas some of the spec- imens from Chang Bai mountain are G. reflexidens, other collections from the same locality (e.g., Cao Tong 199, ALTA) are G. longirostris. The Yunnan specimen (Zhu & Wu 64082, ALTA) is G. elongata. All authors except Deguchi (1978: 207) have mistakenly ascribed me authorship of Grimmia la- mellosa to Ms ill. Hal." (e.g., Greven, 1995; Wijk et al., 1962: 390; 1969: 655). In fact, Müller = scribed с. lamellosa. The reason for this mistake may be that in the original publication the genus Guembelia follows Grimmia, and Müller wrote out the full generic name only for the first species of each genus. For the remaining species he used the contractions “С.” and “Gr.” and this could have escaped the notice of later authors. UGANDA. Ru- Selected specimens examined (533). wenzori, Mijusi Valley, E slope of Mt. Speke, O. Hedberg 571 ( ANTARCTICA. Antarctic Penins Andrée Island, Charlotte Беу, Lewis-Smith pum 8 ү 7365), Mel- : Fysted Mod, Water Fanding, Siple nunatak Malenkiy, Konovalov s.n. (IBA-7381). Ross Sec- tor: Cape Sastrugi, Walton 227 (IBA-7382). South Shet- land Islands: King George Island, Admiralty Bay, Skua Cliff above Petrified Forest Creek, W of Arctowski Station, Ochyra 627/80 (NY). Victoria Land: Cape Adare, Terra a Expedition s.n. (IBA-7379). SOUTH GEORGIA. Grytocken, “Dammen,” 4 Apr. 1933, Jréim s.n. (BM). AUSTRALIA. New South Wales: Mt. Kosciusko, Mer- rits Camp, Maiden & Forsyth 203 (H-BR). NEW ZEA- LAND. North Island: p е Mt. Ruapehu, Feb. 1942, 4. s.n. (BM). S Island: Tasman Gla- e La Beche Hut, Cosi ын 756 (BM). AFGHANISTAN. Parwan: Panjskir v Oct. 1948, Кале s.n. (ALTA). ANDORRA. Ríu de la Coma del Forcat, Casas s.n. (BCB-21723). AUSTRIA. Carin- thia: in alpe “Fasihaunernik,” pr. Malta, 1880, Breidler rmark: auf Glimmerschi Schóder,” Breidler s s.n. . (BP- 26598). m i r the Dussel- dorferhutte, Townsend 92/547 (ВА. 5280). BELGIUM. Spa, promenade Annette et Lubin, Cornet s.n. (BP-36284). BULGARIA. Sofia: Rila-Geb., Musalla, Richard s.n. (GLM-2851). CHINA. Jilin: vicinity alon 953267 (MO). CZECH REPUBLIC. Tatra 15 іш valle Kistarpatak-vólgy, Маја Studená dolina, Boros 112930). FRANCE. Ari Culmann s.n. (Z). Haute 21 July 1926, Culmann s.n. (2). Isére: La Lauirtel, 17 Aug. 1894, Thériot s.n. (TRH). Pyrénées Centrales: au sommet du port d'Oo, 5 Sep. 1856, Zetterstedt s.n. (TRH). Haute-Savoie: Chamonix, Payot s.n. (TRH). GEORGIA. Imeretia, in alpe Chrshein, Brotherus 232 (H-SOL); distr. 398 Annals of the Missouri Botanical Garden Sukhumi, in vicinitate pagi Omarishara, in valle rivi Klich, 20 June 1986, Vasák s.n. (NY). GERMANY. Suy- thal, Sendtner s.n. (GLM-12812). INDIA. Himanchal Pradesh: Bara-Lacha-La, Lahul, Kangra, Punjab, Koelz 6774 (MO). ITALY. Trento: Pedergnone di Trento, Oct. 1919, Sbarbaro s.n. (BM). | К shima, Mt. jr inda с 59 (NICH- 116232 p. р). KAZAKSTAN. Alm Alma-Atynskoye Gorg and Lake, shore of La iie Alma P Zalilisliy Alatan, Allen 10636 (MO). MONGOLIA A. Chobd-aimak: Erdene- n es Ден M196 (MO). NORWAY. Opland: Dovre pr. Ко. Kaurin s.n. (BP-36229). ondre 2. s.n. (ТЕН). PAKISTAN. Baltistan, Rimochagma, s.n. (Н). POLAND. Dolinka - s.n. (BP-36671). supra lacu Zenoga, montes Retyezát, 3 Aug. 1969, Vajda s.n. (BP-74854). Maramures : in alp Pop-Jocac, ad Fo- chesány, Margittai s.n. (BP-112952). RUSSIA. Dagestan: Kasi-kumuk, Tschulty, July 1898, Alexander s.n. (H-BR). Gorno Altayskaya Autonomous Oblast: Altai Moun- “Wahlenberg See," Degen s.n. (BP-86539); Hóhe Tatra, Kleines Kohlbachthal auf der Lecwald, 10 Aug diaciones de là Laguna балы Rupidera s.n. (IBA-3985). Gerona: Coll dels Tres Pics, Lloret s.n. (BCB-25784). Huesca: Alta Ribagorça, pic de Salenques, Ballesteros s.n. (BCC-1276). Lérida: Vall de Bof, Cemaloforno, Bal- lesteros s.n. (BCC-1277). SWEDEN. Tornetrüsk-omrádet isko Nationalpark, Bergvegg. ovenfor Ridopakte, 11 Aug. 1945, Gjárevoll s.n. (ТАН). SWITZERLAND. Bern: 1. Trautmann s.n. (BP-36682). Graubün- den: 10 Sep. 1930, Amann s.n. (FH). Rhaetia: m: July 1887, Du. s.n. (TRH). Tessin: au lac de Muz prés de Lugano, 6 Sep. 1930, Amann s.n. (PC). TURKEY. Malatya: Taurus Cataonicus, In convallibus subalpinis prope vicum Bekikara inter urbem Malatja et vicum Kjachta, Handel-Mazzetti 2422 (FH). ADA. Br itish Columbia: ~ shoulder of Storm as Table-top Mt., (FH). U.S.A. Alaska: Kenai (A7) Quadrangle, Chisik Is- land, along shoreline about 0.5 mi. N of Cannery, Talbot 407 (NY). Arizona: Coconino Co., Point Sublime, on Fos- sil Shell Ridge, Young 20 (MO). California: San Bernar- dino Mts., South Fork of Santa Ana, Munz 6250 (FH). Colorado: Mesa Verde, Lutz 4371 (FH). Montana: Flathead Co., Glacier National Park, along Sperry Glacier Trail, % mi. E of S C ew mit of Gothics, Miller 12045 (IBA). Oregon: Mt. Hood, 1 Aug. 1871, Hall s.n. (FH). Washington: Mason Co., near the summit of Mt. Ellinor, 13 June 1940, Meyer s.n. (G). Wyoming: Park County, Beartooth Lake Meadow, 20 Aug. pon Conard s.n. oma ARGENTINA. Santa z: in alpinis pr. Rio Tarde, Halle i Sd CHILE ү locality given]. Póppig s.n. (BM, JE). La Araucanía: Lonquimay, Guenckel 1768 (PC). . Ottoshóhe, S Ufer des Lago Nahuel Huapi, Schiller 25 (PC). Santiago: San Gabriel, Río Maipo, Loos- er 1037 (IBA). 9. Grimmia ungeri Jur., in Unger & Kotschy, Ins. Cypern. 169. 1865. Grimmia alpestris subsp. ungeri (Jur.) Kindb., Bih. Kongl. Svenska Ve- tensk.-Akad. Handl. 7: 112. 1883. Grimmia alpestris var. ungeri (Jur.) Husn., Muscol. Gall. 129. 1887. TYPE: [Cyprus.] In Olimpo Cypri, Vere [spring] 1862, Unger s.n. (lectotype, here designated, BM; isolectotypes, BM [2 repli- cates], GOET [2 replicates], H-SOL, S). ir — Austin, Bull. Torrey Bot. Club 6: 45. 18 montana, var. Sulliv. & Leaqx. Куй: Ed. 2, pro parte" [Sullivant & Lesquereux, Musci americani ed. 2, | 215 (1865 [1866)) here designated, cates], С, H, Grimmia jamesii Ай б, Bull. Torrey Bot. 2 6: 43. 8 Grimmia montana var, boreali- . H; isolectotypes, ЕН |4 г « James, U.S. А. ] т 4. Watson 1412 іе here designated, isolectotypes, FH [2 replicates ]). Grimmia 4... Müll. Hal. & Kindb., іп Macoun & indb., Cat. Canad. Pl. 6: 70. 1892. TYPE: [Cana- | | British Columbia: Revelstoke, 7 890, Ma- oun s.n. (lectotype, here designated, S; ураннан САММ-198087). Grimmia montana var. idahensis Renauld & Cardot, Bot. Gaz. 30: 18. 1900. TYPE: [U.S.A.] Idaho: lac Pend d'Oreille, 1892, Leiberg s.n. (lectotype, here desig- nated, PC; isolectotypes, CANM-197540, FH [2 rep- licates], NY). Grimmia pseudomontana Cardot & Thér., Bot. Gaz. 30: 18, Pl. IV fig. 2. 1900. TYPE: tg pre near Moscow, 24 Mar. 1894, Henderson s.n. (lectotype, here designated, P А Grimmia canadensis Н. Winter [Karl Herman Winter, 1933; as canadensis Hedwigia 55: 102, fi 4. 1914, nom. illeg., non Kindb., "1897 Grimmia ovalis subsp. Sand (H. Winter) Podp., Musc. Eur. 279. 1954. TYPE: [Spain. Santa Cruz de Tenerife:| Tenerife Island, Cafiadas del Teide, 2000 m, Apr. 1912, Winter s.n. (lectotype, here Кына. JE; isolectotypes, JE [5 replicates ]). Illustration. Figure Autoicous. Plants in tufts, olive-green to black- ish. Stems to 1.5 cm tall, with central strand weakly developed; axillary hairs 5-8-celled, 105-175 дт long. Leaves erect and appressed when dry, patent and rigid when moist, 1-1.7 X 0.3-0.5 mm, ovate, acute, weakly keeled in the distal half, plane; mar- gins plane proximally and incurved distally; costa semi-terete, prominent on the dorsal surface, al- though only weakly so above, slightly to clearly de- limited; lamina 2-3(4)-stratose in the distal half, smooth; distal cells 5-7 um, isodiametric, plane or slightly bulging on dorsal surface, smooth; proximal Volume 85, Number 3 1998 Mufioz 3 Revision of Grimmia Subgenus Orthogrimmia Figure 1 Grimmia ungeri. —a. Ө: x И A |} 7 ж: У ap Pos CAR < Leaves. —b. Capsule. —c. Transverse sections of leaf. —d. Proximal leaf cells. 8. 28. —e. Medial exothecial cells. —f. Peristome teeth. [a-f, Unger s.n. (BM).] paracostal cells 20-35 X ca. 16 jum, isodiametric to rectangular (1-2: 1), with straight walls uniform- ly thickened or the transverse walls thicker than longitudinal walls; proximal marginal cells 10-25 O шп, isodiametric to rectangular (1-2: 1), with the transverse walls thicker than the longitu- dinal walls; hyaline hair-points terete, straight, to 0.7 mm long, smooth. Perichaetial leaves 1.5-2.5 X 0.5-0.9 mm, convolute and larger than vegeta- tive leaves (2-3X); hyaline hair-points terete, straight, to 1.5 mm long, nearly smooth. Androecia terminal. Setae erect and straight, ca. 2 mm long. Capsules exserted, ovoid or ellipsoid, symmetric, smooth, stramineous to brownish, lacking stomata; exothecial cells 17-65 X 13-45 ши, isodiametric and rectangular intermingled, thin-walled; annulus simple and persistent; peristome teeth 40-50 ¡um wide at the base, entire or moderately cribrose at the apex, orange-brownish, concolorous with the urn; opercula mammillate; calyptrae cucullate. Spores 9-13 um. Diagnostic characters. (1) Costa slightly prom- inent on the dorsal surface. (2) Laminar cells not bulging or only slightly. (3) Plants autoicous. (4) 400 Annals of the Missouri Botanical Garden Distribution of Grimmia ungeri. Figure 19. Capsules lacking stomata. (5) Opercula mammil- ate. Distribution (Fig. 19). | North America, Europe, Cyprus, and Canary Islands; coniferous forests and open areas between 600 and 3900 m elevation, on metamorphic and igneous rocks, (either basalts or granites). Mature sporophytes were present in 89% of the specimens studied. Grimmia ungeri is the only species in the genus so far known to combine ап autoicous sexual con- dition with capsules lacking stomata. Unfortunately, the perigonia usually arise on different branches than the perichaetia and are usually difficult to find. Grimmia ungeri has been included in the syn- onymy of G. alpestris almost since its original pub- lication. However, it is closer to G. reflexidens or G. montana and is only remotely related to G. alpestris. From G. montana it is distinguished by its auto- icous (rather than dioicous) sexual condition and its obtusely mammillate opercula. Grimmia reflexi- dens, also autoicous, has stramineous urns contrast- ing strongly in color with the orange peristome teeth, and the capsules have stomata at the base. The remaining species in section Montanae, G. al- pestris, G. caespiticia, and G. nivalis, differ in hav- ing bulging laminar cells, whereas members of sec- tion Donnianae are distinguished by their proximal laminar areolation. Grimmia ungeri exhibits a rather anomalous dis- tribution pattern. In North America, it is common along the Pacific Coast, but very rare on the Atlan- tic Coast, with only one locality known (in Québec). In Europe, it is known only from single localities in Scotland and Sardinia. From Africa Grimmia un- geri is also known from just one locality in the Ca- nary Islands, from where it was described as G. canadensis. lt also occurs in Cyprus, where it is common. Greven (1995) considered this species as endem- flexidens (as G. sessitana). However, the European collections are morphologically identical with Cyp- riot and North American plants. The protologue of G. brachyodon states *Grimmia montana, var. Sulliv. & Lesqx. Exsic. Ed. 2, n. 215 pro parte" as the type of the taxon. The label of the specimen n* 215 of Sullivant and Lesquereux's exsiccata reads: "California: montis Diablo (Bolan- der) etiam in alpibus Sierra Nevada montium (Brewer),” but in none of the studied sets is the material separated according to either the prove- nance or the collector. Selected specimens examined (135). CYPRUS. In Olympo (Troodos) Cypri, Unger s.n. (S). ITALY. Sardi- nien: Ostseite der Pta. La Marmara (Mt. Gennargentu), 2 June 1906, Herzog s.n. (JE). SPAIN. Santa Cruz de T. ~ rife, Я п. ). UNITED KINGDOM. Scotland: Aberdeenshire, Ballater, Mt. Curtin, July 1870, Fergusson s.n. (FH). NADA. British Columbia: head of Baker Creek, Jakmia River, Cascade Mts., 17 Oct. 1880, Watson s.n. FH). Québec: Gaspé Co., Mt. Albert, Collins 3961 (FH). MEXICO. Baja California: Las Cuevitas, Sierra Juárez, ca. 10 mi. 5 of Laguna Hanson, Wiggins 9177 (FH). Méx- ico: Mt. Popocatépetl, Kiener 18591 (FH). U.S.A. Cali- fornia: Luyo Co., Rock Creek, MacFadden 17386 (FH). Colorado: Tolland South Boulder Canyon, 18 July 1923, ` Id Clearwater-St. Joseph divide, Leiberg 1692 (S). Montana: Columbia Falls, 24 June 1895, Williams s.n. (S). Nevada: Carson City, 1868, Watson s.n. (FH). Oregon: Klamath Co., Odell Lake, 23 June 1931, Howell s.n. (FH) ington: Ellesburg, 23 July 1916, Bailey s.n. (FH). Wy- oming: Park Со., Shoshone Natl. Park, at “Three Mile Campground" area, off Hwy. 16, Churchill 5845 (G). PuBLISHED NoMINA МОРА Grimmia alpestris Schleich., Neues J. Bot. 1: 196. 1806[1805] (= G. alpestris). rimmia caespiticia f. epilosa Pilous, Musci če- choslovenici exsiccati n^ 483 (— G. caespiti- cia). Grimmia holmiensis Lindb. ex Hartm., Handb. Skand. Fl. Ed. 7: 374. 1858 (7 G. montana). Grimmia intermedia Fergusson, in Braithw., J. Bot. 10: 198. 1872 (= С. ungeri). Literature Cited Abramov, 1. 1. & A. L. Abramova. 1983. Konspect flori mkhov Mongol'skoj Narodnoj Respubliki. Nauka, Len- ingrad. Volume 85, Number 3 Mufioz 401 Revision of Grimmia Subgenus Orthogrimmia Afonina, O. M. 1986. Additamenta ad bryofloram Pen- insulae Czukotka. 4. Novosti Sist. Nizsh. Rast. 23: 222- 228. Allen, B. 1995. Eight neglected species of Grimmiaceae m x North America. Fragm. Florist. Geobot. 40: e. 23 1954. Hoyer's solution as a rapid p manent snare medium for bryophytes. Bryologist : 57. 242-244. & H. Crum. 1958 [1959]. Cytotaxonomic studies on mosses of the Canadian Rocky Mountains. Bull. Natl. Mus. 1. 160: 9. Вагапоуа, M. A. 1987. Historical development of the pre- sent classification of morphological types of stomates. Bot. Rev. pue aster) 53: 53-79. Bardunov, L. V. 1974. Listostebel'nye mkhi Altaya i Sa- n. Nauka, о Bartram, E. В. 1957. Mosses from the United States Ant- arctic Service Expedition, 1940-41. Bryologist 60: 43. Bednarek-Oc ћуга, H., J. Muñoz & R. Ochyra. 1992. The identity of 5. ругепаіса (Musci, Grimmiaceae). Fragm. Florist. Мау нен 37: 389-393. E es H. H. 1996. A revision of the Schistidium apocar- compe i in | Norway and Sweden. Bryophyt. Bib- lioth. 49: 1-333. Brodo, 1. M. s V. Alstrup. 1981. The lichen Bryoria sub- ағаға (Dahl) Brodo & D. Hawksw. іп Greenland and 2 Bryologist 84: 229-235 R kay P. & W. Pene 1845. Ca In: P. Bruch, W. Р. $c 'him W. T. von Gümbel, Bryologia europaea 3: 89-147. uh. 230-271. E. Schweizerbart, Эшан. Јао, ; D. H. Уш. 1986. А taxonomic revision and AA analysis of Grimmia and Schistidium (Bryopsida; Grimmiaceae) in China. J. Hattori Bot. Lab. 247. 123- Chałubiński, T. 1882. — tatrenses. Pamietn. Fi- zyogr. 2: 1-118, tab. Correns, C. 1899. 1. hungen die Vermehrung der Laubmoose. Gustav Fischer, Jen Crum, H. A. & L. E. Anderson. 1981. “Mosses of Eastern North America. Columbia Univ. Press, New York Deguchi, H. 1978 [1979]. A revision of the genera Grim- mia, Schistidium and Coscinodon (Musci) of Japan Sci. Hiroshima Univ., Ser. B, Div. 2, Bot. 16: 121-2 56. 1984. Studies on some Patagonian species of Grimmiaceae (Musci, Bryophyta). Pp. 17-72 in H. In- oue (editor), Studies on Cryptogams in Southern Chile. Kensei-sha, Tokyo 87. Studies on some Peruvian species of the Elbe (Musci, Bryophyta). Pp. 19-74 in H. In- oue (editor), Studies on Cryptogams in Southern Peru. Tokai Univ. Press, Tokyo. Dixon, Н. N. 1904. The Student's Handbook of British Mosses, 2nd ч. УТ. поно, Eastbourne, U K. Fritsch, R. 1991. Index to bryophyte 'ount Ee Biblio. 40: 1-3 152. Geissler, P. Maier. 1995. Lectotypifications of Cen- tral European Grimmia species (Musci, Grimmiaceae). Candollea 50: 495—514. Grant, V. 1981. Plant Speciation, 2nd ed. Columbia Univ. Press, New York. ;reuter, W., F. R. Barrie, H. M. Burdet, W. С. Chaloner, V. Demoulin, D. L. Hawksworth, P. M. Jørgensen, D. H. э Р. с. Silva, P. Trehane & J. McNeill (вон). =~ 994, International Code of Botanical Nomenclature (Tokyo Code). Regnum Veg. 131: [i]-xviii, 1 Greven, H Grimmia arenaria Hampe by the 994. Mawddach estuary in North Wales. J. Bryol. 18: 196- T ———, Grimmia Hedw. (Grimmiaceae, Musci) in mi. Back Publishers, Leiden Habeeb, 1950. Nomenclatural and other notes on mosses. "Rio: 52: 72- Hagen, 1. 1909. ak: эе til en Norsk Lévmosflora. IX Grimmiaceae. X ceae. XI Schistostegaceae. XII i Nao Vidensk. Selsk. Skr. 114 1978. Spore morphology of bryophytes ob- served by scanning electron microscope, IV. Kee" aceae. Bull. Natl. Sci. Mus., Tokyo, B 4: 33-42, pl. Ignatov, M. & O. M. Afdpina. 1992. Check-list of mosses of the former ps Arctoa 1: 4 994. Bryophytes of the Altai Moun- tains. IV. The deis Grimmiaceae. Arctoa 3: 67—122. Jóhannsson, B. 1993. Islenskir mosar. Skeggmosaett. Fjólrit Меке 24: 1-116. Kawai, І. 1965. Studies on the genus Grimmia, with ref- erence to E affinity of a dui. 5 ci. Rep. Kana- тама Univ., Biol. 10: 79-132. . 1968. Taxonomic studies on bs ciem in Mus- ci. Sci. Rep. pese a Univ., Biol. 27-157. Khanna, K. R. 1964. Cytology of some mosses from the — Mountains. Bryologist 67: 343-350. . Sex chromosome in Bryophytes. Nucleus (Cale P»: 14: 14-23. Kindberg, N. C. 1898. European and N. American Bryi- neae. 2. err Aktiebolag, L асай Кис, М. osses from an Antarctic oasis. Rev o EA n.s. . 36: 655—672. Kuta, E., L. Przywara € R. Ochyra. 1984. Chromosome studies gr Polish eee I. Bryol. 27 21. Lazarenko, 1. Vysotskaya & N. Lesnyak. 1971. а ui idis of en mosses "€ the SR]. Naukova poa I [cnn K. G. 1890. pou Deutschlands, . und der hares Abth. 1. E. Kummer, Leipzig. ie L. 1913. Die Laubmooose Europas I. Grimmi- aceae. Max Lande, Berlin-Schóneberg. 30. Monographie der MEM then Grimmi- aceen. 1. Biblioth. Bot. 101: [i]-ix. Maier, E. & P. Geissler. 1995. colat in o m Ein Bestimmungsschldsse. Herzogia 6. Bryophytes of the тогы Area, Mogensen, G. S. 1983. The Spore. /n: R. M. Schuster (editor, New Manual of Bryology 1: 325—342. Hattori Botanical Laboratory, Nichina Müller, C. 1887. Beitrüge zur Bryologie Nord-Amerika's. Flora 70: 219-225. ----- 1898. Bryologia provincia Schen-si sinensis ex collection giraldiana III. Nuovo Giorn. Bot. Ital. n.s. 5: 158- Muñoz, T 1 998. The correct name of Grimmia alpestris (Musci, roc wen ryologist 100: 517—519. Noguchi, A. 1 strated Moss Flora of Japan, Part 2. Hattori Bau Laboratory, Miyazaki-ken. Nyholm, E. 1956. ШУ Moss Flora of Fennoscan- dia. II. Musci. Fascicle 2. CWK Gleerup, Lun Ochyra, R. & A. J. Sharp. 1988. Results of a bryogeo- 402 Annals of the Missouri Botanical Garden graphical expedition to East Africa in 1968, IV. J. Hat- tori Bot. Lab. 65: 335-377. Ortiz, S. 1989. On the correct use of the term “holotype.” Taxon 38: 635—636. Pfeiffer, L. 1874. Nomenclator botanicus 1(2). T. Fischer, assel, Germany. Podpéra, J. 1954. Conspectus Muscorum Europaeorum. eskoslovenské Akademie Véd, Prague. Ramsay, Н. P. 1969. Cytological studies on some 2” from the British Isles. Bot. J. Linn. Soc. 62: 8 . Cytology of mosses. Pp. 1 2. іп R. M. Schuster (editor), New Manual of Bryology, Vol. Hattori Botanical eu Nichinan. Savicz-Lyubitskaya, L. I. & Z. N. Smirnova. 1969. Grim- mia doniana Sm rete ii Orientali. Novosti Sist. 2. соті bryologiae europaeae. E. Sc жн МА pubes нен then, С. E T md ies ipsis Frondo- m Supplementum 1(1). J. A. Barth, Le Smith. A. J. 78. Cytogenetics, biosystematies and evolution in te Bryopsida. Advances Bot. Res. 6 276. M. E. Newton. 1966. Chromosome studies on some British and Irish mosses. 1. Trans. Brit. Bryol. Soc. -130. Chromosome studies on s psp T Irish mosses. ІП. Trans. Brit. Bryol. Soc "ics Е E L. E. Anderson & V. S. Bryan. 1954. Chro- mosome studies on Californian mosses. Mem. Torrey Bot. ws 20: 1-75. Touw, A. & W. Rubers. 1989. De Nederlandse Blad- mossen. Stichting Uitgeverig Koninklijke Nederlandse Natuurhistorische Vereniging, Utrecht. 1949. family Grimmiaceae. Portugaliae Acta Biol, Sér. A, Volume R. B. Gold- schmidt: 47-78. Vitt, D. Н. 1984. Classification of the Bryopsida Pp. 696— 759 in R. M. Schuster (editor), New Manual - Bryology, Vol. 2. Hattori gr Laboratory, Nichina Vysotskaya, Е. l. N. Lesnyak. 1984. [C Шш study in кај mosses s of Khybiny (Kolsky Peninsula)]. ev) 69: 1399-1402. Wijk, R. van - W. D. Margadant & P. А. Florschütz. Index Muscorum (D-Hypno). Regnum Veg. 26: ———, . Index Muscorum (T- 7, Appendix). eem Yes si Gui, 1-922. Wyatt, В. & L. E. A cca dha 1984. M. а | bryophytes. Pp. 39—64 in A. F. Dyer & Ј. С. Ducket (editors), The 2.0 Biology of ык сы | ademic Press Zander, R. H. 1983 A тінің microscopic mounting me- dium for delicate beyophytes. Taxon 32: 618-620. INDEX TO SCIENTIFIC NAMES (accepted species in roman font) Campylopus pulvinatus var. alpestris .................. 384 Coscinodon eribrosus ...........................388 ИПИНЕ RP eee ea tee ade cn r 1 od. donnianus ..........................3718 etOSlS coo Ray aues pulvinatus var. alpestris ..................: 384 Grimmia alpestris . sidus Spr eX ERGO ETC EY e a ҰН alpestris f. hybrida vodka e ER s 395 alpestris f. sessitana ....................394 alpestris f. subsulcata ................... 394 alpestris subsp. subsulcata ................ 394 alpestris subsp. ungeri .................. 398 alpestris var. caespiticia ..................387 pestris var. eualpestris ..................: 384 alpestris var. holzingeri ..................: 384. stris var. hybrida ...................395 alpestris var. manniae ...................387 lpestris var. sessitana ...................: 394. alpestris var. stomata ...................394 alpestris var. iris paris 9 901 2... — е енто о ЗЭ estris var. ипдеп .................... 398 GhCeps oy eis aces a's Ой doy m nw Said ie 994 агепапа ............................376 ATIC: arcu ene к хы aco ERN brachyodon ..........................398 brachyphylla ......................... 39] НІНЕ wig cii exque e] caespiticia |. subimberbis .................: 387 caespiticia var. subimberbis ...............: 387 canadensis 2......................... 398 ME а йын аи айу А OME комны Beh a en Ge Bute Ge итал ал и ноо. BF OU donniana ........................... 378 donniana f. antarctica ..................395 donniana f. sudetica (Schkuhr) Loeske ....... 378 donniana ма arenaria ................376 donniana su eudonniana ..............: 3 78 donniana var. УИС: donniana var. arenaria .................. 376 donniana var. bohemica .................379 onniana var. breviseta ..................38 donniana var. curvula ................... 376 donniana var. 7. саланың WR 378 donniana var. holzin geri ................. 384 donniana var. таппіае .................. 387 donniana var. obtusa ...................378 donniana var. sudetica ..................384 inia ar. triformis ..................: 381 Donnianae E EEEE TEE E EAE ET 376 exann E oper E a A ТАЙ: E жш к-ка кале ЕЛДЕН ae ЭО funalis var. caespiticia ..................387 "Lori уза уз TR n беа ai жуки ec a OOD 2. лына ты АЕ ДЕНЕ Ұй жы ЗЕЯ holzingeri ........................... 384. IRCUPUR. unu = 64 н онен i BIO n var. subimberbis ................387 jam AA aeos eer КЕТТ EJE ОЗ ludin PPP 94. аха A luu spp каиа куле ени VERAT RE 389 таптае ...........................387 microtricha ..........................: 398 Мапа еа а OOO montana f. longifolia ...................: 390 a mutica ................... 390 montana var. абпоба ................... 390 тотапа уаг. уде ii етее о ВОВ montana var. frag наа Ears юз ЗВО montana var. és quif cp. 398 montana var. longifolia .................. 390 Volume 85, Number 3 Mufioz 403 Revision of Grimmia Subgenus Orthogrimmia montana var. truncata ..................398 subsúlcala: E usu diues tue exe ed nu EE x 394 Montanae (sect.) ......................383 sided банан КК: лына ЧИ у а eee ee 378 Divas oia we ле къа ЕИ E 392 ТАЙ Spreng. ех Schkuhr .............. 378 MMSE D pei bee BS UY RU RR dre ac ERE CA udo ха ҚҰТ OPE ка а 387 obtusa var. donniana ................... 378 ues var. subimberbis ................. 387 Orthogrimmia (subg.) ...................375 tenella 5e nnne 390 ovalis 7 canadensis ................. 398 а а а ЉЕТА Че Ado e eoa ecd . 395 ——— E" 392 triformis DE eT cp 381 кынан CLERI 3 ¿OE AAA REN н Ыы ARA A 398 тепаїс . ...887 Guembelia ee T ee eee ыы Н alpestris ы ех као оа a ИА reflexidens 22.2%: 22.5% эше enie ваја си жыз 394. кер ) саезрій бі. за СЕ ца» poe SORS PRS 387 HE ысы сала be meds ue RO CU ES 394 ana A! ООШ. MIRAS 378 sessitana f. longifolia ................... 895 ¡A Ж No Oe rA 394 sessitana f. subsulcata ...... ..........: 394 нана эы E ы МИНИ E O sessitana var. subsulcata .................. 394 поса ВИЗИИ T О ТИЗЕТ ЧИ 387 sinensianodon ........................ 389 шей sow akon КЗ ee ior e С ЕН 390 PA ООРОО Bio REVISIÓN SISTEMÁTICA Y ANÁLISIS CLADÍSTICO DEL GÉNERO CHAETIUM (POACEAE: PANICOIDEAE: PANICEAE) Osvaldo Morrone,' Fernando O. Zuloaga,' Mirta О. Arriaga,? Каш Pozner! y Sandra S. Alisciont? RESUMEN Chaetium (Poaceae: Panicoideae: Paniceae) comprende tres jos ies, distribuidas desde México y las Antillas hasta el norte de An asimismo un callo piloso en la base y muestran que Chaetium constituye un PH monofilético, he г base los caracteres morfoló fotografía de antecios superiores, fotomic с de transcortes de hoja y tallo forífero, mapas de esente revisión se incluye una descripción de descripciones e ilustraciones de las especies mérica del Sur. Este да se caracteriza por pose dins gui res s con artic mi ión п oblic ‘ua, las s que tienen S Ават, se acteres morfológicos y anatómicos. 14 resultados de los caracteres antes mencionados. En la s y anatómicos, una clave de las especies, distribución, 1COS ABSTRACT Chaetium (Poaceae: Panicoideae: iue 'eae) is a neotropical genus of three species, ranging from Mexico and the sing Antilles to northern South America. The i. considered in the present treatment; two anatomical cladistic analysis was carried out 1 us is mainly defined by having an oblique articulation of mie spikelet, a hairy callus at the base of the spikelet, ae a linear hilum on the cary characters er the leaves. горѕіѕ. 2. and anatomica patterns were found in the transverse sectio sing 29 morphological characters; the results have ti that Chain forms a 2. clade, based оп c ud d and leaf anatomical data. ecie an anatomical description of the genus and photomicrographs of the species, morphologic a dese 11 Шаан, an a distribution map аге presentec Chaetium Nees (Poaceae) es un género ameri- cano perteneciente a la tribu Paniceae, de la sub- familia Panicoideae. Comprende tres especies de distribución neotropical, las que se hallan desde éxico, América Central y Cuba hasta el norte América del Sur, en Colombia, Venezuela y noreste de Brasil. Si bien sus especies han sido tratadas parcialmente en floras regionales de Méxi 1983), las islas del Caribe ес 1909, 1936; Hitchcock & Chase, 1917) y Sudamérica (Doell, 1877; Smith & Wasshausen, 1981), no existe una revisión sistemática de con- junto de las especies del género, por lo que no han sido propuestas ^ im filogenéticas para las es- pecies de Chaetiu En el presente jm se estudian las es- pecies de Chaetium desde el punto de vista exo- morfológico, anatómico, se aportan nuevos datos y mediante un análisis cladístico, la monofilia y delimitación de Chaetium. se discute, MATERIALES Y MÉTODOS ESTUDIO ANATÓMICO Se realizaron cortes transversales a la altura del tercio medio del limbo foliar y empleándose ma- terial procedente de ejemplares de herbario, previ- amente tratado con etilenglicol durante 24-48 ras. Los cortes se realizaron a mano alzada y los preparados de epidermis se obtuvieron siguiendo el método de Metcalfe (1960). Para el estudio de los tallos se realizaron cortes en la zona próxima al último nudo de la caña florí- fera o en la zona media del entrenudo del estolón. Algunos cortes se observaron sin vaciar para de- terminar la posición relativa de los cloroplastos en ? División Anatomía Vegetal, Capital Federal 1405, Argentina .. de Botánica de Labardén 200, Casilla de Correo 22, San Isidro 1642, Argentin Museo Argentino de Ciencias Naturales “Bernardino Rivadavia.” pea Gallardo 470, ANN. Missouni Bor. GARD. 85: 404—424. 1998. Volume 85, Number 3 1998 Morrone et al. 405 Revisión de Chaetium la vaina Kranz. El resto fue vaciado y coloreado con Alcian Blue-safranina (Cutler, 1978). La presencia y ubicación del almidón de síntesis se detectó mediante la prueba de lugol (Johansen, Se aplicó microscopia de fluorescencia para di- ferenciar por autofluorescencia, los cloroplastos co- munes, ubicados en el clorénquima, y los cloro- plastos especializados, localizados en la vaina Kranz (Elkin & Park, 1975; O'Brien & McCully, 1981). Para ello, algunos materiales cortados, sin vaciar, se montaron en agua destilada y se obser- varon con un microscopio de fluorescencia con ex- citación por luz reflejada IVFI (Epifluorescencia) Zeiss Standard, equipado con filtro de excitación azul BP 450—490; divisor cromático FT 510 y filtro supresor LP 520; la fuente de iluminación consistió en una lámpara de alta presión de mercurio HBO 50W perteneciente a la División Anatomía Vegetal del Museo Argentino de Ciencias Naturales (MACN), Buenos Aires, Argentina. Las fotomicrografías, que se obtuvieron con un equipo Zeiss, fueron realizadas en materiales va- ciados. ANÁLISIS DE ANTECIO SUPERIOR Para el estudio de la epidermis abaxial de la lemma y pálea se empleó un microscopio de barri- do Zeiss 940A, perteneciente al Instituto de Botán- ica Darwinion (IBODA), Buenos Aires, Argentina. ESTUDIO EXOMORFOLÓGICO Este estudio fue realizado sobre materiales per- tenecientes a los herbarios BAA, G, LIL, MEXU, MO, SI, UC y US. Con un asterisco (*) se señalan los ejemplares empleados en el estudio histofoliar y con dos as- teriscos (**) aquellos especímenes utilizados en el análisis de la lemma y pálea superior: Black 55- 18634***. Davidse & D'Arcy 10412***, Díaz Luna 5010*, Eiten & Eiten 4713*, Ekman 7816*-**, 15699%, 16827**, 16474*, Hinton 2418*, Hitchcock s.n.*, 6856**, 6985**, Пиз & Nee 1048*, Пиз et al. 588*, Lyonnet 2447*, Moraes 2217***, Pickel 1714*, Pittier 14926*, Pringle 11736*, Reeder & Reeder 2334*, Swallen 4635**, 4797**, Weatherwax 158*, Weston 2140*, 4145*. ANÁLISIS CLADÍSTICO El análisis de las relaciones filogenéticas de las especies de Chaetium se realizó siguiendo los prin- cipios de la sistemática filogenética propuestos por Hennig (1966). Para el análisis filogenético se em- plearon 29 caracteres morfológicos. Cinco de éstos (1, 2, 3, 24, 29) son caracteres multiestados y fue- ron tratados como no aditivos (no ordenados). La lista de los caracteres y estados usados para el aná- lisis se muestra en la Tabla 1. La Tabla 2 contiene la matriz de datos (especies por caracteres). El análisis cladístico fue llevado a cabo utilizan- do la opción de enumeración implícita (ie*) del Hennig86, versión 1.5 (Farris, 1988). Los árboles fueron re-enraizados al grupo externo más cercano Nixon & Carpenter, 1993). Para la construcción del árbol de consenso estricto se usó la opción Nel- sen del citado programa. Los caracteres fueron ana- lizados sobre la base del estudio morfológico y ana- tómico de materiales citados en el apéndice 1 y complementados con datos bibliográficos. Para el análisis de la distribución de los carac- teres en los cladogramas y la generación de los mismos se empleó CLADOS (Nixon, 1993). „=“, SELECCIÓN DE LOS GRUPOS EXTERNOS Diversos autores relacionaron a Chaetium con varios géneros de Paniceae, por lo que las relacio- nes de este género dentro de la tribu son poco cla- ras. Como no existe hasta el momento un estudio filogenético de las Paniceae se siguió el criterio sustentado por Nixon y Carpenter (1993) para la elección de los posibles grupos externos. A tal efec- to, se consideraron las referencias bibliográficas con el siguiente criterio: cuando un autor citó uno o más géneros relacionados, se tomó la especie tipo de cada uno de ellos; cuando la publicación men- ciona una especie en particular, se incluyeron tanto dicha especie como la especie tipo del género co- rrespondiente. Los grupos externos utilizados en este análisis son: Echinochloa (Kunth, 1833; Steudel, 1853-1855; Fournier, 1886; Hitchcock & Chase, 1917; Clayton & Renvoize, 1986; Beetle, 1987): den- tro de este género se analizaron Echinochloa crus-galli (L.) Р. Beauv., E. holciformis (Kunth) Chase y E. oplismenoides (E. Fourn.) Hitchc. Louisiella (Clayton & Renvoize, 1986): L. fluitans C. E. Hubb. & J. Léonard Melinis (Presl, 1830): M. minutiflora P. Beauv. Panicum (Zuloaga & Soderstrom, 1985): P. milia- ceum L. y P. aristellum Doll Oplismenopsis (Clayton & Renvoize, 1986): O. na- jada (Hack. & Arechav.) Parodi Oplismenus (Nees, 1829; Presl, 1830; Kunth, 1833; Steudel, 1853-1855): О. hirtellus (L.) P. Beauv. Oryzidium (Hubbard & Schweickerdt, 1936; Clay- 406 Annals of the Missouri Botanical Garden Tabla 1. Lista de los caracteres y estados de caracteres usados para el análisis cladístico de Chaetium. l. Orden vx O elemental: 0 = de 3er. orden; 1 de orden; de ler. orden. o abajo] 2, Número de la por nudo: 0 = = dos; 2 ariable. ов caracteres 1 y 2 corresponden a ан terísticas de la inflorescencia. La inflorescencia de las Poaceae es variable y compleja. Para establecer las . se tuvo en cuenta el idein máximo de paracladios presentes en la base de la in - оге ла. Я i-em de las espiguillas: 0 = unilateral no unilateral; 1 = 4. Desarticulación de la espiguilla: O = transversal; 1 = oblicua El punto de desarticulación es uno de los caracteres más es perpendicular al eje mayor del pedicelo y oblicua cuando es diagonal. 5. Compresión de la espigulla: О = dorsiventral; 1 = late- s un carácter válido para 2. géneros dentro terminación de la bido a que las espiguillas — io Hite defor- maciones de su estructura en el material herbor 6. Pelos de la base de la. Е 0 = adn ausentes; l= pe 08 ni es. Se consideró la presencia o ausencia de un denso mechón de pi (macropelos unicelulares) en = base de la espi- guilla junto a la articulación de la mis 7. Callo: 0 = ausente; 1 = pres El término callo es aplicado a 4. ier modificación del disemínulo presente en el punto de desarticulación (Web- ster, 1988). En este estudio se definió callo al estípite for- mado por la expansión en Pid dh de la base de la gluma ~ inferior y el nudo donde esta gluma se inserta. 8. Base de la gluma 4, 0 - foliosa; 1 - reducida. 9. Arista de la gluma inferior: 0 = ausente; 1 = presente. 10. Nümero de venas de la gluma inferior: 0 — 5—7-nervia; 1 3-nervia = hendido. 11. Apice de la gluma superior: О = entero; 1 12. Arista de la gluma superior: 0 = ausente; 1 = presente. 13. Flor inferior: О = presente; 1 = ausente. 14. Consistencia de la lemma inferior: 0 = herbácea; 1 — oriácea 15. Apice de la lemma inferior: О = entero; 1 = hendido. 16. Arista de la lemma inferior: О = ausente; 1 = presente. 17. Pálea inferior: O = Los caracteres 8, 9, 11, 12, 15, 16 PAR diversas es- presente; 1 = ausente. por Troll (1967: 1248-1 1290), donde la porción inferior de las glumas y lemmas es homóloga de la vaina foliar y la arista de las mismas, a la lámina foliar. Se consideró como una arista gula. Con estas pautas se Стану que ane eee son siempre pesci 18. Compresión del antecio superior: O = dorsiventral; 1 = lateral. 19. Apice del antecio superior: 0 — pálea, y abierto cuando este 1000 se halla libre. Ambos 4. ве observaron cuando e 20. Consistencia del antecio superior: О = crustáceo; 1 = cartilaginoso. Como la consistencia varía a lo largo del desarrollo de la espiguillas, se tuvo en cuenta que el antecio superior se hallara maduro y encerrara una cariopsis desarrollada. 21. Bordes de la lemma superior: 0 = curvos; 1 = planos. s márgenes de la lemma superior tienen gran valor diag- nóstico a nivel се en las Poaceae (Webster, 1988). En este estudio los márgenes de la lemma pueden ser involutos (curvados hacia la cara prr al de la Ted o planos cuan- do los | están aplicados sobre la cara abaxial de la pálea superi 22. Apice P if it - 0 = sin diferenciación mor- heal 2 líneas verdes. Los caracteres ie. EN 20, а 22 implican diferentes еѕ- шош del antecio superior. Las características morfoló- icas anteci 1 ili portantes a las Paniceae para establecer relacio ster, 1988; Zuloaga s Soderstrom, 23. ш del hilo: rci 1 о; 2 = linear w del largo. La Pa rma del hilo en la tribu es ampliamente variable, ve nes (Chase, 1911; Web- 85). = linear 1/1 del ),5 endulsky el al. (1987) y 991) « а 1. En este estudio se cons cuando no alcanza el 1/4 de la longitud de la cariopsi considera linear 1/1 cuando ocupa toda la longitud FH la i dd y cuando el hilo presenta una forma intermedia, ocupando el 50% de la longitud de la cariopsis, se trató como ur 1/2 24. Papilas ке 0 = ausentes; 1 = presentes. 25. Papilas abaxiales: O = ausentes; 1 = presentes 26. Células parenquimaticas asociadas a las buliformes: O = ausente; 1 = presente. 27. Vaina cS О = sin cloroplastos; 1 = con clo- roplast 28. Vaina o О = con cloroplastos; 1 = sin cloroplastos 29. Vaina parenquimática: О = presente; 1 = ausente. s caracteres 27, 28 y 29 están P aspice con el tipo fotosintético y el Síndrome de Kranz. acteres fueron utilizados para delimitar los géneros i ені :eae por Brown 1985), quienes consideraro 4. MS es homó lantas non-Kranz, y la vaina Kranz del tipo PS es lodi. a la vaina parenquimática de las gramíneas C,. Volume 85, Number 3 Morrone e 407 Revisión ей ЕНЕ “ж” se indican los caracteres polimór- Tabla 2. Matriz de datos usada para el análisis cladístico de Chaetium, con ficos, con “—” los estados de caracteres no codificables. 1 2 12345678901234567890123456789 Panicum milia Oryzidium аа 10000001010100000011100000000 Melinis minutiflora 11001001011110111111100001000 Lousiella it 10000001010010001011101110000 ismenus Oplismenopsis najada Echinochloa crus-galli Echinochloa holciformis Echinochloa oplismenoides aetium bromoides Chaetium festucoides Chaetium cubanum 12001000110100000001100000010 211010001*01*0001001100110010 20100000110100010011101000010 211000000*01000100100101101-1 221000000101100100100101101-1 221000000001100100100101101-1 21010110100111011011102000000 210101111101100110111020001-1 210101101101100010111020001-1 n & Renvoize, 1986): O. barnardii C. E. Hubb, & Schweick. HISTORIA TAXONOMICA DEL GENERO Nees von Esenbeck en el afio 1829 fundó el gé- nero Chaetium sobre la base de una especie, C. festucoides, y relacionó al mismo con Oplismenus. Nees von Esenbeck distinguió a Chaetium princi- palmente por tener inflorescencias espiciformes, pedicelos con articulación oblicua, gluma inferior setiforme, m superior rostrado-setosa y antecio superior cartác 4. Presl (1830) estableció el género Berchtoldia para México, describiendo e ilustrando una ünica especie, B. bromoides. Este autor rela- cionó a Berchtoldia con Panicum, Melinis y Oplis- menus. Kunth (1833) transfirió a Chaetium festucoides al género Oplismenus, ubicándolo dentro de la sección Echinochloa, junto con especies actualmente con- sideradas en los géneros Echinochloa y Anthae- nanthia por tener “glumae inaequales, flos sterilis aristatus, spicae paniculatim, rarius racemosim dis- positae." En la misma obra, Kunth trató a Berch- toldia como un género válido, citando al final de la descripción del género “Charact. gen. ex Presl.” Por ello se deduce que posiblemente Kunth no hu- biera visto la especie descripta por Presl, recono- ciendo a Berchtoldia como un género independien- te de Oplismenus sobre la base de la descripción e yo de Pre 1 1853, Steudel consideró a Chaetium festu- 54% como sinónimo de Panicum chaetium Steud. Este autor incluyó a P. chaetium en la sección Echi- nochloa, con especies actualmente consideradas en Echinochloa, Oplismenus y Panicum. En la misma obra Steudel consideró a Berchtoldia con rango ge- nérico basándose en la obra de Presl. Doell (1877) trató a Chaetium como una sección de Panicum, incluyendo una unica especie, P chaetium. Bentham (1881) reconoció a Chaetium como un género distinto de Oplismenus, sobre la base de di- ferencias morfológicas de la espiguilla e inflores- cencia. En 1883 Bentham considera por primera vez a Berchtoldia sinónimo de Chaetium. Posteriormente, Fournier (1886) trató a Berch- toldia (erróneamente citado “Berchtholdia”) como un género válido para México, caracterizando al mismo por tener “gluma inferiore remota," posible- mente haciendo referencia al entrenudo desarrolla- do de la raquilla entre la gluma inferior y superior. ournier reconoció tres especies en - В. bromoides, В. holciformis (Kunth) E. Fourn. y В. oplismenoides E. Fourn., siendo estas dos pt consideradas actualmente en Echinochloa. Hackel (1887) reconoció a ү, e incluyó en la sinonimia del género a Berchto Chase (1911), en su sinopsis de ie US de Paniceae de América, reconoció a Chaetium y lo distinguió de Echinochloa por tener espiguillas lan- ceoladas, con un largo callo en la base y glumas largamente aristadas. Hitchcock y Chase (1917) relacionaron a Chae- tium con Echinochloa por tener espiguillas esca- brosas. Hsu (1965) trató a Chaetium como un género válido por el callo barbado presente en la base de la espiguilla y el patrón epidérmico del antecio su- perior con costillas longitudinales y depresiones echinadas. 408 Annals of the Missouri Botanical Garden Beetle (1977) citó C. bromoides para México y lo ubicó, sin aclarar las razones, dentro de la tribu Andropogoneae. Posteriormente, en 1987, este au- tor trató C. bromoides dentro de la tribu Paniceae y lo relacionó con Echinochloa, separando a este último género por tener espiguillas globosas, dor- siventralmente comprimidas, con la base obtusa y glumas desprovistas de largas aristas. A la vez se- ñaló que C. bromoides tiene espiguillas angostas, lateralmente comprimidas, con callo basal agudo y ambas glumas con largas aristas. Clayton y Renvoize (1986) relacionaron a Chae- tium bromoides con Echinochloa, distinguiendo a este ültimo género por tener la gluma inferior corta y aristas rígidas Webster y Valdés Reyna (1988) reconocieron a Chaetium por tener el ápice de los pedicelos obli- cuo y la base de la espiguilla prolongada en un callo oblicuo y piloso. Estos autores indicaron que las evidencias que sostienen la relación de Chae- tium con Echinochloa son dudosas y su relación con otros géneros dentro de la tribu es poco clara. RESULTADOS CARACTERES MORFOLÓGICOS Y TAXONÓMICOS Forma biológica. Chaetium incluye especies pe- rennes, con rizomas de entrenudos cortos. ae- tium bromoides posee cañas erguidas, simples a ra- mificadas hacia la base; se pueden hallar en esta especie cañas estoloníferas, de entrenudos cortos y arraigados, los que originan nuevas plantas. Chae- tium cubanum tiene cañas simples a profusamente ramificadas, que pueden doblarse hacia el suelo, arraigar y propagar la mata a cierta distancia. Chaetium festudoides posee cañas simples, erguidas a geniculadas. Las cañas en las especies del género tienen en- trenudos comprimidos y macizos, con médula es- ponjosa. Las lígulas son membranáceo-pestañosas en C. cubanum y C. festucoides y pestañosas en C. bromoides, y las láminas son lineares, planas a ple- gadas o involutas, con pilosidad variable, desde glabra a híspidas en ambas caras. Inflorescencias. Las inflorescencias son contrai- das, espiciformes a subespiciformes, parcialmente incluidas en las vainas foliares en C. cubanum a exsertas en el resto de las especies. Espiguillas. | Chaetium presenta el plan básico de espiguilla de las Paniceae, la que se caracteriza por ser biflora, con dos glumas y articulación en la base de la espiguilla, junto a la inserción con el pedi- celo. Las espiguillas son largamente ovoides, com- primidas dorsiventralmente, escabrosas, en pares, sobre pedicelos desiguales. La base de la espiguilla posee un callo como resultado del desarrollo de la raquilla entre las glumas y el antecio superior. El callo es agudo, híspido y con su extremo oblicuo. La espiguilla posee en el nudo basal una gluma erior, con un cuerpo principal que varía de lan- ceolado a ovado, terminando en ambos casos en una arista; está reducida a una estructura setácea en C. festucoides. La gluma superior es herbácea y aristada. La lemma inferior es herbácea en C. cu- banum y C. festucoides y coriácea en C. bromoides, siendo corta a largamente aristada en C. bromoides y C. festucoides, respectivamente, a acuminada en C. cubanum. La pálea y flor inferior se hallan au- sentes. La cariopsis es largamente elipsoide, dorsiven- tralmente comprimida, con hilo linear, la mitad del largo de la cariopsis. Textura y ornamentación del antecio superior (Fig. ) Elantecio superior es largamente ovoide, dor- siventralmente comprimido, cartilaginoso, papiloso, pajizo y escabroso hacia el ápice; la lemma posee los márgenes membranáceos y planos, dejando li- bre el ápice de la pálea. La epidermis abaxial de la lemma y pálea posee células largas rectangula- res, agrupadas en hileras longitudinales, tres veces más largas que anchas en Chaetium festucoides y C. cubanum y más de tres veces más largas que anchas en C. bromoides. Las células largas poseen las paredes anticlinales longitudinales onduladas. La epidermis abaxial de la pálea y lemma posee aguijones hacia el ápice y micropelos bicelulares, fusiformes, en toda su superficie, con mayor ná- mero hacia el ápice de la lemma y pálea. Se ob- servaron papilas simples en la lemma y pálea de Chaetium cubanum y C. festucoides, distribuidas re- gularmente, una por célula larga, en toda la super- Figura 1. A-F. F Porc ins ves del antecio visto desde la pálea. — oides. La flecha indica el А pice del antecio visto desde la rn —F. ише de E, con a de inserción oblicu — otomicrografías МЕВ de antecios superiores de especies de Chaetium. A, B. C. bromoides. —A. ;. festucoides. e la 2. с de la lemma. E, F. C. aguij . Base de la espiguilla —H. Apice del ората de c 2. (А, В, С, H, Davidse & D'Arcy 10412; C, D, Swallen 4635; E, F, Ekman 7816.) Volume 85, Number 3 Morrone et al. 1998 Revisión de Chaetium 410 Annals of the Missouri Botanical Garden ficie. La pared tangencial externa es marcadamente convexa en C. bromoides, con costillas longitudi- nales y surcos asociados a las paredes anticlinales longitudinales; en esta especie, las papilas son compuestas, asociadas a la pared transversal distal de las células largas, y se hallan formadas por una papila simple con 2-6 proyecciones equinadas. ambién posee papilas simples distribuidas a lo largo de los surcos y costillas. ANATOMIA FOLIAR Caracteres histofoliares en corte transversal (Figs. 2, . Transcorte: variable de plano а conduplicado o convoluto; láminas simétricas a ambos lados de la costilla central, con costillas bien marcadas en ambas caras y ápice redondeado en C. bromoides o con ambas caras sublisas, con costillas ligeramente insinuadas en las restantes especies. Todas las cos- tillas se hallan limitadas por grupos de células buli- formes. Costilla central: de estructura variable, conspicua, con un haz vascular primario y dos ha- ces vasculares terciarios, estructuralmente distin- guibles de los restantes haces vasculares, asociado o no a células parenquimáticas incoloras hacia la cara adaxial o inconspicua, con un haz vascular primario solitario sin células parenquimáticas in- coloras. Distribución de los haces vasculares: haces vasculares primarios y secundarios equidistantes de ambas epidermis y asociados a las costillas; ha- ces vasculares terciarios dispuestos por debajo de las células buliformes y próximos a la cara abaxial. Estructura de los haces vasculares: haces vasculares primarios de contorno subcircular a elíptico, con vasos de metaxilema de contorno angular y diá- metro igual o ligeramente mayor que las células ranz; haces vasculares secundarios de contorno angular, con xilema y floema distinguible; haces vasculares terciarios de contorno angular, con xile- ma y floema pobremente desarrollado. Vaina de los haces vasculares: compuesta por una ünica vaina mestomática con cloroplastos de posición centrífu- ga, en C. cubanum y C. festucoides, o compuesta por dos vainas en C. bromoides, la externa paren- quimática con cloroplastos de posición centrífuga, la interna mestomática, compuesta por células de paredes uniformemente engrosadas; vaina perifloe- mática presente. Células distintivas Kranz: ausen- tes. Esclerénquima: pobremente desarrollado, en grupos discontinuos, subepidérmicos, adaxiales y abaxiales. Mesofilo: clorénquima formado por célu- las raquimorfas, imperfectamente radiado alrededor de los haces vasculares en C. cubanum y C. festu- coides, y radiado en C. bromoides. Células epidérmi- cas: células buliformes presentes en los surcos ada- formando grupos células epidérmicas pequeñas y regulares en forma, de pared tangencial externa muy engrosada. Macro- pelos: unicelulares, de base bulbosa, asociada a cé- lulas epidérmicas isodiamétricas, de paredes más engrosadas. xiales, pequeños; Epidermis en vista paradermal. Células largas, en ambas epidermis, 3 a 5 veces más largas que an- chas, de paredes sinuosas, interrumpidas por célu- las cortas alargadas en sentido transversal, a veces silicificadas, en ocasiones cuadrangulares en C. fes- tucoides. Células buliformes: isodiamétricas a róm- bicas. Aparatos estomáticos: rómbicos a subrómbi- cos, distribuidos en dos hileras en la zona intercostal, flanqueados por una hilera de ganchos; células largas interestomáticas de extremos exca- vados. Micropelos. _ bicelulares, tipo *panicoide." Macro- pelos: unicelulares, distribuidos en las zonas inter- costales, de paredes engrosadas y asociados a célu- epidérmicas sobreelevadas. Cuerpos de sílice costales: halteriformes de eje corto, ocasionalmente cruciformes en C. festucoides; células suberosas costales de contorno sinuoso, de igual o mayor lon- gitud que las silíceas; pares sílico-suberosos oca- sionales en C. cubanum — £ Autofluorescencia. Los cloroplastos de la vaina Kranz de Chaetium bromoides autofluorescen como los típicos cloroplastos de las especies PS, mientras que los cloroplastos de la vaina Kranz de C. cu- banum y C. festucoides autofluorescen como los tí- picos de las especies MS (Elkin & Park, 1975). Corte transversal de tallo florifero (Figs. 2G, Н, 3C, D). édula formada por grandes células paren- quimáticas sin contenido, con grandes espacios in- tercelulares, limitada externamente por un anillo de esclerénquima constituido por 3-4 hileras de fi- bras. Zona medular con haces vasculares primarios, distribuidos en 2—4 cíclos, con vaina perifloemáti- ca, rodeados por una vaina esclerosada. Haces vasculares periféricos primarios y terciarios, dis- puestos en dos ciclos. Haces vasculares primarios trabados, los terciarios libres. En C. cubanum y C. festucoides los haces vasculares periféricos se en- cuentran rodeados por una sola vaina mestomática Kranz, con cloroplastos especializados de posición centrífuga y depósitos de almidón de síntesis; en C. bromoides los haces vasculares periféricos po- seen una vaina interna esclerenquimática y externa parenquimática Kranz, la que se extiende entre los haces vasculares contiguos, con cloroplastos espe- cializados de posición centrífuga y depósitos de al- midón de síntesis. Volume 85, Number 3 Morrone 411 998 Revisión pa е esit ura 2. А, B. Cha um cubanum. — А. Transcorte foliar а la altura de la costilla central. —B. Transcorte а de una porción de la lámina. С-Е. C. festugoides, —C. Transcorte foliar a la altura de la — central. —D. Transco foliar de una porción је la lámina. —E. Epidermis abaxial en vista topográfica. —F. Detalle de macropelo unic ar con base en cojinete. G, H. Chaetium cubanum. —G. Haces vasculares periféricos en transcorte de tallo florífero. —H. Detalle паша де la fotomic кз а anterior. Las flechas indican restos de una vaina externa, en hoja y tallo. (А, B, G, H, Ekman 15699; C, Pittier 14926; 0-Е, Moraes 2217) La única diferencia observada entre la anatomía ción en el tamaño y calibre de las fibras de dicho del tallo florífero y del estolón es un aumento en la — anillo que se observa en el transcorte del estolón. cantidad de ciclos en que se presentan los haces vas- culares internos, del námero de capas de fibras que En la Tabla 3 se resumen los caracteres anatómi- componen el anillo esclerenquimático y una disminu- — cos distintivos de las tres especies de Chaetium. 412 Annals of the Missouri Botanical Garden tre em" 3 " » A 9; ЕЯ г $e ^ AME > “229, Figura 3. central. —B. Tran iscorte foliar de una apnd de la lám Detalle aumentado de la fotomicrografía anterior. —E. F vainas parenquimáticas en los 5010; С-Е, Hitchcock s.n. — Los caracteres anatómicos observados en Chaetium bromoides concuerdan con lo previamente sefialado por Montiel (1972), siendo esta una especie C,, con vaina Kranz de tipo PS, tanto en hoja como en tallo; en dicha vaina se localizan los cloroplastos espe- cializados, los que autofluorescen de la misma for- laces vasc ie беп д Че 2. de tallo (А, Лиз et a 588; B. , B. Chaetium bromoides. —A. Transe orte digas a la altura de la costilla central; se señala la нар a. —C. Zona 5. а de transcorte a tallo florífero. —D. 1. abaxial en vista topográfica. Las flechas indici an las 3, Díaz Luna ma que los de otras especies PS. A su vez, del estudio anatómico de Chaetium cubanum y Chae- tium festucoides se desprende que son especies C,, con vaina Kranz de tipo MS, tanto en hoja como en tallo, con cloroplastos de posición centrífuga, los que autofluorescen de igual forma que los de otras Volume 85, Number 3 Morrone et al. 413 Revisión z Chaetium Tabla 3. Comparación de los caracteres anatómicos de las especies de Chaetium. C. bromoides C. festucoides C. cubanum Aspecto del transcorte moniliforme subliso subliso Forma del transcorte plano a conduplicado conduplicado convoluto Ancho del transcorte muy ancho ancho angosto Costilla central inconspicua, con un haz conspicua, con 3 ha- conspicua con 3 haces vasc. 1° ces vasc. (1° y 39) гаѕс. (12 y 3? Células parenquimáticas inco- ausentes presentes ausentes loras en la costilla central Tipo de haces vasculares py 2 16 09767 1,27 y y Vainas completas alrededor de dos una una os haces vasculares Extensiones de la vaina presentes adaxialmente ausente ausente Vaina d incompleta completa completa Tipo anatóm 5-РСК 5 ) E o p los cloroplas- como en especies PS- como en especies MS como en especies MS tos NAD-me Vinculación de los haces vas- 1° y 3? vinculados a las 1? vinculados a las 1° vinculados a las culares costillas costillas y 3? a cél. costillas y 3? a las buliformes cél. buliformes especies MS, de tipo bioquímico NADP-me. La longitud de 58 pasos, un índice de consistencia de presencia de dos vainas alrededor de los haces vas- 0.55 y un índice de retención de 0.65. Chaetium culares, en Chaetium bromoides, la externa con clo- ве comportó como un taxón monofilético en todos roplastos centrífugos, tradicionalmente ha sido aso- — los casos (Fig. 4A), estando sustentado por cuatro T a al tipo fisiológico PEP-ck (Gutiérrez et al., ^ sinapomorfías: desarticulación oblicua de la espi- 976). Sin embargo, dicha correlación debe tomar- — guilla (4), pelos presentes en la base de la espi- se con precaución, dadas las excepciones halladas рца (6), callo presente (7) e hilo linear % (23). hasta el presente dentro de las Paniceae (Ohsugi En ambos árboles originales Chaetium presenta la & Murata, 1980; Ohsugi et p^ 1982; Oguro et al., misma topología interna; el clado Chaetium festu- 1985; Prendergast & Stone, 1987). coides y C. cubanum forma un grupo monofilético En Chaetium cubanum y C. festucoides puede ob- рог Ја presencia de vaina mestomática con cloro- servarse, por fuera de la vaina Kranz, unas pocas plastos (27), que corresponde al subtipo fotosinté- células globosas de paredes algo engrosadas, vacías |, C, MS y por la ausencia de vaina parenqui- de cloroplastos; estas células representan restos de mática (29). 2. festucoides se distingue рог una segunda vaina, confirmando lo observado pre- la base reducida de la gluma inferior (8) y C. cu- viamente por Renvoize (1987). La presencia de res- b lal a ЕКА ip Ch tos de una vaina externa en especies MS fue an- re ea пон B ed tium bromoides está definido por la nerviación de teriormente citada para especies de Panicum de las Ld inferior 5-7-nervia (10) v la 1 infe- sects. Agrostoidea Hitchcock & Chase, Discrepantia ed caia кзн " A айнаны с R2 4. e gs " бо. à eie El nümero básico de cromosomas fue excluido nero Anthaenantiopsis (Morrone et al., n Че! as debido a que de las tres especies del especies de Aristida (Brown, 1977), ale género sólo se conocen recuentos cromosómicos Neurachne y Paraneurachne (Hattersley, pid pres А ЕЕ 57 Ж _ E encuentra una vaina externa a la vaina Kra z MS, , ; feeder, 510 avıase, , bien desarrollada, vacía o con unos pocos soins Montiel, 1972). El número básico x = 13 es poco plastos, de naturaleza parenquimática. Brown frecuente en las Poaceae, especialmente en las Pa- (1977) y Dengler et al. (1985) definen este último nicoideae, por lo que deberían realizarse estudios carácter como intermedio entre una vaina non- cromosómicos en G cubanum y C. festucoides para Kranz y una vaina Kranz MS. confirmar este carácter. De confirmarse este número básico de x = 13, el mismo agregaría un carácter adicional para la monofilia de este género. El análisis cladístico arrojó dos árboles igual- Brown (1977), sobre la base del estudio anatómi- mente parsimoniosos (Fig. 4B, C), ambos con una co en géneros de Poaceae propuso, que los tipos ANÁLISIS CLADÍSTICO 414 Annals of the Missouri Botanical Garden Oplismenopsis najada Oplismenus hirtellus Echinochloa crus-galli _ [г Echinochloa holciformis Echinochloa oplismenoides Panicum miliaceum Oryzidium barnardii Melinis minutiflora Louisiella fluitans Panicum aristellum Chaetium bromoides M E us Chaetium festucoides Chaetium cubanum -m—»- Oplismenopsis najada Oplismenus hirtellus Echinochloa crus-galli Echinochloa holciformis Echinochloa oplismenoides Panicum aristellum Panicum miliaceum Oryzidium barnardii Louisiella fluitans Melinis minutiflora Chaetium bromoides Chaetium festucoides Chaetium cubanum Oplismenopsis najada Oplismenus hirtellus Echinochloa crus-galli Echinochloa holciformis Echinochloa oplismenoides Chaetium bromoides Chaetium festucoides Chaetium cubanum Melinis minutiflora – пао SE T Louisiella fluitans Oryzidium barnardii Panicum miliaceum Panicum aristellum Volume 85, Number 3 1998 Morrone et al. Revisión de Chaetium "0 100 30 “үе ҒЫ - nt Е] ие ө ү] жай a %. 2 ө: 10 € Chaeti bromoides Я | Ж Chaetium cubanum “ес = —4 A » —.—| ж Chaetium festucoides ~ 7 ` py | = s Á < у MS j ^ yi 3 0+- у, =. URS о uc a R X 2 | — | 70 60 | 40 Figura 5. Distribución de las especies de Chaetium. anatómicos PS y MS evolucionaron independien- temente en la tribu Paniceae y correlacionan ambos tipos con los subtipos fisiológicos NAD-me у NADP-me respectivamente. Consecuentemente, Brown зећајб que Chaetium, que posee especies y MS, debería segregarse en dos géneros. Más recientemente, Hattersley y Watson (1992), indi- caron que existe una marcada diversificación foto- sintética en las Poaceae y postularon la hipótesis de que las gramíneas NADP-me (= MS) podrían haber evolucionado a partir de las NAD-me (PS) El presente estudio corrobora el estatus monofilé- tico de Chaetium y sustenta la hipótesis antes men- cionada por Hattersley y Watson (1992), pues el análisis cladístico muestra que las especies MS, festucoides y C. cubanum, son derivadas de un an- tecesor hipotético PS. A la vez, se descarta la po- sibilidad de segregar al género de acuerdo a la ana- tomía foliar que presentan las especies. DISTRIBUCIÓN GEOGRÁFICA Chaetium es un género neotropical cuya área se extiende aproximadamente desde los 23° de latitud Norte hasta los 10? de latitud Sur (Fig. 5). Sus es- pecies poseen una distribución alopátrica: Chae- tum bromoides, crece desde el centro y sur de México hasta el sur de Panamá. Chaetium cubanum sólo se la halla en Cuba. Chaetium festucoides ha- bita en el norte de América del Sur, presentando un área discontinua: se encuentran poblaciones en el noreste de Brasil (en los estados de Ceará, Mara- te Figura 4. —A. Cladograma de consenso estricto obtenido a partir de dos cladogramas B parsimoniosos B y C. Los caracteres indicados con barras negras señalan sinapomorfías, las barras vacías homoplasia 416 Annals of the Missouri Botanical Garden nháo, Paraíba y Rio Grande do Norte) y otras cre- cen en el norte de Colombia (en el estado de Bo- lívar) y Venezuela (en el estado de Guárico). De esta distribución alopátrica se puede deducir que la posibilidad de que los grupos fotosintéticos representados por Chaetium bromoides y cubanum—C. festucoides (MS) al quedar aislados geográficamente pudieran haber evolucionado en forma vicariante (Hedges et al., 1994) TRATAMIENTO TAXONÓMICO Chaetium Nees, in Mart., Fl. Bras. Enum. Pl. 2: 269. 1829. Panicum sect. Chaetium (Nees) Dóll, in Mart., Fl. Bras. 2(2): 149. 1877. TIPO: Chaetium festucoides Nees. dert Presl, in C. Presl, Reliq. gaan 1: 323. 30. TIPO: Berchtoldia bromoides J. Pre Plantas perennes, cespitosas, en ocasiones es- tolonfferas, con cafias de entrenudos macizos, con médula esponjosa. Lígulas membranáceo- pestafiosas a pestañosas. Láminas lineares. Inflo- rescencias espiciformes a subespiciformes, con ramificaciones ascendentes, adpresas, terminan- do en una espiguilla desarrollada; espiguillas en pares, con desarticulación oblicua en la base de la espiguilla junto al pedicelo. Espiguillas lar- gamente ovoides, dorsiventralmente comprimi- das, plano-convexas, glabras; callo puntiagudo, formado por la base de la gluma inferior y el entrenudo de la raquilla, híspido. Gluma inferior largamente lanceolada a setácea, aristada, no abrazadora, separada de la gluma superior por un entrenudo conspicuo. Gluma superior aristada. Lemma inferior acuminada a aristada. Pálea in- ferior y flor inferior ausentes. Antecio superior lar- gamente ovoide, cartilaginoso; papiloso, escabro- so junto al ápice; lemma con los márgenes membranáceos, planos, dejando libre el ápice de la pálea; pálea binervia; estambres 3; lodículas 2; estilos libres desde la base; estigmas plumo- sos. Cariopsis largamente elipsoide, dorsiventral- mente comprimida, pálida; hilo linear, Y del largo de la cariopsis; embrión %—% del largo de la cariopsis. Género americano con tres especies, desde Méx- ico e islas del Caribe hasta Brasil. CLAVE PARA LA IDENTIFICACIÓN DE LAS ESPECIES la. Espiguillas de 8-11 mm de largo (sin las aristas); gluma inferior oe lanceolada, cubrié ndo totalmente el dor via, con З nerv distantes; lemma in vainas aqui- lladas; lígulas pestafiosas [México hasta Panamá] РОР REE teresa ТЫТ aetium ш 1Ь. эе аы шша de 4.8–9.2 тт de largo (sin las a dorso redondeado; lígulas membranáceo- -pesta- ñosas. 2a. Gluma inferior ovada; lemma inferior acu Iminada; plantas erectas a apoyantes, con láminas de 4-12 X 0.1—0.2 cm, poco diver- gentes del caule; Cuba ___ Chaetium cubanum . Gluma inferior reducida a una seta; lemma inferior aristada, la arista de 0.6-1.5 ст де ~ c E = = 4 z n о ~ c e = n w Uia c Б. E a: & Ф о 5-І ст, divergentes del ЖК Colombia, бы y Brasil .. A ы ы ы е deus айн 1. Chaetium bromoides (J. Presl) Benth. ex Hemsl., Biol. сеп!.-атег., bot. 3: 503 Berchtoldia bromoides J. Presl, in C. Presl, Re- liq. haenk. 1: 324, pl. 43. 1830. TIPO: Mé- xico. Sin localidad, 1836, Haenke s.n. (holóti- no visto; isótipos, MO-1837508, US- 865574 no visto). Figuras 5, 6. Plantas cortamente rizomatosas, con cañas esto- loníferas, arraigadas y ramificadas en los nudos ba- sales a erectas, simples o ramificadas hacia la base; porción erecta 40-120 x 0.2—0.3 cm; entrenudos de 15—30 cm de largo, comprimidos, longitudinal- mente surcados, glabros, pajizos; nudos glabrescen- tes a barbados. Vainas de (3-)5-15 cm de largo, menores que los entrenudos, glabras a esparcida- mente papiloso-pilosas, aquilladas, las basales flo- jas, los márgenes pestañosos, más densamente ha- cia la porción distal. Lígulas pestañosas, de 1.6— 2.6 mm de largo; porción membranácea de 0.1-0.3 mm de largo, pestañas formada por un arco de den- sos pelos blanquecinos de 1.2-2.5 mm de largo; cuello híspido. Láminas 10-30 cm X 0.4—0.9 mm, ascendentes, planas a plegadas, esparcidamente papiloso-pilosas en ambas caras, de base redon- Figura 6. Chaetium bromoides gluma inferior. —D. Es Lemma inferior. —H. —J. Lodículas. —K. Cariopsis, vista hilar. —L. continuación de las aristas. (Weston 4145.) => . Hábito. —B. Detalle de la región ligular. —C. Espiguilla vista del ms de la piguilla vista Fe lado de la gluma superior. —E. Glu F. Glu Antecio superior visto del lado de la lem hil Cariopsis, vista escutelar. En c', uma inferior. — uma super 5. Antecio ues visto en lado de le ЕРІН 4, d”, егу Ғ se indica la ma. — 1. 417 Revisión de Chaetium Morrone et al. Volume 85, Number 3 1998 * | о қ 5 AA ee ho] PA тен алын тәж. ОООО РАНИ І С sna T cum eue pp m poorer сс см | LTDA ed — 55 N SENS “Soo SS Р SA NS SONS E. ЭМ, E > = WSs У RN = EE Қ => LAZO HS LS = NN =, xN SSS SS SSAA SSS ESG ЗУ po TP >, Е == at 22 TA 272 ж | Vi Wie “+ 205% А | CN і 418 Annals of the Missouri Botanical Garden deada y ápice agudo, con los márgenes escabrosos, pestañosos. Pedúnculos subincluidos en las vainas foliares a exsertos, hasta de 45 cm de largo, com- primidos, glabros a esparcidamente poor Inflo- rescencias subespiciformes, 9-26 X cm, ter- minales, multifloras; eje principal y pu de las ramificaciones triquetros, glabros, escabrosos; pul- vínulos glabros a pilosos; pedicelos en pares, de- siguales, triquetros, escabrosos. Inflorescencias axi- lares en ocasiones presentes, similares a las terminales. Espiguillas de 8-11 mm de largo (sin las aristas), 1-1.2 mm de ancho, escabrosas, verdes a púrpuras. Gluma inferior herbácea, largamente lanceolada, de 7-10 mm de largo (sin la arista), 7- nervia, con 3 nervios centrales aproximados, los restantes submarginales, escabrosa sobre los ner- vios, espacios internervales escabriúsculos, cara interna escabrosa hacia la porción distal, luego lisa, lustrosa; arista de 1.8-2.5 cm de largo, pálida a púrpura, escabrosa. Gluma superior herbácea, de 6. mm de largo (sin la arista), el dorso ligera- mente concavo, 7-nervia, con 3 nervios centrales, los restantes submarginales, escabrosa sobre los nervios y espacios internervales, la cara interna es- cabrosa hacia el ápice, lisa y lustrosa hacia la base; arista de 1-2 cm de largo, pálida a púrpura, es- cabrosa. Lemma inferior de 6-8 mm de largo, coriácea, lisa, el ápice acuminado a cortamente aristado, arista de 1-1.5 mm de largo, escabriús- cula, 3(-5)-nervia. Antecio superior largamente ovoide, 6-8.8 X 1.2 mm, cartilaginoso, pajizo, pa- piloso, escabroso hacia el ápice; lemma aristada, arista de 1.2-2 mm de largo, escabrosa; lodículas 0.4—0.6 mm de largo, hialinas, conduplicadas, abrazando los bordes de la lemma; anteras 1.5-2 mm de largo. Cariopsis 2.6-2.8 X 1 mm; embrión 2 del largo de la cariopsis. Nombres vulgares. tilla” (en Costa Rica). Distribución y ecología. Habita desde México hasta Panamá desde el nivel del mar hasta 2100 m de elevación. Se halla en laderas montaíiosas en bosques de Quercus y Pinus; también en áreas mo- dificadas junto a malezas o en bordes de caminos. *Granillo" (en México); *se- Material representativo citado. COSTA RICA. Alajue- : In and around Zarcero, Cantón Alfaro Ruíz, Highway 15. 1700-2000 m, Weston et al. 3073 (UC). Cartago: 6 km SW of Pacayas, 1700 m, Pohl & Davidse 10891 (UC); San Ramón, Е of San José, 1400 m, Pohl & Gabel 13677 (MO); W side of Cartago, 1450 m, Pohl & Calderón 10238 (MO). Puntarenas: forest along trail between ge к and Cotonsito, along the Río Cotón, 8°56'N, 82°48'W, 1400 m, Davidse 24656 (MEXU, MO). San - 3 km SW of Tarbaca, 1650 m, Weston et al. 2659 (UC); : N of Buenavista, 1450 m, Pohl & Davidse 10972 EL SALVADOR. San Andrés, 1700 ft., Mena 45 (US). GUATEMALA. Quezaltenango: in untains near Santa María, just S of Quezaltenango, Weatherwax 158 2 (UC). MÉXICO. Chiapas: Mun. Motozintla de Mendoza, 45— 50 km NE of Huixtla along road to Motozintla, 1900 m, Breedlove 40213 (US); Mun. е del Valle, 14 Кт у a а hwy. to Comitán, 2100 m, Davidse et al. 29802 XU). Colima: mg Comala, Rancho El Jabalí, 20 km p NNW of Colima in the SW foothills of the Volcán de Colima, ca. 19°26.2’N, 103°41.8'W, 1450 m, Sanders et al. 10284 (MO). Distrito Federal: Pedregal de San Angel, Lyonnet 1661 (MEXU, US); Cerro Xochi- tepec, cerc ochitepec, Azedowski 23292 (MEXU). Guanajuato: 5. Andrés, S. Miguel, Liebmann 595 (US); ана 5. 2. et 5. Miguel, Liebmann 866 (US) a de Cacahuamilpa, Piedras Ne Matuda 29718 US). Jalisco: 2.00 Jones 27442 (UC); Guadalajara, Jones 27662 (US); Mun. T. Barranca de San Juan de Dios, cerca га Los Sierra de Los els Rzedowski 17381 (US); banks near Guadalajara, 5000 ft., Pringle 11736 (L IL, MEXU, US); about 9 mi. S of Guadalajara Pis eed right of way, 1765 m, Reeder & Reeder 2334 (MEXU); plains of Guadalajara, Pringle 2331 (LIL, or US); Barrio of Ciudad Granja, 8 km W of nn ca. 100 m S of Avenida Vallarta, 1560 m, /ltis & Nee 1048 (MEXU, US), forming dense clumps up to % m diam, ca. 1-1.2 m tall; Club de Golf Santa Anita, km 7 del Peri- férico, carretera a Morelia, 1550 m, González & Carvajal 559 (MEXU); 1.5 km E of El Depósito, ca. 10 km WSW of Ciudad аи, 19*39'N, 103°32'W, 1800 m, /ltis et Е 588 (MEXU, US); Lomas del Valle, Guadalajara, 1610 . Díaz Luna 5010 (M 2 Мех xico: T 1750 m, Hinton 2418 (BAA, 05); ' Pire 13 Oct 1934, Hinton 6748 (MO. cán: Morelia, Arséne 5445 (MEXU, US); Uruápan, 5600 ft., Finite 6985 (US). Morelos: Duenae vaca, Woronow & Juzepczuk 948 (US); 5 mi. N of Cuernavac ` Gould 7018 (UC); Valle del Tepeite, Lyonnet 2447 (MEXU, US); Cuer- navaca, 4500 ft., Hitchcock 6856 (US). Nayarit: 2 mi. W of Guadalajara on Highway 15 towards Tepic, 5000 ft., Soderstrom 596 (US). Veracruz: Cerro Macuiltepec, just outside Jalapa, 5300 ft., Reeder & Reeder 5982 (MO); Río Blanco near Orizaba, Hitchcock s.n. (LIL-39361, m US); Río Blanco, June-Oct. 1886, Palmer 619 (MEXU, US); Orizaba, Botteri s.n. (US-976996, -821224); 52. hillo pres ner Bourgeau 2597 (US PANAMA. Chiriquí: ca. 3 km NE of El Hato del Vol- Figura 7. Chaetium cubanum. —A. Hábito с. Antecio superior visto del lado de la pálea Cariopsis, vista hilar. —L. (А, Ekman 16474; B-L, Ekman 7816.) — . —B. Región ligular. —C. Espiguillas vista del lado de la gluma inferior. —D. Espiguilla vista del lado de la gluma superior. —E. Gluma inferior. —F. ( Glum а a inferi ior. las per ior. ›. —]. Anteci io superior visto del en ide la pacem =. Lodículas Cariopsis, isa esc ife ‘lar. En e”, c", d', d", e' y f' se indica la continuación de las aristas. Volume 85, Number 3 Morrone et al. 419 1998 Revisión de Chaetium 420 Annals of the Missouri Botanical Garden cán at base of Volcán de Chiriquí (Barú), 3 km E of high- way, 1800 m, Davidse & D'Arcy 10412 (MO, US). Chaetium bromoides es exomorfológicamente similar a C. festucoides; se distingue de esta especie por la gluma 7-nervia, lígulas pestañosas y espi- guillas de mayor tamaño. Un ejemplar con caracteres morfológicos discor- dantes de Chaetium bromoides es Matuda 29718, coleccionado en el estado de Guerrero, México, el que posee la gluma inferior y superior esparcida- mente pilosas en los bordes y panojas más laxas, con menos espiguillas. 2. Chaetium cubanum (C. Wright) Hitchc., Contr. U.S. Natl. Herb. 12: 232. 1909. Perotis cubana C. Wright, Anales Acad. Ci. Méd. Ha- bana 8: 288. 1871. TIPO: Cuba. Región ori- ental, 1856-1857, Wright 735 (isótipos MO- 2095218, -1837473). Figuras 5, 7. Plantas con rizomas nodosos, de entrenudos cor- tos, con cafías largamente estoloníferas, arraigadas y con numerosas innovaciones extravaginales a erectas a apoyantes con ramificaciones prolíferas; porción erecta 40—70 х 0 .3 em; entrenudos de 4—12 cm de largo, cilíndricos, glabros, finamente estriados; nudos pubérulos. Vainas de 2–3.5 cm de largo, menores que los entrenudos, de dorso redon- deado, glabras, con pelos esparsos junto a la unión con la lámina, los márgenes ciliados. Lígulas de 0.2-0.6 mm de largo, membranáceo-pestañosas, castañas, glabras; cuello castaño, cortamente pilo- so. Láminas 4-12 X 0.1–0.2 cm, planas a involu- tas, ascendentes, cortamente híspidas en la cara adaxial y con cara abaxial glabra o con ambas caras cortamente híspidas, de base atenuada continuán- dose con la vaina y ápice largamente agudo, los márgenes cartilaginosos, glabros, esparcidamente papiloso-pilosos hacia la base. Pedúnculos subin- cluidos en las vainas foliares. Inflorescencias ter- minales, subespiciformes, paucifloras, subexsertas, parcialmente incluidas en las vainas foliares, 3-9 X 0.5 cm; eje principal y ejes de las ramificaciones triquetros, escabrosos; pulvínulos pilosos; pedicelos en pares, triquetros, escabrosos. Espiguillas de 7— .2 mm de largo (sin las aristas), 0.8-1 mm de ancho, escabrosas, glabras, verde pálidas o con tin- tes purpúreos. Gluma inferior ovada, de 1.8-3 mm de largo (sin la arista), herbácea, 3-nervia, esca- brosa sobre los nervios; arista de 1.4—3 cm de largo, pálida a púrpura, escabrosa. Gluma superior de 5.6-6.5 mm de largo (sin la arista), herbácea, 9— 13-nervia, los nervios escabrosos, cara interna es- cabrosa hacia la porción distal, luego lisa, lustrosa; arista de 2-2.5 cm de largo, escabrosa. Lemma inferior glumiforme, de 6.4—7.2 mm de largo, her- ácea, 7-9-петуіа, de ápice acuminado, cortamente hispídula entre los nervios, escabrosa hacia la por- ción superior de la cara interna, lisa en el resto de a superficie. Antecio superior largamente ovoide, 4.4—6.4 X 0.8 mm, 27 pajizo, papiloso, escabroso junto al ápice; lemma 5-nervia, corta- mente aristada, la arista hasta de 0.8 mm de largo, escabrosa; pálea de igual largo que su correspon- diente lemma o ligeramente mayor, lodículas de 0.6—0.8 mm de largo, hialinas, no conduplicadas, glabras; anteras de 0.8 mm de largo. Cariopsis 3.2 X 0.6 mm de ancho; embrión % del largo de la carlopsis. Distribución y ecología. Endémica de Cuba, crece desde el nivel del mar hasta los 800 m, en aderas secas. Material adicional examinado. CUBA. Sin Provin- a: Región oriental, Wright 734 (MO). Matanzas: al S de Сиа, ЕКтап 16474 (US). peo de Cuba: El Co- JS); El Cobre, near the mine, Ekman 7816 (US). Villa Clara: Santa Clara, Motembo, Ekman 16827 (US). Chaetium cubanum se distingue de las restantes especies del género por comprender plantas lar- gamente estoloníferas, con cañas profusamente ra- mificadas y apoyantes en la vegetación; sus láminas son ascendentes, poco divergentes del caule. Esta especie ha sido escasamente coleccionada en la isla de Cuba existiendo especimenes en los her- barios de más de 70 años de antiguedad. 3. Chaetium festucoides Nees, in Mart., Fl. K O: Brasil. “Habitat in graminosis et in cultis ad lumen S. Francisci, ad Joazeiro, etc. Provinciarum Pernambucanae et Bahien- sis," Martius s.n. (holótipo, M no visto; isótipo, US-3049472). Figuras 5, 8. — Figura 8. Chaetium festucoides. —A. Hábito. —B. Región ligular. —C. Espiguilla vista del lado de la gluma superior. —D. E: spiguilla vista del lado de la gluma inferior. —E. Gluma inferior. —F. Gluma superior. —С. Lemma inferior. bu cio superior visto del lado de la lemma. —I. Antecio superior visto del lado de la раје: Lodículas. — riopsis, vista hilar. —L. Cariopsis, vista esc clan En сс", d'-d" y е” se indica la continuae ión de las aristas. (A, Pus 14926; B-L, Pickel 1714.) 421 Morrone et al. Volume 85, Number 3 1998 Revisión de Chaetium $4 T ee l D sd =. | ) oi t dee y oe vc IT ——— 422 Annals of the Missouri Botanical Garden Plantas cortamente rizomatosas, de 70—110 cm de alto, con cañas erectas a geniculadas, simples; entrenudos 9-22 X 0.2-0.3 cm, cilíndricos, gla- bros, los basales más cortos; nudos castaños, gla- bros. Vainas de 5-11 cm de largo, usualmente me- nores que los entrenudos, de dorso glabras a papiloso-hirsutas hacia la porción distal, con un margen ciliado, el restante glabro. Lígulas de mm de largo, membranáceo-pestafiosas, redondeado, con la cara externa esparcidamente pilosa; cuello hirsuto. Láminas 10-35 x base redondeada y ápice largamente agudo, papi- loso-hirsutas en ambas caras, con largos pelos rí- gidos junto a la lígula, hasta de 5 mm de largo, los márgenes escabrosos. Pedúnculos exsertos a subin- cluidos en las vainas foliares, hasta de 30 cm de largo, cilíndricos, glabros. Inflorescencias espicifor- mes, 5-20 X principal y ramificaciones triquetros, glabros, es- cabrosos; pulvínulos con un mechón de pelos blan- quecinos; pedicelos delgados, triquetros, glabros a esparcidamente pilosos, escabrosos. Espiguillas de 4.8-7.2 mm de largo (sin las aristas), 0.8—1 mm de ancho, glabras, escabrosas, verde pálidas o con tin- tes purpüreos. Gluma inferior setácea, reducida a una arista escabrosa de 1.2-3.5 cm de largo, ex- céntrica, pajiza a pürpura, enerve. Gluma superior mm de largo (sin la arista), herbácea, 3- nervia, con los nervios manifiestos, escabrosos, uno central, los restantes submarginales, espacios in- ternervales hialinos, delicados; arista de 1-1.7 cm de largo, escabrosa. Lemma inferior glumiforme herbácea, de 4-5 mm de largo (sin la arista), 5- nervia, con los nervios manifiestos, uno central, los 0.6-1.5 cm de largo, escabrosa. Antecio superior largamente ovoi- de, de 44.8 mm de largo (sin la arista), 0.8-2 mm de ancho, cartilaginoso, papiloso, escabroso hacia la porción superior; arista hasta de 0.8 mm de lar- go, escabrosa; lodículas de 0.5 mm de largo, hia- linas, truncadas, no conduplicadas, con una cara pilosa; anteras de 0.6-0.8(-1.6) mm de largo. Cariopsis 2.8 X 0.7 mm, pálida; embrión % o un poco más del largo de la cariopsis. 0.5-1 cm, planas, de 0.5-2 cm, terminales, exsertas; eje restantes equidistantes; arista de Distribución y ecología. Se encuentra en el no- reste de Brasil y en Colombia y Venezuela, desde el nivel del mar hasta los 500 m de elevación, en lugares abiertos, secos o en bordes de monte. Material d examinado. BRASIL. Ceará: Mun. Parangaba, Tapereóba, Black 55-18634 үе US): Martinópolis Swallen 4635 (US). Maranhão: Mun. Lo- l Rios Balsas and araíba: sin loca US). Piauí: sin n e alidad, Gardner 2346 (US). Pernam- buco: Tapera, Pedras de Коро, Pickel 1714 (US). Rio Grande do Norte: Estremoz to Natal, Swallen 4797 (US). COLOMBIA. Bolívar: N of Arjona, 30—50 m, Killip & Smith 21194 (MO EZUELA. Guárico: Llanos de Calabozo, entre la Encrucijac ada y la Misión de Arriba, Pittier 14926 (US); Estación Biológica de los Llanos, Trujillo 8639 (MO). Doell (1877) cita erróneamente la presencia de Chaetium festucoides en Cuba, sobre la base del ejemplar Wright 735, correspondiendo este último al tipo de C. cubanum. Se observaron ores cleistógamas en los ejem- plares Eiten & Eiten 4713, Black 55-18634, Swa- llen 4797 y Pittier 14926 los que poseen pram desarrolladas y anteras pequeñas, de 0.6—0.8 m de largo, presentes en el ápice de la cariopsis. El único ejemplar que posee anteras de 1.6 mm de largo es Gardner 2346, pero no se halló cariopsis desarrollada. La distribución geográfica de Chaetium festucoi- des es disyunta; la mayoría de sus colecciones son de la región nordeste del Brasil habiéndose hallado ocasionalmente en savanas de Venezuela y en zonas costeras del estado de Bolivar en Colombia, junto al mar Caribe. Nuevas colecciones permitirán com- probar si esta especie crece en áreas intermedias del norte del Brasil y las Guyanas. En el herbario US se examinó un duplicado del tipo de esta especie, correspondiente a un frag- mento del holótipo depositado en Munich. Cabe destacar que si bien en el protólogo de la especie se cita “Provinciarum Pernambucanae et Bahien- sis,” lo cual podría indicar que existe más de un ejemplar tipo de C. festucoides, los datos antes men- cionados del protólogo coinciden exactamente con la información de un único ejemplar de herbario coleccionado por Martius en el Brasil. Esta situa- ción se repite en otras especies de Poaceae colec- cionadas por Martius y descriptas por Nees von Esenbeck, en géneros como Panicum, Paspalum y otros. Literatura Citada Beetle, A. A. 1977. Noteworthy grasses from Mexico. V. Phyiologia 37: 317-407. Las Gramíneas de México, tomo II. Se- с iban de Кале 'ultura y Recursos Hídraúlicos, México. Bentham, С. otes on Graminae. J. Linn. Soc., Bot. 19: 14—134. . 1883. Pe ig Pp. 1074-1215 en G. Ben- tham & J. D. Hooker, Genera Plantarum. J. Cramer, Weinheim б 1965). Brown, W. V. 1977. The Kranz syndrome and its 2 in in systematics. Mem. Torrey Bot. Club 23: 1-97. Chase, 11. oi a ad of Paniceae. D Proc. Biol. ish Wash. 2" Clayton, W ‚А. пате 1986. Genera Grami- num. Kew Bull. z^ 1-389 Volume 85, Number 3 1998 Morrone et al. Revisión de Chaetium Cutler, D. 1978. Applied Plant Anatomy. Academic Press, London. Dengler, N. G., R. E. Dengler & P. W. Hattersley. 1985. Differing ontogenetic origins of PCR (“Kranz”) sheaths in leaf blades of C, grasses (Poaceae). Amer. J. Bot. 72: 284-302. Doell, J. C. 1877. e 3, Paniceae. En: C. F. P. von Martius (editor), Fora Hral inia 2(2): 33-340. Mu- nich, 2. Leipz Elkin, L. & R. B. Park. 1975. Chloroplast fluorescence of C, plants. I. Detection with infrared color film. Planta 127: 243-250. Ellis, R. P. 1977. Distribution of the Kranz Syndrome in the Southern African Eragrostoideae and Panicoideae according to xc sheath anatomy and cytology. Agro- xu 9: - Farris. Ј. S Eu Reference, version 1.5. Port Selleni 4. J. S. Farris, New York. lai. T. S. 1986. O conceito do fruto em gramineas o na taxonomia da família. Pesq. Agropecu. Brasil. 21: 93-100. 886. Mexicanas Plantas. Pars secunda Gra- Chromosome numbers of some Mexican grasses, Canad. J. Bot. 44: 1683—1696. Gutiérrez, M., G. E. Edwards & W. à Brown. 1976. PEP side 2. species in the Brachiaria group al Pos СЧ: а Biochem. Syst. & Ecol. 4 Hackel, + 7. Gramineae. Еп: A. Engler & K. Prantl (editores), ru natürlichen Pflanzenfamilien 2(2): 1-97. Engelmann, v кыены j ses , P. W. riations on | photosynthetic path- p. 49-64 1" R strom, K. W. Hilu, C. S. Campbell & M. E Barkworth (editors), Grass System- atics and Evolution. Smithsonian Institution Press, le са ——&L о. 1992. Di a of photosyn thesis. Pp. 38-116 en G. P. Chapman (editor), pi Evolution and tios мыкы Univ. Press, + S.B s & L. R. Max 1994. Re aba a caia of the 1 Cladistics 10: 43-53. Hennig, W. 1966. Phylogenetic Systematics. Univ. Illi- nois Press, pers Hitchcock, A 1909. Catalogue id zà grasses of Cuba. Contr. pud Nal Herb. 12: 183-2 The grasses of 5. America. Contr. U.S al p 24: 557-762. 1936. Manual of the pue of the West Indies. U.S.D.A. Bur. Pl. Industr. Misc. Publ. 243: 1–439. & A. Chase. 1917. nO А the West Indies. Contr. US. Natl. Herb. 18: 261—4 Hsu, C. C. 1965. The classification pr Panicum (Grami- neae) and its allies, with special reference to the char- acters of lodicule, style- farsi а lemma. J. Fac. Sci. v. Tokyo, Sect. 3, Bot. 9 Hubbard, C. E. & H.G. E 1936. Oryzidium, a new genus from South West Africa. Bull. Misc. Inform. . 1940. Plant Microtechnique. McGraw- Y 833. тни Plantarum: 1—606. Ј. С. Cottae, Stuttgart & Tübin McVaugh, Н. 1983. Gramineae. En: W. R. Anderson (ed- itor, Flora Novo-Galiciana. A Descriptive Account of the Vascular Plants of Western Mexico 14: 1-436. Univ. 960. Anatomy of Monocotyledons. I. Gramineae. gere: Univ. Press, Oxford. Montiel, M. 972. Determinación taxonómica de la especie с bromoides (Presl) Benth. basada en Sireptostachys (Poaceae-Panicoideae) su posición sis- mática dentro р и Paniceae. Ann. Missouri Bot Gard. 78: 359— T. S. Filgueiras, F. O. Zuloaga & J. Dubcovsky. 1993. Revision of Anthaenantiopsis (Poaceae: Panicoi- En: C. F. P. von Martius (editor). Flora brasiliensis se enumeratio AME rum, Vol. 2(1). J. G. Cottae, Stuttgart & Tübin Nixon, K. C. 3. Clados, versión 1.38 [32 bit] beta version. Programa de computación distribuido por el autor. & J. M. Carpenter. 1993. On outgroups. Cladis- tics 9: 413—4. O'Brien, T. P. & M. E. McCully. 1981. The Study of Plant Structure Principles and Selected Methods. Termacarp- hi, аа pru" а 4 5. үка 1985. Comparative о Panicum dichotomiflorum. Pl. Cell Physiol. 21: 1329— 1333. — € N. Chonan. 1982. C, syndrome of the species in the d LAE of the genus Pan- icum. Bot. Mag. (Tokyo) 95: 339-347. Pohl, R. W. 1980. #15, 2. Еп: W. Burger с tor), Flora costaricensis. Fieldiana, Bot., n.s. 4: 1-60 -----. 1994 tium. Pp. 330-331 еп С. ета M. Sousa & A. 0. Chater (editors), Flora Mesoameri- Botanical т St. Louis; The Natural History Mu- seum, Lon & G go 1971. Chromosome numbers of Costa Rica grasses. Brittonia 23: 293-324. Prendergast, H. D. V. & N. E. Stone. 1987. New struc- tural/biochemical associations in leaf blad grasses (Poaceae). Austral. J. Presl, J. 1830. Gramineae. En: C. Pre mn 207—356. A. Asher, Amsterdam (reimpresión, 973). TOC J. R. 1967. Notes on mexican grasses. VI. cellaneous chromosome numbers. Bull. Torrey т. Club. 94: 1-17 Renvoize, S. A. 1987. A survey of leaf-blade anatomy in grasses ХІ. Paniceae. Kew Bull. 42: 739-768. ра Institution Press, Washington, D Smith, L. B. & D. Wasshausen. 1981. Chave pabi os gêneros dis Gramíneas Brasileiras. Bradea 3: 1-3 Steudel, E. G. 1853-1855. Synopsis Plantarum а & 424 Annals of the Missouri Botanical Garden nearum. Еп: Synopsis ARI Glumacearum 1: 1– 475. J. B. Metzler, Stuttg Swallen, J. R. 1955. e of Guatemala, part II: Grasses of Guatemala. Fieldiana, Bot. Troll, W. 1967. Vergleichende Morphologie der höheren Ser йы 2 Teil, IV Abschnitt. Gebrüder Borntrager, erlin Pi в. D. 1988. Genera of the North American Pa- niceae (Poaceae: Panicoideae). Syst. Bot. 13: 576—6 — ——— & J. Valdés Ке nera of Mesoame- псап Paniceae (Poaceae: Panicoideae). Sida 13: 187— Zuloaga, F. O. & T. R. Soderstrom. 1985. e ан of de outlying species of New World Panicum (Poace Paniceae). Pius sonian Contr. Bot. 59: 1—63. one & J. Dubcovsky. 1989. Exomorp- ho logical: м... and cytological studies іп Pani- cum validum (Poaceae: Panicoideae: Paniceae): Its sys- tematic position within the genus. Syst. Bot. 14: 220- 230. ÍNDICE DE COLECCIONES NUMERADAS Cada especimen es citado por el primer nombre del especie: Chaetium bromoides (1); C. cubanum (2); C 2. (3). Arsene Bro. 5445 (1). lack, С. A. y den (3); Bourgeau, M. 2597 (1); Breedlove, D. E 40213 (1). Davidse, G. ПУТ (1), 24656 (1), 29802 (1); Díaz Luna 5010 (1). Eiten, G. 4713 (3); Ekman, E. L. 7816 (2), 15699 (2), rhe e» 16827 (2). r, G. 2346 (3); González 559 (1); Gould, F. W. 7018 41), 11697 (1). Нітоп, G. B. 2418 (1), 5043 (1), . S., Amer Gr. Nat. Herb. 307 (1 7114 (1), 7287 (1). Iltis, Н. Н. 588 (1), 1048 (1). Jones, M. E. 27442 (1), 27662 (1). Killip, E. P. 21194 (3). Liebmann, F. M. 594 (1), 595 (1), 866 (1), 12912 (1); Lyonnet, E. 1661 2447 (1). Martínez Pérez 149 (1); Matuda, E. 1591 (1), 25920 (1). 29718 (1); Moraes 2217 (3); Müller 2126 (1). Nee, M. 2 (9. Palmer, E. 619 (1); Pickel, B. J. 1714 (3); Pittier, H. 14926 (3); Ро M. К. W. 9908 (1), 10238 (1), 10364 (1), 10891 (1), wine бу; 11040 (1), 13677 (1); Pringle, C. 6748 (1); Hitchcock, ). 6856 (1), 6985 (1), Reeder, J. - р, 4169 (1), 5982 (1); Rzedowski, J. 17381 (1), 23292 ( 2. 10284 (1); Soderstrom, T. R. 596 (1); Stevens, W. D. 13645 (1); Swallen, J. R. 4635 (3), 4797 (3). Trujillo, B. 8639 (3 Vázquez 1064 (1); Vera Santos, J. 310. Walkins, J. M. 45 (1); Weatherwax, P. 158 (1); Weston, A. S. 2140 (1), 2659 (1), 3073 (1), 3341 (1), 3555 (1). 4145 (1); Woronow, G. 948 (1); Wright, C. 734 (2). APÉNDICE 1. MATERIAL ADICIONAL EXAMINADO Echinochloa crus-galli. ARGENTINA. Buenos Ai- res: Pergamino, Boelcke 506 (SI). Córdoba: Castellanos 465 (SI). Entre Ríos: Concepción del Uruguay, Sorarú 168 (SI). PARAGUAY. Central: Па, Ramírez 69 (BAA). RUGUAY. Sin localidad, е 5466 (51). Echinochloa d рај CO. Jalisco: Lake Chapala near Tuxcueca, ы. f Leavenworth 1848 (US); Orozco, Hitchcock 7375 (US). Michoacán: Cerro Santa María, Feddema 96 (US); Cerro Potrerillos, King & 04 (US); 3 mi. E of Morelia, Soderstrom 546 Morelia: Morelia, Arséne s.n. (LIL-39358). Echinochloa oplismenoides. MÉXICO ascalien- tes: Aguascalientes, Hitchcock 7441 (US), Archer 3991 (US). Jalisco: 36 km Ojuelos-Aguascalientes, Xolocotzi X- 2503 (US). Puebla: 6 mi. E of Puebla, Soderstrom 394 b Arsène 5444 (US). Louisiella fluitans. REPÚBLICA CENTRAL AFRI- Manovo Gounda-St. Floris National Park, Fay 6160 (MO), 7353 (MO). Melinis minutiflora. ARGENTINA. Buenos Aire La Plata, Burkart 12484 Misiones: El Dorado, Mar- tínez Crovetto 10032 (51). BRASIL. Paraná: Porto d . VENEZUELA. Cojedes 16166 (SI). Miranda: Miranda, Burkart 16012 (SI). Oplismenoides najada. ENTINA. Buenos Aires: Delta del Paraná, Burkart 4509 (SI), 7598 (SI); Médanos, Burkart 3564 (51). Corrientes: La Cruz, Burkart 8105 (SI). Santa Fe: Los Amores, Lewis & Pire 815 (SI). Oplismenus hirtellus. ARGENTINA. Entre Ríos: La Paz, Burkart 21063 (SI). Jujuy: El Fuerte, Kiesling et al. 5500 (SI). Salta: Rosario de Lerma, Venturi 8219 (SI). SI) МА: near Mboma 15- land "adi in n Mons mi i Wildlife Reserve, Smith 1944 (MO). Panicum milia A . Buenos Aires: Ensenada. A. 189 O (SI); San Isidro, Pastore 849 (SI). BRASIL. Rio Grande do Sul: Sáo Leopoldo, Rambo 1937 (BAA Panicum . aristellum . Paraná: Foz do Rio Taquaral, Hatschbach 33 7 MUS. кы е. ндөн = ge Tu- paceretan, Araujo 338 (US). Santa Catarina: 6 km of Porto Unido, Smith & Klein 15279 (US) Sao Paulo: Mun. Salesópolis, Boraceia, Kuhlmann 2774 (SI). REVISIÓN DEL GÉNERO CUCURBITELLA (CUCURBITACEAE)! Raúl Pozner? RESUMEN De acuerdo con Ja literatura, el género sudamericano Cucurbitella comprende seis especies cuyos caracteres diag- entre los especímenes. La variación de esto i imórfica: Cucurbitella asperata (Gillies ex Hook. intermedios estas razas no reciben n ejemplar de Posadaea sphaerocarpa e estudio demuestra que la estructura s caracteres es continua y no correlacionada. Por ello & Arn.) Walp. La distribución geográfica de las frecuencias de los primordios foliares sugiere la existencia de tres razas ecológicas. Pero en vista de los ombres científicos formales. Por la misma itella ecuadorensis Cogn. queda excluida de Cucurbitella porque su holótipo es en realidad un Cogn. razón no se incluye una c ABSTRACT According to previous authors, the South American genus Cucurbitella (Cucurbitaceae) includes six species, osten- sibly separated by several diagnostic characters of dubious value ресін the degree of stamen exsertion, and the shape and size of the petals and hypanthium as the onl gical characters with significant variation among specimens. These . This study establishes the morphology of foliar id any internal partitions would be arbitrary. Thus, Cucurbitella consists of only one polymorphic species: C. asperata (Gillies ex Hook. & Arn.) races, but due to intermediate spec em is not include Posadaea sphaerocarpa Cogn Walp. = аша ары distribution of foliar primordium characters suggests three ecological s, these races do no d. Cucurbitella Boden Cogn. is removed from Cucurbitella because the holotype represents El nombre Cucurbitella Walp. (Cucurbitaceae) está basado en Cucurbita asperata Gillies ex Hook- er & Атон (1833). Fue publicado como “Curcu- bitella” (Walpers, 1846) y corregido posteriormente por el mismo autor (Walpers, 1847: 769). Cucurbita asperata había sido combinado anteriormente bajo Schizostigma Arn. (1840), homónimo ilegítimo de Schizostigma Arn. ex Meisn. (1838, Rubiaceae). Prasopepon Naudin (1866), con sus dos especies P. durieui Naudin y P. cucumifolius Griseb., fue trans- ferido a СЫА por Cogniaux (1878). A estas tres especies se agregaron Cucurbitella integrifolia ogn. y Cucurbitella ecuadorensis Cogn. Más tarde Jeffrey (1978) transfirió Cucurbita urkupinana Cár- enas, de modo que, hasta hoy, Cucurbitella se con- sidera un género de seis especies: Cucurbitella ecuadorensis, endémica de Ecuador, C. urkupinana, endémica de Bolivia, y las especies restantes, con su área principal de distribución en la Argentina. Cucurbitella durieui se ha citado también para Paraguay, sur de Brasil y Uruguay, C. integrifolia, para Bolivia y Paraguay, y C. asperata, para Chile. Los caracteres utilizados para identificar las su- puestas especies de Cucurbitella son: la consisten- cia, la pubescencia y el grado de división de las hojas (Cogniaux, 1916; Martínez Crovetto, 1965, 1974; Cabrera, 1993), la forma del fruto (Cogniaux, 1916; Cabrera, 1993), el їашайо de las hojas (Ca- brera, 1993), la presencia o ausencia de brácteas en las inflorescencias estaminadas, la forma del hi- panto y la presencia de pelos en la garganta del hipanto (Cogniaux, 1916). La revisión de casi 600 ejemplares de Cucurbitella, en su mayoría de la Argentina, mostró que los caracteres utilizados para reconocer las especies, no permiten la identifica- ción precisa de más de la mitad de los ejemplares. E (1978, 1990 y com. pers.) duda de la iden- ad de C. cucumifolia, C. urkupinana y de los ER entre C. durieui y C. asperata. Por otro lado, la identidad de C. ecuadorensis plantea serias du- a. ! El autor agradece a los curadores de los herbarios de K, Br r su ы лалы. personal sobre los problemas del Аа Cucurbitella, y a Fernando Zuloaga por su guía Charles Jeffrey po y lectura crítica P de „| manu y P. por los s préstamos de material tipo y fotografías, a ? [nstituto de Botánica idm CC 22, 1642 San Isidro, Buenos Aires, Argentina. ANN. MISSOURI Bor. GARD. 85: 425—439. 1998. Annals of the Missouri Botanical Garden das pues su distribución está alejada y aislada del área de distribución del género, y sólo se cuenta con la colección del tipo. Por todos estos motivos, y a raíz de la preparación de las Cucurbitaceae para el proyecto Proflora de Argentina, se realizó la revisión de este género. MATERIALES Y MÉTODOS Se estudiaron las colecciones de Cucurbitella de los herbarios BAA, BAB, BBB, CORD, CTES, LP, MERL, SI, SRFA, material fijado en FAA, y ejem- plares cultivados a partir de semillas para conocer la sucesión foliar del tallo principal. Para el estudio de los primordios foliares se eligió el primer pri- mordio en dirección apical que tuviera su zarcillo no elongado (no sensible en el momento de la her- borización). Los cortes anatómicos fueron hechos a mano alzada y algodón. cas se realizó con rojo de rutenio (Jensen, 196 Todas las ilustraciones son originales, han sido rea- lizadas por el autor y corresponden a material de herbario, fijado, cultivado o fotografías. Para los es- quemas anatómicos se utilizaron los símbolos de Metcalfe y Chalk (1950). Las medidas de las flores y sus partes corresponden a flores totalmente abier- tas. coloreados con safranina o azul de нонун de las sustancias pécti- MATERIAL FIJADO Pozner 62, 63, 65 . Cocucci s.n. ( ; J. H. Hunziker 13116 y 1308 4 (SD. Forma bis “cucumifolia”: Pozner 50 y 88 (BAB); Hoc 63 (BAFC); A. A. Co- cucci s.n. (CORD). Forma foliar intermedia entre “asperata” y “durieut”: A. A. Cocucci s.n. (CORD). Forma foliar “durieui”: Pozner 82 (ВАВ) Forma foliar “asperata”: MATERIAL CULTIVADO Forma foliar “asperata”: Pozner 62, 63, 65 (BAB) y 106 (SI). Forma foliar “cucumifolia”: Pozner 87 (BAB). ANALISIS DE LOS CARACTERES RAIZ Todo el sistema radical de Cucurbitella es de ori- gen primario y forma una estructura reservante jun- to con el hipocótilo, los primeros nudos del epi- cótilo y la porción basal de las ramas de los años siguientes (año dos en adelante), por medio de su desarrollo secundario en diámetro. La raíz principal y sus primeras ramificaciones son napiformes. resto del sistema radical tiene porciones no tube- rosas y tuberosas (tubérculos radicales) que pueden alcanzar hasta 20 cm de diámetro. El ritidoma po- see numerosas lenticelas pulviniformes, notorias y en general alineadas horizontalmente, de modo que semejan un repliegue como los que se observan en las raíces contráctiles, aunque no es éste el caso. Un corte transversal de una porción no tuberosa de la raíz muestra un súber de 6-7 capas celulares, felógeno, una felodermis de 4-5 capas con grupos, esclereidas subyacentes, el floema secundario, el xilema secundario con vasos aislados o en grupos, con parénquima vasicéntrico. Los tubérculos radi- cales presentan la misma estructura secundaria pero con un gran desarrollo de los radios paren- quimáticos, ricos en granos de almidón compuesto, y la formación de cámbium supernumerario (Fig. Este sistema radical combina el desarrollo na- piforme de la raíz principal como en Bryonia y la capacidad de desarrollar tubérculos radicales como Thladiantha dubia Bunge (Troll, 1967). Según Ruíz Leal (1975) el sistema radical de C. asperata puede formar unas 8 a 10 tuberosidades de más de 1 kg, y son éstas más abundantes cuanto más árido es el lugar donde crece esta especie. De acuerdo con Cárdenas (1945), Cucurbitella asperata (sub Cucur- bita urkupinana) posee un tubérculo radical for- mado por la raíz principal. El desarrollo del parén- quima del xilema tejido reservante se conoce en los tubérculos radicales de Coccinia engleri Gilg y en los rizomas de Melothria argyrea A. Zimm. (Zimmermann, 1922). No se ha observado variación en la estructura de la raíz. secundario como TALLO La porción basal del vástago, que junto con la raíz, forma el sistema reservante, es perenne y porta las yemas hibernantes para la próxima estación de crecimiento. Todo el resto del vástago es anual y dura sólo el verano y el otoño. La porción anual de los tallos suele desarrollar un crecimiento secun- dario incipiente (Fig. 1H, I) sin formar felógeno. La epidermis, el colénquima y el clorénquima subepi- dérmicos асотраћап el crecimiento en diámetro. Los tallos con estructura primaria son cilíndricos, carecen de costillas y tienen 10 haces vasculares anficribales: cinco internos mayores y cinco exter- nos menores (Fig. 1G). La actividad del cámbium comienza primero en los cinco haces internos, y en esta etapa los tallos son generalmente 5-angulados o con cinco costillas (Fig. 1H). Cuando el cámbium comienza su actividad en los cinco haces externos, los tallos muestran un número mayor de costillas (Fig. 1I). Esta estructura del tallo es uniforme en Volume 85, Number 3 Pozner 427 1998 Revisión de Cucurbitella A IZA? AV YA VAR IZDA Figura 1. A. B. Tricomas glandulares foliares (Hoc 63). —C, J. Tricomas uncinulados (Hoc 63). —D. Plántula del tipo foliar “asperata” (Pozner 106). —E. Detalle de la sección transversal de un tubérculo radical (Pozner 87) F. Dos células basales de un tricoma uncinulado y su corona de células adyacentes (Pozner de un tallo joven (Hoc 63). —H. Idem С pero de un tallo más desarrollado (Pozner 87). —1. D transversal de un tallo hacia el final del verano (Cabrera 34534). —K, L, M, N. Sucesión foliar del t catafilo (К), hojas de transición (L, M), nomofilo (М) (MERL 42789). —O, P. 0, К. Sucesión foliar del tipo folia i,” catafilo (О), hojas de transición (P. Q), nomofilo (R) (Zuloaga 1305). En los esquemas С, H, I no se representa el colénquima y el clorénquima subepidérmico por razones de escala. 5). —G. Sección transversal e una secció e соп ipo foliar “asperata,” 428 Annals of the da Botanical Garden todo el material estudiado y relativamente frecuente en las Cucurbitaceae (Zimmermann, 1 PUBESCENCIA La superficie de los tallos y las hojas varía desde casi glabra (especialmente la cara adaxial de las hojas) hasta densamente hirsuta (en la abaxial de las hojas). Los tricomas más abundantes son uni- seriados, 3—5-celulares, con la célula apical unci- nulada (Fig. 1C, J) y las células anexas dispuestas en una corona que rodea la célula basal del tricoma (Fig. 1C , F), y sobresale por encima del nivel del resto de la epidermis. Estos tricomas uncinulados varían entre 60 y 200 шт (Fig. 1C, J), y son los responsables del tacto áspero de esta pubescencia. Por otro lado, y con menos frecuencia, hay tricomas glandulares, sin células anexas diferenciadas, con un pie l—4-celular y una cabezuela 2-8-celular (Fig. 1A, B). Por tratarse de las únicas estructuras secretoras de los tallos y las hojas, y aunque no se ha identificado la naturaleza química de su secre- ción, son los probables responsables del olor fétido típico de estas plantas. De hecho estos pelos glan- dulares faltan en el las flores, que son inodoras. HOJAS Con el objetivo de facilitar la descripción de la variación de la estructura foliar, se distinguen tres formas básicas de nomofilos a las que se hará re- ferencia como: Tipo “cucumifolia”: nomofilos enteros, cordiformes (incluye las formas foliares de C. cucumifolia y C. integrifolia, Fig. 2A, B); Tipo “durieut”: nomofilos 3-5-palmatipartidos con 22. enteros (incluye las formas foliares de C. dur . urkupinana, Fig. 2D, E); Tipo * са. ale 3-5- ва ти ние са соп óbulos pinnatifidos hasta casi disectos (inclu- ye las formas foliares de C. asperata, Fig. 2H, I La variación dentro de un mismo individuo se debe a la sucesión foliar y afecta la forma y el grado de división de la lámina Variación intra-individual. (los detalles de la variación de las brácteas y de las hojas florales se analizan bajo los títulos de In- florescencias y Flores). La sucesión foliar del eje principal parte siempre de cotiledones elípticos, trinervados, epígeos, y catafilos enteros 5-angula- dos (Fig. 1D). En los individuos con nomofilos tipo “cucumifolia” se suceden hojas de transición más o menos cordiformes o imperfectamente y levemen- te trilobuladas. En los individuos con nomofilos tipo “asperata” se suceden hojas de transición paulati- namente más divididas hasta los nomofilos profun- damente palmatipartidos con segmentos pinnati- partidos (Fig. 1K-N), y en los individuos con nomofilos tipo “durieut” las hojas de transición son 3—5-palmatipartidas y culminan con nomofilos 5- palmatipartidos con lóbulos enteros de ápice obtuso o redondeado (Fig. 10–К). En general, en los bro- tes hibernantes se repite la misma sucesión de ca- Пов y nomofilos observada en el eje principal, pero en las ramificaciones de estos brotes hiber- nantes casi no hay diferencias en la forma de los catafilos y los nomofilos. En algunos ejemplares con nomofilos tipo “durieui” o “cucumifolia” la suce- sión foliar es al revés y lleva de hojas lobuladas a enteras. La sucesión foliar observada en Cucurbitella es una de las más frecuentes en las Cucurbitaceae Zimmermann, 1922). A través del cultivo de al- gunos ejemplares (Pozner 62, 63, 65, 87, 106) en condiciones hídricas diferentes a las naturales se ha observado que la forma de la hoja de un indi- viduo no cambia con el régimen hídrico. Variación interindividual. La variación inter- individual de los nomofilos afecta la forma, el grado de división de la lámina, el margen, la consistencia y la pubescencia. =“ Entre los individuos con nomofilos tipo “сиси- mifolia," y aquéllos con nomofilos tipo “asperata,” existen individuos con todas las formas foliares intermedias, que se con nomofilos tipo “durieut” corresponden con las respectivas estructuras inter- medias de sus primordios foliares (Fig. 2A—I). La ontogenia foliar de los tres tipos foliares principales muestra a los primordios más jóvenes enteros y cor- diformes (Fig. 2a, d, р). Los nomofilos tipo “сиси- --» Figura 2. Primordios foliares. --А, B. Tipo foliar * 'cucumifolia" (Cristóbal 2100, 2” 3133 respectivamente), — C. Forma intermedia entre los tipos foliares * oo y “durieui” (A. T. Hunziker 4709). —D, E. Tipo foliar “durieui” ep vara 2884, Schulz 6323 respectivamente L —К, G. Formas intermedias entre los tipos foliares “durieui” y “asperata” (Venturi 7734 y 7665 а. —H. I. Tipo foliar “asperata” (Pedersen 15211, Krapovickas 14631 respectivamente e). —a, b, c. Tres е 2. e la ontogenia de un primordio del tipo foliar * gu cel 2. —d, e, f. Idem a, b, с pa ra un ы. del про foliar “durieui” (BAB 752730). —g. h, . b. € para un поа У del tipo foliar * ы > (Piccinini 1721). 429 Volume 85, Number 3 1998 430 Annals of the 2. Botanical Garden mifolia" mantienen esta forma inicial durante todo su desarrollo (Fig. 2а-с, A). Еп los nomofilos lo- bulados surgen, en la base del primordio foliar, pri- mero los dos lóbulos medianos y luego los dos lóbu- los basales (Fig. 24-4). Los nomofilos tipo “durieui” y “asperata” comparten más etapas en común du- rante su desarrollo que con los nomofilos *cucu- mifolia." Quizás por esta razón los individuos con tipos foliares intermedios entre “asperata” y “du- пеш” tienen una sucesión foliar poco marcada, mientras que los individuos con tipos foliares in- termedios entre “durieui” y “cucumifolia” una sucesión foliar marcada que combina ambos tienen tipos de hojas. Existe una gran variación en el mar- gen foliar, que puede ser desde casi liso con dien- tes remotos hasta serrado con dientes de bordes rectos, convexos o cóncavos (margen obcrenado en este último caso). Con excepción de las hojas casi disectas de algunos individuos, donde los segmen- tos de los lóbulos se corresponden siempre con los dientes, existen individuos con todas las combina- ciones posibles de formas foliares y tipo de margen. tamaño de los nomofilos de Cucurbitella varía entre 1 X 1 cm hasta 18 X 16 cm, y los ejemplares estudiados sugieren que el tamaño de los nomofilos depende del ambiente. En efecto, ejemplares de los suelos arenosos de la Salina de Mascasín, La Rioja (Piccinini 1721 y 1744), poseen nomofilos de sólo l cm de longitud y anchura. Algo similar ocurre con algunos ejemplares de suelos arenosos y pe- m de Salta (Novara 2884), con nomofilos de 1.5 cm. Incluso un mismo ejemplar puede combinar hojas muy pequeñas y grandes. La consistencia y la pubescencia de los hojas están afectadas por el ambiente. Individuos colec- cionados en Córdoba (600—700 mm de precipita- ción media anual, Pozner 62, 63, 65) con hojas rígidas y pubescentes, desarrollaron hojas delgadas y casi glabras cuando fueron cultivados en Buenos Aires (900-1000 mm de precipitación media a- nual) ZARCILLOS En la mayoría de las Cucurbitaceae el zarcillo es una estructura compuesta por una porción linar (unifacial), denominada portazarcillo (“Ran- kentrüger"), y una o más partes foliares (bifaciales y con la misma filotaxis que las hojas normales) que forman las ramas del zarcillo. La primera rama de estos zarcillos compuestos corresponde al profilo de la yema vegetativa lateral, que es la hoja tectriz del portazacillo. Los zarcillos de Cucurbitella son simples con prefoliación recta. De acuerdo con Ku- mazawa (1964), son de naturaleza foliar (con es- tructura bifacial en toda su longitud), sin portazar- cillo, y homólogos al profilo anódico de la yema vegetativa lateral. Estos zarcillos simples no son raros en las Cucurbitaceae, se conocen también en Cucumis y Trichosanthes (Kumazawa, 1964). cau- ESPORIDAD El período de floración se extiende desde no- viembre hasta abril, y su comienzo varía entre no- viembre y febrero según el inicio de la época llu- viosa en cada región. El material estudiado vivo y herborizado incluye ejemplares con flores carpela- das, con flores estaminadas o con ambos tipos de flores dentro de cada tipo foliar. Asimismo, indi- viduos que en condiciones naturales producen am- bos tipos de flores, una vez trasplantados sólo for- man flores estaminadas (cfr. Naudin, 1866). Los mecanismos que determinan la esporidad (“sex ex- pression”) en las Cucurbitaceae son complejos y variados (Condon & Gilbert, 1990; Delesalle, 1989; Roy « Saran, 1990). Por ahora es posible afirmar que Cucurbitella es siempre monoica. Pero no se sabe si las poblaciones combinan individuos per- fectos con individuos estaminados y/o carpelados. INFLORESCENCIAS Las flores carpeladas y las inflorescencias esta- minadas se forman generalmente en nudos dife- Figura 3. cículo sésil (B, ¿a 82), monocasio racimif Esquema en planta de (Более 82). —L, esquemática (N), (Pone 88). Q, К, S. Flor estaminada tipo * ‘asperata,’ H. Hunziker 1308 65). —V, X con saco Анја но (Pozner 106). —Y, 7. Fruto, vista later. B, C, D, los pedicelos triangulares indican que ~ — ve; Is . . Inflorescencia estaminada 2. 82). —B, . N. Flor cd tipo "nucum aba, * as glandulares de los pétalos (Pozn ; " vista lateral (Q), vista superior (R) y sección longitudinal esquemática (5) (J. --Т. Tricomas ampuliforme conectival (P. ті i . Flor carpelada, vista lateral (V), sección 2 esquemática (X) (J. Н a flor se ы һас > , D. Variación de la inflorescencia estaminada, fas- me (C, Agric 20592), Pe к ‘ulado (D, d 1, riación de las brácteas a inflorescencia estamin rien lateral (L), vista superior (M) y sección longitudinal er 65, Pozner 88 respecti ozner ricoma de la garganta del hipanto al (Y), sección сее (2), (Pozner 106). En los esquemas nte о hacia atrás (A) respecto del plano del esquema (se trata de la misma convención utilizada en los esquemas de las fórmulas moleculares). Pozner 431 Volume 85, Number 3 1998 Revisión de Cucurbitella 9000000009 0000000000 432 Annals of the Missouri Botanical Garden rentes de una misma rama. Las flores carpeladas son solitarias o excepcionalmente geminadas. En este ültimo caso las dos flores carpeladas carecen de brácteas y tienen distinto grado de desarrollo. Si se compara con la inflorescencia estaminada de este género y con las inflorescencias cimosas típi- cas de las Cucurbitaceae, estos casos de flores car- peladas geminadas pueden interpretarse como mo- nocasios bifloros. Las flores estaminadas se agrupan en monocasios (bóstrix) 3-10-floros (Fig. 3A—E), donde algunas veces sólo desarrolla una sola flor. Estos monocasios poseen un pedünculo de longitud variable entre 1 y 30 mm, y los nudos interflorales pueden estar reducidos (fascículo, Fig. 3B, D) o desarrollados (monocasio racimiforme, Fig. 3C). En general, cuanto más largo es el редипсшо del mo- nocasio, más cortos son los pedicelos florales y vi- ceversa (Fig. 3B, D), de modo que la longitud total de la inflorescencia estaminada siempre es igual o menor que el pecíolo de su hoja tectriz. Las flores estaminadas marchitas se desprenden de su pedi- celo, que persiste y se engruesa sobre la inflores- cencia. En los ejemplares con monocasios racimi- formes, los entrenudos que separan a las flores desarrollan a medida que las flores se abren y mar- chitan (Fig. 3C). Por esta causa un mismo ejemplar puede tener fascículos (monocasios que recién co- mienzan a abrir sus flores) y monocasios raci- miformes (monocasios que están abriendo sus úl- timas flores). Además de este caso particular, la longitud del pedúnculo, de los entrenudos y los pe- dicelos florales suele ser variable dentro de un mis- mo ejemplar. Las brácteas de las flores estaminadas pueden estar presentes o ausentes en distintas in- florescencias del mismo ejemplar. En general las brácteas son pequeñas, subuladas a lineares (Fig. 3F, G), y pueden pasar desapercibidas. En algunos ejemplares las brácteas son foliosas (Fig. 3H-K). FLORES El cáliz tiene prefloración abierta (Fig. 3A). Los sépalos varían entre lineares a triangulares y, del mismo modo que la cara externa del hipanto, están cubiertos por pelos uncinulados. La corola tiene prefloración coclear distal (sensu Weberling, 1992), os pétalos son orbiculares, elípticos и oblongos (Fig. 3M, R), están libres entre sí o levemente sol- dados en su base. Los pétalos tienen vernación pla- na en su porción basal y media, y vernación invo- luta en el ápice. Su color varía entre el blanco, blanco-verdoso, amarillo, amarillo-verdoso y amari- llo-anaranjado. La cara adaxial de los pétalos está cubierta de tricomas glandulosos, moruliformes, con un pie 1–2-зепадо 2-8-celular y una cabezue- la con 3 a 6 células de diámetro, ricas en almidón compuesto (Fig. ЗО, P). La base de los estambres y los estaminodios en las flores carpeladas) y la garganta del hipanto sobre la cual se insertan, está rodeada por tricomas uniseriados, 6—10-celulares ~ tarostegio que cierra la porción del hipanto donde se acumula el néctar. El nectario es mesenquimá- tico, formado por la base del hipanto engrosada. En las flores estaminadas el androceo es claramente 2+2+1. Cada estambre conserva su hacecillo vas- cular y puede ser separado de su par por simple tracción. El conectivo de las anteras posee pelos ampuliformes explosivos (Fig. 3T), cuyo producto de liberación actáa como un aglutinante del polen, como ocurre también en otros géneros de Cucur- bitaceae (Zimmermann, 1922; Vogel, 1981; Dierin- ger & Cabrera, 1994). Los granos de polen son tri- colporados reticulados, con un diámetro ecuatorial de 42 a 54 um y polar de 61 a 71 шт (Marticorena, 1963). No hay pistilodio. Las flores carpeladas tie- nen un hipanto campanulado, cinco estaminodios mameliformes, y un gineceo 5-carpelar (Fig. 3V, X, ). El ovario contiene numerosos rudimentos se- minales horizontales, el estilo es recto y el estigma está ubicado al nivel de la garganta del hipanto (Fig. 3V, X) y formado por 10 ramas cilíndricas (2 por carpelo) unidas en pares entre carpelos vecinos (estigmas comisurales, sensu Weberling, 1992). La superficie del estigma está cubierta de emergencias papiliformes. Tanto las flores estaminadas como las carpeladas se abren desde la mafiana temprana y sólo duran abiertas un día. Así como la estructura floral carpelada es uni- forme, existe una amplia variación del hipanto y la posición de los estambres en la flor estaminada (Fig. 3L-N, Q-S). En general el hipanto es cam- panulado, raramente es tubular y excepcionalmente urceolado. Los estambres pueden insertarse sobre la garganta del hipanto y estar exsertos, o bien se insertan en la mitad superior del hipanto y están completamente insertos (Fig. 3N, S). Existen indi- viduos con flores con todas las formas descriptas de hipanto e inserciones y exserciones intermedias de los estambres. FRUTOS El fruto de Cucurbitella es una baya esférica, levemente deprimida o alargada, con epicarpo gla- ro o pubescente, verde con manchas blancas o verdoso-blanquecinas, alineadas en vetas longitu- dinales (Fig. 3Y). El meso- y el endocarpo son mucilaginosos y de color verde intenso. Los frutos Volume 85, Number 3 1998 Pozner Revisión de Cucurbitella alcanzan su madurez, generalmente después de que los tallos y las hojas de esa temporada se han mar- chitado. Segán Martínez Crovetto (1965, sub C. du- rieui), los frutos son amarillentos a la madurez, ras- go se ha observado en ninguna de las plantas o. Definitivamente se trata de una observación errónea pues el material citado por Martínez ula (1965) corresponde a Apodan- thera sagittifolia (Griseb.) Mart. Crov. var. dissecta (Cogn.) Mart. Crov. ue no SEMILLAS Las semillas son ovoideas, comprimidas, lisas y de color castaño. Están envueltas por una porción del endocarpio, denominada saco ariloideo (Font Quer, 1982), que se separa del resto del tejido del ruto y rodea a cada semilla como si fuera un arilo (Fig. 3W). El saco ariloideo es mucilaginoso, rico en sustancias pécticas, de color verde y adhesivo. Por tales causas habría que considerar la endozoo- coria y la epizoocoria por mixospermia en la dis- persión de las semillas. RELACIONES RELACIONES INTERGENERICAS De acuerdo con la clasificación más reciente (Jeffrey, 1990), el género Cucurbitella está inclui- do dentro de la subfamilia Cucurbitoideae, tribu Melothrieae, subtribu Dendrosicyinae. No ha opiniones publicadas sobre las relaciones entre Cucurbitella y los géneros restantes de las Dendro- sicyinae. En el presente estudio se considera que Cucurbitella está definido por el conjunto de las siguientes características: flor carpelada con gi- < la. Androceo formado por dos estambres dobles (A 2+2) neceo 5-carpelar, estigmas bisectos (en total 10 ramas estigmáticas lineares), y 5 estaminodios; pedúnculo de la inflorescencia estaminada igual o menor que el pecíolo de la hoja tectriz; y flor es- taminada con estambres dobles separables por tracción en sus componentes y conectivo con pelos ampuliformes explosivos. Ninguno de estos carac- teres es una autapomorfía de Cucurbitella. Toma- dos individualmente, estos caracteres relacionan a ucurbitella con Apodanthera, y con un grupo de géneros afines a Apodanthera, como Wilbrandia, Guraniopsis y Melothrianthus, cuyos límites y re- laciones deben ser revisados (cfr. Martínez Cro- vetto, 1954a; Jeffrey, 1978). El género Apodant- necesita una revisión integral, particularmente las especies de la sect. Apodant- hera, para algunas de las cuales se desconoce la información de las flores carpeladas. La relación de Cucurbitella con Apodanthera parece ser muy estrecha, particularmente con A. ferreyrana Mart. Crov. (sect. Apodanthera), por sus cinco estigmas ífidos, y con A. lasiocalyx Cogn. (sect. Apodan- era thera), por sus cinco estaminodios (carácter que está relacionado con la posibilidad de separar los presentes en A. undulata А. Gray topsis) pero son frecuentes también en otros gé- neros de Cucurbitaceae (Zimmermann, 1922). este contexto, Cucurbitella se distingue de Apo- danthera sects. Apodanthera y Cucurbitopsis por sus monocasios cortamente pedunculados, y de Apodanthera sect. Pseudoapodanthera por sus flo- res carpeladas solitarias. А continuación se agrega una clave para reconocer los taxones principales americanos de las Dendrosicyinae: Guraniopsis lb. Androceo formado por dos estambres dobles y uno simple (A 2+ 2a. Flores estaminadas sésiles, en monocasios espiciformes; 1. circinados еп la Wilbrandia 2b. Flores estaminadas pediceladas, en monocasios racimiformes o en fascículos жч кон solitarias); zar- cillos rectos en la yema. За. Pétalos bipartidos, los lóbulos circinados en el capullo ..... 3b. Pétalos enteros, no и en el capullo. di _ Ceratosanthes Melothrianthus 4b. Anteras dorsifijas 5a. Ovario y fruto t el tegumento escrobiculado 5b. Ovaro y fruto lo comprimidas, con el tegu 6a. Flores carpeladas en Кемеш f ransversalmente oblongos; rudimentos seminales 8; semillas piriformes, con Halosicyos ее oblongos; rudimentos seminales м ó más; semillas a nto liso o con excrecencias eun к. = sect. Pseudoapodanthera 6b. Flores earpeladas solitarias (raro geminadas eceo 5-carp nada menor o igua elar, los estigmas 5, ее pedúnculo de la inflorescencia estam ue el pecíolo de la hoja ен. Cu кек Tb. Gineceo bi- o tricarpelar, los estigmas 2 6 3, bisectos о de formas variadas; pedün culo de la inflorescencia estaminada mayor que el pecíolo de la hoja tectriz (ex- cepciones: Apodanthera ferreyrana, con gineceo 5-carpelar y 5 estigmas bífidos; 434 Annals of the Missouri Botanical Garden Apodanthera anatuyana (Mart. igual al pecíolo de la hoja tectriz). estaminada Ва. Hojas membranáceas 8b. Hojas suculentas RELACIONES INFRAGENERICAS La revisión de los caracteres morfológicos de Cu- curbitella muestra que la estructura de la raíz, del tallo, del fruto y de la semilla es uniforme entre los ejemplares estudiados. El tamafio, la pubescencia y la consistencia de las hojas están estrechamente relacionados con los factores ambientales. La inflo- rescencia estaminada, en cuanto al desarrollo del pedünculo, los entrenudos, los pedicelos florales y las brácteas, varía dentro de un mismo individuo. Sólo la forma del primordio foliar de los nomofilos, la exserción de los estambres, la longitud del hi- panto y de los pétalos son los caracteres constantes en cada ejemplar y más variables entre ejemplares. En todos los casos la variación de estos caracteres es continua y no correlacionada (—0.34 < índice de Spearman « 0.23 y —0.35 « índice de Kendall .18). La distribución geográfica de las frecuen- cias relativas de las formas foliares primordiales muestra que los individuos con primordios foliares de про “asperata” (Fig. 2H, I) son más frecuentes en el centro, oeste y una pequeña parte del este de la Argentina, dentro de las provincias fitogeográfi- cas del Monte, Espinal y el sur del distrito Cha- quefio Serrano de la provincia Chaquefia (Fig. 4). En esta misma área son más frecuentes los indi- viduos con estambres exsertos. Hacia el norte son más frecuentes los individuos con primordios de transición entre los tipos “asperata” y “durieui,” y cuyas frecuencias dominan en las provincias fitogeográfi- con primordios foliares del tipo “durieui,” cas Prepuneña y Puneña, Pampeana y el distrito Chaqueño Occidental de la provincia Chaqueña (Fig. 4). Dentro de estas mismas áreas fitogeográ- ficas predominan los individuos con estambres semiexsertos. Los individuos con primordios de tipo “cucumifolia” dominan en las provincias fitogeo- gráficas de las Yungas, Paranense y en el norte del distrito Chaqueño Serrano de la provincia Chaque- ña (Fig. 4). Asimismo predominan los individuos con estambres insertos. El distrito Chaqueño Ori- ental de la provincia Chaqueña combina frecuen- cias similares de individuos con primordios foliares “cucumifolia” y “durieui.” La longitud del hipanto tiene su máxima frecuencia entre los 5 y 7 mm en toda la distribución del género, pero el rango de variación aumenta de sur (4-6 mm) a norte (4-12 mm). Aunque ciertas formas foliares predominan en Crov.) Pozner; con pedúnculo de la inflorescencia ens Apodanthera sect. Apodanthera Apodanthera sect. Cucurbitopsis cada una de las áreas fitogeográficas mencionadas, casi todas las formas foliares restantes están tam- bién presentes (Fig. 4). Dentro de esta variación, los tipos nomenclaturales de C. asperata, C. cucu- mifolia y la ilustración de C. durieui publicada por Naudin (1866, pl. 2, sub Prasopepon durieui), co- rresponden a formas foliares bien diferenciables. El tipo nomenclatural de C. urkupinana coincide con la ilustración de C. durieui publicada por Naudin (1866, pl. 2, sub Prasopepon durieui), y el tipo no- menclatural de C. integrifolia con el de C. cucu- mifolia. El tipo nomenclatural de C. ecuadorensis queda excluido del género (véase más abajo). Este análisis de caracteres permite establecer un único taxón con un amplio rango de variación con- tinua, donde es arbitrario colocar límites internos. La distribución geográfica de las frecuencias de los caracteres sugiere la distinción de tres grupos o razas: 1. Estambres exsertos, pétalos largos (8—12 mm), hipanto corto (4—5 mm), primordios foliares H y I (Fig. 2). Se incluye aquí el tipo de C. asperata. Estambres semiexsertos, pétalos medianos (6—8 mm), hipanto mediano (6-7 mm) y primordios foliares D y E (Fig. 2). Se incluyen aquí el tipo e C. urkupinana y la ilustración de C. durieui publicada por Naudin (1866, pl. 2, sub Praso- pepon durieui). 3. Estambres insertos, pétalos cortos (4—6 mm), hi- panto largo (8-12 mm), primordios foliares J y в. 2). Se incluyen aquí los tipos de С. cucu- mifolia y C. integrifolia. N Los diferentes tipos de razas poseen una distri- bución geográfica de sus poblaciones y una distri- bución de las frecuencias de las variantes polimór- ficas dentro de cada población, que les son características (cfr. Grant, 1989). El presente es- tudio de frecuencias de las formas foliares primor- diales se ha basado en ejemplares de herbario y no en censos de poblaciones. A pesar de esta limita- ción, la distribución de las frecuencias de las for- mas foliares primordiales de este género tiene un patrón en mosaico (Fig. 4) característico de la dis- tribución de las razas ecológicas (Grant, 1989), con la particularidad que existe un cambio paulatino de frecuencias en las direcciones norte-sur y oeste- este. En general, cuanto más árido es el ambiente, son más frecuentes los ejemplares con hojas tipo "asperata," con menor tamafio de las hojas, y pu- Volume 85, Number 3 1998 435 Pozner Revisión de Cucurbitella Chaco Occidental (IX) Yungas (1) Chaco Serrano Norte (X) 188 13,64 13,64 y, E П 45500 C D E ған |! 200 S— Prepuneña - Рипећа (II) ған |! ($ | 7 ~ ' j 30 ) 2 FA i x г /f VI) 3 o "n у A a 0 РОД | VIII „7 ! { 30° S— ; & | i VAL Chaco Serrano Sur (ІІ 3 t 17” Paranense (VII m | \/ (URUGUAY 7 — 1 90 / | ARGENTINA ю ш { 60 a, 50 I о \ IV 20 f Е оо 0 o { E FG H |! b | Monte (IV) 609 W 509 уу Espinal = Pampeana (VI) 30 20 10 0 BRASIL У Chaco Oriental (VIII) 29092727 25,45 Frecuencias relativas de los tipos de p apri ilustrados en la Figura 2, calculados para las a 4. 4 B um provincias y distritos fitogeográficos donde se dist e el género Cucurbitella. Los límites de las provincias 2. fitogeog están marcados con línea Пећи у зе nh basado en Cabrera y Willink (1980), Morello y ] 3). Cada gráfico de frecuencia está асотраћадо por el nombre 1968), емін et al. (199 área en el mapa. Las provincias r de la provincia Altoandina para кш. ar la representación gráfica. Los ун políticos internacionales están indi- as foliares ilustradas cados con línea interrumpic en la Figura 2, de modo que la ba ida. Las abreviaturas A, B, C, D, arra AB corresponde a la frecuencia и de las formas foliares А y Bc , E, М rresponden а las form e la Figura 2 para ans una de las áreas fitogeográficas en cuestión, y así sucesivamente bescencia más densa. A pesar de que los indivi- duos con primordios foliares tipo son dominantes en el centro y oeste de la Argentina (clima regional árido o semiárido) su presencia, y la de individuos con formas de transición, alcanza el norte de la Argentina (clima regional subtropical) en hábitats localmente áridos por sus condiciones edáficas (salares, bosque xerófilo, médanos y pedre- ales). Y viceversa con los individuos con primor- dios foliares de tipo “илеш” y “cucumifolia.” La variación continua de los caracteres en este género coincide con una variación paulatina de la fre- cuencia de hábitats áridos regionales con ambien- tes locales más hámedos a subtropicales regionales con ambientes semi-áridos locales, según las con- diciones del suelo. El cultivo de individuos en condiciones hídricas distintas a las de su lugar de origen mostró que la forma foliar tiene una base genética y no es una respuesta morfogenética al ambiente. La distribu- ción de las frecuencias de los caracteres morfoló- gicos sugiere que el género Cucurbitella comprende una ünica especie polimórfica, C. asperata (Gillies ex Hook. & Arn.) Walp., con tres posibles razas Annals of the Missouri Botanical Garden ecológicas. Según Stace (1993), las razas ecológicas suelen denominarse bajo la categoría de variedad. En este caso particular se considera que la deno- minación formal de estas razas de Cucurbitella as- perata no tiene aplicación práctica debido a los nu- merosos individuos con variantes intermedias. Por el mismo motivo carece de sentido proponer una clave para distinguir las razas. En consecuencia se establece el siguiente tratamiento taxonómico. TRATAMIENTO TAXONÓMICO Cucurbitella Walp. [“Curcubitella”], Керегі. Bot. Syst. 6: 50. 1846. TIPO: Cucurbitella asperata (Gillies ex Hook. € Arn.) Walp. (Cucurbita as- perata Gillies ex Hook. & Arn.) Schizostigma Arn., Madras J. Lit. Sci. 12: 50. 1840, non rn. ex Meisn. 1838. TIPO: d asperatum (Gillies ex Hook & Arn.) Arn. (Cucurbita asperata Gillies ex Hook & Arn ки Naudin, Ann. Sc п. Nat., Bot ‚ 5: 26. 1866. TIPO: Prasopepon dac Naudin i 2. Нетісгіріб ов diclimonoicos, fétidos. Raíz re- servante, napiforme, con tuberosidades esféricas u ovoideas, ritidoma con lenticelas pulviniformes, en general alineadas horizontalmente. Vástago anual con pubescencia áspera de tricomas uncinulados y glandulares. Porción basal de las ramas perenne, portadora de las yemas hibernantes. Hojas simples, suborbiculares, enteras a 3—7-palmati- o pedati- partidas, base cordada, segmentos enteros o pin- natipartidos, hasta disectos; margen serrado, obcre- nado, o con dientes remotos. Zarcillos simples, de vernación recta. Flores estaminadas en monocasios (bóstrix) sésiles o cortamente pedunculados 2-10- floros, fasciculiformes o racimiformes; brácteas su- buladas o lineares, foliosas o ausentes. Hipanto tu- bular a infundibuliforme; garganta vellosa. Cáliz abierto; sépalos triangulares a subulados. Corola imbricada (coclear distal); pétalos amarillos, ama- rillo-anaranjados, amarillo-verdosos, blancos o blanco-verdosos, cara adaxial con pubescencia glandulosa. Androceo 2+2+1; estambres exsertos, insertados en la garganta del hipanto, hasta inclu- sos, insertados en la mitad superior del hipanto; anteras dorsifijas, levemente adheridas entre sí; te- cas rectas; filamentos breves; conectivo con pelos ampuliformes explosivos cuya secreción actúa como aglutinante del polen; polen 3-colporado re- ticulado; pistilodio ausente. Flores carpeladas so- litarias, raramente geminadas. Hipanto, cáliz y co- rola como en la flor estaminada. Estaminodios 5, mameliformes. Gineceo 5-carpelar; ovario ovoideo o fusiforme, con numerosos rudimentos seminales horizontales; estilo recto; estigma divido en cinco ramas bisectas, segmentos lineares cubiertos de emergencias papiliformes. Baya subesférica, verde con vetas longitudinales formadas por máculas blanquecinas. Semillas aovadas, comprimidas, par- das, lisas, envueltas en un saco ariloideo verde, mucilaginoso y adhesivo. l. Cucurbitella asperata (Gillies ex Hook. & Arn.) Walp., Керегі. Bot. Syst. 6: 50. 1846. Cucurbita asperata Gillies ex Hook. & Arn., ot. Misc. 833. Schizostigma aspe- ratum (Gillies ex Hook. & Arn.) Arn., Madras J. Lit. Sci. 12: 50. 1840. TIPO: Argentina. Mendoza: “in travesia or uncultivated places,” Gillies s.n. (lectótipo, aquí designado, GL; fo- afía SI) togr Prasopepon durieui Naudin [“duriaei”], Ann. Sci. Nat Bot. sér. 5, 5: 26. 1866. Cucurbitella durieui (Nau- din) Cogn., en Mart., Fl, bras. 6(4): 40. 1878. TIPO: "Hab. In regione uruguayensi Americae australis. Р? ~ m mie viva in hortum parisiensem transmigravit" no visto алерт cucumifolius Griseb., Pl. lorentz. 98, Abh. Kö- igl. Ges. Wiss. Góttingen 19: 98. 1874. gregi Aer (Griseb.) Cogn., en ii Ph oras. 6(4 70. 1878. TIPO: Argentina. Tucu “freque ens in fruticetis et sepibus pr. Siambon,” olo 309 (ho- lótipo, GOET; fotografía de F-8987, S е a 1. ^ A. DC. &( . DC., Mo- лорт. phan TIPO: Patios: ^ TAs- RU Jars i "А Balansa 1112 (lectótipo, aquí designado, K; pacis afía, SI). Cucurbita РА шеи», Revista Agric. (Cocha- 76. М a: Cerro del Calvario near Qui- llacollo, 2560 msnm, ск colinas secas con arbus- los y piedras, XII-1945, Cardenas 3600 (isótipo, SI). Tubérculos radicales hasta de 20 cm de diáme- tro. Tallos plurimetrales. Pecíolo de 1.5-10 cm; lámina de 1-18 X 1-16 cm. Inflorescencia esta- minada con pedúnculo de 1-30 mm; pedicelos de mm. Flor estaminada con hipanto de 4-12 X 3—6 mm; sépalos de 2-6 mm; pétalos de 7-12 X 3-5 mm; anteras de 3 mm. Flor carpelada con pe- dicelo de 10-55 mm; ovario de 8-15 X 3-6 mm y estilo de 4 mm. Baya de 30-35 х 30-40 mm. Se- millas (sin saco ariloideo) de 5-6 x 2-3 x 0.75- Nombres vulgares. Sandillo del Campo (Men- doza: Hooker € Arnott, 1833), Sandía del Campo (Catamarca: Jórgensen 1247), Sandía de la Zorra Mendoza: Ruiz Leal 8530, San Juan: Spegazzini 344), Sandía del Zorro (San Juan: Haene 162, La Rioja: Krapovickas 5931), Sandía del Diablo (Bue- — Volume 85, Number 3 1998 Pozner Revisión de Cucurbitella nos Aires: Schulz 9372, Mendoza: Kurtz 1339, Tu- cumán: Venturi 321), Angola de Zorro (Jujuy: Krapovickas 17619), Sandía de la Víbora (Salta: Sa- ravia Toledo 1532), Zapallito (Santiago del Estero: Pire 1115), Zapallito de la Víbora (Catamarca: Troncoso 1865), No'otalán (Paraguay: Arenas 1750). Distribución. Argentina, Uruguay, Paraguay, Bolivia y Sur de Brasil. El género Cucurbitella no está incluido en la Flora Chilena (Мићог Pizarro, 1959; Navas, 1979). La cita de Arnott (1841) para Chile se considera dudosa, basada en una confu- sión en la ubicación de la provincia de Mendoza, pues Arnott sólo menciona los ejemplares de Gil- lies. Asimismo la presencia de este género parece dudosa en Brasil. Aunque se mencionó en la Flora brasiliensis (Cogniaux, 1878), no figuró en la Flora Ilustradada Rio Grande do Sul (Porto, 1974). Hasta el momento, los únicos ejemplares citados para Brasil son los de Engler, Arechavaleta (Augusto, 1946) y Sello 896, 897 (Cogniaux, 1878). El ma- terial citado para Uruguay fue publicado por Cog- niaux (1878), Herter (1930) y Martínez Crovetto Material representativo examinado. ARGENTINA. Buenos Aires: San Pedro, 29 dic. 1945 (fl, fr), 3550 (SI). Catamarca: Andalgalá, La Playa, 25 feb. 1916 (fr). ш! 1247 (SI). : Punilla, falda oeste de la Sierra Chica, ruta 1961 (i) A. E. Cocucci "к (CORD). a das, San Lorenzo, 19 ene. 1983 (fl, fr), purs: 13490 (CTES). Chaco: 1го. de Мо, Colonia Benítez, 20 ene. adf (8), Schulz 16497 (51). Entre Ríos: a bajada nde al sur de la ciudad de Paraná, 4 feb. 1973 (fl), Burkar 29608 (51). Еогтова: Pilcomayo, co Nacio- . 1988 (fl, fr), Guaglianone 2291 ~ S 5 me е = æ С A 2 WN e Б, П Tala y Chamaico, 25 Rioja: Chilecito, Valle à D Talas, : ene. pe Sosa s.n. (MERL 1 endoza: Las Heras, Parque aborigen, 4 dic. A 2 iv Leal a (MERL). Salta: Orán, Tabacal, 4 feb. 1943 (fl, A. T. Hunziker 2778 — fi, fr), n: Jac Roque y Fuerte, 3 dic. 1937 (fl). Spegazzini 344 (BAB). San Luis: Ayacucho, 5 ene. 1944 (fl), Burkart 13957 (SD). Santa San Jerónimo, Arroyo Colastiné, ruta 11, 15 km al sur de Coronda, 27 ene. 1971 (fl, fr), Krapovickas 17801 (BAA). Santiago del Estero: Carlos Pellegrini, Cerro del ic. 1927 (fl), Venturi 5681 (Sh. T Tafí, Yerba oeni, 3 ene. 1919 (fl), Venturi 321 (51). Cruz: Andrés Ibáñez, 3 km al sudoeste de Angostura, 25 ene 5. (fl. fr), Nee 38825 (СТЕ5); Cor- ta Cru al sur del Río Grande, 12 mar. 1981 (fl, fr), Beck pes (51); dise. 10 km al nornoroeste de Maratal a San Juan del Potrero, 30 ene. 1994, Nee & 83 (SI); Florida, Maidana. 31 ene. 1984 (fl, fr), a Bermejo, 22 nov. 1986 (fl), Ehrich 242 (SD; Gran Chaco, Villa Montes, 11 abr. 1977 (fr), povickas & Seal 31180 (51); entre Narbáez y Entre Ríos, 24 oct. 1980 (fl), Zuloaga et al. 1305 (SI). PARA- GUAY. Alto Paraguay: Puerto Diana, 6 km al norte de Bahía Negra, 7 ene. 1974 (fr), Arenas 300 (SI). Boque- 16 1 (fl), Arenas 1750 rans-Chaco, 21?26'S, 61%25'W, 7 mar. 1979 (fr) Sehinini & Bordas 16402 (SI). Presidente Hayes: Co- nia menno, 28 ene. 1976 (fl), Arenas 1450 (SI). Iconografia. Cogniaux, 1878: tab. 19; Cog- niaux, 1916: 232, fig. 51; Martínez Crovetto, 1974: 71-72, figs. 31-32; Roig, 1981: 143, fig. 86; Ca- brera, 1993: 504—505, figs. 205-206. Conviene aclarar aquí que en la ilustración de C. asperata publicada por Martínez Crovetto, 1965: 395, fig. 130A y Cabrera, 1993: 501, fig. 204A, se ha ba- sado en un ejemplar de Apodanthera sagittifolia Griseb.) Mart. Crov. var dissecta (Cogn Crov. (probablemente Cabrera 2062, LP) En la publicación original de Cucurbita asperata no se citaron explícitamente ejemplares de herbario sino una breve nota de campo: “uncultivated places in the province of Mendoza, J. Gillies, nom. vernac. Sandillo del Campo." En dicha publicación se acla- гб además que Gillies encontró una segunda varie- dad: “... that growing in travesia, or more arid places, has the segments of the leaves less deeply sinuated than the other," pero sin asignarle un nombre. Se han encontrado tres ejemplares de Gil- lies identificados como Cucurbita asperata: ~ 1. Un primer ejemplar (CL) cuya etiqueta dice “Cucurbita asperata n. sp. Gillies. In Travesia or arid uncultivated places in province of Men- doza. J. Gillies," con la anotación B/H 751471 en la cartulina, y cuyas hojas responden a la descripción de la variedad con los segmentos menos profundamente sinuados. 2. Un segundo ejemplar (K) cuya etiqueta dice *Cucurbita asperata n. sp. var. dissecta Gillies, Sandillo del Campo nom. vernac. Jarillal or un- cultivated places. Mendoza. J. Gillies" y cuyas hojas tienen segmentos más profundamente si- nuados que el ejemplar anterior. La cartulina de este ejemplar también posee la inscripción B/H 751471 e incluye una descripción latina ma- nuscrita en el ángulo superior derecho. . Un tercer ejemplar (GL) con ambas etiquetas (1 y 2) pegadas en la misma cartulina, y cuyas ho- jas se ajustan mejor al ejemplar 2. Segün las etiquetas de los ejemplares 1 y 2, la variedad típica de C. asperata corresponde al ejem- lar 1. Y aunque tales variedades no tienen valor (nomenclatural ni taxonómico), permiten conocer la intención del autor. Por este motivo se designa como lectótipo de Cucurbita asperata Gillies ex 438 Annals of the Missouri Botanical Garden Hook. & Arn. al ejemplar 1, pues coincide con la variedad típica en el sentido de Gillies. En cuanto a Cucurbitella integrifolia Cogn., se designa como lectótipo el sintipo Balansa 1112 (K) por ser más completo que Gibert 73 (K). El material tipo de Prasopepon durieui Naudin fue solicitado al Muséum National d'Histoire Na- turelle (P), pero no pudo ser localizado con certeza . Aymonin, com. pers.). Sin embargo, la ilus- tración publicada por Naudin (1866: pl. 2) permite identificar sin dudas las características del material tipo de Prasopepon durieui. NOMBRE EXCLUIDO Cucurbitella ecuadorenis Cogn., en Engl, Pflanzenr. V. 275 I (Heft 66): 233. 1916. = Posadaea sphaerocarpa Cogn., Bull. Acad. Roy. Sci. Bel- gique sér. 3, 20: 477. 1890. TIPO: Ecuador. “Felsen am ев Pilaton 900 m. diro 516 (holótipo, BR). и. M.,” So- Literatura Citada Arnott, G. A. W. 1840. Remarks on the no E the nat- ural order Cucurbitaceae. Madras J. Lit. . 12: 43- 5 ----- 1841. On the Cucurbitaceae. J. Bot. (Hooker) 3 271-280. Augusto, Hermano. 1946. Flora do Rio grande do Sul. Oficinas Grafic ps da Imprensa = ‘ial. P orto Ale egre. Cabrera, A. L. 1993. Cucurbitaceae. En: A. L. Cabrera eeen Flora la provincia de pom С iden | i. Inst. Nac. ecnol. Agropecu. 13(9): 472-515. А Willink. 1980. Biogeografía de América La- tina. Colección de Monografías Científicas de la Organi- zación de Estados Americanos, Secretaría General de la OEA, serie Biología, monografía nro. 13. Cárdenas, M. Notas sobre taxonomía de plantas económicas de Bolivia. Revista Agric. (Cochabamba) 3: 76-1 76–77. Cogniaux, A. 1878. Cuc ric» ‘eae. En: C. F. P. Martius, Flora perde d x F. Fleiac he Leipzig. . 19 a 2. t Melothrieae. En: A. E nales dio Das Pflanzenreic У IV 275 I (Heft 66): 1- d eipzig Condon, M. A Gilbert. 19€ 90. Reproduc tive bi- ology jua: natural bu of the Aur 'al vine rania and Psiguria. Pp. 150-166 en D. M. рајем | W. Robinson & C. Jeffrey, Maid an Utilization atthe Cucurbitaceae. Cornell Univ. Press, Ithaca Delesalle, V. 1989, Та и changes in phenotyp- ic gender in a monoecious cucurbit Apodanthera un- : 30–39. p e 1994. Sexual selection of anther tric omes and sexual Шш іп /bervillea lindheimeri (Cucurbitac с Мал hrieae). Amer. J. Bot. 1: 111-118 bes bani Н 1982. Diccionario de Botánica, 8 ed. Labor, ‘elon Сов V. 1989. Evolución Vegetal, 2 ed. Limusa, Méxicc Herter, G. 1930. Florula Uruguayensis. Plantas а 'u- lares. Imprenta Nacional, Montevideo. Troll, Hooker, W. J. . W. Arnott. 1833. Contributions towards a ns ot у! America and the Islands of the Pacific. Bot. Misc. 3: 129-211, 303-367. Jeffrey, С. 1978. Further notes on Cucurbitaceae: IV, Some New World Taxa. Kew Bull. 33: 347—380. 990. An outline classification of the Cucurbi- taceae. Pp. 449—463 en ates, R. W. Robinson & C. Jeffrey, Biology and Utilization of the Cucurbita- ceae. Cornell Univ. Press, Ithaca. Jensen, W. A. 5 Histochemistry. W. Н. Freeman, San 2. Kumazawa, М. 1 ее ТИН interpretations of ax- illary organs in the Cucurbitaceae. Phytomorphology 14: 287-298. Marticorena, C. 1963. Material para una monografía de la morfología del polen de Cucurbitaceae. Grana Paly- nol. 4: 78-91. artínez Crovetto, R. 1954a. Sur les organes femelles de quelques espéces du genre Apodanthera (Cucurbita- ceae). 1. Syst. (Paris) 15: 41-43. 954b. Synopsis des Cucurbitacées de l'Uru- guay. end. Syst. (Paris) 15: 47-55. 1965. Cucurbitaceae. En: A. L. Cabrera, le de la: provincia de Buenos a Colecc. Ci. Inst. Nac Tecnol. Agropecu. 4(5): 390-407 74. Cucurbitaceae. En: А. trada de la provincia de Entre Ríos. Nac. Tec nol. 2... 6(6): 63-94. Meisner C. F . Plantarum vasc d genera, Vols. .L 1. 2. Leipz © Es C. R. & L. Chalk. 1930. т Шашлы of the Di- cotyledons. Clarendon Press, Oxford. Morello, J. & J. Adámoli. 1968. Las grandes unidades de vegetación y ambiente del Chaco argentino. Primera parte: objetivos y E Ser. Fitogeogr. Minist. Agric. Ganad. 10: 1-12 . 1959. [m de la Flora Chilena. тн de Chile. айй, M. Ch. 1866. Cucurbitacées nouvelles cultivées au Muséum d'Histoire Naturelle еп 1863, 1865. Ann. Sci. Nat., Bot. зет. 5, 5: 543, pl. 1-6. s Bustamante, L. E. 1979. Flora de la Cuenca de Santiago e 2” Ed. Universidad de Chile, 3 vols. Porto, M. I . Cucurbitaceae . R. Schulz (ed- itor), Flora 1. do Rio Conde de Sul. Bol. Inst. Centr. Bioci. Univ. Fed. Rio Grande do Sul, Sér. Bot. 31: 1-64 Burkart, Flora Ilus- Colecc. Ci. Inst. 64 et Prado, D. 993. What is the Gran Chaco vegetation in South reves 'a? 1. A review. Contribution to the study i 2” and vegetation of the Chaco. Candollea 48: 145- @ e A. 1981. Flora de la Reserva Ecológica de Na- cufián. Cuad. Técn. Inst. Argent. Invest. Zonas Aridas 3-80. Roy, R. P. & S. Saran. 1990, Sex 22. in the Cu- LES MANN Pp. 251-268 en D. M. Bates, R. W. Rob- inson & C. Jeffrey, Biology and ШЕШ of the Cu- d не Cornell Univ. Press, Itha Ruiz Leal. 1975. 3: 1-299, | Stace, C. А. 1993. Plant Taxonomy and Biosystematics, 2nd ed. Cambridge Univ. Pr .K. Spichiger R Flora Popular Meine Deserta a R a Ramella. 1995. Origin, affinities and diversity of hot € of the dx pe dendrofloras. Candollea 50: 515 1967. Vergleichende Morphologie de hóhren Volume 85, Number 3 Pozner 439 Revisión de Cucurbitella Pflanzen. Band I, Teil 1—3. Gebrüder Borntraeger. Ber- lin. Vogel. S. 1981. Die Klebstoffhaare an den Antheren von Cyclanthera pedata (Cucurbitaceae). Pl. Syst. Evol. 137: 291-316. Walpers, G. G. 1846-1847. 4. Botanices Systematicae, Vol. 6. F. Hofmeister, Le Weberling, F. 1992. 4... of а pm Inflores- cences. Cambridge Univ. Pres Zimmermann, A. 1922. Die йы Beitrüge zur Anatomie, ішек Morphologie, Biologie, Patho- logie und Systematik. Heft 1 und 2. Gustav Fischer, pin E ÍNDICE DE EXSICCATAE . Abbiatti 211, 517; O. Ahumada 3408, 4425, 4677 iu 4974, 5064; M. M. Arbo 540, 559; P. Arenas 300, 1450, ке ay J. L. Argafiarás 57; L. Artico z 29; H. H. Bartlett 19839, 20252; 8. Back 6490: M. Birabén 30, 229; К. Biurrum 1407, 3055. 3137. 3304, 3323, 3371. 4000; О. Boelcke 74, 6629, 14630, 14631, 14632; Bridarolli 639, 3224; А. Brown 1694; А. Burkart 12586, 12588, 12747, 13245, 13370, 13957 20116, 20134, 20166, 22316, 23514, 25304, 25307. 25308, 25478, 25764, 27386, 1" 29536, 29608. А. Сађга! 253, 965, 1146; L. Cabrera 1133, 4359, 5195, 5975, 12850, 13701, 16201. 16742, 17247, 17898, 20648, 20992, 21768, 21826, 22357, 22413, 23350, 23420, 24635, 27227, 29852, 29894, 31387, 31389, 31424, 33994, 34534, 34580, 34582, 34614, 34628, 34629a, 34660; C. Calderón 1260; E. Cano 4379, 4626, 5149; P. Cantino Ser. 597, 613; E. Carette 104, 153, 181; В. Б. Carnevali 2092; J. A. Castiglioni 6872, 7189, 7301, T 7980; E. C. Clos м А. E. Cocucci 352; A. Co- a 319; N. Correa 4300; N. E. Crespo 159; S. 2943; C. Cristóbal 2160; L. Coast 3603; S. Chalukian 1784; A. Charpin 20438. G. 2. 3326; N. Deginani 6, 810; E. Dinelli 757; R. Duré 4 R. “Чеп 242; С. Ezcurra 1403. Н. A. Каһп 5036, 6254, 6430; A. Flossdorf s. nov. 1908 (BAB-25200), s.n. 20 nov. 1908 (ВАВ- dene 1571, . 2024; Franceschi 505, uelli 529. Giardelli 766%, 1262; Gibert 73; E. R. Guaglia- none 1369. 1375, 1818, 2291 е 162; С. М. Wicken 25306, 25309, 25310; G. Hieronymus аА de F-8988); Hoffmann 1863; С. . Hosseus A. T. Hunziker 1055, 2778, 4709, 5774, 9144. ed vp 13584, 14081, 17669, 17830, 18516, 18645, 20557, 24105; J. Н. Hunziker 11253, 9535, 9547, dy Job 826, 18 OP Jórgensen 165, 273, 1241, 1247, T qos 2 2 Jozami 874; Juárez de Varela 1723. R. Kiesling 148, 1196b, 4256, 4468, : 5354. un 5846, 5947, 6577; R. King (LP 896296); A. Krapovickas 1626, 5008, 5136, 5931, 11503, 11897, 13819, 13829, 14631, 14761, 15375, 17503, ло 17801, 18414, 20592, 21827, 24566, 27663, 27836, 27837, 28375, 28625. 30591, 30885, 31180, 32970, 46327, 46363; F. 200, 1339, 2409, 3354, 11862, 13219, 13245. H. Lagiglia 1963; A. Lanfranchi 1182; P. Legname 5058; Leguiza 51; J. P. Lewis 1350; P. С. Lorentz 750, 8 17619, 17725, 24565, ~ с A Kurtz 1248. — 236, 1145, (LP 46736); L. Malmierca 2065: rtínez Crovetto 3694, 9880, 10302; V. Maruñak 469, 517; D. Medán 284; A. M. Molina 2985, 2897, 3197, 3216, 3426, 3454, 3953; Morici 8; M. Múlgura 683. M. Чез 33825, 44783, 44809; E. Nicora 3550, 9058; L. Novara 2884, 2887, 4044, 4082, 4097, 4686, 5557. т. 8900, 9317, 9501, 10086. . Palací 391; Palán 385, 391; L. Tun 6033, 14865, us T. M. Pedersen 13490, 15211; J. Pensiero 4589; J. B. Pesce 122; Petersen 2589; C. E Petetin ond 1359, 1531, 1664, 1831; B. С. Piccinini 1134, 1721. 1744, n маг S. M. Pire 1115, 1503; R. ы 50, er Prado in у. 2324, 3038, 3250 Ragonese 7131, 7139, 7987, 8744, 9655; F. Rial Alberti 190, 203; Rodríguez 269; A. Rodrigo 2548; F. A. Hoig 562, 1521, 2253, 3501, 3558, pout 5449, 6436, 6598, 8512; T. Rojas 2354; A. Rotman 8 . Ruiz 113; . Ruiz Leal 904, 5761, 6472, 8010, 8530, 8809, 9011, 9096, 9237. 11348, 13418, 15908, 16487, 16736, 17203, 18434, 21124, 21369, 22559, 27023. C. Saravia Toledo 1532; A. Scala (LP); F. Schickendantz 116. 288; A. Schinini 8709, 10988, 12136, 12465. . 12584, 13021, 13702, 13863, 16402, 19547, 19554, 19565, 20023, 22248, 22446, 22480, 22508, 24175; A. С. Schulz 101, 2863, 2865, 2866, 2868. 2871. 2874, 4079, 4086, 5916, 6323, 6353, 6358, 6375, 6425, 6429, 7400, 7404, 8263, 8271, 8466, 8648, 8985, 8988. 9372, 14556, 15240, 15650, 15828, 15830, 15832, 15853, 15868, 16257, 16497, 17064, 17352, 17999; H. Schwabe 357; C. Spegazzini 3131, s.n. 12 Mar. 1937 (BAB-5801 1), s.n. 5 Mar. 1904 (BAB-11229), s.n. 11 Mar. 1905 (ВАВ-13841), s.n. 18 Mar. 1905 (BAB-14235). s.n. 9 Feb. 1906 (BAB-15471), s.n. 17 Jan. 1908 (BAB- 22248) 17 Jan. 1908 (BAB-22250), s.n. 17 Jan. (ВАВ-22276), s.n. 1 Feb. 1908 (ВАВ-22664), s.n. 1 Feb. 1908 (ВАВ-22675), s.n. 1 Feb. 1908 (BAB-22676), s.n. 1 Feb. 1908 (BAB-22682), s.n. 1 Feb. 1908 (BAB- = s.n. 1 Feb. 1908 (BAB 22773), P. L. Spegazzini s.n. Dec. 1909 (BAB 28732); R. A. puo 344; P. „ы 3886, 3921, 4058, 5468, 6544, = C. M. Taylor 11366; S. С. тз 3212; Troiani 4129, 7246, 7249, 7878, 7954, 9537, 0538, 9547, 9871, 6750b; N. Troncoso 1865, 190 E. Ulibarri 305, 1401. Varela 242, 306, 345; S. Venturi 321, 1159, 1602. 4405, 5680, 5681, 7665, 7687, 7734; Vignati 244, 526; C. Villamil 7678. J. Williamson 3233. Zapata 92; К. Zuloaga 300, 352, 1305, 2615, 2704, 2890, 2891, 3514, 3650, 3652, 3663, 3698, 3700, 3762. A TAXONOMIC REVISION OF DICOMA (ASTERACEAE: CICHORIOIDEA EE: MUTISIEAE) FOR THE HORN OF AFRICA! Santiago Ortiz,? Juan Rodríguez- Oubiña,? and Mesfin Tadesse? ABSTRACT A revision of the genus Dicoma (Asteraceae: Cichorioideae: Mutisieae) in the Horn of Africa (Djibouti, Eritrea, f Ethiopia, and Somalia) is provided. This area is probably wo authors. For each species, data are presented on 2 e species recognized have been described recently al characteristics, area of distribution (of particular interest in view of the inc orrect information previously published for several of the species), ecology, and local names. One new species, D. thuliniana, is described proposed, and D. шлш. is lectotypified. and illustrated, one new combination, D. schimperi subsp. cinerea, is The genus Dicoma Cass. (Asteraceae: Cichorioi- deae: Mutisieae) consists of about 50 species of herbs, shrubs, and small trees. Most species are from tropical and southern Africa and Madagascar, though one species reaches the Arabian Peninsula and India and Pakistan. Traditionally, Dicoma has been included in the tribe Mutisieae (Jeffrey, 1967; Cabrera, 1977). Re- cently, Hansen (1991) has suggested its exclusion from this tribe, largely because (a) the surface mor- phology of the corolla cells does not show the typ- ical Mutisieae pattern, (b) the corolla is clearly di- vided into a narrow tube and a broad limb, and (c) in species with bilabiate flowers the upper lobes are absent or short and uncoiled. However, recent cladistic analyses of the subfamily Cichorioideae have suggested that Dicoma should be included in the Mutisieae despite these differences (Karis et al., 1992; Bremer, 1994). Horn of Africa (Djibouti, Eritrea, Ethiopia, and Somalia), which is probably a major center of radiation for Dicoma, is one of the least known ar- eas of the continent floristically (cf. Hedberg & Ed- wards, 1989; Thulin, 1993; Edwards et al., 1995). Until recently, only four species of Dicoma were known from this area; Cufodontis (1967) cited five species, but these included D. gnaphaloides, which is merely a synonym of D. tomentosa. More recent work, and particularly study of herbarium material collected from the 1970s onward (especially in ETH, K, and UPS), has revealed the presence of twelve species, most of which have not been de- scribed previously (Ortiz & Rodríguez-Oubiña, 1994, 1995, 1996a, b; Rodríguez-Oubifia & Ortiz, 1995). We have also recently typified D. bangueo- lensis (Ortiz & Rodríguez-Oubiña, 19962), thus re- solving the nomenclatural problems and confusion arising from the erroneous reports of Buscalioni and Muschler (1913): as a result of this confusion, all the herbarium material of this species examined by us had been previously identified incorrectly or un- identified, despite this being one of the commonest species in Somalia. It is also necessary to clear up a confusion de- riving from Cufodontis (1967), who stated that D. bangueolensis, D. somalensis, and D. schimperi ex- tended into Ethiopia, when in fact these species are known only from Somalia. In response to these recent developments, we here present a revision of Dicoma for the Horn o Africa, accounting for all the species previously known together with those described recently. MATERIAL AND METHODS A total of 120 herbarium specimens from BM, ETH, FT, G, K, OXF, P, PAL, PI, S, UPS, and W, ' Our thanks go to Manuel Laínz for the Latin diagnosis, to Alfredo "Tokio" López for the Hes to G. Norman for the English translation, and to the keepers of the herbaria mentioned for the loan ol 2. erial. ? Laboratorio de Botánica, Facultade de Farmacia, Universidade de Santiago, 15 оре in. 1 И И Саћала, xartment of Plant Biology. College of Biological Sciences, The Ohio State University, 1735 Neil Avenue, Colum- S.A. ы Ohio 43210-1293, U.S ANN. Missouni Bor. GARD. 85: 440—459. 1998. Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa Table 1. present in the Horn of Africa). Composition of the sections of Dicoma proposed by Hoffmann (1893) (considering only those species sect. Eudicoma sect. Hochstetteria sect. Psilocoma sect. Pterocoma D. bangueolensis D. schimperi D. tomentosa D. sessiliflora орооо ово + Y 8 5 % thuliniana collected in Eritrea, Ethiopia, and Somalia, were studied, and all type material was examined by the authors; it was not possible to obtain herbarium ma- terial on loan from EA and MOG. We saw no Di- coma collections from Djibouti, although we know of one literature reference for the presence of D. schimperi subsp. schimperi in that country (Audru et al., 1994). All material was studied with the aid of a light microscope. We also studied other micro- morphological and anatomical characters with a compound light microscope; for this part of the study, floral parts were first boiled in water and placed in Hoyer's solution (Anderson, 1954) for ob- servation. Testa morphology was classified following Grau (1980), and epidermal cell surface, twin hairs, and superficial achene gland morphology following Karis et al. (1992). In the species descriptions, an- ther length includes the length of the apical ap- pendage and the anther tails. PHYLOGENY Very little is known about the phylogenetic re- lationships of Dicoma. A cladistic analysis of the subfamily Cichorioideae indicated that Dicoma is closely related to Pleiotaxis and Erythrocephalum (Karis et al., 1992). This conclusion was supported by Bremer (1994), who assigned these three genera plus the African genera Pasaccardoa, Achyrothal- amus, and Gladiopappus to the "Dicoma group," which he considered to be monophyletic. We are currently performing the first cladistic analysis of Dicoma in which the species considered include not only members of Dicoma but also mem- bers of the other genera of the Dicoma group sensu Bremer. Our preliminary conclusions indicate that Dicoma is paraphyletic with respect to Pasaccar- doa, and that Pleiotaxis, Achyrothalamus, and Er- ythrocephalum are sister to Dicoma plus Pasaccar- doa. Іп general, the most primitive species of Dicoma appear to be those of southern Africa and Madagascar. Most of the species of the Horn of Af- rica form part of more advanced lineages resulting from radiation into more arid regions. Dicoma ses- siliflora appears to form part of a lineage that un- erwent secondary adaptation to shadier, wetter sites. If we accept Hoffman's (1893) division of Dicoma into eight sections, the species of the Horn of Africa fall into four sections: Eudicoma DC. (Dimorphae F. C. Wilson), Hochstetteria (DC.) О. Hoffm., Рзи- ocoma Harv. (Barbellatae F. C. Wilson), and Pter- ocoma DC. (Plumosae F. C. Wilson), as detailed in Table 1. However, Hoffman's division of the genus does not appear to be a good reflection of phylog- eny. А more useful approach might be to consider all species present in the Horn of Africa, except D. sessiliflora, as members of a single group. DiAGNOSTIC CHARACTERS In what follows, we discuss the principal diag- nostic characters used in the systematics of Dico- ma, and, in particular, those relevant to the species of the Horn of Africa. Habit. Dicoma includes annual and perennial herbs, shrubs, and (Madagascar only) small trees. Some species, such as D. aethiopica, are variable in habit, with some individuals being annual herbs and others biennial or perennial. Other species, such as D. sessiliflora, are rootstock perennials. Some shrub species form dense cushions. tem. Gross stem morphology is of little diag- nostic value: stems range from scarcely to highly branched, and the branches range from straight to highly twisted. Stem pubescence (or lack thereof) is a more useful character, ranging from glabrous or glabrescent to very densely tomentose, with in- terlaced, long, flexuose, simple hairs. Stem color ranges from greenish or stramineous to whitish or grayish white, or in some cases purplish. Leaf arrangement is always alternate, and spacing ranges from well separated to subfas- ciculate. Leaf shape ranges from linear to subor- aves. 442 Annals of the Missouri Botanical Garden bicular, and the leaves may or may not be condu- plicate. The leaf margin is generally entire or slightly serrate, and is in some cases highly revo- lute. A pseudopetiole may or may not be present. The leaf apex ranges from acute to obtuse, with or without a spine or a mucro. Leaf pubescence and color are generally similar to those of the stem, and leaves may be concolorous or discolorous, depend- ing on whether the pubescence of the lower surface is markedly denser than that of the upper surface. Capitulum. Тһе capitulum has a variable num- ber of subtending leaves. The shape and size of the involucre, ranging from narrow cylindrical to broad- ly campanulate, is useful for diagnosis. he number of phyllaries and phyllary rows is also important, and whether the inner phyllaries are longer or shorter than the adjacent outer phyllaries is of particular value. The phyllaries can be erect, patent, or squarrose. Reflexed phyllaries are an im- portant diagnostic characteristic, particularly in section Pterocoma, but Dicoma rarely has phylla- ries of this type in the Horn of Africa (only the outermost phyllaries in D. schimperi). Phyllary shape is variable, ranging from linear or linear-lan- ceolate to deltate, while the apex may be acuminate or aristate. The phyllaries may be scarious at the margins or over their entire surface. The presence or absence of a conspicuous midrib is important. Pubescence ranges from glabrous to densely to- mentose, and is generally similar to that of the stem and leaves. Florets. course related to capitulum size, shows great vari- ation among species. The corolla may be white, cream, yellowish, violet, or lilac. The corolla lobes are recurved apically in all except one species in the Horn of Africa, and the veins along the edge may be slender and submarginal, or thick and mar- ginal. The epidermis of the corolla shows either an “intestine-like” surface (i.e., a rugose longitudinal bands) or is slightly transversely un- dulate-striate or smooth. The indument is made up of short or long, twin, glandular hairs (see Karis et al., 1992: 418, figs. 4I, H, respectively). The sizes of the different parts of the stamens and the shape of the anther appendages are like- wise variable, but of little diagnostic value; the presence or absence of antrorse hairs on the tail apex is, however, important. The most useful characteristics of the style are (a) the extent of the area occupied by sweeping hairs (hairs that brush the pollen through the anther tube for pollination), and (b) whether or not the basal sweeping hairs are longer than the rest. Fruit. The achenes have (5-)8-10 prominent The number of florets, which is of attern of ribs, or are not conspicuously ribbed. The achene hairs are situated between the ribs or all around the achene; in the latter case, they have a conspic- uous bulbous, glandular base. Most species have epidermal glands and superficial biseriate glands, always located between the ribs. The testa may be of two types: the Dicoma type, with the external layer of the testa prosenchymatic and strengthened with ribs; or the Gochnatia type, with the lateral and basal walls of the testa epider- mis strengthened and u-shaped in cross section (Grau, 1980). The pappus may be isomorphic or dimorphic. In the latter case, there is an inner row of scales in addition to the various outer rows of bristles. In species in which the pappus is isomorphic, it may comprise a single row of about 10 rigid flattened bristles or bristle-like scales (as in D. schimperi), or several rows of bristles that may be either bar- bellate or plumose. In general, the number of pap- pus rows and the length of the pappus bristles are important diagnostic characters. TAXONOMIC TREATMENT Only synonyms relevant to species occurring in the Horn of Africa are included. Likewise, the ge- nus description includes only those species present in the Horn of Africa. Dicoma Cass., Bull. Sci. Soc. Philom. Paris 1817: 12. 1817. TYPE: Dicoma tomentosa Cass. Hochstetteria DC., Prodr. 7: 287. teria schimperi DC. 1838. TYPE: Hochstet- Annual or perennial herbs and shrubs. Stem simple to branched, to lanate, the branches generally striate. Leaves alternate, often sericeous with a pseudopetiole, the margins entire to ser- rulate. Capitula obconic to campanulate, sessile or pedunculate, discoid and homogamous (or radiate and heterogamous outside the Horn of Africa), of- ten with subtending leaves; phyllaries multiseria- te; acuminate to aristate, coriaceous, sometimes cream, yellowish, violet, or lilac, actinomorphic, deeply 5-lobed, with short-glandular or long-glan- dular twin hairs. Anthers with lanceolate, acute to acuminate apical appendages and long-tapering, retrorse-pilose tails. Style swollen at the base, the branches connivent, with obtuse apex and short, generally subapical, acute, sweeping hairs. chenes obovoid to turbinate, with (5-)8-10 ob- scure to prominent ribs, often glutinous between Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa 443 the ribs, with ascending hispid hairs inserted be- tween the ribs or all around the achene, the basal hairs shorter and spreading; testa of the Dicoma or Gochnatia type. Pappus of barbellate to plu- mose bristles arranged in several rows with or without one internal row of scales; in one species (D. schimperi) the pappus is only of scales ar- ranged in one row. Ккү TO THE SPECIES OF DICOMA OCCURRING IN THE HORN OF ÁFRICA Pappus conspicuously dimorphic, with external capillary bristles and ca. 10 innermost scales . D. ban gueolensis ). tomentosa la. 2a. Shrub; leaves broadly elliptic to suborbic E not conduplicate; phyllaries with short (0. 2-0.5 mn acuminate apex; achenes 1.3-1.5(-2.5) mm long .____. 2b. Annual pe leaves coca to linear-oblane it often conduplicate; phyllaries inh long 7 mm), acuminate-aristate apex; achenes 1.6-3 m 12 Ib toward the base. 3a. Pappus uniseriate, of ca. 10 rigid, flattened bristles | 3b. Pappus multiseriate, of more than 10 bristles 4a. Phyl ong Pappus bone bie. made up Ms "ut bristles, those of as innermost series sometimes somewhat broader AAA Е EA 7. D. schimperi yllaries without median vein, the ene shorter than the adjacent series; Bp of pen bristles |. sessiliflora 4b. Phyllaries with a conspicuous median vein, the innermost longer than the other series; pappus arbeite bristles. 5а. yllaries squarrose hyllaries 35— 70 per capitulum, pubescent; corolla ca. 5 mm long; achenes 2.7-3 mm long, narrowly штале; pappus of 60-70 bristles, 4.5—5.2 mm long ..... О per capitulum, glabrous: corolla ca 6b. Phyllaries 100-14 4. D. Pal dl mm long; po do 1: .9n mm long, broadly turbinate: pappus of 100—130 bristles, 7-9 mm long ..................... ob. Phyllaries not squarro 7a. Leaves skies ta often with an apical spine. 8a. Branches twisted, forming dense whitish cushions; leaves often subfasciculate, 7-17 mm long, slightly curved; capitula sessile to subsessile, apex of the phyl- laries acuminate ong ТЬ. Leaves s linear-elliptie to elliptic, oblanceolate or spathulate, with flat Ma s an apical spine 9a. 4. ог > perdniiial herb; achenes 1.5-2.5 mm long; pa 2-17 mm; involucre 7-11 X 10a. Lea ves 15-90 X , 0.2-1 mm 8b. Branches erect, straight; leaves solitary, (5-)15-22(-26 dn on leafy peduncles, apex of the phyllaries acuminate, 1.2-3.5 = lor n s 3-5 mm lon . paivae ) mm long. ailes ca- long; pappu thuliniana s 4—6 mm long. 12-20 mm; corolla (5-)6.5- 2 mm long; pappus of 40-100 bristles arranged in 3(—4) Me e LOb. ek 15-25 X 2.8-3.7 m 9b. Shrub: d la. Leaves narrowly elliptic. 10-1 30 x 23 mm, broadly campanulate; pappus of 100-120 bristles ........ 11р. Le eaves Ж, кошы р. to 3 X 1. Dicoma aethiopica S. Ortiz & Rodr. Oubiña, Nordic J. Bot. 16: 279. 1996. TYPE: Ethiopia. Bale: at the Sof Omar Caves, 6°55'N, 40*50'E, . 1400 m, 31 Oct. 1985, Früs, Gilbert & Vollesen 3686 (holotype, K; isotype, ETH). Annual or perennial herb to 40 cm high. Taproot slightly ramified. Stem ramified; the branches stri- ate, greenish to purple, moderately grayish white- tomentose, with simple hairs and sessile to subses- sile glands. Leaves 15-90 X liptic to oblanceolate, attenuate at base on a de- current pseudopetiole 2-20 mm long; the margins serrulate, slightly callose; apex acute; both surfaces —17 mm, linear-el- 2-3 mm; involucre 6-7 X 7—10 mm; s conl (4-5 n g; pappus of 25-40 bristles arranged in 1–2(–3) ro 6. D. . D. aethiopica nm Кет рореапа ) mm long. mm; involucre 13-15 X 22- 3. D. gillettü 5 mm, narrowly 8. D. scoparia m long; pappus 6.5—7(-8 2-12.5 х 1 mm; jo ге 6.5-8 X greenish to purple, rugose, slightly to moderately tomentose, with sessile glands in some cases on the upper surface slightly darker and slightly more gla- brescent. Capitula numerous per plant, on erect- patent peduncles, 30— m long, with 3—5( subtending leaves; ia 1-11 x 12-20 mm, campanulate, with 70— ae arranged in 4-5(-6) rows, stramineous to purple, with a darker stripe on either side of the midrib, acuminate, pun- gent, glabrous to glabrescent, the margins entire to shortly serrulate principally toward the apical part; not scarious or with very narrow scarious margins; outermost phyllaries 3.5—4.5 X .8 mm, ovate- lanceolate, erect-patent, with an acuminate apex 444 Annals of the Missouri Botanical Garden 1.5-2 mm long; middle phyllaries 6-8 X 1.2-1.5 mm, lanceolate, erect-patent, with an acuminate apex (3-)4-5 mm long; innermost phyllaries 9-11 X 1.5-1.7 mm, longer than the outer phyllaries and projected ca. 2 mm beyond the pappus or shorter than the pappus, oblong-lanceolate, erect, with an acuminate apex 1.2-3 mm long; receptacle con- cave, alveolate, pits surrounded by а membrane with an irregular dentate margin 1-1.3 mm high. Florets 15-25 per capitulum. Corolla (5-)6.5-8.2 X 1.8-3 mm, white, with epidermal cell surface “intestine-like” and with short glandular twin hairs; tube 2.8-4.8 X 0.4—0.9 mm; lobes 3.3-3.5 X 0.4- 0.5 mm, becoming recurved, with slender submar- ginal veins. Stamens exserted for 0–2.5 mm beyond the corolla; filaments 1.3-2 mm long; collar 0.4— 0.5 mm long; anthers 3.9—4.1 mm long; apical ap- pendages ca. 1 mm long, conspicuously apiculate; anther tails 1.2-1.3 mm long, with retrorse hairs 0.2-0.4 mm long and some shorter antrorse hairs at the apex. Style 6.5-9 mm long, stylar branches 2.3 mm long, with sweeping hairs forming a sub- apical ring, covering a surface 0.4 mm long, the basal ones longer than the others. Achenes 1.5-2.5 X 0.7-1.2 mm, turbinate, 10-ribbed, hispid, with ascending hairs 0.2-2.5 mm long, inserted between the ribs from the base to the top of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pappus iso- morphic, of 40—80(-100) barbellate bristles ar- ranged in 3(4) series, the innermost bristles broad- ened toward base, the longest bristles 4—6 mm long, the shortest ca. 1 mm long; barbellae 0.1-0.2 mm long. Habitat and distribution (Fig. 1). Known only from Ethiopia, at 1400—1500 m. Limestone or gra- nitic outcrops; dense to open Acacia- Commiphora bushland and open areas ne umm molle, Terminalia brownii, etc. Endem Additional specimens examined. ETHIOPIA. Bale: 13 km from Ginir on road to Robi via Sof Omar, 7%5'N, . 1500 m, Gilbert, Ensermu & Vollesen 7870 (ЕТН, К, UPS). Sidamo: Bitata, 20 km from Negele on road to Kibre Mengist, 5°29’N, 39?28'E, 1450 m, Gilbert, Enser- u & Vollesen 7781 (ETH, K, UPS). The principal diagnostic characters of this spe- cies are its concolorous, generally rugose, elliptic to oblanceolate leaves with more or less long pseu- dopetioles, and the pappus of 40-100 barbellate bristles that are slightly broadened at the apex ar- ranged in 3(—4) rows. In the Horn of Africa, the most closely related species is undoubtedly D. po- peana, which has smaller leaves (15-25 x 2-3 mm), a smaller involucre (6-7 Х 7-10 mm), and a Distribution of Dicoma “= а (А) апа Figure 1 D. bangueolensis (8) in the Horn of Africa =~ — smaller corolla ((4-)5 mm long), as well as fewer pappus bristles (< 40) arranged in 1-2(-3) rows. In the genus as a whole, the most similar species morphologically is D. anomala Sond., one of the most widely distributed species of the genus. This species can be distinguished from D. aethiopica by its sessile or short-pseudopetiolate leaves, these discolorous with the lower surface densely tomen- tose, and by the 7-12-mm-long pappus made up of more than 100 bristles that are conspicuously nar- rower at the apex The color of dried D. aethiopica specimens shows great variability: some specimens (such as the holotype) are basically greenish to stramineous, while Gilbert, Ensermu & Vollesen 7781 (K) exhibits a strong purplish coloration of the stem, leaves, and phyllaries. 2. Dicoma bangueolensis eom & Muschl., Bot. Jahrb. Syst. 49: 514. 1913. TYPE: So- malia. Between Hobyo and Wuarandi and in Merehan, Robecchi-Briquetti s.n. (lectotype, designated by Ortiz & Rodríguez-Oubiña (1996), K). Dicoma Су О. 2. ех Engl., Sitzungsber. Кбп- igl. Pre Akad. Wiss. 1904: 371. 1904, nom. il- leg., non "n Moore, 1899 TYPE: Somalia (without recise locality or collec loc see Ortiz & Rodríguez- Oubiña, 1996). Vernacular names. Darran (Beckett 242), Ged- aad (Kazmi, Celmi, Mahmund & Sulaiman 5), Ghedhad, uadad (Cufodontis, 1967). Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa 445 Shrub or shrublet to 60 cm high. Taproot slightly ramified. Stem often highly ramified; the bran nearly quadrangular in section, covered with a ches grayish white tomentum of simple hairs, especially on young branches, with sessile to subsessile glands. Leaves (6-)12-17(-23) х (3.5-)5-7(-10) mm, alternate, sometimes fasciculate, elliptic to broadly elliptic, sometimes suborbicular, with a pseudopetiole (1.5-)3-5(-8) mm long; the margins entire to denticulate; apex with a mucro up to 1 mm long; both surfaces with sessile glands, densely grayish white tomentose. Capitula numerous per plant, arranged in corymbiform synflorescences, on erect-patent peduncles 1-3(-9) mm long with lin- ear bracts at the base, without subtending leaves; involucre 6-7(-9) X 4—6(-7) mm, obconic, with 0—30 phyllaries arranged in 4—5 rows, strami- neous, with a darker stripe on either side of the midrib, acuminate, pungent, densely tomentose on the whole surface of the outer phyllaries and on the distal part of the median and innermost phyllaries, the margins shortly serrulate principally toward the apical part; outermost phyllaries 1-1.5(-2) X 0.7 mm, deltate, erect-patent, without acuminate apex, the margins not scarious; middle phyllaries 4—5 X 1-1.7 mm, deltate to oblong-lanceolate, erect, with an acuminate apex 0.2-0.3 mm long, the scarious margin 0.2–0.5 mm wide; innermost phyllaries 5—6 X 1.5-2(-2.5) mm, longer than out- er phyllaries, projected 2-3(-4) mm less than the pappus, oblong-lanceolate, erect, with an acumi- nate apex 0.2-0.5 mm long, the scarious margin 0.3-0.8 mm wide; receptacle concave, alveolate, the pits surrounded by a membrane with a scarcely dentate margin 0.05-0.1 mm high. Florets 3-7 per capitulum, whitish pink. Corolla 7 X 1.5-1.7(-2) mm, purple or bluish, with epidermal cell surface “intestine-like” and glabrous or with short glan- dular twin hairs, especially toward the basal part; tube 2-3 X 0.4-0.6(-0.8) mm; lobes 4—4.5 X 0.4— 0.5 mm, becoming recurved, with slender submar- ginal veins. Stamens exserted for 0.7-1.5 mm be- yond the corolla; filament 1.5-2 mm long; collar 0.2-0.5 mm long; anthers 4—5 mm long; apical appendages ca. 1.3 mm long, conspicuously apic- ulate or not apiculate; anther tails 1.2-1.5 mm long, with retrorse hairs ca. 0.3 mm long and 3—5 antrorse hairs at the apex. Style 6—7.5 mm long, stylar branches 1.3-1.7 mm long, with sweeping hairs forming a subapical ring covering a surface 4—0.7 mm long, the basal ones conspicuously lon- ger than the others. Achenes 1.3-1.5(-2.5) X 0.8- 1.7(-2.5) mm, )10-ribbed, 1 covered with ascending, white hairs 0.2-2. long, these inserted between the ribs from the и turbinate, (8— to the top of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pappus dimorphic, of 110—130 barbellate bristles arranged in 2-3 external series, the longest bristles (5-)6.5-7.5 mm long, the short- est 1 mm long; barbellae ca. 0.05 mm long; ca. 10 scales arranged in a single internal series, these 5.5—6.5 X 0.5-0.8(-1) mm, barbellate toward the apex. Habitat and distribution (Fig. 1). Known only from northeastern Somalia, at 130—1300 m. Prin- cipally on gypseous soils, but also on limestone soils, stony or not; іп Acacia-Commiphora bush- and. Endemic. peewee specimens examined. SOMALIA. Mudug: 23 km E of Ba pes 5°71'N, 47?26'E, 145 m, Beckett 242 (ЕТ, mre 6 km NW of Gaalkacyo, 6°49'N, 47%22'E, 300 m, Beckett 397 El 23 km E of Baxdo on road to Hobyo, ?48' 30 m, Gillett, Hemming & Watson 22381 (K); 8 km E al Bacaadweyn on road to Baardale, 7°09'N, 479377Е, 360 m, Gillett, Hemming & Watson 22028 (K); 1 km from ет toward 2. Кагті, Celmi, Маһтипа & Su an 5 (FT); 10 km from Gaal- kacyo along road to Belet eh. 300 m, T. = (K, UPS); Nugaal: 1 mi. N of Laascaanood, 2400 ft., Bally B10873 (K); ca. 8 km NE of Xalin on road to pe nd 9°05'N, 48°41'Е, 675 m, cuña ie Beckett 23352 E 3 km E of Laascaanood, 8°25'N, 47?20'E, Hansen & H stra 6322 (К); Sanaag: 10°37'N, jeu E, 1300 m, Gillet & Watson 780 (K). The principal distinguishing morphological char- acters of Dicoma bangueolensis are its concolorous densely grayish-tomentose, elliptic to broadly ellip- tic, sometimes suborbicular leaves, obconic invo- lucres, and dimorphic pappus of 110—130 barbel- late bristles in 2-3 rows with an inner row of 10 scales. In the Horn of Africa, D. tomentosa is the only other Dicoma species with a dimorphic pap- pus, of similar characteristics; however, this species is very different as regards habit, leaf morphology, and the size and morphology of its phyllaries and achenes. The only other Dicoma species with an obconic involucre is D. dinteri S. Moore (from Na- mibia); however, this species has linear leaves and an isomorphic pappus Wilson (1923) cited D. bangueolensis as a mem- er of his section Dimorphae, characterized by a dimorphic pappus (though he stated that this ме" cies occurs in the Tanganyika region, as err ously reported by Buscalioni & Müschler, 1913). Cufodontis (1967) reported this species from So- malia, probably on the basis of Volkens (1915), and also from Ethiopia, though we have yet to see a specimen from the latter country. е have recently (Ortiz € Rodríguez-Oubiña, 1996a) resolved the nomenclatural problems and Annals of the Missouri Botanical Garden confusion arising from the erroneous reports of Bus- calioni and Muschler (1913). In fact, all specimens of D. bangueolensis seen by us in the course of our herbarium studies of Dicoma had been incorrectly identified or left unidentified. 3. Dicoma gillettii Rodr. Oubiña & S. Ortiz, Bot. J. Linn. Soc. 119: 59. 1995. TYPE: Somalia. Shabeellaha Dhexe: 30 km S of Aadan Yabaal, 3°27'N, 46"12'Е, 170 m, 6 June 1981, Gillett & Beckett 23263 (holotype, K; isotype, UPS). Shrub to 40 cm high, ramified; the branches stri- ate, covered with a dense grayish white tomentum of simple hairs covered externally with a membra- naceous pellicle, with sessile to subsessile glands. Leaves (10-)30-110(-130) x (2-)4-9(-12.5) mm, narrowly elliptic, attenuate at base and decurrent on a slightly stem-clasping pseudopetiole 7-25 mm long; the margins irregularly callose denticulate; apex acute; upper surface greenish, glabrous and covered with a membranaceous pellicle, with ses- sile glands; lower surface densely grayish white- tomentose, with sessile glands. Capitula several per plant, solitary on erect to erect-patent peduncles, 40-70 mm long, with 3-5 subtending leaves; in- volucre 13-15 x 22-23 mm, broadly campanulate, with 65-80 phyllaries arranged in 5—6 rows, stra- mineous, rigid, acuminate, pungent, scarcely to- mentose to glabrescent, the margins shortly serru- late-fimbriate toward the basal part; outermost phyllaries 4-5 X 1.5-2 mm, ovate, widening sharp- ly and becoming scarious in their basal third, pat- ent to erect-patent, with an acuminate apex 3—5 mm long; middle phyllaries 10-12 X 2-2.5 mm, ovate-lanceolate, erect-patent, with an acuminate apex 2-3 mm long; innermost phyllaries 9-12 X 2—3 mm, longer than the outer phyllaries, as long as the pappus, lanceolate, erect to slightly patent, with an acuminate apex 0.5-1.5 mm long. Florets (25—)30—35 per capitulum. Corolla 9-12 x 1.2-2.5 mm, yellowish, with epidermal cell surface “intes- tine-like" and with short glandular twin hairs; tube 5-6 X 0.7-1.5 mm; lobes 5-6 X 0.4-0.7 mm, be- coming recurved, with slender submarginal veins. Stamens exserted for 0.5-1 mm beyond the corolla; filaments 2-2.5 mm long; collar 0.5—0.7 mm long; anthers 5.8—6.5 mm long; apical appendages 1.5— 2 mm long, conspicuously apiculate; anther tails 2.3-2.5 mm long, with retrorse hairs 0.3-0.5(-0.8) mm long and some shorter antrorse hairs at the apex. Style 9–11.5 mm long, stylar branches 2.5— 2.7 mm long, with sweeping hairs forming a sub- apical ring, covering a surface 0.3-0.7 mm long, the basal ones conspicuously longer than the oth- Figure 2. Distribution of Dicoma gillettii (8) and D. hindiana (А) in the Horn of Africa. ers. Achenes 2.8-3.1 X 1.7-1.9 mm, turbinate, 8- to 10-ribbed, hispid, with ascending hairs 0.3-2 mm long, these inserted between the ribs in the basal У;-У, of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pappus isomorphic, of 100— 120 barbellate bristles arranged іп 2—3(—4) series, the innermost bristles broadened toward the base, the longest bristles 7(—8) mm long, the shortest 0.1— 0.5 mm long; barbellae 0.1—0.2 mm long. Habitat and distribution (Fig. 2). Known only from the type locality, at ca. 170 m. On yellowish level sand; Acacia- Commiphora bushland. Endem- 16; The most important distinguishing characteris- tics of Dicoma gillettii are the presence of a mem- branaceous pellicle entirely covering the tomentum of the stem, and the broadly campanulate involucre with the outermost phyllaries widening abruptly and becoming scarious in their basal third. None of the other Dicoma species of the Horn of Africa exhibit a close morphological relationship with D. gillettii. Considering the genus as a whole, D. an- omala Sond. is somewhat similar in leaf morphol- ogy and the shape and size of the involucre. How- ever, D. anomala lacks a membranaceous pellicle covering the tomentum of the stem, and has con- siderably fewer phyllaries. Furthermore, the outer- most phyllaries are deltate-lanceolate, without the widened, scarious, basal third that characterizes D. gillettii. Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa 4. Dicoma hindiana S. Ortiz & Rodr. Oubiña (“hidana”), Nordic J. Bot. 15: 187. 1995. TYPE: Somalia. Mudug: near Hobyo, 14 Oct. 1912, Drake-Brockmann 960 (holotype, K). Shrub to 30 cm high. Taproot slightly ramified. Stem ramified; the branches striate, moderately grayish white-tomentose, with simple hairs, with sessile to subsessile glands, often with glabrescent to glabrous areas. Leaves (2-)8-18(-19) х (1.2—)2— 5(-7) mm, elliptic to obovate, attenuate at base and decurrent on a pseudopetiole 0.5—4 mm long; the margins serrulate, often somewhat revolute; apex mucronate; upper surface greenish, tomentose, with sessile glands; lower surface generally densely grayish white-tomentose, with sessile glands. Ca- pitula numerous per plant, 1—5 arranged in lax sub- corymbose synflorescences, on erect-patent pedun- cles, 3-30 mm long, with 1-3(-4) subtending leaves; involucre 7-9 X 7-13 mm, campanulate, with (35-)45-60(-70) phyllaries arranged in (4-)5- 7 rows, stramineous, with a darker stripe on either side of the midrib, acuminate-pungent, slightly to- mentose, especially in the apical third, the margins entire to shortly serrulate-ciliolate principally to- ward the apical part, not scarious; outermost phyl- laries 1-1.5 X 0.5-0.7 mm, deltate to ovate, erect- patent; middle phyllaries 2.54 X 0.7-1 mm, lanceolate, squarrose, with an acuminate-aristate apex 0.2-0.6 mm long; innermost phyllaries 5.5— 7(-8) X 0.7-1.1 mm, longer than the outer phyl- laries and projected (0-)1.5-3(-4) mm beyond the pappus, lanceolate to oblong-lanceolate, squarrose, with an acuminate apex 0.5-1 mm long; receptacle concave, alveolate, the pits surrounded by a mem- brane with irregular denticulate margin 0.2—0.3 mm high. Florets 10—17 per capitulum. Corolla 5 X 1.6 mm, with epidermal cell surface “intestine- like” and with short glandular twin hairs; tube 1- 1.5 х 0.6-І mm; lobes 2.5-3 X 0.4 mm, becoming recurved, with slender submarginal veins. Stamens exserted for 1 mm beyond the corolla; filaments ca. 1 mm long; collar 0.4 mm long; anthers 3.2—4 mm long; apical appendages ca. 1 mm long, conspicu- ously apiculate; anther tails 1 mm long, with re- trorse hairs 0.2—0.4 mm long and some shorter an- trorse hairs at the apex. Style 4.8-5 mm long, stylar branches 1.3 mm long, with sweeping hairs forming a subapical ring, covering a surface 0.2-0.3 mm long, the basal ones conspicuously longer than the others. Achenes 2.7-3 X 0.8 mm, turbinate, 10- ribbed, hispid, with ascending hairs 0.3-1.5 mm long, these inserted between the ribs from the base to the top of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pappus isomorphic, of 60—70 bar- bellate bristles arranged in 4—5 series, stramineous, the innermost bristles broadened toward the base, the longest bristles 4.5—5.2 mm long, the shortest 0.2–0.3 mm long; barbellae 0.05 mm long. Distribution (Fig. 2). Known only from Mudug province in Somalia at ca. 90 m. Endemic. Additional specimen fiw SOMALIA. Mudug: on road to Hobyo, 4 S of Jirriiban, 90 m, Gillett, Hemming & Watson 22104 (K). The principal diagnostic characteristics of Di- coma hindiana are its squarrose phyllaries, narrow- ly turbinate achenes, and pappus of 60—70 barbel- late bristles, these 4.5—5.2 mm long. The most morphologically similar species in the Horn of Af- rica is D. somalensis; indeed, all of the specimens of D. hindiana at К were labeled as D. somalensis. However, there are clear morphological differences between the two taxa. Dicoma somalensis has more phyllaries per capitulum ((100-)120-140); these are glabrous, and the innermost are erect. Further- more, D. somalensis has a longer corolla (ca. 9 mm), shorter (1.5-2 mm), broadly turbinate achenes, and longer (7-9 mm) and more numerous (100—130) pappus bristles. Considering Dicoma as a whole, the only other species with consistently squarrose phyllaries as in D. hindiana and D. somalensis is D. squarrosa Wild, endemic to Angola (though some of the outermost phyllaries may be squarrose in some specimens of D. anomala). However, D. squarrosa can be clearly distinguished from D. hin- diana by its elliptical leaves up to 50 mm long and its dimorphic pappus with both bristles and five broadly alate inner setae. Note that we here use the epithet “hindiana,” despite having originally described this species as *hindana"; this change is in accordance with the spelling recorded in Index Kewensis (Davies, 1996: 95) 5. Dicoma paivae S. Ortiz & Rodr. Oubiña, Can- ad. J. Bot. 72: 1478. 1994. TYPE: Somalia. Nugaal: about 80 km М of Eyl, 8°30'N, 50?7'E, 110 m, 11 June 1981, Gillett & Watson 23375 (holotype, K). Woody perennial, highly ramified; old branches very thick, young branches striate, covered with a whitish tomentum of simple hairs, old branches glabrous to glabres- cent, without or with very scarce sessile to subses- sile glands. Leaves sessile, subfasciculate, 7-17 X 0.7–1 mm, linear, slightly curved; leaf margins rev- olute, entire; apex with a short spine 0.5-0.8 mm twisted, forming cushions, Annals of the Missouri Botanical Garden 448 4s* + Figure 3. Distribution of Dicoma paivae (8) and D. popeana (A) in the Horn of Africa. long; both surfaces densely whitish-tomentose. Ca- pitula numerous per plant, solitary, sessile or sub- sessile on up to 10-mm-long peduncles, with 1—3 subtending leaves; involucre 6—6.5 X 3—4 mm, ob- conic-campanulate, with 25-30 phyllaries arranged in rows, stramineous, with a darker stripe on either side of the midrib, rigid, slightly acuminate, pungent, tomentose, the margins entire to shortly serrulate principally toward the apical part, not scarious or with a very narrow scarious border; out- ermost phyllaries 0.8-1 X 0.4—0.5 mm, deltate, patent to erect-patent, with an acuminate apex 0.2— 0.3 mm long; middle phyllaries 3 x 1.3 mm, del- tate-lanceolate, erect-patent, with an acuminate- aristate apex 0.5—0.7 mm long; innermost phylla- ries 4-5 X 1.2-1.5 mm, longer than the outer phyl- laries and shorter than the pappus, deltate-lance- olate, erect, with ап acuminate-aristate apex 0,7—1 mm long. Florets unknown. Achenes 1.5-1.6 х 1- 1.2 mm, turbinate, 8- to 10-ribbed, hispid, with ascending hairs 0.5-2 mm long, these inserted be- tween the ribs from the base to near the top of the achene, with epidermal glands and superficial bis- eriate glands between the ribs; testa of Dicoma type; pappus isomorphic, of 80—100 barbellate bristles arranged in 3—5 series, the innermost bris- tles broadened toward base, 0.1—0.5 mm wide, the longest bristles 3—5.5 mm long, the shortest 1—1.5 mm long, barbellae 0.05-0.1 mm long. Habitat and distribution (Fig. З). ^ Known only from the type locality at ca. 110 m. Low gypseous limestone ridges, very stony and windswept; com- mon with prostrate species of Commiphora, Cae- salpinia, Ochradenus, Acacia, and cushions of Old- enlandia saxifragoides. Endemic. The most important distinguishing characteris- tics of Dicoma paivae are its twisted branches form- ing dense white cushions, frequently subfascicu- late, linear-revolute, slightly recurved leaves with an apical spine, and sessile or subsessile capitula. The most similar species in the Horn of Africa is D. thuliniana, which can be distinguished from D. paivae by its erect, straight branches, straight, sol- itary leaves longer ((5-)15-22(-26) mm) than those of D. paivae, long, leafy capitular peduncles, and longer pappus (6-7 mm). In the genus as a whole, D. cana Вай. f., from Socotra, shows some similar- ities to D. paivae, particularly in leaf morphology and general habit (it likewise forms dense, white cushions). However, the capitula of the former spe- cies are very different: all phyllaries lack a midrib, the outer phyllaries have a long, acuminate-aristate, patent apex, and the inner phyllaries are white and membranaceous-papery (Ortiz & Rodríguez-Oubi- йа, 6. Dicoma popeana S. Ortiz & Rodr. Oubiña, Nordic J. Bot. 16: 277. 1996. TYPE: Somalia. Hiiraan: 12 km from Сее] Baraf, 3°21'N, 45"49.5' Е, 190 m, 14 Dec. 1987, Kuchar 17653 (holotype, K; isotype, UPS). Vernacular name. Reexaan (Kuchar 17653). Annual herb to 25 cm high. Taproot slender, slightly ramified. Stem highly ramified; the branch- es quadrangular in transverse section, moderately grayish white-tomentose, with simple hairs and ses- sile to subsessile glands. Leaves 15-25 X 2-3 mm, linear-elliptic, attenuate at base on a pseudopetiole 2-4(-5) mm long; the margins serrulate; apex with a mucro 0.2-0.5 mm long; both surfaces greenish, rugose, slightly to moderately tomentose, in some cases the upper surface slightly darker and slightly more glabrescent. Capitula numerous per plant, on erect-patent peduncles, 10-30 mm long, with 3-5 subtending leaves; involucre 6-7 X 7-10 mm, campanulate to broadly cylindrical, with 40-60 phyllaries arranged in 4—5 rows, stramineous, with a darker greenish to brownish stripe on either side of the midrib, acuminate, pungent, glabrescent to slightly tomentose, minutely strigulose, the margins entire to shortly serrulate toward the apical part; outermost phyllaries 2.5—4 X 0.4—0.7 mm, lance- olate, erect-patent, with an acuminate apex 0.7-2 mm long; middle phyllaries 6-7 X 1-1.3 mm, ob- long-lanceolate, erect-patent, with an acuminate apex 1.5-2 mm long; innermost phyllaries 6.5-7. Volume 85, Number 3 Ortiz et al. Dicoma in the Horn of Africa 449 X 1-1.4 mm, longer than the outer phyllaries and projected 0-1.5 mm beyond the pappus, oblong- lanceolate, erect, with an acuminate-aristate apex 1-1.7(-2) mm long, and а scarious margin 0-0.4 mm wide; receptacle concave to flat, alveolate, the pits surrounded by a membrane with a very irreg- Шаг dentate margin 1.5-2 mm high. Florets 8-12 per capitulum. Corolla (4—)5 X 1.2-1.5 mm, yel- lowish, with epidermal cell surface “intestine-like” and with short glandular twin hairs; tube 2.2-2.5 X 0.3-0.4 mm; lobes 2.8-3.5 X 0.4—0.5 mm, be- coming recurved, with slender submarginal veins. Stamens exserted for 0-1 mm beyond the corolla; filaments 1-1.1 mm long; collar 0.3-0.5 mm long; anthers 3.3-3.6 mm long; apical appendages ca. 1 mm long, not conspicuously apiculate; anther tails 1 mm long, with retrorse hairs 0.3 mm long and some shorter antrorse hairs at the apex. Style 4.6– 6 mm long, stylar branches 1.3-1.5 mm long, with sweeping hairs forming a subapical ring, covering a surface 0.2–0.3 mm long, the basal ones соп- spicuously longer than the others. Achenes 1.5—2 X 1 mm, turbinate, 10-ribbed, hispid with ascend- ing hairs 0.2-1.8(-2.5) mm long, these inserted be- tween the ribs from the base to the top of the achene, with epidermal glands and superficial bis- eriate glands between the ribs; testa of Dicoma type; pappus isomorphic, of 25—40 barbellate bris- Џез arranged in 1-2(-3) series, the longest bristles 4—5 mm long, the shortest ca. 1 mm long; barbellae 0.15-0.2 mm long. Habitat and distribution (Fig. 3). Known only from the type locality, at 190 m. In gently rolling orange sandhills; bushland with Commiphora cf. ve- lutina and Indigofera ruspolii. Endemic. The principal distinguishing characteristics of Dicoma popeana are its concolorous, generally ru- gose, linear-elliptic leaves with a more or less long pseudopetiole and а pappus of 25-40 barbellate bristles, broadened slightly at the apex and ar- ranged in 1-2(-3) rows. The most morphologically similar species in the Horn of Africa is D. aethiop- ica; in the latter species, however, the leaves (15— 90 x 2-17 mm), involucre (7-11 X 12-20 mm), and corolla ((5-)6.5-8.2 mm long) are all larger, and the pappus comprises 40—100 bristles arranged in 3(—4) rows. Considering the genus as a whole, the most similar species is D. anomala Sond., which differs from D. popeana in the same ways as D. aethiopica. 7. Dicoma schimperi (DC.) Ваш. ex О. Hoffm., in Engl. & Prantl, Nat. Pflanzenfam. 4(5): 339. 1893. Hochstetteria schimperi DC., Prodr. 7: 287. 1838. TYPE: Saudi Arabia. “In Montibus ad Vallem Fatmensem," Schimper 864 (holo- type, С). Annual herb to 45 cm high. Taproot slightly ram- ified. Stem scarcely to highly ramified; the branches striate, glabrous to densely grayish white-tomen- tose, tomentum of simple hairs, with sessile to sub- sessile glands. Leaves (8-)15-50 x (2-)5-16(-18) mm, elliptic-lanceolate to broadly ovate or subor- bicular, with a pseudopetiole (2-)4-8(-15) mm long and auricles at the base; the margins serrulate, with teeth bulbous or not; apex with a 0.3-1 X 0.2-1- mm mucro; both surfaces green to grayish white, glabrous to densely tomentose, with sessile glands. Capitula numerous per plant, generally solitary or 2-12 in lax corymbiform synflorescences, on erect- patent peduncles, 0.5-10 cm long, with 1—4 sub- tending leaves; involucre (4-)5-12(-13) X (10—) 15-20(-23) mm, obconic-campanulate, with 40— 70(-100) phyllaries arranged in 4—5(—6) rows, lin- ear-deltate, stramineous to whitish colored, with a greenish stripe on either side of the midrib, rigid, with an acuminate apex 0.5-1 mm long, pungent, glabrescent to densely tomentose, with broadly scarious margins 0.2-0.5 mm wide, serrulate prin- cipally toward the apical part; the outermost phyl- laries 2-5 X 0.3-1 mm, patent to reflexed, the mar- gins not scarious; the middle phyllaries (3-)5-7(- 9) X 0.5-1(-1.3) mm, erect-patent; innermost phyl- laries 8-11 X 1-1.5 mm, erect, longer than the outer phyllaries and projected 2—4 mm beyond the pappus, all of the phyllaries becoming reflexed when the achenes are ripe; receptacle flat, alveo- late, the pits surrounded by a membrane with a highly irregular dentate to fimbriate margin 0.2-0.3 mm high. Florets (10-)20-40(-50) per capitulum. Corolla 5.5-9 х 0.5-2(-2.5) mm, greenish yellow or greenish white to bluish, with epidermal cell sur- face "intestine-like" and with short glandular twin hairs; tube 2.6-5.5 X 0.4—1 mm; lobes 2.8-3.7 X 0.5 mm, becoming recurved, with thin slender sub- marginal veins. Stamens exserted for 0.5-1.5(-2.2) mm beyond the corolla; filament 0.8-2.5 mm long; collar 0.4—0.5 mm long; anthers 3.4—4.7 mm long; apical appendages ca. 1-1.7 mm long, apiculate; anther tails 1-1.3 mm long, with retrorse hairs 0.3— 0.5 mm long and some shorter antrorse hairs at the apex. Style 5—8.3 mm long, the branches 1-1.5 mm long, with sweeping hairs forming a subapical ring, covering a surface 0.4—0.6 mm long, the basal ones longer than the others. Achenes 1.5-2.5 X 1-2.2 mm, turbinate, (8—)10(—12)-ribbed, hispid, with as- cending hairs 0.3-2 mm long, inserted between the ribs from the base to near the top of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pap- pus isomorphic, of (4-)8-10(-11) rigid, barbellate bristles, broadened toward base (or bristlelike Annals of the Missouri Botanical Garden 450 45° +. e 4. Distribution of Dicoma schimperi in т TA qe Africa: D. schimperi subsp. schimperi (8). D schimperi subsp. cinerea (A). scales), arranged in one series, each bristle ( 7 X 0.2-0.5(0.6) mm. 4-)5- The most characteristic feature of Dicoma schim- peri is its pappus, comprising (4-)8-10(-11) rigid, flattened bristles (or scales). Other characters wor- thy of note are its phyllaries, which flexed when the achenes are ripe, and the relatively narrow depressions between the ribs of the achenes. Most other characters show considerable intraspecific variation. No other species of Dicoma shows particular morphological similarities with D. schimperi. Some authors, such as Cabrera (1977), have con- sidered the above-mentioned characters, particu- larly those relating to pappus morphology, to be suf- ficient grounds for placing this taxon in a separate genus, namely the monotypic Hochstetteria. Nev- ertheless, more recent treatments (Hansen, 1991; Bremer, 1994) have maintained Hochstetteria as a synonym of Dicoma, in accordance with Hoffman (1893), who considered Hochstetteria as a section of Dicoma. Cufodontis (1967) cited D. schimperi from Ethi- opia, though we have not seen any specimens of this taxon from that country. Habitat and distribution (Fig. 4). In the Horn of Africa, known only from Djibouti (Audru et al., 1994) and the north of Somalia at 714—2000 m. Principally in gypseous and limestone soils; in Aca- cia-Commiphora bushland. Djibouti, Egypt, Sudan, Somalia, and the Arabian Peninsula. ecome re- In our description of Dicoma cinerea (Ortiz & Rodríguez-Oubiña, 1994), we stated that it was closely related to D. schimperi. Subsequently, how- ever, we have become aware of the existence of intermediate specimens [Yemen. Hadhramaut, Po- pov GPH610 (BM), 11 Dec. 1947, Thesiger s.n. (BM)], suggesting that D. cinerea should be consid- ered a subspecies or variety of D. schimperi. We favor the subspecies option, since D. schimperi con- sistently exhibits a number of distinctive features (glabrous or glabrescent stems, leaves, and phyl- laries; elliptic-lanceolate leaves with non-pustulate teeth; etc.) throughout its geographic range (Dji- bouti, Egypt, Sudan, Somalia, and the Arabian Pen- insula). Individuals that deviate from this pattern are found only in two small areas, in the northeast of its range in the Horn of Africa (see Fig. 4), and in the region of Hadhramaut in Yemen. These two areas were contiguous until their separation at the end of the Miocene (Mattauer, 1967), by which time the Asteraceae were already fully differentiated (Graham, 1996). Individuals of D. schimperi from Hadhramaut are densely tomentose and have some ovate leaves, and appear to be intermediate be- tween this taxon and D. cinerea. In our opinion, the treatment of D. cinerea at subspecies rank is jus- tified purely by the geographic distribution of the two taxa. In addition, the specimens from north- eastern Somalia that we described as D. cinerea are sharply defined by their densely tomentose stems, leaves, and phyllaries, and broadly ovate to sub- orbicular leaves with pustulate teeth. KEY TO THE SUBSPECIES OF DICOMA SCHIMPERI la. Plants glabrous or glabrescent; leaves elliptic- lanceolate with non-pustulate teeth: apical mucro 0.3-0.5 X 0.2-0.3 mm _ m schimperi subsp. schimperi . Plants densely ота leaves broadly ovate to и ‘ular with pustulate teeth: apical mu- 0.5-1 X 0.3-1 mm — — - . D. schimperi subsp. cinerea 7a. Dicoma schimperi subsp. schimperi Plant glabrous to glabrescent. Leaves (8—)15—50 x (2-)5-13(-18) mm, elliptic-lanceolate; leaf mar- gins serrulate, with non-pustulate teeth and an api- cal mucro 0.3-5 X 0.2-0.2 mm; both surfaces greenish, bee ous to glabrescent. Involucre with 40-)70-100 phyllaries, glabrous to glabrescent. Corolla 2. mm long. Achenes 1.8-2.5 х 12- 1.8 mm; pappus of (5—)9—10 bristles, 4-5 mm long. gm Habitat and. distribution (Fig. 4). In the Horn of Africa, known only from Djibouti (Audru et al., 1994) and the north of Somalia at 714—1000 m. Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa Principally in gypseous and limestone soils, stony or not; in Асасіа-Соттірһоға bushland. Djibouti, Egypt, Sudan, Somalia, and the Arabian Peninsula. Spe wies. eie SOMALIA. Without exact local- ity: Goda, 1, Chedeville 411 (ЕТ), 518 (FT). Bari: 3 km жың ж on road | Boosaaso, Bally & Melville 15857 (К); Gardo, 9°32'N, 49*07' E, 714 т. Gillett 23405 (K). Togdheer: Wagga Minis 1897, Lort Phillips s.n. (BM, K). 7b. Dicoma schimperi subsp. cinerea (S. Ortiz & Rodr. Oubiña) S. Ortiz € Rodr. Oubiña, comb. nov. Basionym: Dicoma cinerea 5. Ortiz & Rodr. Oubiña, Canad. J. Bot. 72: 1479. 1994. TYPE: Somalia. Sanaag: Einand, 11?02'N, 48°55’E, 2000 m, 21 Aug. 1957, Newbould 1013 (holotype, K). Plant densely tomentose. Leaves 5-10 X 6-16 mm, broadly ovate to suborbicular; leaf margins serrulate, with pustulate teeth and an apical mucro 0.5-1 X 0.3-1 mm; both surfaces grayish white, densely tomentose. Involucre with 40—60 phylla- ries, densely tomentose. Corolla 7.5—7.7 mm long. Achenes 1.5-2.5 X 1-2 mm. Pappus of (4—)8—10(— 11) bristles, 4—7 mm long. Habitat and distribution (Fig. 4). Known only from the type locality in the north of Somalia at ca. 2000 m. Limestone scree and boulders. Endemic? There are morphologically intermediate specimens between this subspecies and subspecies schimperi in the Arabian Peninsula. e Dicoma scoparia Rodr. Oubiña & S. Ortiz, Bot. J. Linn. Soc. 119: 61. 1995. TYPE: So- malia. Mudug: Central Massif, about 18 km from the sea, 6%01'N, 48°47'E, 35 m, 27 May 1989, Gillett, Hemming & Watson 22175 (ho- lotype, Shrub to 25 cm high. Taproot slightly ramified. Stem highly ramified; the branches striate, with a grayish white tomentum of simple hairs, principally along the grooves. Leaves (0.8—)1.1-1.9(-3.1) х 0.5—0.9(-1) mm, sessile, deltoid to spathulate; the margins entire; apex acute; upper surface greenish, glabrous to tomentose; lower surface greenish, gla- brous, with 3 conspicuous veins. Capitula numerous per plant, solitary or two on erect to erect-patent peduncles mm long, with 1-2 subtending leaves; involucre 6.58 X (3.8-)4-4.5(-5.5) mm, narrowly cylindric, with 20—30 phyllaries arranged in 5-6 rows, stramineous, with a broad greenish stripe on either side of the midrib and brownish pur- ple at the apex, rigid, acuminate, pungent, glabrous to scarcely tomentose, the margins entire to shortly serrulate-fimbriate, principally toward the apical part, not scarious or with very narrow gins; the outermost phyllaries 0.5-1.5 X 0.8-1.5 mm, broadly deltate, patent to erect-patent; middle phyllaries 3-4 X 1.3-1.5 mm, deltate-lanceolate, erect, with an acuminate-aristate apex 0.2-0.3(-0.5) mm long; innermost phyllaries 5—5.5(-6) X (1.3-) 1.5-1.7(-1.9) mm, longer than the outer phyllaries апа 3-4(-5) mm shorter than the pappus, oblong- lanceolate, erect, with an acuminate-aristate apex 0.4-0.5(-0.6) mm long, the scarious margin 0-0.1 mm wide; receptacle concave, alveolate, the pits sur- rounded by a membrane with an irregular dentate margin 0.2—0.5 mm high. Florets 5—7 per capitulum. Corolla 6.5 Х 1.5 mm, pale lilac to purple, with epidermal cell surface “intestine-like” and with short glandular twin hairs; tube 2.5-3 X 0.4-1 mm; lobes 3-3.8 X 0.5-0.6 mm, becoming recurved, with slender, submarginal veins. Stamens exserted for 1— 1.5 mm beyond the corolla; filaments 2 mm long; collar 0.5—0.6 mm long; anthers 3.9—4.2 mm long; apical appendages ca. 1 mm long, conspicuously long apiculate; anther tails 1.3-1.5 mm long, with retrorse hairs О antrorse hairs at the apex. Style 7-7.5 mm long, sty- lar branches 1-1.1(-1.2) mm long, with sweeping Scarious mar- 4—0.6 mm long and some shorter hairs forming a subapical ring, covering a surface 0.4 mm long, the basal ones 2. longer than the others. Achenes 3.5—3.7 mm, turbinate, 10- ribbed, hispid with 41 1. (0.3-)0.4-2 mm long, these inserted between the ribs from the base to the top of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pappus isomorphic, of 65—75 barbel- late bristles arranged in 3—4 series, the innermost bristles broadened toward the base, the longest bris- tles 6.5—7 mm long, the shortest bristles 0.3—0.5 mm long; barbellae 0.1-0.15 mm long. Habitat and distribution (Fig. 5). Known only from the Mudug region near the sea at 35-200 m. Limestone; low grassy vegetation. Endemic. датан ep examined. SOMALIA. Mudug: E of Gawe 3 km on road from Hobyo to Wisil, 150— 200 m, Thulin 3 Abdi Dahir 6620 (K). Dicoma scoparia is one of the most morphologi- cally distinctive of all species of the genus. Impor- tant characters include the very small (generally less than 2 mm long), deltate to spathulate leaves with three longitudinal nerves, and the one or two long-pedunculate capitula with narrowly cylindrical involucres conspicuously overtopped (by (3-)4-5 mm) by the pappus. The achenes (3.5-3.7 X 1 mm) are turbinate, but considerably narrower than is 452 Annals of the Missouri Botanical Garden Distribution of Dicoma scoparia (Ф) and D. sessiliflora subsp. stenophylla (À) in the Horn of Africa. typical for the genus. No other species of Dicoma shows particular similarities with D. scoparia. 9. Dicoma sessiliflora Harv. subsp. stenophylla G. V. Pope, Kew Bull. 46: 701. 1991. TYPE: Nigeria. Lokoja, Mt. Patti, Dalziel 33 (holo- type, Perennial herb to 95 cm high, from woody root- stock. Stems annual, simple, solitary, with 4-6 sides in transverse section, covered with a tomen- tum of long, mixed, grayish white, simple hairs which form a sheath around the stem, with sessile to subsessile glands, often with glabrous to gla- brescent areas. Leaves 110-130 X 15-37 mm, el- liptic to oblanceolate, sessile to subsessile; the mar- gins sparsely serrulate; apex acute to subobtuse; upper surface green, glabrous to glabrescent; lower surface with sessile to subsessile glands, densely grayish white-tomentose. Capitula solitary, terminal (in the sole specimen known from the Horn of Af- rica) or few per plant, sessile (in this specimen) or short-stalked, with 0—1 subtending leaf; involucre 33 X 33 mm, obconic-campanulate, with 100-120 phyllaries arranged in 10-12 rows, greenish-stra- mineous, rigid, acuminate, pungent, glabrous; the margins slightly revolute, entire to shortly serrulate principally toward the apical part, with an acumi- nate apex 1-1.5 mm long; outermost phyllaries 6.5-7 X 2.3-2.5 mm, deltate, patent to erect-pat- ent, the margins not scarious; middle phyllaries 20-23 X 3-5 mm, lanceolate, erect to erect-patent, the scarious margin 0-1 mm wide; innermost phyl- laries 18-20 X 4—5 mm, shorter than the adjacent outer phyllaries, as long as the pappus, lorate, erect, white scarious; longest phyllaries projected (3-)5-10( concave, alveolate, the pits surrounded by а mem- brane with an irregular dentate-fimbriate margin 1- 2 mm high. E 30-50(-65) per capitulum. Co- rolla 9.3-9.6 X 1.5-1.6 mm, yellowish, with epi- dermal cell б slightly transversely undulate- striate to smooth and long glandular twin hairs; tube 3-4 X 0.6-1.2 mm; lobes 6 X 0.5-0.6 mm, straight, with thick marginal veins. Stamens little or not exserted; filaments 1.5-2.5 mm long; collar 1 mm long; anthers 8.5-9 mm long; apical append- ages ca. 1.7-2 mm long, not apiculate; anther tails 4.5—4.6 mm long, with retrorse hairs 0.3-0.5 mm and without antrorse hairs at the apex. Style 9-9.1 mm long, stylar branches 2.2-3 mm long, with sweeping hairs forming a subapical ring, covering a surface 0.5 mm long, all the sweeping hairs of a similar length. Achenes (immature) 3 X 2-2.5 mm, obovoid, not conspicuously ribbed, hispid, with as- cending hairs 0.3—4(—5) mm long, with a bulbous, glandular base, inserted all around the achene from its base to near its top; without epidermal glands and superficial biseriate glands; testa of Gochnatia type, with lateral and basal walls of the testa epi- dermis strengthened, u-shaped in cross section; pappus isomorphic, of 70-80(-90) plumose bristles arranged in 3—4 series, the longest bristles 9-1 mm long, the shortest 2-3 mm long; plumose bar- bellae 0.5-1 mm long. Habitat and distribution (Fig. 5). Known from Ethiopia from only one locality, at 1500 m. Savan- na. Senegal, Guinea Bissau, Ghana, Togo, Nigeria, Central African Republic (Pope, 1991), Sudan, Ethiopia, and Uganda —15) mm beyond the pappus; receptacle Specimen examined. ETHIOPIA. Welega: Apiye near bridge crossing Didessa river, 1500 m, W. J. J. O. de Wilde 8941 (K). The principal diagnostic characters of Dicoma sessiliflora subsp. stenophylla include its elliptic to oblanceolate leaves, very large by comparison with those of other species of our area (110-130 х 15- 37 mm), together with its very large involucres (33 X 33 mm in the single capitulum examined), phyl- laries without a midrib, white, scarious innermost phyllaries shorter than the adjacent outer phylla- ries, florets with straight corolla lobes, slightly transversely undulate-striate to smooth epidermal cell surfaces, long anthers (8.5-9 mm), anther tails without apical antrorse hairs, stylar sweeping hairs of uniform length, achenes not conspicuously ribbed, without epidermal glands or superficial bi- Volume 85, Number 3 1998 Ortiz et al. 453 Dicoma in the Horn of Africa seriate glands and with hairs with a bulbous, glan- dular base inserted all around the achene, Goch- natia-type testa, and pappus of plumose hairs. Most of these characters are shared with other species of section Pterocoma, all of which have very similar morphology; differentiation is largely on the basis of vegetative characters, particularly leaf morphol- ogy (see Pope, 1991). Of the taxa of this section, the most similar to D. sessiliflora subsp. stenophylla is obviously D. sessiliflora subsp. sessiliflora; how- ever, this taxon has phyllaries with a conspicuousl scarious border, while the outermost phyllaries are reflexed. Dicoma elliptica Pope is very close to D. sessiliflora, and can be differentiated from D. ses- siliflora subsp. stenophylla by the same characters that distinguish subspecies stenophylla; in addition, its leaves are generally linear-elliptic and subco- riaceous with a conspicuously revolute margin (see Pope, 1991). We know of only one sheet of Dicoma sessiliflora subsp. stenophylla from Ethiopia, though one of us (Mesfin Tadesse) considers that it is probably not uncommon there. Pope (1991) did not cite this sub- species from Sudan, Ethiopia, or Uganda. The single Ethiopian specimen has only one ca- pitulum and this is immature, so it was not possible to assess the relative length of the pappus ensem- ble, the shape of the receptacle pits, the number of florets per capitulum, etc., without damaging the specimen. These data were thus evaluated from specimens from other countries. 10. Dicoma somalensis S. Moore, J. Bot. 37: 60. 1899. TYPE: Somalia. Togdheer: Wagga Mountain, above Upper Shiikh, 6000 ft., 1 Feb. 1897, Lort Phillips s.n. (holotype, BM; isotype, K). Figure 6. Vernacular names. Adah (Gillett 4002), Gundre (Godding 87), Ma-adadi (Gillett 4620, Cufodontis (1967)), Bastireh (Drake-Brockmann s.n.). Shrub 50 cm high, highly ramified; the branches striate, glabrescent or covered with a grayish white tomentum of simple hairs, with sessile to subsessile glands. Leaves (3-)8-20(-30) x (0.8-)1-1.5(-2.5) mm, linear to linear-oblanceolate, attenuate at base and decurrent on a pseudopetiole 0-2(-3) mm long; the margins somewhat revolute, scarcely serrulate to the apical part; apex mucronate; upper surface green, glabrescent to tomentose, rugose, with ses- sile glands; lower surface generally densely grayish white-tomentose, without sessile glands. Capitula numerous per plant, on peduncles erect to erect- patent, 2-10 cm long, normally without subtending leaves; involucre 5-13 X (10-)15-20 mm, broadly campanulate, with (100-)120-140 phyllaries, these mainly squarrose, arranged in -1 mineous, with a darker stripe on either side of the midrib, acuminate-pungent, glabrous, minutely stri- gose, the margins entire to shortly ciliolate or fim- briate principally toward the apical part; outermost phyllaries 1.5-2 X patent, with an acuminate apex ca. 1 mm long, the margins not scarious; middle phyllaries 5-7 X 1.5- 2 mm, oblong-lanceolate, squarrose, with an acu- minate apex 2-2.5 mm long, the scarious margin 0.2-0.3(0.5) mm wide; innermost phyllaries (7-)8- 10(-13) X 1.5-2 mm, linear-lanceolate, erect, lon- ger than the outer phyllaries and projected (0-)1- 2 mm beyond the pappus, with an acuminate-aris- tate apex 0.5-2 mm long, the scarious margin 0.3— 0.5 mm wide; receptacle flat, very slightly alveo- late, pits surrounded by a membrane 0—0.15 mm high. Florets (13-)16-20 per capitulum. Corolla 7— 1.5 mm, white to purplish white, with epider- mal cell surface “intestine-like” and with short glandular twin hairs; tube 4-5 X 0.5-0.6 mm; lobes 3-3.5(-4) 0.5-0.7 mm, becoming ге- curved, with slender submarginal vein. Stamens ex- serted for 1–2.5 mm beyond the corolla; filaments 1.7-1.8 mm long; collar 0.6–0.8 mm long; anthers 5.5-7 mm long; apical appendages ca. 1.2 mm long, narrowly apiculate; anther tails ca. 2 mm long, with retrorse hairs 0.1-0.2 mm long and some shorter antrorse hairs at the apex. Style 9-10 mm long, stylar branches 2-3 mm long, with sweeping hairs forming a subapical ring, covering a surface 0.3—0.5 mm long, all hairs of similar length or the basal ones longer than the others. Achenes 1.5-2 X 1-1.3 mm, turbinate, (8—)10-ribbed, hispid, with ascending hairs 0.3-2.5 mm long, these inserted between the ribs on the inferior third of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pap- pus isomorphic, of 100—130 barbellate bristles ar- ranged in 4—5 series, the innermost bristles broad- ened toward the base, the longest bristles 7-8(-9) mm long, the shortest 1 mm long; barbellae 0.1— 0.2 mm long. rows, stra- 1 mm, deltate, patent to erect- Habitat and distribution (Fig. 7). Known only from northwestern Somalia at 1250—1868 i cipally on rocky slopes in limestone areas. Endem- ic. m. Prin- Additional specimens examined. SOMALIA. exact locality: 26 Feb. 1913, Drake- or s.n { ка Brockmann (K). naag: у m, Gillett r Watson 23761 (K); Medishe Valley icto, т & Gilliland 942 (BM, K).Togdheer: 9°57'N, 4 m, Gillett & Watson 23616 (K); Upper Sikh. 2000 К. p рн 75 (BM); Without . (К); 454 Annals of the Missouri Botanical Garden SoH АЛАЙДА У | j Figure 6. Dicoma somalensis (Gillett 4620 (K)). —A. Habit. —B. Capitulum. —C. Leaf. —D. Floret. —E. Achene with pappus. Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa ~ ure 7. Distribution of Dicoma ааш (9) and D. 2. (A) іп the Horn of Afric Shiikh, 4500 ft., Ironside ie 5/73/2 (K). Woqooyi Gal- b Abassa, 1 Boorama to Djibouti, 4900 ft., Bally B9967 (K); сане 9°33'N, 44%0'E, 4300 ft., Gillett 4002 (K); Andoba, 106 N, 43?0' E, 5400 ft., Gillett 2 , К); 35 km NW of Hargeysa, 941'N, 44719'Е, 1250 m, Gillett 22845 (К); Dardode, 10°07'М, 2°58'К, 1868 m, Godding 87 (K). The principal diagnostic characters of Dicoma somalensis are its linear to linear-oblanceolate leaves, involucre with numerous ((100-)120-140), glabrous, mainly squarrose phyllaries arranged in 6—8(-10) rows, and pappus of 100-130 barbellate bristles 7-9 mm long. The most morphologically similar species in the Horn of Africa is D. hindi- ana; indeed, as mentioned above, all specimens of D. hindiana at K had been labeled as D. somalen- sis. However, there are marked morphological dif- ferences between the two species: D. hindiana has fewer phyllaries per capitulum (35—70); these are pubescent, and the innermost are squarrose. In ad- dition, D. hindiana has a shorter corolla (ca. 5 mm), longer (2.7-3 mm), narrowly turbinate achenes, and fewer (60—70) and shorter (4.5—5.2 mm) pappus bristles. Outside the Horn of Africa, the only spe- cies with clearly squarrose phyllaries (like D. so- malensis and D. hindiana) is the Angolan endemic D. squarrosa; as noted for D. hindiana, however, D. squarrosa shows a number of clear differences. Cufodontis (1967) cited D. somalensis from Ethi- opia, but we have not seen any specimens from that country. 11. Dicoma thuliniana S. Ortiz, Rodr. Oubifia & Mesfin, sp. nov. TYPE: Somalia. Bari: 20 km from Dhurbo on road toward El Gal, 11?33'N, 50?19'E, 900 m, 22 Nov. 1986, Thulin & War- fa 5975 (holotype, UPS). Figure 8. iffert а simili Dicoma paivae caulibus erectis, foliis rectis atque (5-)15-22(-26) mm longis, bracteis involu- cralibus longis acuminatis, acumine (1.2-)1.5—3.5 mm longo, atque pappo 6-7 mm lon Shrub to 20 cm high, ramified; the branches stri- ate, covered with a dense grayish white tomentum of simple hairs, with sessile to subsessile glands. 15-22(-26) x 7-10 (-12) mm, linear; the margins revolute, entire; apex with a short spine 0.2-1(-1.2) mm long; both surfaces densely grayish white-tomentose, with sessile to subsessile glands. Capitula numerous per plant, solitary, on leafy, erect to erect-patent peduncles, mm long, with (0-)1-3 subtending leaves; involucre 10-11 X 11-12 mm, campanulate, with 40—50 phyllaries arranged in neous, with a darker stripe on either side of the midrib, long acuminate-aristate, pungent, tomen- tose, the margins entire to shortly serrulate prin- Leaves sessile, (5— cipally toward the apical part, not scarious; outermost phyllaries 2-2.5 X 0.8-1.2 mm, deltate- oblong, erect-patent, with an acuminate apex 1.2- 2 mm long; middle phyllaries 3-6 X 0.8-1.2(-1.5) mm, ovate-lanceolate, erect-patent, with an acu- minate-aristate apex mm long; innermost phyllaries 6-8 X 1.5-3 mm, longer than the outer phyllaries and projected 1.5-2 mm beyond the pap- pus, oblong-lanceolate, erect-patent, with an acu- minate-aristate apex 2.5-3.5 mm long. Florets un- known. Achenes (2-)3-3.5 X 1.5-2 mm, turbinate, 10-ribbed, hispid, with ascending hairs 0.3-2.5(- 3) mm long, inserted between the ribs from the base to the top of the achene, with epidermal glands and superficial biseriate glands between the ribs; testa of Dicoma type; pappus isomorphic, of 100-130 barbellate bristles arranged іп (4—)5—6 series, the innermost bristles broadened toward base, ca. 0.4 mm wide, the longest bristles 6-7 mm long, the shortest ca. 1 mm long, barbellae 0.05-0.15 mm ong. Habitat and distribution (Fig. 7). Known only from the type locality, at ca. 900 m. South-facing limestone slopes; vegetation of sparse low bushes. Endemic. The most distinctive characters of Dicoma thu- liniana are its solitary, linear leaves with revolute margins and short, apical spine, capitula on leafy, erect to erect-patent, peduncles 20-50 mm long, large (10-11 X 11-12 mm) campanulate involucres 456 Annals of the Missouri Botanical Garden DRAE AU O а ај ш “Cee Figure 8. Dicoma thuliniana (Thulin & Warfa 5975 (UPS)). —A. Habit. —B. Capitulum. —C. Leaf. —D. Phyllary apex. —E. Achene with pappus. Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa 457 with 40—50 phyllaries, and deltate to oblong-lan- ceolate phyllaries with an acuminate-aristate apex 1.2-3.5 mm long. The most morphologically similar species in the Horn of Africa is D. paivae, which can be readily distinguished by its twisted branches forming dense whitish cushions, slightly recurved and often subfasciculate leaves, sessile or subses- sile capitula, obconic-campanulate involucres (6— 6.5 X 3-4 mm), 25-30 deltate to deltate-lanceolate phyllaries with pungent, acuminate apices 0.2-1 mm long (this feature being particularly distinc- tive), and shorter pappus (3-5 mm). Of the species of Dicoma from outside our area, D. cana Balf. f., from Socotra, has leaves showing certain morpho- logical similarities, though in other respects ex- hibits clear differences (mostly similar to those observed between D. cana and D. paivae). Fur- thermore, D. thuliniana does not appear to form the dense cushions characteristic of D. paivae and D. cana. This species is named after Mats Thulin (UPS), one of the collectors of the type, one of the most knowledgeable experts on the Somalian flora, and editor/author of the Flora of Somalia. 12. Dicoma tomentosa Cass., Bull. Sci. Soc. Philom. Paris. 1818: 47. 1818. TYPE: Sene- gal, Adanson s.n. (holotype, P). Dicoma 47 DC., Prodr. 7: 36. 1838. TYPE: “East- * Wallich s.n. (lectotype, here designated, С). Dicoma ==. Mattei, Boll. Reale Orto Bot. ig mo 7: 112. 1908. TYPE: Somalia. Goscia, Torda, 2 jan 1907, M. 96 (holotype, PAL). Vernacular names. Віѕсіаг (Cufodontis, 1967), Gaiocum (Cufodontis, 1967), Tebe-mro (Mateos Er. 68), Yset-melase (Soda Ash Project 2). Annual herb to 60 cm high. Taproot slender, lightly ramified. Stem scarcely to highly ramified; the branches striate, generally reddish purple, cov- ered with a grayish white tomentum of simple hairs, with sessile to subsessile glands, often with gla- brous to glabrescent areas. Leaves (10—)20—50(—80) X (2-)4-10(-16) mm, linear to linear-oblanceolate, rarely linear-elliptic, conduplicate, attenuate at base and decurrent on a pseudopetiole 0—5 mm long; the margins callose-serrulate; apex 0.5-1.5 mm long, apiculate to aristate with a short spine; upper surface greenish, glabrescent, with sessile glands; lower surface with sessile glands, generally densely grayish-white-tomentose, rarely greenish and glabrescent. Capitula numerous per plant, gen- ) subtending leaves, sessile or on peduncles, erect-patent, 1—5 (-8) mm long; involucre (11-)12-15(-17.5) X (10—) erally solitary in leaf axils, with 1(—4 12-20(-25) mm, obconic-campanulate, with 30- 45(-50) phyllaries arranged in rows, strami- neous to purple, with a darker stripe on either side of the midrib, rigid, long-acuminate, pungent, gla- brescent to densely tomentose, minutely strigose, the margins entire to shortly serrulate-ciliolate, principally toward the apical part; outermost phyl- laries 224.5 X (0.5-)0.7(-1) mm, deltate to ovate- oblong, patent to erect-patent, with an acuminate apex 1—3.5 mm long, the margins not scarious; mid- dle phyllaries 9-17 X 1.5-2.5 mm, oblong-lance- olate, erect-patent, with an acuminate-aristate apex 4-7 mm long, the scarious margin 0.2-0.3(-0.5) mm wide; innermost phyllaries 11-14 X 2-3 mm, generally of similar length to or shorter than the adjacent outer phyllaries and projected 2-5(-6) mm beyond the pappus, oblong-lanceolate, erect, with an acuminate-aristate apex 2-5 mm long, the scarious margin up to 0.5 mm wide; receptacle con- cave, alveolate, the pits surrounded by a membrane with a highly irregular dentate margin 0.7 mm high. Florets (7-)10-12(-15) per capitulum. Corolla 6.5— X 1.5-2 mm, white, lilac, or violet, with epider- mal cell surface “intestine-like” and with short glandular twin hairs; tube 3.4—3.7 X 0.7-1 mm; lobes 3-3.7 X 0.3-0.5 mm, becoming recurved, with slender submarginal veins. Stamens exserted for 0.2-1.5 mm beyond the corolla; filaments 1.7— 2 mm long; collar 0.6—0.7 mm long; anthers 3-3.8 mm long; apical appendages ca. 1.2-1.3 mm long, not conspicuously apiculate; anther tails 1–1.3 mm long, with retrorse hairs 0.2-0.5 mm long and some shorter antrorse hairs at the apex. Style 7.5-8 mm long, stylar branches 1-1.5 mm long, with sweeping hairs covering a surface 0.5—0.8 mm long, the basal ones conspicuously longer than the others. Achenes 1.6-3 X 1-1.5 mm, turbinate, (8—)10-ribbed, his- pid, with ascending hairs 0.3-2.5 mm long, these inserted between the ribs from the base to the top of the achene, with epidermal glands and superfi- cial biseriate glands between the ribs; testa of Di- coma type; pappus dimorphic, of 50—80 barbellate bristles arranged in 1-3 external series, the inner- most bristles broadened toward the base, the lon- gest bristles 5-6(-7.5) mm long; the shortest 2 mm long; barbellae 0.1—0.2 mm long; ca. 10 scales ar- ranged in a single internal series, these (5—)6—6.5 (7.5) X 0.5-1 mm, barbellate toward the apex. Habitat and distribution (Fig. 9. | Widespread in Eritrea, Ethiopia, and Somalia at 230-1700 m. Sandy, gravelly, and rocky granitic or calcareous soils, lava flows; in open Acacia woodland, Acacia— Commiphora bushland, Acacia grassland, open habitats, roadsides. Widespread in north Africa 458 Annals of the Missouri Botanical Garden 45* Figure 9. Distribution of Dicoma tomentosa in the Horn of Africa (Egypt) and tropical and southern Africa; also in Socotra, Pakistan, and India Additional specimens examined. ERITREA. Between Galeb Zagle and Malekte, 17%7'N, 38°45’E, 850 ft., Bally 6786 (ETH, K); Cheren, Do-Longoroc, Gandussio 79 (FT); ~ m Tellini 781 (FT). ETHIOPIA. Without exact local- : 1853, Schimper 426 (С, К, P, У). Сато Сога; with- 2: exact locality, 610 m, Sebsebe & Amha 653 (ETH): Nechsar National Park, Arba Minch, Mateos Er. 68 (ETH): between Lake Сато and Lake Abaya, 6%30'N, 37?38' E, n m, Gilbert, Thulin & Aweke 295 (ETH, К, e: Errer Valley, 4 km E of the river, 92147 Х, | 1300. m, Burger 3566 (К); about 30 km N of ide crossing Awash River along road to Gewane, 800 Seegeler 2794 (ETH). Shewa: Sodere, 8°25'N, 30°24’ E, N bank of Awash River, ca. 28 km S of Nazreth, Ash 2106 (К); Erer, ТЕСАМА H-9 (K); Gala Aruss. 1909, Negri 911 (FT); Awash National Park, 950 m, aon 77 (ETH): 1 km W of Вита, on track to Koye, са. 10 km from Мека, 8714"М, 38%55'E, 1700 m, Gilbert & 2. Abate 3121 (ETH, K); Awash National Park, N, 39°55'E, Gilbert & Gilbert 1256a (ETH, K); 1-2 ҚА W of Metahara, 8°55'N, 39%54'E, 1050 m, Gilbert & Thulin 219 (ETH, UPS); ака. 1650 m, Soda Ash Project 2 (ЕТН). nec near Murle Lake, no 1 (FT); 2. : Haramat, Cian- ne noe prope Dsc helad- eze Valley, 3 Nov. 1839, Schimper s.n. (С, К, OXF, P, PI, S, W); "in montibus prope Gageros,” Agow, 3500 ft., 18 hide 1854, Schimper s.n. (G, K, . № Temben Awraja, 11.7 bü along Yec 'hila-Abiy Adi road. T. & Tewolde 8520 (ETH). S Baydhabo, : 259'N, 44°16'E, 230 m, Thulin & Bashir Mohamed 7105 (К, UPS). Gedo: 20 km along road between Garbahaarrey and Qansaxdeheere, 300 m, Thulin т Bashir Mohamed 6917 (UPS); 5 km along road between Garbahaar ж and Luuq. 250 m, Thulin & Bashir uL 6924 (K, UPS). Hiiraan: 5.5 km S of Dharyo Beckett 1361 (К); dade. Y km SW of Aborey, 4°0'N, 45°41'E, 195 m, Kuchar 17601 The principal diagnostic characters of Dicoma tomentosa are its linear-elliptic to linear-oblanceo- late often conduplicate leaves, apically apiculate to aristate with a short spine, its sessile or short-pe- dunculate, generally solitary axillary capitula with obconic-campanulate involucres, deltate to oblong- lanceolate phyllaries with long (1-7 mm) pungent acuminate-aristate apices, style with sweeping hairs covering a surface 0.5-0.8 mm long, and dimorphic pappus of 50-80 barbellate bristles and ca. 10 scales arranged in a single internal series. Of the Dicoma species in the Horn of Africa, only D. ban- gueolensis has a dimorphic pappus of similar com- position; as noted elsewhere, however, that species shows marked differences as regards habit, leaf morphology, and the morphology and size of the phyllaries and achenes. A number of other species, including D. cana and D. spinosa (L.) Druce, and to a lesser extent D. latifolia DC. and D. relhan- ioides Less., have phyllaries with a long-pungent, acuminate apex; however, all of these are shrubs, and in addition present other clear differences from D. tomentosa, including the absence of a phyllary midrib, straight corolla lobes, achenes without prominent ribs and without epidermal glands or su- perficial biseriate glands between the ribs, and an isomorphic pappus. ithin the study area, Dicoma tomentosa shows relatively little morphological variability, though some variation is observed in growth habit (more or less prostrate), degree of ramification, pubescence and coloration of the stems and phyllaries (which may in some cases be purplish), leaf width, and length of peduncles, and number of capitula. The type of Dicoma gnaphaloides Mattei, a name accepted by both Wilson (1923) and Cufodontis (1967), is indistinguishable from D. tomentosa, an we thus include the former name here as a syno- nym. Literature Cited Anderson, L. E. 1954. Ноуегз solution as a rapid E. manent mounting medium for bryophytes. Bryologist 57 -241. Audru, J., n & P. Lebrun. 1994. Les Plantes Vasculaires de la Вене de Djibouti. Flore Шиѕ- a и aris. Bremer, K. 22 4... & Classification. Timber pres Portland. Or Buscalioni, L. & R. Muse ш. 191 3. Beschreibung der Volume 85, Number 3 1998 Ortiz et al. Dicoma in the Horn of Africa 459 von Ihrer Kóniglichen Hoheit der Herzogin Helena von Aosta in Zentral-Afrika gesammelten neuen Arten. Bot. Jahrb. Syst. 49: 457-515. Cabrera, A. L. 1977. Mutisieae-Systematic review. Pp. 1039-1066 in V. H. Heywood, J.-B. Harborne & B. L. Turner (editors), The Biology а oe of the Com- positae. ІІ. Academic Press, Lo Cufodontis, G. ee HL Aethiopae — ње a paie ot Bull. Jard. Bot. Natl. Bel- gique 37, Suppl.: 115-11 Davies, R. A. (editor). 1996. p Kewensis, suppl. 20. Hoyal = Gardens, Kew Edwards, n T. & I. Hedberg (editors). 1995. Flo- ra of Ethiopia ges Eritrea, Vol. 2(2). The National Her- barium of the Addis Ababa University and the Depart- ment of Systematic мйнез of the Uppsala University. Addis Ababa and Орр Graham, A. 1 A она ВИ to the p history of the € pisito: Pp. 123-140 in D. J. N. Hind & H. J. Beentje (editors), Compositae: 2. Proc eed- ings of the International Compositae Conference, Kew Royal UM Gardens, Kew. Grau, J. 198 ie Testa der Mutisieae und ihre syste- matische н Mitt. Bot. Staatssaml. München 69-332. . 1991. Phylogenetic studies in Compositae tribe Mutisieae. Opera —50. Hedberg, 1. & S. Edwards (editors) 1989. Flora of Ethi- i : al Herbarium of the Addis Ab- aba University and the Department of Systematic Bot- Jppsala University, Addis Ababa and Upsala. Hoffmann, О. 1893. Tubiflorae-Mutisieae. In: ingler & К. A. E. Prantl, Die natürlichen al ilien 4(5): 333-350. W. Engelmann, Leipzig. Jeffrey, C. 1967. Notes on Compositae: 2. The DP in ++ DE Africa. Kew Bull. 21: 177-22: Karis, . M. Killersjó & mer. 1992. B yloge- netic 2. of the Cichorioideae (Asteraceae), with emphasis on the Mutisieae. Ann. Missouri Bot. Gard. 79: 416-427. Mattauer, M. 1967. Les deformations de materiaux de l'ecorce terrestre. Herman Editeurs des Sciences et des Art Ortiz. % & L Rodríguez-Oubiña. 1994. Dicoma paivae апа Dicoma cinerea (Asteraceae), two new species from Somalia. Canad. ый Bot. 72: 1478-1481 & Dicoma hindana [rr a new species s Somalia. Nordic J. Bot. 15: 187- 189. ——. 1996а. The identity pF "iot ban- gueolensis (Asteraceae). m 45: 519-5 — а w species ы Dicoma (Asteraceae) боп боа jide Ethiopia. Nordic J. Bot. 16: 277-281. Pope. G. V. 1991. Notes on Dicoma Cass. (Compositae). Kew Bull. 46: Me ES nio cum -Oubiña, J. & S. Ortiz. 1995. Two new species the genus Dicoma тика from Somalia. Bot. J. Linn. ж 1 T hulin, M. (editor) 1993. Flora of Somalia, Vol. 3. Royal dank ›. L. А. 1915, чон, zu den von P uschler in Engl. Jahrb. XLIII. (1909), (1911), XLIX (1913) - L. Suppl. (1914) vee chten Diagnosen afrikanischer Pflanzen. II Ihrer Kóniglichen Hoheit der Herzogin Helena von P ta in Zentral-Afrika po neuen Arten," Com- positae. Bot. Jahrb. Syst. 53: 373—375. Wilson, F. C. 1923. Revision. of the genus Dicoma. Bull. Misc. Aaa Kew 1923: -388. A REVIEW OF THE GENUS PARAGONIA (BIGNONIACEAE)!-? Warren D. Hauk’ ABSTRACT Paragonia (Bignoniaceae) is a genus of two species, P. brasiliensis and P. pyramidata, the latter containing two varieties (var. pyramidata and var. tomentosa). Both s species are lianas with subulate-appressed pseudostipules, lavender to magenta, tubular-campanulate corollas, linear-oblong fruit, and winged seeds. Paragonia pyramidata var. pyramidata is distributed from southern Mex to southern Brazil. Paragonia brasiliensis is known fruiting material, maps of s pyramidata var. pyramidata are provi xico to southern Brazil and Uruguay, whereas P. pyramidata var. tomentosa is restricted only from a few sta species distributions, graphs of flowering and fruiting phenology, and an illustration of P. ided. ates in eastern Brazil. A key to flowering and Paragonia Bureau (Bignoniaceae) is a ditypic genus of lianas with lavender to magenta, tubular- campanulate corollas, linear-oblong fruit, and winged seeds (Fig. 1). It is distinguished from other genera of the liana tribe Bignonieae by a combi- nation of characters that includes stems with four phloem arms in cross section, subulate-appressed pseudostipules, bifid or trifid tendrils, moniliform- puberulent corolla tubes, psilate 3-colporate pollen, the absence of interpetiolar glandular fields (Gentry, 1973, 1977, 1978, 1982a, b; Gentry & Tomb, 1979). Paragonia is generally found in low- land portions of Central and South America and is а common component of tropical moist forest, trop- ical wet forest, and premontane wet forest environ- ments. Paragonia brasiliensis (Baill. A. H. Gentry is a poorly known species restricted to portions of east- ern Brazil (Fig. 2). Paragonia pyramidata (Rich. Bureau var. pyramidata is more wide-ranging (Fig. 3) and morphologically variable than the geograph- ically restricted P. pyramidata var. tomentosa Bu- reau & K. Schum., of south-central Brazil (Fig. 2). This treatment attempts to compile all informa- tion available on Paragonia, notably that obtained by the late Alwyn H. Gentry during his extensive investigations of Bignoniaceae. The maps of geo- graphic distribution and graphs of flowering and fruiting phenology presented here were derived Мм from a database initiated during Gentry's studies of the family HISTORY Paragonia was described by Bureau in 1872 based on Bignonia lenta Mart. ex DC. (1845). How- ever, Bignonia lenta is considered synonymous with a previously described species, Bignonia pyrami- data Rich. (1792), and thus the epithet y ipu takes precedence. A second species, Paragon brasiliensis, was originally described by Baillon i in 1888 as the sole member of the genus Sanhilaria. Paragonia was monotypic until 1976, when Gentry transferred Sanhilaria brasiliensis into Paragonia. Gentry (1976a) evaluated the type of P. brasiliensis and concluded that it was specifically distinct from P. pyramidata because of its softly puberulous, short-petioled leaves, trifid tendrils, narrower inflo- rescence, acute corolla lobes, costate calyx, and compressed fruit that lack the sandpaper-like sur- face of fruit of P. pyramidata (Table 1). However, the puberulence of the type specimen of P. brasiliensis is not manifest in all collections (Gentry, 1976). SYSTEMATICS According to Gentry and Tomb (1979), the gen- era Paragonia, Leucocalantha Rodr., Spathicalyx J. C. Gomes, Manaosella J. C. Gomes, Ceratophytum ! This paper is number 5 of the GE ме шл SERIES, in acknowledgment of contributions to the study of the Bignoniaceae made by Alwyn H. Raven and the Mis project. Financial support arium specimens. valuable contribution to this cial thanks to Linda Wellsey for hee volunteer support of the was provided by the National Science Foundation Sena DEB-9509270). * Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166-0299, U.S ANN. Missouni Bor. GARD. 85: 460—474. 1998. Volume 85, Number 3 Hauk 461 1998 Review of Paragonia 20M Tes, 3. More. Paragonia pyramidata var. pyramidata. —A. Inflorescence and leaves (after Steinbach 428). —B. Seed Te ша 3580). —C. Fruit (after Martínez 15747). 462 Annals of the Missouri Botanical Garden - ; | ia 2 0 pa b 47; on "| ES 55 = | 105 РА М ® о 155 M У У . к "205 -- r LG “255 sf ‘80W 50W 40% g f tosa gun and collections of anomalous specimens of P. pyramidata (squares). —Figure 3 (bottom). Geographic distributions. — Figure 2 (top). Paragonia brasiliensis (circles), P. pyramidata var. tomen- Paragonia = pyramidata var. pyramidata. Volume 85, Number 3 Hauk Review of Paragonia 463 able 1. Features used to differentiate Paragonia py- ramidata and P. brasiliensis (after Gentry, 1976) P. pyramidata P. brasiliensis 1. Tendril tip minutely bi- — Tendril tip minutely trifid fid (rarely trifid) 2. Petioles and petiolules Petioles and petiolules re- well developed duced, obsolescent 3. Leaflets elliptic ог Leaflets narrowly elliptic ovate-elliptic, the to oblanceolate, the apex o minate btuse to acu- apex obtuse Calyx conspicuously ri 4. Calyx ecostate ec I nflorescence racemose- 5. Inflorescence broadly paniculate paniculate 6. Capsule subterete Capsule strongly com- sandpaper-surfaced, pressed, smooth-sur- moderately lepidote faced, densely lepidote (when immature) 7. Corolla lobes rounded Corolla lobes acute 8. Mexico to southern Eastern Brazil (Bahia and Brazil Minas Gerais) Pitt, Tynanthus Miers, and Periarrabidaea А. Samp. may form a natural group because they share pubescent corolla tubes, 2-3(multi)-fid tendrils, and “more or less psilate 3(—4)-colpate pollen." Paragonia pyramidata has psilate, microperforate, 3-colporate pollen with narrow colpi (Tomb & Gen- try, unpublished), whereas the pollen of P. brasi- liensis is unstudied. Simmonds (1954) reported a chromosome count of 2n — 40 for Paragonia pyramidata. Of the 23 genera of Bignonieae cited by Goldblatt and Gentry (1979), only 2 (Mansoa and Pachyptera) have dip- loid chromosome numbers other than 2n — 40. The near uniformity of chromosome numbers in Bignon- ieae supports the monophyly of this lineage (Gold- blatt & Gentry, 1979), but provides little informa- tion regarding relationships among genera of the DISTRIBUTION Paragonia pyramidata is wide-ranging through- out the Neotropics (Figs. 2, 3), typically below 1000 m, although collections extend to 2066 m. Common through all of Central America and the northern half of South America, P. pyramidata var. pyrami- data extends southward to the eastern Andes in Peru and Bolivia, and across Brazil to the eastern shore of South America. The northernmost collec- tions are from Mexico, in Colima and the Yucatán Peninsula. The southernmost collection examined was from Uruguay (adjacent to Buenos Aires, Ar- gentina), with other collections from the Brazilian states of Paraná and Sáo Paulo. Gentry (1973, 1977) reported that P. pyramidata occurs in Argen- tina, but no collections from Argentina were seen in the present investigation. Gentry (1973, 1977, 1978, 1982a, 1982b) reported collections from Guadeloupe (West Indies), but other collections from the Caribbean are not documented. Paragonia pyramidata var. tomentosa is more restricted than variety pyramidata and is found only in south-cen- tral Brazil (Fig. 2). aragonia brasiliensis is more restricted geo- graphically than P. pyramidata var. pyramidata and occupies higher (500-1000 m) and drier portions of eastern Brazil (Fig. 2), i.e., the states of Ceará, Bahia, and Minas Gerais. It is likely that P. brasi- liensis occurs in Pernambuco, Piauí, Rio Grande do Norte, and Paraíba, but collections from these states were not seen. PHENOLOGY Large bees are the primary pollinators of Para- gonia pyramidata, and flower production follows strategy (Gentry, 1976b). “Cor- nucopia" species produce numerous flowers over a oe es s cornucopia period of several weeks, and a wide range of pol- linators are attracted during this period. The cor- nucopia strategy is the most widespread and gen- eralized of the five flowering patterns typical of Central American Bignoniaceae (Gentry, 1974). Gentry (1976b) documented the cornucopia polli- nation strategy for P. pyramidata in tropical moist forest, tropical wet forest, and premontane wet for- est environments. Graphs of flowering and fruiting phenology for P. pyramidata var. pyramidata show that flower and fruit production occur throughout the year (Figs. 4, 5). Peaks in the number of flowering and fruiting collections were in February and May, respectively. However, the wide geographic range of P. pyrami- data var. pyramidata (Fig. 3) may obscure more localized phenological patterns. Figure 5 presents flowering and fruiting phenology for collections from Panama, Colombia, and Venezuela only. Mean monthly precipitation in centimeters for Cristóbal, Panama, and Caracas, Venezuela, was plotted to as- sess floral and fruiting phenology relative to pre- cipitation. А marked peak in flowering occurs in ebruary during the dry season, with a smaller peak in fruit production occurring in May during the first part of the wet season. However, numbers of collections from south of the equator did not peak during the dry season (not graphed). Thus, the floral and fruiting phenology of P. pyramidata var. 464 Annals of the Missouri Botanical Garden No. of Specimens 6 7 Month = Flower - Fruit No. of Specimens or Precipitation 6 7 Month = Flower y Fruit — Caracas >< Cristobal No. of Specimens or Precipitation 6 7 Month Ф Flower w Fruit >e Belo H. Fig —Figure 4 (top). Flowering and fruiting phenology of P. d var. Ж for all collections. Figure. 5 la. e and fru ruiting phenology of P. pyramidata var. pyramidata from Panama, Colombia, and | = Mean monthly precipitation in cm is plotted (1X) for кш. 5. апа (2X) for Cristóbal, Panama. re 6 (bottom). Flowering and fruiting phenology of P. pyramidata var. tomentosa. Mean monthly precipitation in cm is ui (0.25 X) for Belo Horizonte, Brazil. Volume 85, Number 3 1998 Hauk 465 Review of Paragonia pyramidata appears to be influenced by regional climatic conditions. There were few fertile collections of P. pyrami- data var. tomentosa (Fig. 6). imens were all collected at the beginning of the wet season, between August and September. Fruiting collections were limited to the latter part of the wet season, from January to April. Although these data are preliminary, they indicate that P. pyramidata var. tomentosa differs phenologically from Р. pyr- amidata var. pyramidata. he six flowering spec- Assessments of flowering and fruiting phenology of P. brasiliensis did not reveal clear trends because of the limited number of fertile collections avail- able; four flowering collections are known from Jan- uary, one from June, and two from November. Of the two known fruiting collections, one is from Jan- uary and the other is from February. Flowering and fruiting probably peak during the first few months of the year, but additional collections are needed to confirm this. ECONOMIC AND ETHNOBOTANICAL USES Reports of uses for Paragonia are limited. Gen- try (1992) cited the use of Paragonia as a treatment for stomach and intestinal problems. Paragonia pyramidata is one of several lianas used by native eoples “para tomar agua” (Gentry, in press). Mac- bride (1961) reported that the stems of P. pyrami- data are used for lashings. MATERIALS AND METHODS Gentry compiled a private database of label in- formation from herbarium specimens he collected and from specimens at other herbaria that he ex- amined personally. Gentry's database has been in- corporated into the Missouri Botanical Garden da- tabase-management system, TROPICOS, which also contains label information for all other Para- gonia specimens housed at MO. All types were as- sumed to have been seen by Gentry unless other- wise noted. Gentry did not siwaya designate types as “holotype,” “isotype,” or “syntype,” and the des- ignations presented here are based upon inferences drawn from Gentry's work and the original litera- ture; these type designations were not based on per- sonal verification of specimens at the various her- baria. Uncertainty of the type designation is indicated by a question mark. ata used for mapping and phenology were downloaded from TROPICOS. For records with no latitude/longitude coordinates in TROPICOS, ap- proximate coordinates were obtained from gazet- teers produced by the U.S. Board on Geographic Names, Office of Geography, Dept. of the Interior. Distribution maps were produced using the com- puter program VERSAMAP 1.51 (C.H. Culberson, Newark, Delaware, 1991-1995). Graphs of flower- ing and fruiting phenology were generated using the computer program Quattro Pro 7.00 (Corel Inc., 1996). Phenology is reported as the number of flow- ering specimens collected during each month of the year; detailed studies of flower production (per plant, per population, per species, or per time pe- riod) have not been conducted. Amounts of precip- itation used in the graphs of phenology were ob- tained from Agroclimatological Data for Latin America and the Caribbean (FAO, 1985). TAXONOMIC TREATMENT Paragonia Bureau, Bull. Soc. Bot. France 19: 17. 1872. TYPE: Bignonia lenta Mart. ex DC. [= Paragonia pyramidata (Rich.) Bureau]. Sanhilaria Baill., Hist. Pl. 10: 27. 1888 [1891], non Le- andro (1838). TYPE: Sanhilaria brasiliensis Baill. [= P. brasiliensis (Baill.) A. H = ry]. garaia Pichon, Bull. Soc. France 92: 228. 1945. TYPE: Sanhilaria doses Ваш. [= P al (Baill.) A. H. Gentry]. Lianas; stems woody with 4 phloem arms in cross section; branchlets terete, lenticellate, with inter- petiolar glandular fields lacking, glabrate to lepi- ote or densely puberulent; pseudostipules subcon- ical, subulate (basally expanded with acuminate tips), curved inward and appressed or nearly ap- pressed to branchlets or angled away from branch- let and nearly appressed to the subtending petiole, eglandular, glabrate to puberulent. Leaves opposite, petiolate, estipulate, 2-foliolate with oppositely ar- ranged simple leaflets and a bifid or trifid (rarely simple) terminal tendril (or tendril scar); petioles and petiolules puberulent, the petiolules sulcate; distal adaxial petiolar glandular fields present or absent; leaflets entire, chartaceous, glabrate to densely puberulent beneath, venation brochido- dromous, the midrib and secondary veins promi- nent, glandular fields in axils lacking, margins slightly undulate. Inflorescences elongate terminal or axillary panicles, many-flowered; rachis and pe- duncles minutely bracteate, the axes minutely scurfy to densely puberulent. Flowers ovoid in bud, the calyx expanding before corolla emergence; ca- lyx cupular-campanulate, minutely and densely lepidote to sparingly lepidote or moniliform-pubes- cent, the calyx apically truncate except for minute, mucronate teeth, costate or ecostate, the margin fre- quently split and/or reflexed, often ciliate; corolla zygomorphic, tubular-campanulate, lavender to ma- 466 Annals of the Missouri Botanical Garden genta, frequently with a white throat, the outer sur- face densely moniliform-pubescent and the inner surface glabrate with a ring of elongate, dense, mo- niliform pubescence immediately below insertion of stamens; corolla lobes 5 (2 upper and 3 lower), short-orbicular, rounded to acute, the inner and outer surfaces moniliform-pubescent; fertile sta- mens didynamous with a single staminode present, stamens and staminode adnate to the corolla; fertile anthers glabrous, with two spreading thecae, in- cluded; disk present; ovary cylindrical, usually densely lepidote; ovules 2-seriate in each locule; stigma bipartite, the divisions laterally flattened or partially fused and appearing hollow, included. Fruit a compressed, woody, linear-oblong septicidal capsule, dark brown to tan, the valves dehiscing parallel to the septum, the midline inconspicuous, and the surface conspicuously tuberculate to nearly smooth, many-seeded; seeds oblong, flattened, bia- late, the body ovoid and frequently bipartite. Paragonia contains two species and ranges from Mexico to Brazil and Uruguay. Collections are also reported from Guadeloupe (Gentry, 1973, 1977, 1978, 1982a, b). KEY TO SPECIES OF PARAGONIA la. Petioles < 10 mm long; petiolules < 6 mm long; petiolar glandular field absent or obscured by pco ‘ence axes glandular-puberulent; calyx ostate; fruit surface nearly smooth кет E brasiliensis . Petioles =10 mm long: petiolules -10 mm long; petiolar glandular fields present and evident; tendrils generally bifid, rarely trifid or simple; pseudostipules appressed or nearly appressed to the branchlet; inflorescence axes lepidote-puber- ulent to densely tomentose-puberulent; calyx smooth; fruit surface tuberculate — — - 2. P. pyramidata 1. pss brasiliensis (Baill. A. H. Gentry, n. Missouri Bot. Gard. 63: | 1976. Зап- hilaria brasiliensis Baill., Hist. Pl. 10: 27. 1888 (1891). Hilariophyton 422 (Baill.) Pichon, Bull. Soc. Bot. France 92: 228. 1945. Brazil. Minas Gerais: St. Hilaire 745 (holotype, P). Lianas; branchlets terete, drying brown, puber- ulent; pseudostipules angled away from branchlet and nearly appressed to the subtending petiole, pu- berulent. Leaves 6-10 cm long, 2-foliolate with а single, minutely trifid, terminal tendril (or tendril scar); petioles 6-8 mm long, lepidote-puberulent to densely puberulent, glandular fields lacking; peti- olules 3—6 mm long, sulcate, lepidote-puberulent to densely puberulent; leaflets 4-9 X 1.5-4.0 ст, elliptic, apices acute with minute mucronate tips lacking, bases acute to obtuse, with 5-8 principal secondary vein pairs, the lamina frequently punc- tate, glabrate above and glabrate to densely puber- ulent below. Inflorescences to 12 cm long, glandular- puberulent, several-flowered; rachis and peduncles minutely bracteate, the bracts linear-triangular, 2- X 1 mm, = persistent, eglandular, puberulent; pedicels 4—9 mm long, densely puberulent. Flowers ovoid in bud; calyx 5-6 X 8 mm, costate, densely lepidote to moniliform-pubescent, apically truncate except for 5 minute, mucronate teeth, dark glands present on distal half of calyx, the margin smooth to ciliate; corolla exserted ca. 45 mm beyond the calyx lip, 3—4 mm wide at the calyx lip, 15 mm wide at the mouth, the outer surface densely mo- niliform-pubescent and the inner surface glabrate with a ring of dense uniseriate pubescence at the level of the calyx lip; corolla lobes 15 Х 12 mm, the apices acute; fertile stamens 12 or 16 mm long, inserted into the inner ring of corolla pubescence, the single staminode 4 mm long, inserted beyond the ring of corolla pubescence; disk 1 X 2 mm; ovary З mm long; style ca. 21 mm long. Capsule 40 X 1 ст, drying dark, the outer surface nearly smooth or minutely lepidote; seeds X 3.5 cm. Paragonia brasiliensis is a poorly known species from the eastern Brazilian states of Bahia, Minas Gerais, and Ceará (Fig. 2). АП collections known are from 500 to 1000 m, typically in the caatinga. Patterns of flowering and fruiting phenology are not evident because only nine fertile collections were available (flowering collections: four from January, one from June, and two from November; fruiting collections: one each from January and February). Peak flowering probably occurs from November to January. However, because a single flowering col- lection is known from June, P. brasiliensis may not have a rigidly constrained flowering period. BRAZIL. Bahia: Mun. Caetite. 20 km E de Cae 4?08'S, 215% 500 m, Arbo et al. 56: E (MO); Rodovia, inbe ) km N da divisa com Minas Gerais, 14°50'S, 39°00’ W, Ps 1196 (CEPEC, H, MO); Rod. BR-116 Na Candido e Hatsc bach & Silva 50026 (MO); Jequie, 13°05'S 10277 (VAN, NY, UB); Serra da Agua de Rega - + М of Seabra, road to Agua de Rega, 12725'5, 4 m, Irwin et al. 31159 (MO, NY, UB); BR 4, km 966, ' Pabat & Pereira 8364 (MO); 6 km antes de Planalto Bahiana. Pereira & Pabst 9539 (MO); 9 km de 2. 72. aracas rumo Caatinga, 13?26'S, 4 ыы. Pereira & Pabst 9705 Lise Ceara: Serra da oca, Sitio Antonio, 03?28' Mer 40°30'W, Fernándes s.n. “(EAC. 1950). The stems and leaves of Paragonia brasiliensis Volume 85, Number 3 1998 Hauk Review of Рагадота 467 are often dark and densely puberulent, particularly on the short petioles and petiolules. The tendrils of P. brasiliensis are trifid rather than bifid as is usu- ally observed in P. pyramidata. Petiolar glandular fields were not observed in P. brasiliensis, and these are a nearly ubiquitous feature of P. pyramidata. The inflorescence axes of P. brasiliensis are glan- dular-puberulent, whereas those of P. pyramidata are lepidote-puberulent to densely tomentose-pu- berulent. The costate calyces of P. brasiliensis are distinct from the smooth calyces of P. pyramidata. Gentry (1976; Table 1) reported that the inflores- cences of P. brasiliensis are narrower than those of P. pyramidata. However, fertile collections of P. brasiliensis are few, and it is difficult to assess whether inflorescence width is a useful character to distinguish the two species. Gentry (1976) reported that the fruit of P. brasiliensis are "strongly com- pressed," whereas those of P. pyramidata are sub- terete. The few fruiting collections of P. brasiliensis that are available possess immature fruit, and any generalizations based on these collections would be somewhat speculative. Despite the immaturity of the P. brasiliensis fruiting collections, the nearly smooth fruit surface of P. brasiliensis appears dis- tinct from the tuberculate surface of P. pyramidata fruit. 2. Paragonia pyramidata (Rich.) Bureau, Vi- densk. Meddel. Dansk Naturhist. Foren. Kjøbenhavn 1893: 104. 1894. Bignonia pyr- amidata Rich., Actes Soc. Hist. Nat. Paris 1: 110. 1792. Tabebuia pyramidata (Rich.) DC., in A. DC., Prodr. 9: 214. 1845. TYPE: French Guiana, Leblond 292 (holotype, P-LA). Lianas; branchlets terete, drying gray, tan, or oc- casionally dark brown, the younger growth glabrate to densely tomentose and the older stems often rough-surfaced. Leaves 10—30 cm long, 2-foliolate with a single, minutely bifid or trifid (rarely simple) terminal tendril (or tendril scar); petioles 10—20 mm, glabrate to lepidote or densely tomentose-pu- berulent, the distal adaxial glandular fields usually present and either evident or obscured by pubes- cence; petiolules 1-2 cm, lepidote to densely to- mentose-puberulent; leaflets 7-26 X 3.5-13.0 cm, narrowly to broadly elliptic, elliptic-orbicular or ovate-elliptic, apices acute with minute mucronate tips present, bases broadly acute to obtuse or rounded, with 4—5(6) principal secondary vein pairs, the lamina punctate, nearly glabrate above and glabrate to sparsely puberulent or densely to- mentose-puberulent below. Inflorescences to 18 cm long, lepidote-puberulent to densely tomentose-pu- berulent, many-flowered; rachis and peduncles mi- nutely bracteate, the bracts linear-triangular, 2 X 1 mm, caducous, eglandular, puberulent to densely tomentose-puberulent; pedicels to 12 mm long, lep- idote or tomentose-puberulent. Flowers ovoid in bud; calyx 5-7 X lepidote, mealy, or densely tomentose-puberulent, occasionally sparsely and minutely puberulent, api- cally truncate except for 5 mucronate teeth, the margin ciliate; corolla tubular-campanulate, exsert- 6—7 mm, ecostate, glabrate to mm above calyx lip, 2-4 mm wide at calvs lip, 15-20 mm wide at mouth, the outer sur- face densely moniliform-pubescent and the inner surface glabrate with a ring of dense uniseriate pu- bescence at the level of the ovary apex; corolla lobes 12-15 X 16-20 mm, the apices rounded; fer- tile stamens 16 or 19 mm long, inserted at inner c ring of corolla pubescence, the single staminode 4 mm long, inserted beyond the ring of corolla pu- bescence; disk 1 X 3 mm; ovary 3 mm long; style 20-25 mm long. Capsule 32-52 X 1.0-1.5 cm, dark to light brown or uniformly tan to silvery-tan, the outer surface tuberculate to finely tuberculate and lepidote; seeds 1 X 4 cm. Figures: Gentry (1973, fig. 24), Gentry (1982a, fig. 19), Gentry ‚ fig. 31), Gentry (1997, fig. 339), Sprague (1903, he 2111, 2172). Paragonia pyramidata ranges from southern Mexico through Central America and South Amer- ica east of the Andes, to southern Brazil and Uru- guay (Fig. 3). Gentry (1973, 1977) included Argen- tina in the distribution of P. pyramidata, but no collections from Argentina were located during this investigation. It typically ranges from О to 1000 m, with collections reported to 2066 m. Paragonia pyramidata is common in tropical and premontane wet forests, and thrives in a diversity of ecological conditions from dry hillsides to swamps (Gentry, 1973). The subulate, appressed (or nearly appressed) pseudostipules, large “lauraceous” leaflets, and distinctive, sweet smell of the freshly crushed leaves are important field characters for P. pyram- idata (Gentry, 1973, 1978). The minutely bifid (ver- sus trifid) tendrils and absence of interpetiolar glandular fields distinguish P. pyramidata from the vegetatively similar Ceratophytum tetragonolobum Jacq.) Sprague & Sandw. (Gentry, 1973). Although Bureau described Paragonia pyrami- data var. elliptica in 1845, and Bureau and Schu- mann described P pyramidata var. tomentosa in 1896, Gentry (1973, 1977, 1982a, b) did not rec- ognize varieties of P. pyramidata, and regarded var- ~ iation in pubescence as “taxonomically unimpor- Annals of the Missouri Botanical Garden tant" (Gentry, 1976a). However, my inspection of specimens from South America revealed forms clearly identifiable as variety tomentosa, and these are restricted to a specific geographic area (Fig. 2). Variety tomentosa apparently grows intermixed with the glabrate variety pyramidata. However, no in- termediates were identified. The characters of the glabrate and pubescent va- rieties differ more in frequency of expression than in fundamental structure, e.g., all characters of va- riety tomentosa are present in variety pyramidata but at different frequencies. The principal differ- ence between the two varieties is in the overall pu- bescence; variety pyramidata is usually glabrate and variety tomentosa is typically densely tomen- tose-puberulent. The leaflets of variety tomentosa are generally wider and more nearly ovate than the elliptic leaflets typical of variety pyramidata. Typ- ically, variety pyramidata has minutely puberulent inflorescence axes, whereas those of variety tomen- tosa are densely tomentose-puberulent. The calyces of variety tomentosa are densely tomentose-puber- ulent, whereas those of variety pyramidata are gla- brate to lepidote (rarely mealy; see below). The fruit surface of variety tomentosa is uniformly tan, whereas that of variety pyramidata varies from dark brown to light tan and is generally less lustrous and more coarsely tuberculate. The fruit surface of va- riety tomentosa is often more finely textured and more lustrous than that of variety pyramidata. nomalous collections of Paragonia pyramidata that do not fit clearly into either variety tomentosa or variety pyramidata are known from the Brazilian states of Pará, Mato Grosso, and Mato Grosso do Sul. These anomalous collections are well removed from the main range of variety tomentosa (Fig. 2). The Pará collection (Prance et al. P25318) has gla- brate-mealy calyces and elliptic leaflets, and inflo- gth. The Mato Grosso do Sul collection (Hatschbach et al. 52475) has tomentose-puberu- lent leaflets (indistinguishable from those of variety tomentosa), short-tomentose inflorescence axes, and glabrate-mealy calyces. The Mato Grosso collection (Prance et al. 26131) has evenly but sparsely short- pubescent leaves (no young inflorescence axes or calyces are present because the specimen is fruit- ing). These anomalous collections were excluded from the variety descriptions and key. Additional collections are needed to assess the taxonomic sta- tus of the anomalous specimens. The correlation among character states (of leaflet shape, leaf pubescence, and fruit surface) for some collections warrants recognition of variety tomen- tosa as distinct from variety pyramidata. However, the absence of character state discontinuities (in individual characters) between the taxa argues against recognition of variety tomentosa as a spe- cies or subspecies. More detailed investigations may provide additional characters to support rec- ognition of this variety at a higher taxonomic level. KEY TO VARIETIES OF P. PYRAMIDATA la. Leaflets glabrate or nearly so, narrowly to broadly elliptic, only occasionally ovate-elliptic or ellip- calyx glabrate to Viet occasionally ак and minutely pubem ROMS a. P pyramidata vi var. . pyramidata lb. Leaflets ы ш densely tomentose-puber- ulent be tuse; calyx densely tomentose-puberulent ». P. pyramidata var. tomentosa 2a. Paragonia pyramidata var. pyramidata Bignonia d Vahl, Eclog. Amer. 2: 44. 1798. TYPE: Trinidad. von Rohr s.n. (holotype, C). Bignonia 22. ат., Linnaea 7: 704—705. 1833 Г = 21.” ЗЫ Brazil. Sellow s.n. (holotype?, В not en by Gentry). Жанда 5. "Cardner, London J. Bot. 1: 179. 1842. TYPE: Brazil. Rio de Janeiro: Gardner 78 (holo- үй type ш Ius Mart. ex DC., . DC., Prodr. 9: 159, 1845. TYPE: Brazil. Amazonas: Martius 2977 (ho- 5. M; isotype, 5. Bignonia martiusiana DC DC., Prodr. 9: 156-157. чек ТУРЕ: Brazil: Pará: 1817, Martius s.n. (holo- R). Mb. ae DC., in A. DC., Prodr. 9: 176. pen PE: Brazil. Rio Sào Francisco, Blanchet 1... G-DC; áo nb d buen DC., in A. DC., Pro dr. 9: 176. 1845. TYPE: French Guiana. Perrottet 2851 (holotype, G- DC). dii oc striata DC., in A. DC., Prodr. 9: 176. 1845. PYPE: Brazil. Sáo Paulo: Lund 783 (holotype?, G- Pachyptera Ж е DC., in A. DC., Prodr. 9: 175— 76. 1 NTYPES: Brazil. Sáo Paulo: Martius s.n. (M © seen by 2. Rio Paraibia, Neuwied n. (M not seen by Ger Pithecoctenium reticulare D 1845. TYPE: Brazil. m C. , Prodr. 9: 197 4. те ality or collec tr 2” surinamensis Miq., Linnaea 18: 250. Т "Ы TYPE: Suriname. Focke 230 2. VU. excluding Tm of Cydista aequinoc- tialis (L.) Miers; isotype, K). Bignonia го Cerón, Bot. Voy. осш 129. 1845. TYPE: Panama. Sinclair s.n. (holo Arrabidaea 2 Donn. Sm., Bot. Gaz. 20: 6. 1895. TY Honduras. San Pedro Sula: Thieme 5393 (iso- types?, NY, US). Paragonia чалда Loes., Bot. Jahrb. Syst. 23: 130. 1897. TYPE: Nicaragua. Matagalpa: Rothschuh 230 otype?, B) Mem. New York Bot. ololy Adenocalymna densiflora Rusby, Volume 85, Number 3 1998 Hauk Review of Paragonia 469 = 7: 355. 1920. TYPE: Bolivia. Cataracts of i River, Rusby 484 (isotypes?, NY, US). | Te c a ee Repert. Spec. Nov. Regni Ve 2, YPE: Brazil. Paraná: Dusén 8633 кима! К) Petastoma macrocalyx Kraenzl., Repert. Spec. Nov. Regni Veg. 17: 59. 1921. TYPE: Brazil. Зао Paulo: Heiner 569 (holotype, S; photo, K Young branchlets glabrate to lepidote; petioles and petiolules glabrate to lepidote, with distal ad- axial petiolar glandular fields usually present and conspicuous; leaflets narrowly to broadly elliptic, infrequently elliptic-orbicular or ovate-elliptic, the leaflet bases acute to obtuse or infrequently round- ed, the surface glabrate or nearly so above, glabrate to sparsely puberulent below; rachis and peduncles glabrate to lepidote or puberulent; pedicels and ca- lyces lepidote, occasionally sparsely and minutely puberulent or glabrate; outer surface of capsule dark to light brown or (less commonly) tan. Paragonia pyramidata var. pyramidata ranges from southern Mexico through Central America and South America east of the Andes, to southern Brazil and Uruguay (Fig. 3). Collections of Paragonia pyr- amidata var. pyramidata are known from 0 to 2066 m. It is common in tropical and premontane wet forests and thrives in a diversity of ecological con- ditions from dry hillsides to swamps (Gentry, 1973). Flowering occurs throughout the year, and collec- tions peak in February (Figs. 4, 5). Fruiting collec- tions increase from January to April and peak in ay. Representative. specimens. MEXICO. Campeche: 5 km 5 de Ulmal, Cabrera 2308 (MO). Chiapas: 6 km al sur de la desviaci ion a Chancala, Cabrera & Cabrera 6216 ma: W of 7 Bay, 5 mi. W of Santiago, ‚ 104?00' W, 90-150 m, McVaugh 15707 (MICH). Уа Mpio. Sta. Мапа Chimalapa, 16°55'00"N, 94?40'30"W, 300 m, Hernández 180 (MO). uintana Roo: 10 km al oeste de La Pantera, Cabrera & Cabrera 4252 (MO). Tabasco: Balancan, Finca la Es- peranza, 17?48'N, 91732", 50 m, Calzada et al. 2651 MO). Veracruz: 10 km N of Sontecomapan, vic. Playa Escondida, 18°35'N, 95°03'W, 100 m, Nee 24741 (MO). Yucatán: Tzucacab, 20%04/М, 89°03'W, Enríquez 645 (MEXU). BELIZE. Belize: N of Hwy. 5 of Altunha, 0 m, wd 8259 (MO). Cayo: Sibun River near Hummingbird 17°26’N, 88^16'W, 66-100 m, Gentry 8432 (MO). Co ады” ] mi. N of Buena ier 7. 88?32' W, Gen- try 8547 (MO). Orange Walk: 10 mi. S of Orange Walk, 17?15'N, 88°47'W, Whitefoord 2599 ) (MO). Stann Creek: Carib Reserve, 16°57'N, 88 bank of Río Gracias a Dios, 15%53'N 13'W, Contreras s.n. (F). Jutiapa: between San José 1. апа Кіо 4е Los Esclavos, 14?15'N, 90%08'W, 900-1200 m, Standley 60621 (F). Petén: Camino para El Remate, km 6 parque Tikal, 17%00'N, 89°42'W, Tun E d peo E talhuleu: between Nueva Linda erico, 14?25'N, 91?49' W, 120 m, Standley 87669 vm EL SAL- VADOR. La Libertad: El Amatalito, 13°29’N, 89?16'W, Villacorta et al. 844 (MO). HONDURAS. Atlántida: be- tween Tela & Pajuiles, 1544'N, 87°27'W, 200 m, Molina & Molina 25719 (F). Colón: Río Guaimoreto, 4.5 mi. NE of Trujillo, 15°57'N, 85^54'W, Saunders 299 (MO). mayagua m of Siguatepeque, 14^25'N 87°37'W, 566 m, Webster et al. 12748 (LL). Cortés: Cerca de Choloma, carretera San Pedro Sula—Cortés, 15° 8800", 100 m, Molina 6667 (F, LL). El Paraíso: sa of Río Dantas, barranco El Muro, 14?10'N, 86°30'W, 733 m, Webster et al. 12048 (MO). Gracias a Dios: Mosquitia, Río Plátano, 0—4 hrs. upriver from village of Ras, 15°30'N, 84°40’ W, 0 m, Gentry et al. 7521 (F. MO). Islas de la Bahía: Isla de Roatan, camino entre Roatan y 16°23'N, 86°30’ W, 0-50 m, Nelson «С js ai 4495 (MO). Olancho: Culmi, 14%45'N, 86%00'W, 5 Nelson € Romero 4634 (MO). Santa Bárbara: уа ид al mineral del Mochita, 15?10'N, 88?20' W, 900 m, Molina 5603 (F). NICARAGUA. Carazo: 1 km E of San Marcos, 11%55'N, 86712", Neill 260 (MO). Chontales: Cerro Ol- uma, Cordillera Amerisque, 750 m, Gentry et al. 43918 MO). Jinotega: below Peñas Blancas via El Тита, 13%15'N, 85%41'W, 1200 m, Neill 7139 (MO). El Zapotal Е of Managua, 12?09'N, 86%07'W, 15 nier 1049 (K). Matagalpa: 7 km al NO de Esquipulas, 12°40' М, 85?43' W, 800 m, Moreno 25421 (MO). Río San Juan: between Río Santa Cruz and Cafio Santa Crucita, 11?03'N, 84725", 50 m, Stevens тя (MO). Zelaya: 12 km SW of Bonanza near Lago Siempreviva, 14°02' №, 84234! ы 300 m, Меш 4037 (MO). "COSTA RICA. Ala- juela: Bord de la route à Carrillo, 09°54' №, 83°33’ W, 300 m, n 2497 (CR, G, US). Cartago: Las Vueltas, Tuc- urrique, 635 m, Tonduz 7481 (BM, CR, СН, К, US). baie: 17 km SW of Nicoya, 12 km SW of Curime, 10%03'N, 8532", 100—300 m, Liesner 5027 (MO). He- redia: Finca La Selva, the OTS Field Station, 100 m, Wilbur 34424 (MO). Limón: Río Colorado between Caño Bravo and Caño Pereira, 10°43'N, 83°42’W, 5 m, Stevens ~ Сепега! Viejo, El General Valley, 09°11'N, 83730", 750 m, Williams et al. 28484 (F, MO). PANAMA. Bocas del oro: Lower Río San Pedro Valley, 08°49'N, 81°33'W, Gordon 20D (MO). Canal Zone: Barro Colorado Island, Fuertes Cove, 09?11'N, 79*57'W, Croat 8136 (MO). Chi- гіди: W of Río Chorchita, 08?22'N, 82°15'W, Gentry 5849 (МО). Cocle: 1 ті. N of El Valle, 08%36'N, 80°33’ W, Gentry & Dwyer 3572 (MO). Darién: Río Balsas tween Manene and Río Coasi, 08?15' N, 77759", Hart- W, Gentry 3129 (MO). Los Santos: 10 mi. N of Tonosí, 07724'М, 80°27'W, Tyson et al. 2941 (MO, SCZ). anamá: Río ji orona, along Pan Am Hwy., 80°01'Ұ, Gentry 2903 шү San Blas: 09°14'N, 78°01'W. . Hammel & D'Arcy 4997 . Veraguas: 2 mi. S т Sakta Fe, 08°31'N, 81°05'W, Gentry 2942 (MO). TRINIDAD AND TOBAGO. Trinidad: Tamana, 10°20'N, 61°05'W, Broadway 5600 (MO). Tobago: Тһе Widow, 11?15'N, 60%44'W, Broadway 4576 (U) таи Amazonas: Puerto Nariño, 03°29'N, 70730", 100 m, Rudas et al. 2023 (MO). Atlántico: Bar- ranquilla, Juanmina, 10%58'N, 74%54'W, 10 m, Dugand о о > 23 470 Annals of the Missouri Botanical Garden 6926 (COL). Boyacá: El Humbo, 1333 m, Lawrance 800 MO). Caquetá: 21-22 km E of D 75^41'W, 260-280 m, Gentry et al. 9074 (MO). Choed: ЗІ km E of Quibdó, ca 14 km E of Tutunendo, 05°45'N, 76°32'W, Gentry & Brand 36887 (MO). Córdoba: Rio Sinu, 09°24'N, 75^49' W, 120-200 m. Cuodios 4175 (MO). undinamarca: Guaduas, 1040—1320 m, García- Barriga 12338 (COL). Guaviare: Río Ranchería. ?35'N, 72°38'W, 100 m, Pop. 5 (COL). Magdalena: Rin- cón Hondo, Allen 412 Meta: Sierra la Macarena, Río Guapaya, 02°45'N, 7: des 475 m, Philipson et al. 1689 (COL). Nariño: Mun. Tumaco, Llorente, 01%49'N, 78°46'W, de Benavides 627 (COL). Putumayo: Río Pu- tumayo opposite mouth of Río Gueppi. 00°30'N, 2. Е 200 m, Gentry et al. 22117 (MO). Santander: Barr, Bermeja (El Centro), 07°03'N, 73%52'W, 100 m, Haugh 2212 (MO). Valle: Río Naya, Puerto Merizalde, 03°16'N, 77°25'W, Cuatrecasas 14296 (COL). Vaupés: Mitu, e r Río Kubiyu, 01°08'N, 70903", Zarucchi 1261 (MO). EC- UADOR. El Oro: Hoad Zaracay-Las Piedras, 250 m, Harling et al. 15624 (MO). Esmeraldas: W of San Mateo. Reserva Forestal de Jardín Tropical, Universidad Técnica pus Vargas Torres, 00°54' №, 79*37' W, 100—130 m, Gen- y & Lajones 73057 (MO). Guayas: 2-4 km W of Bucay. 02°10' S, 7906'W, 170 m, Gentry 12287 (MO). Los Ríos: 12.5 km E of Patricia Pilar, Centinela, 02%45'S, 80%33'W, 466 m, Hansen et al. 7784 (MO). Manabí: Cuchilla Seca above Estero Perro Muerto, Machalilla National Park. 01%36'S, 80742", 480 m, Gentry & Josse 72645 (МО). Napo: Соса, Coca-Yuca road 15 km SE of Coca, 03%03'5, 79°40'W, 250 m, Harling et al. 19877 (MO). Pastaza: Río Capihuari, 02°30'S, 76°50'W, 285 m, Øllgaard et al. 35079 (AAU, MO). Pichincha: 35 km N of Santo Do- mingo de los Colorados, 00*15'S, 79*09' W, 250 m. Gentry 9593 (MO). PERU. Amazonas: 65 km N de Pinglo, Río Santiago, 04%26'S, 77°39'W, 200 m, Huashikat О). Cusco: Quispicanchis Province, 1321375, 70?45' W, 643 m, Núñez 13813 . Huánuco: San Martín-Río Sion, 07°40'S, 76746", Schunke 2359 (COL, MO). Ju- піп: Е де La Merced, 11%03/5, 75?19' W, 1000 m, Schunke 6213 (LA). Loreto: Alto Amazonas, Río uu. oi Ri- machi, 04?20'S, 76°35'W, 200 m, Díaz & кш 6 (MO). de Dios: Maná National Park, 1124575, 71900", Emmons 1025 (MO). Pasc pa, Palcazu valley, on Río Palcazu, 10^10'S, 75713 m, Smith 3929 (MO). Puno: ridge between Río С ЖМК and Río Guacamayo, 132305, 69*50' W, 400—600 m, Gen- try et а! 77002 (MO). San Martín: Puerto Pizana, Mar- iscal Cáceres, Tocache Nuevo, 0821175, oe 350 m, Schunke 6872 (MO). Ucayali: Yarinacocha (C a Pu- callpa), 250 m, Vásquez & Jaramillo 1542 (МО), '"BOLIV- IA. Beni: Cercado Province Trinidad, 14%49'S, 64?48' W, 150 m, Gentry & Perry 77504 (MO). Cochabamba: Todos Santos-Chapare, 17°30'S, 65740", 300 m, Steinbach 428 (F, MO, NY, U, WIS). La Paz: 4. 17 km NW of Apolo near Río Marchariapo, 1493475, 68°28'W, 1000 m, Gentry 71118 (MO). Pando: Nic KP 442 Río Та- huamanu, 112065, 67736' W, Fernández & Susanna 8498 Cruz: Parque о 17°42'5, 63*35' W, . Seidel 3045 (MO). NEZUELA. Amazonas: . Atabapo, Río EU gos 03°40'N, 65%45'W, 180-210 m, Steyermark et al. 126165 (MO). Anzoátegui: Río León by Quebrada Danta, UM 042137, 500 m, Steyermark 61076 (У . Apure: Distr. Muñoz, 5 km W nns Fernando Hwy., 07%45'N, 6921777, 70 m, 5 . Aragua: Chuao, 50 m, Pittier 12121 (M, VEN). Bolí- мї о: Охарат- ‚ 300 un Б 7s - S 10°13'N, 67°33'W, var: Mpio. Raul Leoni, 04°18'N, 62°05'W, 490 m, Del- gado 83 (MO). Delta Amacuro: Eo Rio Grande and El Palmar, 08°20'N, 61°40'W, Gentry & Berry 14975 (MO). between La S 10°37'N, 66°23'W, Berry 924 (MO). Falcón: Cerro So- copo, 10?30'N, Eli NA lid m, Liesner et al. 8295 (MO). Lara: Serranía de Terapaima, S de Barquisimeto, 10*10'N, 69%30' W. 800-1000 m, S 443 (VEN). Ma- racay: a 'N, 67°36'W, Vogl 217 (м). Miranda: 5 of Santa Cruz, 10 km W of Cupira, 10°09'N, 65 548! W, 18- 20 m, Sd & Davidse 116416 E Monagas: Ке- erva Forestal de к-нө аш ÉS 09°53'N, 62°53'W, 10 m, Castillo 719 (MO). tu ¿T Ne mazonas, Dpto. ies alto E dp ta 30 I" al SE de La Esmeralda, 03?05'N, 65*5 ‚ Аутаг . Sucre: os Benitez, ош. , 10°30'N, 63°07'W, 4 50 m, Steyermark et al. 121402 (MO). Yaracuy: entre San Felipe & Marín, 22. 68°44'W, Pittier 12093 VEN). Zulia : Dito. Mara, Río Coc ш 10%52'N, 72°29' W, Hayward 201 (MO). SURINA AME. Nickerie: area of Ka- balebo Dam project, 03°34'N, 55°59'W, 30-130 m, Lin- deman et al. 15 (MO). “какай иа Saramacca River, Toe- koemoetoe (me К, 5593 W, Maguire 24918 IAN, pn MO). FRENCH GUIANA. Cayenne: 2 ième saut de Marouini prés d'Antecume Pata, 03?18'N, 54^04' W, Cre emers 4999 (MO). Saül: 03%38'N, 53712", 220 m, Gentry et al. 63076 (MO). BRAZIL. Acre: Km 60 from Rio Branco on Rio Branco-Brasileia Rd., 68700", Lowrie et al. 425 (MO). Amapá: pue BR 156, 109 km SSE of Oiapoque O-Calcoene, 03°00'N, 51930", Mori et al. en (MO). ista о de Barcelos, 00°58'S, 62°57'W, Silva et al. s.r 38180) (MO). Bali is trada Сапа сто білгіш 142005, 42°00'W, 2 E (CEPEC). Ceará: Pico Alto, Pacoti, 04%13/S, 56' W, Alas s.n. (EAC- 712). Distrito Federal Сойсо ns 17 еі al. 11172 (MO). Espirito Santo: Reserva Florestal da CVRD, > а 41?07'W, Peixoto et "al 3354 (MO). Goiás: Estrada Alto Paraiso-Teresina, 17%52'S, 51°48’ W, Heringer et p 2400 (MO). Maranhão: Engenho, Mun. de Vitoria do Arari, 0422575, 44°45'W, Rosa с (MO). M ;rosso: MT. BR 1 158, 1. na Rod. voado de Ман Кіса, 10°S, , Cid et с 5. ES ~ ~ ^4 JEN ж- ~ inas | S x. de Minas, 19?15'S, 44°37'W, 760 m, Davidse & arena 10808 (MO); km downstream from Bela a, io Mocoes, 03%22'S, 51?50' W, Sobel et al. 4859 (MO). Paraiba: Ar- eias, 01%21'N, 53°15'W, Moraes 1539 (MO). Paraná: Parque Marumbi, 25°28'5, 48%52'W, Gentry & Zardini 49763 mbuco: Cabo, 08?17'S, 35?02'W, Lima 61—3725 (MO). Piauí: R. Napuera, abaixa do Ta- boleirinho, or S, c W, pes s.n. (MG-9134). Rio Grande do S z. do 2 p. Osorio, 2995475, 50*16' W, Rambo 15133 (B). Rio de Janeiro: Petropolis Mata do Judau, 22?31'S, 43?10' NW. 700 m, Suc raga 4255 (MO). Rondénia: Km 16 on road to Saldar na close to че а Mirin, aia is Lleras 2710 (MO). Santa Catarin sla Santi ina, Saco - 279365 48930” W, 200-400 m, [ein 2343 (MO). 8ао Pa шо: Cananeia, Parque Estadual da Ilha do Cardoso, ЕГЕТ 46°39'W, Kirizawa & Romaniuc 1259 (МО). URUG UAY. Playa S. Domingo, Río Uruguay, 34°12'S, 58°18'W, Twee- die 1347 (P). 2b. Paragonia pyramidata var. tomentosa Bu- reau & K. Schum., in Mart., Fl. Bras. 8, pt. 2, t 118: 182. 1896. TYPE: Brazil. Minas erais: Uberabá, Formigas, Regnell Ш-48 (ho- 2. UPS). Volume 85, Number 3 1998 Hauk Review of Paragonia Young stems moderately to densely tomentose- puberulent; petioles and petiolules densely tomen- tose-puberulent, the distal adaxial petiolar glan- dular fields absent or present but obscured by pubescence; leaflets elliptic-orbicular to ovate- elliptic, infrequently narrowly to broadly elliptic, the bases rounded to broadly obtuse, or infrequent- ly acute, the lamina nearly glabrate above and mod- erately to densely puberulent or tomentose-puber- ulent below (especially along veins); rachis and peduncles moderately to densely tomentose-puber- ulent; pedicels and calyces densely puberulent to tomentose-puberulent; outer surface of the capsule uniformly tan to silvery-tan or (less commonly) ark. Paragonia pyramidata var. tomentosa is known from the Brazilian states of Goiás and Minas Ger- ais, and the Distrito Federal, as well as Paraguay (Fig. 2). Collections are known from 410 to 950 m. The few reports available indicate that P. pyrami- data var. tomentosa grows on rocky forested slopes or in forested areas associated with streams or meadows. The few fertile specimens were collected at the beginning of the wet season, between August and September (Fig. 6). Fruiting collections were limited to the latter part of the wet season, from January to April. Additional specimens examined. COUNTRY UN- KNOWN. Without exact locality, Macedo 5450 (US). BRAZIL. Distrito Federal: Brasilia, Heringer et al. 1172 (MO); Brasilia, bacia do Rio бао Bartolomeu, Heringer et al. 5990 мо vicinity of Sobredinho, Prance & Silva 59085 2. 25 km N of Brasilia, /rwin et al. 13999 (MO, N ү, US): 3 na margem do Rio das Salinas, Kirkbride 3580 (NY), 3639 (NY). Goiás: ca. 20 km S of Caiaponia, Anderson et al. 9440 (MO, NY); between Jataí and Caia- ponia, 40 km from Саларопа, Hunt & Ramos 6272 (NY). Minas ла ais: са. 15 km W of Рага de Minas, Davidse & Ramamoorthy 10808 (MO, NY); km 618 Rodovia Uber- aba-B. Horizonte. Duarte 44873 (MO): entre Lagoa Santa e Murs do Стрб, Duarte 6389 (МО); Rio Doce, Mun. Ја- boticatubas, Hatschbach 35255 ia. Beira do corregodo Carmo, Ituiutaba, Macedo 586 (NY. US); Fundas, fad: taba, Macedo 2608 (US): Uberabá, Rognel s.n. (US). PAR- AY. In regione cursus superioris fluminis Apa., Hase ler 8418 (NY). NOMINA NUDA Bignonia striata DC., in A. DC., Prodr. 9: 174. | , nomen nudum Шаа elliptica Mart. ex DC., in A. DC., Prodr. 6. 1845, nomen nudum Temnocydia lenta Mart. ex DC., in A. DC., Prodr. 9: 159. 1845, nomen nudum. Literature Cited e АА 5 1888 [1891]. Bignoniacées. Histoire des : 1-112. Dessins de Faguet. Paris. B reau, gi t 1872. Valeur des caracteres tirés de la structure de la tige pour la classification des Bignoni- Bull. Soc. Bot. France 19: T P. de. 1845. Шекенішседі. In: A.L.P.P. de , Prodromus 9: 142-248. Treuttel et Würtz. g. London. FAO (Food and [үй cater Organization of the United Nations). 1985. Agroclimatological Data for Latin America and the Caribbean. Rome. ox A. H. 1973. Bignoniaceae. /n: R. E. Woodson, . W. Schery Agen il ne of Panamá. Ann. Missouri Bot. Gard. 60: 781- 19 Flowering oot; and diversity in ыла Т bares Biotropica 6: 1976a. Studies in Bignoniaceae 19: eneric —, and new species of South American Bun Ann. Missouri Bot. Gard. 63: 46—80. 1976 6b. Bignoniaceae of southern Central Amer- ica: Distribution and ecological specificity. Biotropica 17-131. 8: 1 aceae. Fam. 178. Bignoniaceae. /n: G. Har "a & B. Spare йш Flora of Ecuador. Opera Bot. 1-173 Я 1978. The botany of the Guyana e Tres” Big- noniaceae. Mem. New York Bot. Gard. 29: 1982a. Bignoniaceae. Pp. 1-222 in X pos Ponies & V. Bos ај Flora de Veracruz. Fasc. 24. Instituto Naciona s ia nvestigaciones sobre Recursos didi ee lapa, Mexico. P In: Z. Luces de Febres & A. uei i (editors), Flora de Venezuela. 8(4): 7 433. Instituto Nacional de Parques, Caracas 1992. A synopsis of Bignoniaceae ethnobotany "m economic botany. Ann. Missouri Bot. Gard. 7 54. $ A Pu әжесі Pp. 403-491 in J. А. Ste 2. Т E. ту & Holst (general MS Flora of the peii nd Сактана, Vol. 3. Missouri Bo- tanical Garden, St. Louis n press. Bignoniaceae. In: Flora de Colombia. . Tomb. 1979 [1980]. Taxonomic impli- cations of Bignoniac eae palynology. Ann. Missouri Bot. . 66: 756-777 Goldblatt P. & A. H. Gentry. E Cytology of Bignon- ae. Bot. Not. 132: 475—4 I 52% do Sacramento, P. in Nova е pro Brasilia. ee Kónigl.-Baier. Akad. T; [229]-244, pl. 12- Macbride, J. F. oa Bignoniaceae. In: Flora of Peru. Field Mus. Nat. Hist., Bot. Ser. 13: 3-103. Pichon, M. 1945. Notes sur 2 "Bignoniac ées. Bull. Soc. Bot. .. 222-229. 1954. Chromosome behavior in some 46. % Simmonds, N. №. нор "а! pianta. Heredity 8: 129— Sprague, T. A. 1903. Ракыйма pyramidata Bur. Big- noniaceae. Tribe Bignonieae. Hooker's Icon. Pl. 28: t. , 2412. INDEX TO NUMBERED EXSICCATAE = Paragonia brasiliensis (Baill.) A. Н. Gentry: 2a = Paragonia pe (Rich.) Bureau var. pyramidata: b = Paragonia pyramidata var. tomentosa Bureau & K. Schum. Collections are listed alphabetically by the prin- 472 Annals of the Missouri Botanical Garden cipal collector. All specimens entered into TROPICOS were assumed to have been examined by A. H. Gentry. Specimens өнд чи by the author were primarily those duplicates housed at MO, although loans of Brazilian specimens from NY and US were also examined. All col- lections examined by the author are indicated by a “!” in superscript. Agostini 1626' 2a; Alencar 47 2a; Allen 412! За, 898 2a, 1751' 2a, 4166' 2a, 5505 2a; Almeida 572 2a; Alston 8715 2a; Amaral et al. 439' 2a; 2” 11979 2a; Anderson et al. 9440! 2b; Antonio 4: Hahn 49224: За; Ara- quistain & Moreno 2384' За; Arbo et al. pim 1; Aronson 856 2a; Asplund 18904 2а, 20398 2а; Austin et al. 7162! 2a; Ayala 2461' 2a, 2496' 2a, 2511' 2a, 3715 2a; Ayala & Arevalo S. 4249' За; Ayala et al. 3380' За, 3382' За, 3925' 2a, 5672' 2a, 5773' 2a; = 6213 2а, 8017 2a; Aymard et al. 6911' 2a, 6942 Balee 1961' 2a, 1963! 2a; Balick e et tal 2218' 2a; Bal- slev «с Madsen 10597 2а; Bangham 339 За; Barbour 5402' 2a; Barclay 2777 2a; Barfod et al. 48413' 2a; Bar- reto 1500 2a; von Bayern 330 2a; Beaman & del Alvarez 6353 2a; Beck 3456' 2а, 15135! 2, 18802' За. 20046: 2a; Belem 1196' 1; Belshaw 3120 2a; Berg et al. BG756 2a; Berlin 378 2a, 673 2а; Bernardi 2914' 2a, 6653! 2a: Berry 681' 2a, 924: 2а; Billiet & Jadin 1610: 2а, 4602 2a; Black 47-1945' 2a, 49-8412 2a, 49-8486 2a: Black & Foster 48-3394 2a; Black & Ledoux 50-10726 2a: Blanchet 2903 2а; Bonifaz 275! 2а; Boyan 217 За; Brandbyge et al. 30552: За; Bravo Н. 1 За; Breteler 3531 2a; Bristan 1157p.p. 2a; Broadway 2273 2a, 4576 2a, 5600' 2a; Brown 86 2a; Bunting 11611' 2a; Bunting & Licht 835 2a; Bunting & Stoddart 8904 2a; Burger et al. 10602 2a. ле 2308: 2a, 6216' 2a; Calzada et al. 2651' 2a, 2692' 2а; Carauta 853' За; Cardona 6 За, 74 За. 504 2a, 1410 2a; Carpio 1675 2a; Castillo 719' 2a; Caval- cante 1636 2a, 1911 2a, 2399' 2a; Cavalcante & Silva 7 2а; Cazalet & Pennington 5182 2a; Cedillo 3638! 2а; Cedillo T. 2611! 2а, 3283' 2a, 3377 2a; Cerón 6052 2a; Cerón et al. 2009' 2a; Cerón & Cerón 4585 2a; Cid & Lima 3675 2а, 107690 2a; Cid et al. 189! За, 369 2а, 632' 2а, 679 2а, 6448' а, 6927 2a; Clewell & Cruz 4136' 2a; Colella et al. 1671 2a; Conrad & Conrad 3013' га; Contreras 2644' да, 2884' За, 9188 2a, 9750' 2a: Conzatti et al. 3070' 2а; Cordeiro 1018: За; Correa 1762' 2а; Cowan 38330: 2а, 38458: За, 38556! 2a; po 193 eel 2. 4999: 2а, 5400: 2а, 7266: 2a, 5' 2а, 5' 2a; Cremers & Feuillet 12430 2a; 2 p 2a, pen 2а, 4805' 2а, 4917 2а, 5591' 2а, 5615 За. 5641' га, 5645! 2a, 5705 2а, 6538' За, 6584: 2a, 7131! 2a, 7695 2а, 7744: 2a, 7895' 2a, 7896: За, 8136' 2a, 83094 2а, 8370: 2а, 9106: 2а, 9227 За, 9510 2a, 10161 За. 10212 p.p. 2а, 10358 2a, 11932 2a, 14626 2a, 14940 2a, 19671' 2a, 44243 2a, 49782' 2a; Cuadros 4175' 2a; Cuatrecasas 8986 2a, 10835 2a, 14296 2a, 16141 2a. 17691 а; Cuatrecasas & Willard 26027 2a; Cuatrecasas & Llano 24117 2a; Cumana 1357 2a, 1546 2a; Cuming 1179 2a. Daly et al. 1408' 2a; Davidse 32184' 2a; Davidse & González 14793' 2a, 1486 60' 2a; Davidse & Ramamoorthy 10808' 2b; Davis 818' 2a; de 2. учи 2a; de Ca- brera 4252' 2a; de Granville 5164 1 2a; de Gran- ville et al. 8161' 2, 9589' 2a, 9648 Pg 10034: 2а; de Mello 38 2a; de Moraes 1539 За; de Morsie 1539 2a: de Nevers 4749 2a; de Nevers et al. 7628' 2a; de Paula 1081 2а; Delascio С. & Guánchez 10988! 2a; Delgado 83' 2а; Deward 158' 2a; Díaz S. 793' За; Díaz S. & Ruiz 871 2a, 936' 2a; Díaz S. et al. 213 2а, 604' 2a, 617 2a; Monista et al. | son et hs 7105' 2b, 44873' 2b; qut & Falcáo 3310 2a; Ducke 2583 2a; Pus and 6926 2a; Dugand & Jaramillo 4078 2a; Dugue- Jaramillo 2017 2а; Duke 14251' 2a; Duke & Nick- erson 14910 ; Dusén 8633! 2a, 12034' 2a; Dwyer 2892' 2a, n 2a, 12924' 2a, 12996' 2a, 13053 2a, 14795' 2a, Dwyer & Dieckman 13018' 2a; Dwyer et al. 13037 2a. E 7 Е. 3340 E Egler 47638! га; Eiten & Eiten 3936' 2а; Elcoro 317 2а; Emmons 51' За, 99' За, 102! 2a; Encarnación 1269: m Enríquez 645 2a, 6762 2a; m 2344 2a; Fagerlind et al. 2487b 2a; тема 673 2а, 1720 2а, 2555! 2а, 8498: 2а; 5. 18508! За; Feuillet 123 2a; Flores 62' 2a; Florschutz & Maas 2465' 2a; Focke 930 2a; Fol- dats 365-App За; Folsom 9456' За; Forero et al. 1878' 2a, 2508' 2a, 3506' 2a, 4149' 2a, 4597 2a, 5015' 2a, 6349' 2a; Fosberg 29110' 2a; Foster 689 2a, 1090 2a; Foster & Terborgh 6112 2a; Foster et al. 3348! За; von Friedrichsthal 517 2a; Froes 11739! 2a, 12537-231 2a, 26674 2a, 34095a 2a, 34095 p.p. 2a, 34267 2a; Fromm et al. 1290 2a; Fuchs & Zanella 21812' 2a. García- Бе ык 10650 2a, 12338 2a, 13844 2a; Gard- ner 78 2a; Garnier 1049 2a; Garwood 1000' 2a; Gentle 1415' 2a, 1943' 2a, 2764 2a, 2831 2a, 3100 2a, 3837 2а, 6583' 2a, 8010' 2а, 8246: 2а, 8666: За, 8669 За, 8906' 2a, 9243' 2a; Gentry 430' 2a, 433 2a, 451' 2a, 719 2a, 721' 2а, 725' 2а, 1037 2а, 1158: За, 1179 2а, 1180 2a, 1210 га, 1237 2a, 1245 2a, 1282 2a, 1307 2a, 1316 2a, 1417' 2a, 4406: 2а, 4501' 2а, 4690' 2а, 5002 2a, 5020 2a. 5246 2а, 5588' 2а, 5849' 2а, 6384: За, 7795' 2a, 8070' За. 8080: 2а, 8223' 2а, 8247 За. 8259 2а, 8286: 2а, 84. 32 га, 8462' За, 8491' 2а, 8547. 2a, 8579 2a, 9348' 2a, 9560' 2а, 9593' 2а, 9666: 2а, 9732' За, 9990! 2a, 10065! 2а, 12287 2а, 12465: За, 12484: 2a, 12501 За, 28195! 2а, 41242! 2а, 55611D' 2a, 69436! За. 70799 9' 2a, 71118 а, 78597 2a, 79281 2a; Gentry & Berry 14975 2a; nii & Brand 36887 2a; Gentry & Dwyer 3572' 2а; Gentry & Estensoro 70516! За, 70591' За: Gentry & Josse 72. 382 2a, 72645! 2a; Gentry & Lajones 73057 За; Gentry & Lott 30799' 2a; Gentry & Mostacedo 75610' 2a; Gentry & Núñez 66001' 2a; Gentry & Perry 77504 н) Gentry & Revilla 16319 2a, 16364 За. 2 20794' 2a; Gentry & Smith 44942' 2a, 45097 2a, 45106 2а; Gentry & Zardini 49763! За, 49839' 2a; Gentry et al. 7521 За, 7527 2а, 9074: 2a, 10440' 2a, 10680' 2a, 15680' 2a, 18442' 2a. 18485! За, 19633: 2а, 21298' За, 21756: За, 22117 2а, 25850: 2а, 26853' 2a, 27160' 2a, 31296a' 2а, 32497 2а, 38143' 2a, 43918' 2a, 44316! 2а, 53813' 2а, 56186: За, 58807 За, 60149 За, 63076! 2а, 68772 2a, 73879 2a, 74306 2a, 77002' За. 77250A' 2a; Gillespie 1420: 2а, 2435' 2a, 2437 2a: Gil- martin 698 2a; Ginés 4255 2a; Glaziou 4702' 2a, 6720' 2a, 12986! 2а, 12971' За, 15159' 2a; Gómez et al. 20363 2a, 21095 2а; Gonggrijp 13042 За; González 4: Davidse 930' 2a; Gordon њи“ 2a, 80С-Ь За, 118C' 2a; Goulding 75A 2а, 1156 2а, 1295 2а, 1325 2a, 1393 2a: Grández 829 2а, 1014: ^ 1037 2a, 1685! За, 2055! 2a; 4. et al. 1088: 2a; Grant 11002 p.p. 2a; Guánchez 773 Volume 85, Number 3 1998 Hauk 473 Review of Paragonia Gudifio 1339' 2a; Gutiérrez 1095 2a; “ati & Schultes 568 2a, 829 2a; Gutte et al. 1627C 2 Hammel & D'Arcy 4997! 2а; Hou et al. 7784: 2а; Harling & Andersson 11948: 2а, 11965' За, 16525' 2а; Harling et al. 15624' 2a, 19877 2a; Hartman 12523' 2a; Harvey 5285 2a; Hassler 8418 2a; Hatschbach 1704: 2a, 7405 За, 8630 За, 25790: За, 33560 2а, 35255! 2b, 39324: 2a, 44498! За, 45753: За, 45978: 2a; Hatschbach & Guimaraes 19067: 2a; Hatschbach & Silva 50026: 1: Hatschbach et al. 52475' да; Haught 2212' За, 2725' 2а, 3599' 2a, 3988 2a, 4023 2a; Hayes 413 2a, 915 2a. 1043 га; Hayward 192' а, 201' 2a; Heringer 7214 2a. 8730. За, 8877 2а, 9483 2а, 10277 1, 10586' 2а, 10635' 2a: Heringer et F 252 2a, 1172! 2b, 2400: 2a, 5990! 2b: Hernández G. 180' 2a; Herrera 4591 2a, 5087 2а; Heyde 419: га, 73 " 2 a; Holm-Nielsen et al. 21080 2a, 21125' 2a, 21451' 2a, 21651 2a, 21925B 2a; Holst 2031 2а, 4371' 2a, 4406' 2а; Holt & Blake 686' 2а; Hoogte & Roersch 3430 2a; Hopkins et al. 650' 2a, 676 2a; Horner et al. 165' 2a; cn 1083' 2а, 1193' 2а, 1239' 2a, 1526' 2a, 1813 2a; Huber 581 2a; Huft 1925' 2a; Hunt & Ramos 6272 2a. IFAT 7783 Ха; Ibarra M. 733: 2а, 1107 2а, 3142' 2а; Irwin & Soderstrom 6851' 2a; Irwin et al. 8145! 2a. 9464! 2a, 13999: 2b, 31159' 1, 55547 2а, 55548' 2а, 57646 2a. Jacobs 2963 2a; Jansen-Jacobs 1661' 2a; Jaramillo & Coello 4149 p.p. За; Jaramillo-Mejía & Palacios 7914 га: Játiva & Epling 931' 2а; Jones 306 2а; Kayap 105' 2а, 116: 2a, 151' 2a; Kennedy & Steiner 2454: 2a; Kerber 178 За; Kernan 119 2а, 130' 2a, 1107 2a; Killeen 2853' 2a; Killip 35078' 2a, 37242 2a, 37531 2a; Killip & Smith 30593' 2a; Kirizawa € Romaniuc М. 1259' За; Kirkbride 3639' 2b; Kirkbride & Kirkbride 3580' 2a; Kirkbride & 3 2710' 2a; Klein 1142' 2a; Klug 1283' 2a, 1676' ‚ 1996: 2а, 2623' 2а, 3409' 2а; Knab 18' 2а; Knapp F3 Alcor 7426' 2a; Kohkemper 931 2a; Krukoff 6213' 2a, 6272! 2a, 8739' 2a; Kuhlmann 594 2a, 1102 2a, 2272' 2a, 6117 2a, 7160' 2a, 41414' 2a. Lanjouw 1169 2a; Lasser & Foldats 3010 2а; Lathrop 6766 2a, 6773 2a; Laughlin 269 2a; Lawesson et al. 43471 2a. 43549 2a; Lawrance 800' За; Lent 3304 га; León 459: За, 719 га; Lescure 241' 2а, 798: 2а, 2385 2a; Lewis 4048: За, 12490 2a, 12719 2a, 12933 2а, 37628' За. 37669: За; Lewis et al. 171' 2a; Liesner 1978 2a, 5020 2a, 5027 2a: Liesner & Carnevali 22766' 2a; T González 9180: 2a; Liesner & Morillo 13974: ' 2a; Liesner et al. 7673! За, 8295' 2a; Lima rx Pini ps Lima & Nelson 755' 2a; Lindeman et al. 15! 2a, 99 За; Lizot 73 За; Londoño et al. 1608' 2а; Long 118 2а; Lourteig 2342 2а, 2343' 2а; Lowrie et al. 425' 2a; Luetzelburg 114 2a, 327 2a; Lugo 2922' 2a, 3010: 2a; Lund 783 га, 2047 2a; Lundell 6463' га, 16060' 2a, 16351' 2a. Maas et al. 5477: 2a, 5546! 2a; Macedo 586 2b, 2608 2b; Maguire 24918' 2a; Maguire & Fanshawe 23366 2a; Maguire & Stahel 25001 2a; Maguire et al. 36759 2a, 53989' 2a; Marcano-Berti 281 2a; Maria 101 2a; Marín 3' 2a, 6 За, 731' За; Martinelli 7151' да; Martínez 13387 2a, 13399 За; Martínez 5. 15066: 2а, 15215' За, 15220! 2a, 15741' За, 15747. За; Martins & Nunes 7660' 2a; Martius 2976 2a, 2977 2а, 20464 2a: Matuda 1477 p.p.' 2a, 16610 2a, 17822 2a; Maxon 4795 2a; McDaniel & Rimachi 17622' За, 26563! За, 26603: 2а; McDowell 3282' 2а; McVaugh 15707 2a; Meave 1277 2a; Melo 597 2a; Mendonca 1014! За; Mexía 6077 га, 6181' 2а, 6369 2a, 6471 p.p.' 2a; Miranda 6762 2a; Mitchell 75 2a; Molina R. 1807 2a, 5603 2a, 6667 2a; Molina & Molina 25719 2a; А Morales 3542 2a, 3867 2a, 29 2a; Moreno 14654A 2a, ines 2a, 23969 2a, 25421' 2a; E reno & ae o E 2a; Mori & Gracie 21969: 2a; Mori & Souz 9' 2a; Mori et al. 17241 га, 2m 2a, 21 029 224 yea & орй — Nee 24741 2а, 34287 2а, 3 ; Nee n Mori 4049' 2a; Nee & Taylor 29338: Aia vols v^ Tyson 10898' 2a; Neill 260: 2а, 3695' За, 4037 2а, 4583: 2a, 7135 га, 7139' 2а, 8719' 2а, 9187 2а, 9661' 2а, 9902' 2а; Neill & Zaruma 7047: 2a; Neill et al. 8331 2a; Nelson 2098' 2a, 7141' 2a; Nelson & Romero 4495' 2a, 4634' 2а; Nelson & Vargas 2706: 2a; Neto et al. 125 2а; Nijen- huis 642! 2a; Núñez 6010' 2a, 6895' 2а, 12101 2а, 12732! 2а, 13813: 2а, 14011' Y 14025: За; Núñez et al. 8019' 2a, 10823: 2а, 11049' 2 Oldeman 1733 2a, 2304' 2a, B- 1236 2a, В-3607 2а, В-4195! За, T-223' 2a, Т-647 2a, Т-736: 2a; Oldeman «с Sastre 114: 2а, 126: 2a; Oliveira 3776! 2a, 4181! 2a; Allgaard et al. 35079: 2a; Opler 602 2а, 603 2a, 713 2a, 804 2a, 1613 2a, 1718 2a, 1877 2a, 1882 2a; Ortiz 1097: 2а, 2122 За. Pabst & Pereira 8364' 1; Pacheco 1496 2a; Palacios 2476: За; Peixoto 3515 За; Peixoto et al. 3354: 2а, 3515' 2a; Pefia 410 2a; Pereira 771' 2a; Pereira & Pabst 9539' 1. 9705' 1; Pereira et al. 4291 2a; Perrottet 2851 2a; Philipson et al. 1689 2a; Pickel 884! 2a; Pinesta 4 2a; Pipoly 4445' 2а; Pipoly et al. 14765: За; Pires 3891' 2a; Pires & Belem 12210 2a; Pires & Silva 10845 2a; Pires et al. 623: За, 16837 2a, 50877. 2а, 51544 2а; Pittier 2497 2a, sH с 6688 2a, 7568 2a, 12093 2a, 12121 2a, 12162 2178 2a; Pizziolo 162 За, 271' 2а; Plow- man et al. е 2a, 9037 2a; Pontual 46-64 2a; Poveda 1103 2a; Prance & Silva 59085 2a; Prance et al. 1139 2a, 2728 2a, 3981' 2a, 4171 За, 6857 2a, 8131' 2а, 8817: 2a, 9760! 2a, 9770' 2а, 10705' 2a, 10732' 2а, 11073! За, 11139' а. 13997 2а, 14022! 2а, 15090' 2a, 15231' За, 16318: За, 24610! 2a, 25601! 2а, 25781! 2a, 25887. 2а, 26131 2, 28749 2a, 28820: 2a, P25318' 2, P25601' За, P25710' За, P25781' га, P25887' га; Pre- vost 1 T € Prevost & Grenand 2010' 2a; Pulle 391 2a, 62 p.p. 2 Rabelo & Nonato 1389: га; Rabelo et al. 1837 За, 2018: За; Rambo 45133 2a, 45309 За; Ramirez 208' 2a; Regnell 111-48 2a; Reitz & Klein 8110! 2а, 8622' 2a, 9383 2a; Restrepo 485' 2a; Revilla 465' 2a, 500' 2a, 511' 2a, 604! За, 615: За, 720' 2a, 731' а, 774' За, 786' 2a, 799! 2a, 1335: 2a, 1797 2а, 1803' 2а, 1850: 2a, 1866: 2а, 2012 2a, 2100' 2a, 2284' 2a; Ribeiro 450 2a, 1573' га; Riedel & Langsdorff 1794 2a; Rimachi Y. 688' За, 4339' 2a, 5794' 2a, 8175' 2a; Riviere 296' 2a; Robleto 223 2a; Rohr 69 2a; R & Jaramillo 1129' 2a; Rusby 485 2a; Rusb 12' 2a; Rutkis & Foldats 87 ы Rutkis & Udris К. 986 2a, 1033 2a. Saddi 7083 2a, 7187 2a; Saddi & de Lamonica-Freire 2859 2a; Saer 443 2a; Sandino 4801' 2a; wu et al. 202! За; dos Santos 352 2a; dos Santos et al. 40' 2а, 78' га, 101' 2а, 139 2а, 168! а, 201' а, 228' 2а, Es 2a, 273' 2a, 436' 2a; Sastre 1792' 2a, 5965' 2a; Sastre & Echeverry 659' 2a; Saunders 188' 2a, 299' 2a, 380' 2a. 449' 2а; Schipp 347 2а, 5-71 2а; Schmalzel & Schupp 959' 2a; Schultes 3990 2a, 5421 2a, 5423 2a, 5498 2a; 474 Annals of the Missouri Botanical Garden Schultes & Cabrera 12825 За, 13242 2a, 14942A За. 16229 2а; Schunke V. ЗГ 2а, 2359 2a, 4885' За. 6213 2а, 6872 2a, 8500' 2a, 12337 2a, 12371' 2a; pig 400 2а; Sehnem 7995 За; Seibert 1513 2a, 1890: За. 2010' 2а, 3045' За; Sendulsky 504: 2a; Serv. s Cay- enne 7783 2a; Sessé & Мосто 2393 За, 2399 За. 2405 2a; Seymour 21404 2a; Silva 445 2a; Silva & Hatschbach 789 За; da Silva 213 2а, 7106' За: s 3884 а. 4133 2a, 4697 2a, 4885 2a; Smith 756' За, 1645' 2a. 3929' 2a, 6656 2a: Sobel et al. 4859' 2a: су нн 45] 2a; Solomon et al. 8159 2a; Sousa & Magallanes 7263 2а; Sparre 18888 2a, 19984 2a; Spruce 2408' 2a, 3192! 2a; St. Hilaire B745 Y; Standley 19915 2a, 40911 2a, 55566 а, 56622 2a, 60621 За, 87669 За. 89136 За. 89232 2a; Stannard de Arrais 683: 2a; Starry 150: За; 2a; 2 et al. 3973' 2a; 2. ' 2a; Stergios 10319 10755 2a, 11001A 2a, T 2a; Stergios & aa) 7633" 2а, 7655! За, 9051 За, 9189' За; Stergios & Del- gado 12927 2a; Stergios & Опера 2453' За; Stergios & rai 4845" За; Stergios et al. 5091' За, 6065' За, у ‚ 9892' 2a: Stern et al. 1875 2a, 33686: 2a: Stevens 7482 2а, 8007 2а, 8278' За, 12031' 2a, 19954: 2а, 23408: 2a, 23840 2a, 24058 За. 24394 2a, 24600 2a; Stevens et al. 18741' 2a, 19504' За, 19707 За. 24860 2a; Stevenson 3 За; Steyermark 38674 За, 44932 За 61076 2a, 61137 2a, 61498 2a, 87784 2a, 89115 2a. 90761 2a; Steyermark & Davidse 116252' За. 116416: 2a; Steyermark & Gibson 95642 2a: Steyermark & Liesner 123272' 2a, 126165' 2a, 131912' За; Strudwick et al. 3365' ма Sucre & Braga 4255' 2a; а 1948' 2a, 3502: 2 Tahe 230 2a; ms Pig 20 6 2а: e 6112 2a; ima 903' 2a, 3 2a; Timaná 745' 2a, 754! 2a. 1182' 2a, 1847 За; Esdr & Rubio G. 2262' 2a; Tonduz 6857 2a, 7481 2a, 8249 2a, 13864 2a; Torres et al. 3066 2a; Triana J. 4124-7 2a; Trujillo 17326 2a: von Tuerck- heim 7648' 2a; Tulleken 323 2a, 402 2a, 415 zs Tun О. 1116 2a, 1214' 2a, 1274 2a, 1885' 2a, 1912 a; Tunqui ! Za, 500 2a, 619' 2a; hors "n 2a; Stau 1347 2a; Tyson et al. 2941' 2a, riae 202' 2a; тай; P. 2a: Ule 5216 2a, 5276 2a. Td 939 га; Vásquez 745' За, 3848' За, 10509 За. 10599 2a, 11262 2a, 11742 2a; 2 | Јататшо 1252' За, 1542 За, 3645' 2a, 7228 2a, 8 а, 8 2a, 9140 iie 9354' 2a; Vazquez et al. asd. 2a. V1560 2a; Versteeg 141 2a, 261 "i Villacorta et al. 844 2 34 2а; Wawra 997 2a; Webster et al. 12048 2a, 12748 2a, 16422' 2a; von 2 1345 Pp. "2а, 2377 2a; Wendt et al. 2518' 2a, 2812' 2a, 4016' 2a: Went 557 2a; Wetmore & Abbe 210 2а; Wetmore б Woodworth 1 2a, 647 2a; Whitefoord 2599' 2a; 0900 2a; ped 34424' 2a; Williams 263 2a, 747 y р.р. 2a, 1144 11509 2a. 11562 2a, 11604 ai 11788 2a. pus a 13656 2a, 15185 2a, 15212 2a, 15673 2a, 15743 2a. 28484 2a; Williams & Molina 14624 2a; Williams et al. 8484 2a; Woodworth & Vestal 454 2а, 576 За; Woyt- kowski 6215' 2a, ! ‚ 35010! 2а, 35164: 2a; Wurdack 225 2а: Wurdack & PERO 43007 За. Yuncker et al. 8498 2a. Zarucchi 1261' 2a INDEX TO SCIENTIFIC NAMES к MAINT Bi oe Sg A A 468 э жн Rusby PA РРА 468 Ne daed аа за ILC EXT Pc EAE 468 Делано SN eos wate адан Ғы 468 Bignonia Luo acr за is 460, 465, 467, 468 ehretioides Cham. ..................... 168 laurifolia Vall ....................... 468 ta Mat. ex рС.................. 465, 468 martiusiana DC. ....................... 108 rupestris Gardner ...................... 108 sinclairii Benth, ...................... 468 striata DC. .......................... 171 Ceratophytum Pitt ................... 460, 467 boi pipe ақы (Jacq.) Sprague & Sandw. ..... 467 Cydista .............................. 408 aequinoctialis (L.) Miers ................. 408 Hilariophyton Pichon ................. 465, 466 brasiliensis (Ва .) Pichon ................ 466 eucocalantha Rodr. ..................... 460 Мапаозе а E C. Сотез................... 400 Mansoa DU. ........................../ 163 Pachyptera ....................... 463, 468 57 НИЕТТЕРІ 468 perrottetii DC. ........................ 468 str a da 468 umbe lliformis Ll —— OE ER 468 Paragonia brasiliensis (Baill. ) А.Н. Gentry . .460, 463, 465-467 LE 460, 463, 465—468 pyramidata var. elliptica Bureau ............ 467 pyramidata (Rich.) Bureau var. py oe . 460 5, 468, 469 pyramidata var. tomentosa Bureau & K. Schum. T PCT И 460, 463, 465, 467, 468, 470, 471 schumanniana Loes. ................... 468 Periarrabidaea А. Samp. .................. 163 Petastoma ............................ 469 leiophyllum Kraenzl. A Nd ee d агала 469 macrocal yx A eS VERIS 469 Pithecoctenium ......................... 408 reticulare DC. ........................ 4 08 460, 465 brasiliensis Baill. ............... 460, 465, 466 Spathicalyx J. C. Gomes .................. 400 Тетпосуйа ........................... 471 elliptica Mart. ех DC. ................... 471 Te 6X ОС сина. 47 22 Мїегз........................ 463 Zeyheria ............................. 468 surinamensis Miq. ..................... 468 SYNOPSIS OF CRATAEGUS SERIES APIIFOLIAE, CORDATAE, MICROCARPAE, AND BREVISPINAE (ROSACEAE SUBFAM. MALOIDEAE)! J. B. Phipps? ABSTRACT This paper revises four monotypic North American series of Crataegus (Rosaceae subfam. Maloideae). Of these, series , Apiifoliae, Cordatae, and Microcarpae all possess short- eries Brerispinar only exhibits this attribute on extension Аз leaves. The both the lobes and the sinuses, while s brilliantly red-fruited C. marshallii of series Api ifolia and is most closely related to European : species, partic salary in foliage characters. The (ser. Cordatae) and C. spathulata (ser. Mic rocarpae) have glossy foliage a shoot and extension shoot leaves with secondary veins to n United States species-pair °C. phaenopyrum nd small, orange-red, orbicular fruits, and are widespread and common in the southea a little less closely related to European species. Crataegus phaenopyrum is mid-Atlantic in range, westward to the Ozarkian area, while C. spathulata is a common southeastern species. Crataegus brachyacantha (ser. Brevispinae) is the most gerer of the species treated, being black-fruited with different short-shoot foliage and restricted to Louisiana an species eters де and re 1a States (vol. 1, Cronquist, 1980). the bordering parts of adjacent states. Line drawings and а maps are presented for each presentative specimens are cited. The selection of taxa the fact r this paper also reflects een екен to treat Crataegus for the now defunct ocular paa of the Southeastern United This paper is a further one of mine revising Cra- taegus (Rosaceae subfam. Maloideae) of North America. The first (Phipps, 1988) was devoted to series Aestivales (Sarg. ex C. K. Schneid.) Rehder and introduced the genus. This was followed by my monograph of northern Mexican Crataegus (Phipps, 1997) and an introduction to the red-fruited Cra- taegus of western North America (Phipps, 1998). The current paper assembles a group of monotypic series all with greater or lesser affinities to Euro- pean subgenus Crataegus. This subgenus was es- tablished by El-Gazzar (1980) on the basis of its deeply lobed short-shoot leaves with veins to the sinuses. The taxa treated here comprise the native North American species with deeply lobed short- shoot leaves, as well as one in which the long-shoot leaves alone are deeply lobed. El-Gazzar (1980) mistakenly held that all Cra- taegus taxa with deeply lobed short-shoot leaves possessing veins to the sinuses were Eurasian. However, there are American series with this attri- bute, including series Apiifoliae (Loudon) Rehder, which fits very comfortably into subgenus Cratae- gus as perceived by El-Gazzar. Its foliage is of the typical monogynoid shape, by which I imply a shape like that of C. monogyna Jacq., characteristic of many European species of section Crataegus. Further American series possessing lobed short- shoot leaves with veins to the sinuses are series сіна (Beadle) Rehder and series Microcarpae (Loudon) Rehder, each of which has small flowers, small fruit (often orange-red in color), and 3—5 nut- lets. These last two series have no very close rel- atives outside North America. In addition to North American Crataegus species with short-shoot leaves lobed to their sinuses, there are also native American taxa of Crataegus that include elements intermediate between subgenus Crataegus and sub- genus Americanae El Gazzar—a reason that I am not using El-Gazzar’s subgeneric taxa in this revi- sion. The intermediate kinds are principally a group of series normally lacking veins to the si- ! The support of the National Sciences and Engineering Research Council of Canada under whose operating M A-1726 this work was conducted, is gratefully acknowledged. Thanks are due x Susan Laurie-Bourque of Hu who drew the plates, and to the curators of numerous herbaria (А, АПА ‚ JSU, KY, LSU, LYN, MARY, T MO, MSC, NCSC, NCU, NLU, NO, TEX, UARK, UNA, US, USCH, в USLH, VBD, VCU, VDB, МИЛА, WVA) DUKE, FLAS, FSCL, NSU, SMU, SRH, TAEM, TAES, TENN, . GAM, GH, IBE ‚ CI, CLEMS, CM, COV, DHL, DOV, whose cooperation enabled the author to study such a wide range of material. ? Department of Plant Sciences, The University of Western Ontario, Па Ontario, Canada, МбА 587. ANN. MISSOURI Bor. GARD. 85: 475-491. 1998. 476 Annals of the Missouri Botanical Garden nuses on the short-shoots, but usually possessing these veins on the extension shoots. It should be noted that nearly all hawthorns can produce at least a few extension-shoot leaves that are both deeply lobed and with veins to the sinuses. Clearly the distinction involved is not nearly as sharp as was once believed. Possessing such intermediate char- acteristics but not closely related to the above three series are series Lacrimatae J. B. Phipps, Virides (Beadle ex Sarg.) Rehder, and Brevispinae (Beadle) Rehder. Series Brevispinae is treated in this paper. Due to taxonomic complexity, however, treatment of the series Virides is reserved for a separate article. The same is true for series Lacrimatae, which is also particularly complex, as evidenced by Beadle's (1903) having included 77 species in this group. The taxonomy of series Lacrimatae is under active revision by the author. Of the species considered here, Crataegus mar- shallii Eggl. (ser. Apüfoliae) is very close to mem- ers of series Crataegus; C. phaenopyrum (L. f.) Medik. (ser. Cordatae) has large, distinctive leaves with rather triangular lobes and veins to the sinus- es, but otherwise very unlike those of other haw- thorns; C. spathulata Michx. (ser. Microcarpae) has curiously lobed, very small leaves, often somewhat blue to dull green, of a shape unique in Crataegus. Finally, C. brachyacantha Sarg. & Engelm. (ser. Brevispinae) is treated here somewhat for conve- nience. With black fruits, short recurved thorns, and entire short-shoot leaves, it is a very distinct American hawthorn. However, even in series Brev- ispinae, the presumably plesiomorphic (Phipps, 1983; Phipps et al., 1991) veins to the sinuses show up in the deeply lobed leaves of the long shoots. Taxonomic difficulties in Crataegus are frequent- ly held to be due to hybridization. This has been extensively documented in Europe, but while it oc- curs in North America, for this continent there is little documentation. Indeed, the four species treat- ed here yield very few examples of putative hy- brids, these being restricted to a handful of appar- ently non-persistent specimens of probable C. marshallii X C. mollis (в.1.) parentage, as discussed after C. marshallii. MATERIALS AND METHODS This study has been made possible by the loan of over 2000 herbarium specimens from 43 differ- ent herbaria. Thereby the great majority of variation within the species studied should have become ev- ident. Typification of all species and their synonyms was attempted, although in a few cases it was too difficult to complete. These exceptions are clearly indicated in the text. e numbering of the species in this paper is continuous with that of Phipps (1988). The depth of lobing of leaves is quantified as the “leaf incision index” (LII), widely referred to in the text. LII is a percentage value lying between unlobed (096) and lobed to the mid vein (100%). The flowering season is given relative to other sympatric species of Cra- taegus and ranges from “early” to “very late.” Dis- tribution maps for each species have been created by computer on the basis of recording latitude/lon- gitude coordinates for the very large number of exsiccatae studied. Due to the few taxonomic prob- lems encountered with the Crataegus species here only a reduced list of exemplary exsiccatae (one per county) is appended, while the maps show all the locations that can be separated at the scale used. TAXONOMIC TREATMENT The taxonomic treatment commences with a key to series and is followed by taxon descriptions with lists of representative exsiccatae. [Note: “leaves” throughout key and taxonomic descriptions refer to short-shoot leaves, unless otherwise specified. Square brackets indicate those series to be treated in other papers by the author.] la. Short-shoot ipn ан lobed (LIT > 35%); veins to sinuses always present. : Leaf blades 2-4 ст sn PM and sinuses ngular taegus ser. Ген . Leaf blades > than 2 cm i de: lobes and sinuses vario 3a. Leaf blades « 1.3 X longer than wide, the sinuses narrow; bark rough, not ex- foliatin 4a. Inflorescences subglabrous; thorns indeterminate, often branchlets; petals orbici I. Crataegus ser. la ыша 4b. Inflorescences tomentose; thorns of determinate origin; petals elliptic RP II. Cra ación ser. Apiifolia 3b. Leaf blades > 1.5 X longer e sinuses relatively “hallow and broad; bark smooth, exfoliating № = E N = са oming X = . Crataegus ser. Microcarpae lb. Short-shoot leaves shallowly lobed (LII = 35%) or unlobed; veins to sinuses absent. за. Fruit black or deep purple when fully rip 6a. bed oot leaves unlobed; оо oot leaves usually deeply lobed, with veins to sinuses; petals turning orange with age; pii = 1.5 ст i spicuously recurved; fruit w bluish bloom unless книг unpitte taegus ser. Brevispinae 6b. Short-shoot Байа lobed or not; exten- sion-shoot leaves not usually deeply Volume 85, Number 3 1998 Phipps 477 Crataegus de usually lacking veins to sinuses; nutlets usually laterally pitted ____ [Cr taegus ser. Douglasii] 5b. Fruit ies red when fully ripe, occasion- ally greenish, yellow, orange, pink or purple (if pink, e de eventually ii ivan red underne oom al mt mt . Crataegus series и (Loudon) Rehder, Man. cult. trees Ed. 2: 1940. Crataegus sect. Apiifoliae jur 4 frutic. brit. 824. 1835-1838. TYPE: Crataegus marshallii Eggl. Bushes or small trees usually 2-8 m tall; usually with crown of somewhat tabulate branches; trunk to 20 cm diam., usually much less, lacking branched thorns; bark grayish, flaking; thorny, with simple thorns. Leaves long-petiolate, small, broad-ovate to deltate in outline, deeply lobed with 3 or 4 main lobes each side; veins to major sinuses and lobes at wide (457-70?) angle with midrib. Inflorescence many-flowered; pedicels pubescent; anthesis sea- son mid—early. Flowers medium-sized; calyx lobes narrowly triangular, margins slightly lobed; petals elliptic; stamens 20, anthers red; styles 1—3. Fruit commonly 6 X 3 mm, ellipsoid to occasionally spherical, glossy, bright red, flesh mealy when ripe; pyrenes 1-3, convex dorsally. One species, southeastern United States; wide- spread and common. Crataegus ser. Apiifoliae has obvious similarities to the European series Crataegus in its distinctive leaf shape, relatively small flowers, and small and few-pyrened fruit. However, it differs from the latter series in only possessing fully developed thorns of definite growth. Differences from series Crataegus are sufficiently large to postulate a moderately long period of separation, a point of view also supported by fossil Crataegus materials from the Pacific Northwest Tertiary with somewhat similar foliage (e.g., Chaney, 1927). However, one cannot discount the alternative possibility of origin by mid-Tertiary long-distance dispersal of ancestral Crataegus with these leaf types across the Atlantic from Europe. 4. Crataegus marshallii Eggl., Rhodora 10: 79. 1908. Mespilus apiifolia Marshall, Arbust. amer. 89. 1785, non Medik., 1793. Crataegus m (Marshall) Michx., Fl. bor.-amer. Ed. . 1803. TYPE: Not located. Parsley Haw. Figure 1 Bushes or small trees usually 2—8 m tall, usually single trunked with larger specimens having a crown of somewhat tabulate branches; trunks to 20 cm diam., usually much less, lacking compound thorns; bark grayish, flaking; branchlets unarmed or sparsely to moderately thorny with usually simple thorns?; extending shoots densely appressed pubes- cent; l-year-old shoots pubescent or glabrescent, gray-brown; older gray; thorns straight or slightly recurved, slender, (1-)2-3(-5) cm long. Leaves de- ciduous; petioles slender, 1-2 cm long, pubescent; blades small (1.5—3 cm long), broad-ovate to deltate in outline, densely pubescent on both sides when young, = glabrescent with age except on the main veins below; deeply lobed with (2-)3 or 4(-5) main lobes on each side, sinuses often closed distally by overlapping ~ Ж the margins toothed; veins extendin major sinuses and lobes at a de (45°-70°) p with midrib. Inflorescence 3— ia pedicels very pubescent; anthesis mid— early. Flowers 12-17 mm diam.; hypanthium gla- brous to slightly pubescent externally; calyx lobes 3—4 mm long, narrowly triangular, pubescent adax- ially, glabrous abaxially, margins glandular-serrate; petals elliptic, commonly 6-8 mm long, white or very rarely pink; stamens 20, filaments ca. 7 mm anthers red, 0.5 mm long; styles 1-2(-3). Риш commonly 4—6 mm long, ellipsoid to occasionally spherical, glossy bright red (rarely dull orange) at mealy when ripe; pyrenes 1-2(-3), convex dorsally. maturity, fles One of the commonest of southern United States hawthorns, C. marshallii has a wide distribution from Arkansas to Virginia (Fig. 2) southward to eastern Texas and central Florida, with a few out- liers in Oklahoma, southern Missouri, and southern Illinois. It is absent from the southern and eastern parts of Florida. It is found in a variety of open wooded habitats, in woodland openings and edges, successional habitats, fence lines, etc., although not normally in dense shade. It occurs both in quite wet and obviously well-drained soils. Crataegus marshallii is known to be hardy to USDA Zone 5, but this may not be true for all provenances. Flow- ering mostly late March to early April, but as early as mid-February in some seasons in central Florida. Crataegus marshallii is the only representative of series Apiifoliae and as such is not very closely related to other native North American hawthorns 3 The young shoots аге sometimes thom like but iu small distal herbaceous appen either elongate or become true в 478 Annals of the Missouri Botanical Garden She W Қ ЖО? Volume 85, Number 3 Phipps 479 1998 Crataegus 90% 80° a 36° + Ран + + 36 + K Ses ta + ++ |+ + pee + mb tH Ы + + +| + + + + 4e H + 4 +++ GMT Y + ЫН t ++ E + tt fet + E: + xt t + + 7 Ts * +. ++ t F + О ғы +. + +t + HE 9 A # + + NE... ++ ++ TM T +++ + + ++ + + i + HH it + 4 4t яс 44 fa i ++ ele qat + + + 4.45% H UE + едт thes T Lar + tha + + * + 30%] + t ++ ++ CRATAEGUS MARSHALLII EGGLEST. + +. + 90% М Figure 2. Distribution of Crataegus marshallii. Based on collated d records for The Vascular Flora of the i boum United States (Cronquist, 1980) area; incomplete northward. + = exsiccatae seen by me, Д = literature (Phipps, 1983). I have been unable to access Hum- апу combination of such characters. Much of the phrey Marshall's herbarium at West Chester Uni- versity, West Chester, Pennsylvania, where there may be type material. However, there is no doubt that Marshall's protologue exactly fits the accepted concept of C. marshallii Eggl. Variation in Crataegus marshallii is not sharply segmented into often readily identifiable local pop- ulations, as can occur in apomictic forms, and the species therefore appears to be sexual, although critical evidence is lacking. As a 20-stamen spe- cies, it is probably also diploid. Variation in length and density of foliar pubescence occurs, and there is substantial variation in plant size and habitat, which could well have a genetic base. There are also slight variations in leaf shape and size, espe- cially in the sharpness of lobes and teeth and the relative width and depth of the sinuses, but this study did not detect infraspecific variants based on variation in leaf shape can be found on one spec- ime n, i.e., Thieret & Williams Tangihapoa Parish, Louisiana. shape of C. marshallii is quite 17369 (LAF) from The elliptic petal unusual in Cratae- gus, as are the small anthers. A form with flowers turning pink at late anthesis has been discovered in northeastern Texas by the late Houston nursery- man Lynn Lowrey, and was being assessed for the retail trade. Representative specimens examined. U.S.A. Alabama: 2. Mobile, Mohr s.n. (М5С-43853, US). Calhoun: . V mi. NE of Nesbit Lake, Morelle . s.n. (JSU-10086). Cherokee: county road 19, ca. 6.5 mi. S Centre, Kral rein (TENN). € 21 near Cadena, RR 38920a (A). Jallas: 1 mi. WNW Cahaba ravine, Kral 45352 (VDB). Etowah: Attalla, lanes H532 (A, US). 11. ғғ { ‚ Mohr s.n. S pi "у Geneva: swa Cho . NW o Highbluff jet. Clari 7589 (NCU). Greene: 5 mi. SE from Eutaw by picnic area Figure 1. Crataegus marshallii. —a—c. pe 5180 (UWO). —g-m. Leaf vari iation from several sou Inflorescence and flower parts from Phipps 5318 Aly is —4. Fruiting ћ Кот Phipps 5180 (UWO). —е. Fruit and parts pis Moreland 991 (NLU). —f. Lea x» surface from ence from left to right "à g, ret & Williams ces (sequ 7286 (FSU); h, Phipps 5180 (UWO); i, perg 99] (NLU) } j, Phipps 5113 (UWO); k, Phipps Ey. NO. 1, Phipps | (UWO); m, Thieret & Williams 17969 (US 480 Annals of the Missouri Botanical Garden of Lock 7, J. L. Thomas et “ T1058 (NCU, UNA). Hale: by Co. road 10, at jct 1, 9 mi. N of Uniontown, Kral 45385 (VDB). Lee: ia Earle et al. s.n. (GH US). Lowndes: near Mt. Sinai Church on N side of Hwy. 6, frig et al. 5318 (UWO). Marengo: right-of-way of Co. road 44, ca. 4 mi. E of Dayton, Clark 13714 (NCU). Mont- сна on Alabama River, Montgomery, Harbison 25 (A). Morgan: Smith s.n. (GH, US). Perry: 3.5 mi. E of n, Clair: 2.5 mi. S Asheville on US 231, е 30380 (САМ, VDB). Sumter: 6 mi. N of York, hwy. ‚ RR crossing, Flatwoods region, Jones 15548 (LSU, , V ). PS Concord Church and Cemetery N of Dadeville, Rutland 1997 (AUA). Ar- kansas: Bradley: P.O. Warren, Demaree 24819 (FSU, GH . Hampton, Demaree 14403 (CH, MO). Clark: N side of 1-40, rest area at Arkadelphia, Phipps 5884 (UWO). Clay: Moark, Palmer 4787 (А, MO, S). Cleburne: 5 prings, Palmer 6971 (А, MO). Со- lumbia: Р.О. Waldo, Demaree 39231 (СН, NCU). Craig- head: Crowleys Ridge. Demaree 28795 с ЄН, NGN, e MO). Franklin: lower end of Devils Hol- Creek, Barber 583 (UARK). Garland: Lake Hamilton island, Demaree 39481 (FSU, GH, NCU). is 4. of Saline River, Demaree 16313 (А, MO, VA). Greene: P.O. Walcott, Crowleys Ridge State ng Demaree 27946 (GH). Hempstead: Fulton, Bush 53 (MO). од Springs: Magnet Cove, Palmer 26613 (А, UARK). oward: Bakers Springs, Kellogg 18 (A, MO). Indepen- dence Batesville, Palmer 29769 (A, MO, UARK). Jack- Letterman s.n. (А). Lafayette: S of Walnut (LSU). Miller: Texarkana, Bush 2226 (A). Montgomery: P.O. Hopper, "iced 55131 (USLH, WVA). Nevada: Prescott, Bush 71 (MO). Ouachita: p: Bayou bottoms, P.O. Camden, иа 14417 (А, MO, US ). Pike: Rosboro, Demaree 9473 (A, GH, NO 8 W from Mena, Patterson 96 (UARK). Pulas- ki: behind à Little Rock University botany greenhouse, Sinclair 1730 (NLU). Scott: near Upper Black Fork Basin of Poteau River, Lynn s.n. (UAR 1-40, 2 km W of Ark. 78, Phipps 5883 (UWO). Warren: W Prairie, Kral 64781 (VDB). White: Bald Knob, Anonymous H2863 (US). Florida: Alachua: Gainesville, Murrill s.n. (DUKE, GAM-2311, UWO). Calhoun: Chipola R., W of Blountstown, Godfrey 63072 (FSU, MSC, USLH VDB). Dixie: Old Town, Harbison 5605 (A). Escambia: E edge of Bluff Springs, Beckner et al. 1086 (DUKE, FLAS Gadsden: River Junction, Curtiss 5983 (DOV, GAM, FLAS, GH, MO, US). Gilchrist: ca. 7 mi. N of Fanning = 5 % = Nw { ES Ф 2 o? Springs, at the end of the road to Hart Springs, County Park, along the Suwannee River, Hansen et al. 3 (USF). Hernando: vicinity of Brooksville, Jones s.n. (US- Bridge Road, Coo- ley et al. 5857 (FLAS, USF). Jefferson: % mi. rise near Girardeau's Camp, 1. Hid 1939 (FLAS). Leon: banks of Ochlochnee R., W Tallahassee, Coker et al. s.n. (NCU-77619). Tied е s.n. (FLAS-43415). Suwannee: S of Luraville, Ar- et al. s.n. (FLAS-45214). Taylor: near кү Ae Harbison 10 (A). Wakulla: near Aucilla R. I 8, Wil- liams s.n. (FSU-15214). Georgia: Baker: n River above junction with Тећа awaynochaway Creek, Thorne 7143 (GAM). Bartow: along S side of Green Pond 6.8 mi. 58” E of Adairsville, Duncan et al. 12769 (САМ). Charl- K). St. Francis: N side of ton: St. Mary's River at Traders Hill S of Folkston, Dunc an 1975 (GAM). Dodge: Ocmulgee Ri highway 280, W of Rhi E Wood et al. 4365 ( (CD. Floyd: Rome, Anonymous (US). Louisiana: Caldwell: clay soil on hillside E of Co- penhagen and La. 849 of the Ouachita River, Sec. 13, T12N, R. D. Thomas et al. 95071 (UWO). Lasalle: North- ern half of Sec. 11 in old Bartram’s Prairie, SW of U.S. 84 and SE of U.S. 165 S, R. D. Thomas 94590 (UWO). Orleans: New Orleans, Drummond 105 (UWO). Union: closed Spencer School along road W of La. 143, R. D. Thomas et al. 87883, (USCH, UWO). Vernon: 6 mi. NW of Leesville, Wolff s.n. (MSC). Mississippi: Clay: Tibbee Creek bottoms S of West point, McDaniel 2328 (IBE). Cov- Ж. 2.5 ті. W of Collins, McDaniel 2960 (ІВЕ). For- : NW Corner Marble St. and 39t Woofer 2707 (FSU). ar M | А (UW 0. Grenada: 5 mi. W of Holcomb, о 2340 (IBE). . Biloxi, Seymour et al. 36 (DUKE, MSC, NO). Itawamba: 17 mi. S of e ек Senter place, PR 11-181 (IBE). Jackson: Ocean Springs, Ske- han s.n. (MO). Lauderdale: Meridian, Anonymous H4156 US). Lowndes: Columbus, Anonymous H4189 (US). Mar- ion: at old Oxbow Lake along Pearl River at Columbia, Hwy. 98, Jones et al. 6496 (GAM). Newton: 5 mi. W of Newton, Jackson Prairie Region, Ray 7785 (FLAS, GH, NCU, USF). Oktibbeha: 15 mi. S of Starkville, Noxubee Game Refuge, McDaniel 1687 (FSU, NO, UNA). Pearl River: Walkiak Bluff, Pearl River ca. 8 mi. NW of Pica- yune, Sargent et al. 11389 (GAM). Simpson: Saratoga, Trach 8697 (MO). Smith: Bienville National Forest ca. mi. NE of Pineville, NE4 S22 ca. 0.5 mi. W of Co. Rd., Mitchell 29 (IBE). Stone: 7 mi. E Wiggins, Ray 7689 (FSU, NSU, USE, VDB). Wayne: Waynesboro, Pollard 1224 (GH, MO, US). Wilkinson: N of Wo ees near Do- loroso, Ray 7962 (GH). Winston: ca. 4.0 m of Lou- isville, Smith 502 (UWO). North Carolina: ‘Bladen: near Clarkton, 1. си (US, MO). Cumberland: 2.8 mi. N of jct. and NC. 24 on river road, Ahles 36666 (NCU). ^ Fel posa Little Creek near NC 54, 2 mi. E of pe Hill, Britt 1213 (NCU). Edgecombe: near Tar River 1.5 mi. NNE of Rock Mount, Radford 32032 NCU). Franklin: along Tar River, about 3 mi. E/SE of Bunn, Ahles et al. 11388 (NCU, VDB). Granville: Camp Butler, Batson 747 (DUKE). Halifax: phone line road at a dn 0.2 mi. NE of intersection of 1327, enda 1192 CU). Hertford: 3.3 mi. NNE of Union, Ahles et al. 5. e AS, NCU). Johnston: bank of Neuse R., Boone е Twp, 9 mi. below Smithfield, Fox et al. 1277 (€ ;AM, GH, NCSC, TENN). Lenoir: bank of Neuse River, near Kinston, Totten s.n. 1. .. 1.3 mi. М 41875 (CM, NCU). Orange: US rte. 70 Hillsborough, Boufford 12037 (CM). Pender: Cape Fear er above Holly Shelter Lodge, Fox et al. 180 (NCSC). Wayne: close to Neuse R., Phipps 5113 (ОМО). Wilson: ~ ә = = а . E of Idabel n Okla. 3. Citty 144 (NL U). South Caroling Bamberg: 5 mi. ENE of Ehrharot on County Rt. 21, Ahles et al. 5. (NCU). Barnwell: Sta. 64 of 1. Enet£y Commission, Savannah River 2. Mn Kelley et al. s.n. (USCH). Berkeley: baw 12 bridge, 6 mi. NW of Mc 1. lale. 14337 (NCU). Charles- Volume 85, Number 3 1998 Phipps Crataegus 481 ton: Ravenel, vicinity 3189 (CLEMS, NCU). Cheraw and Societ of Charleston, Bear Swamp, Hunt Chesterfield: highway between y Hill, 2 mi. from Cheraw, Coker et al. s.n. (NCU). Colleton: 5 mi. S of Ruffin, on US 21, Batson s.n. (USCH). Da wy slopes of Lynches River W of Hartsville, Smith 953 (NCU). Fa infield: mi. NE of гаїв, Етеетап a (NCU). Florence: 4 mi. E of Olanta of Byrds Crossroads, Bell 6090 (NCU), Jasper: 5.5 mi. "s of US Rt. 321 on Co. Rt. 90, Ahles et al. 10341 (FSU. GAM). Kershaw: S of Lugoff i US 601 about 1 ч. with Co. Rt. 28-47, Wo s (AUA, . MARY, NLU, TENN, UNA, WVA). бсо Congre River, rand 04 i. N of jet. of Co. Rts. 164 € 42 on Co. 42 (5 of Bowman), Ahles et al. 21674 (NCU). Richland: Beach, Philson s.n. (А, DUKE, USCH). erty, Sharp et al. 15923 (TENN). tation—1 пи. SW Ed. McKinney house, Hebb 26958 (TENN). Shelby: bottoms of Wolf River, NE of German- town Sharp et al. 6592 (TENN). Texas: Anderson: 2.3 mi. NW of Palestine, Shinners 12976 (SMU). Angelina: 6 mi. W of Lufkin, Parks 8012 (TAEM). dde N-side 1-20 5 km E of Texas 8, Phipps et al. 5255 (UWO). Galveston: 1/2 mi. E of I.H. 4S, Dickinson, oa et al. 3370 (TEX Grimes: on Rt. 30, 0.5 mi. E of Navasota R., ryx 2883 (SMU, ТЕХ, UWO). Hardin: SW of “ жне Lundell & Lundell 10900 (TEX). Harris: Buffalo Bayou, a mi. S of Memorial Drive, Smith 35 (SMU). Henderson: off 175, NW of РЕ Lundell & Lundell 11111 (SMU). Jasper: about 8 mi. N of Jasper on US 96 r.h.s., Phipps et al. 6074 (UWO). Lamar: 2.5 mi. W of Paris, oo 57 (SMU). Madison: 3 mi. S of Normangee on Farm Rd. 39, Clark 378 (TAEM). Е. Nacogdoches, Parks 52 (ТАЕ5). Newton: 4 25 ті. S of Newton, along Creek, Cory 52598 (SMU, TEX, UWO). Polk: at e “of % Жы». e pine-hardw qo 2 1171 (ТАЕМ). Shelby: 7.7 mi. NW o rell & Correll 29070 (TEX). Van ha Silver ie: near sex: along Nottoway R., SW of Homeville, Fernald et al. 10280 (GH). Putative hybrids: Crataegus Xnotha Sarg., J. Arnold Arbor. 3: 9. 1922. TYPE: U.S.A. Arkansas: Hempstead Co., hills abt. 5 mi. NE of Fulton, 26 Sep. 1921, E. J. Palmer 20646 (A) [suspected C. marshallii Eggl. X C. brachyphylla Sarg. (ser. Molles)] Five specimens were known in 1922. Crataegus lacera Sarg., Bot. Gaz. 33: 123. 1902. SYNTY PES: U.S.A. Arkansas: Fulton, 2 Oct. 1900, С. S. Sargent s.n. (А); 23 Apr. 1901, W. M. Canby, B. F. Bush & C. S. Sargent s.n. (А); Aug. and Oct. 1901, B. F. Bush s.n. (A) [pos- sibly C. marshallii Eggl. X C. mollis (Torr. & A. Gray) Scheele]. This rare taxon is more similar to C. mollis than is С Xnotha. Neither of these putative hybrids has been ob- served since their original collections. III. Crataegus series Cordatae (Beadle) Rehder, Man. cult. trees Ed. 2: 367. 1940. Crataegus МЕС | Cordatae Beadle [without rank], in mall, Fl. s. e. U.S. Ed. 1: 532. 1903. Cratae- gus sect. Cordatae (Beadle) Eggl., in ја b B. L. Rob. & Fernald, Manual Ed. 6. 1908. TYPE: Crataegus cordata ay a. [= C. phaenopyrum (L. f.) Medik Small trees, 4—8 m tall; trunks with branched thorns or unarmed, bark fibrous-shredding; young branches with simple thorns 2—5 cm long. Leaves + glabrous, * deltate, palmately 3—5-lobed, veins ex- tending to sinuses. Inflorescence 20-30-flowered; anthesis season very late. Flowers small, glabrous; calyces triangular, small; petals small, circular; sta- mens 20, anthers ivory; stigmas, styles, and carpels 3-4. Fruits 5-8 mm diam., + orbicular, glossy, ver- milion; calyx remnants present; pyrenes 3, dorsally grooved. One species, Missouri and Arkansas to North Carolina. Found in woodlands on moist soil, gen- erally in moderate but not very heavy shade. This distinctive series shows some relationships, as discussed at the beginning of this paper, to series Microcarpae, Apiifoliae, and Virides in its leaf ve- nation and small fruits. 5. Crataegus phaenopyrum (L. f.) Medik., Gesch. Bot. 84. 1793. Mespilus phaenopyrum L. f., Suppl. pl. 254. 1782. TYPE: Ehrhart s.n. (holotype, GOET). a pe Mill., Gard. dict., Ed. 8., 1768. Cratae- ordata (Mill) Aiton, Hort. kw. 2: 168. 1789, Roem., Farn. nat. tab. 179 in Mill., Fig. pl. Gard. Dict. Ed. 1., vol. 2. 1760. ? = Crataegus acerifolia Lodd. ex Moench, Verz. auslünd. Báume 31. 1785. Mespilus acerifolia (Lodd. ex Moench) Pi. in Lam., Encycl. 4: 442. 1798. Crataegus populifolia Walter, Fl. carol.: 147. 1788. TYPE: BM? Crataegus youngii ‚ J. Arnold Arbor. 4: 105. 1923. TYPE: idm '6028а (holotype, А) Washington Thorn. Figure 3. Annals of the Missouri Botanical Garden гү» Q ы САЗ > Y yA exp 5:07 Џ А ESE M X C ху A M» N D Volume 85, Number 3 Phipps 483 Crataegus Ї 90° O 000 o o 9 © oe сво M + [-36P + + + + + + -309 CRATAEGUS PHAENOPYRUM (L-F-) MEDIC. 90% 809 + + + + Ho + * " 2| t + yet +++ + P + "s + + + + + + р Distribution of C. phaen nopyrum from collated herbarium records. Essentially — for The Vascular d. Boc of he Southeastern United States area; incomplete nort tail = Steyermark records, centered by county. octagon with re-entrant tai Small trees, 4—10 m tall, usually with a single trunk; trunks with compound thorns, bark fibrous- shredding; extending shoots subglabrous, becoming dark purple-brown; l-year-old shoots deep purple or brown; older grayer; young branches with simple + straight thorns 2-5 cm long. Leaves deciduous; petioles mainly 1.5-2.5 cm long, slender, glabrous; blades usually 3-6 cm long, broadly to narrowly deltate, = glabrous, palmately 3-5(-7)-lobed, sometimes very shallowly so, base truncate to more rarely cordate or cuneate, or leaves smaller and ovate in outline (but still lobed) in some plants from central North Carolina; veins extending to sinuses. Inflorescence 20—30-flowered; branches glabrous; anthesis very late. Flowers 10-12 mm diam.; hy- panthium glabrous externally; calyx lobes broad-tri- angular, 2 mm long, margins entire, glabrous adax- ially; petals circular, white; stamens 20, anthers ivory; styles 3(—4). Fruits numerous, globose, glossy, vermilion, 5-8 mm diam.; or abscissile; pyrenes usually 3, dorsally grooved. calyx remnants present war xsiccatae seen by m Occurring in a broad band from Missouri to North Carolina (Fig. 4), with outliers in all states southward except Alabama, from which earlier re- cords (Clark, C. viridis. Many other Crataegus species may pos- sess, particularly in vegetative shoots, a more or 1971) have proven to be vegetative less deltate leaf shape, and extreme caution is re- quired in identifying such material. Possibly non- native in Delaware, D.C., and Maryland. Found in woodlands on rich soil, generally in moderate but not very heavy shade. Flowering very late, after all congeners. This distinctive species shows some relation- in its leaf venation and sm sprout shoots of C. Ка cum are often different from the mature leaves described above, tending to lower left). Also the terminal lobe is Forms differentiated as C. youngii Sarg. on the ba- sis of an abscissile fruiting calyx and narrower and Figure 3. from Barrows 2 (UWO). —а, e. Infructescence and fru is illustrated by: f, Godfrey 81202 (UWO), lower left; Crataegus phaenopyrum. --а-с. Inflorescence, flower section and calyx lobe, and single leaf (upper right). it parts from ч 803 (BM). —f, g. and g, Skean 3 Variation in leaf shape 7 (UWO), lower right. The Godfrey specimens are typical for sprout shoots, and the Skean specimen illustrates the Aa tendency (see text) with mature leaves. 484 Annals of the Missouri Botanical Garden smaller leaf blades cannot be sustained because of intermediates. These forms are found in central North Carolina, where they are quite common, with several records from Greensboro, Guilford Co., and Chapel Hill, Orange Co., as well as a few records from South Carolina and southern Virginia. Crataegus phaenopyrum is an important woody ornamental valued for its tree-like habit, glossy, ivy-shaped leaves, fine fall color, and brilliant, per- sistent vermilion berries. It is widely cultivated in the central and northeastern United States and through the southern Great Lakes area. It is hardy to USDA Zone 5. I have not considered the species C. acerifolia Lodd. ex Moench or C. populifolia Walter worth typ- ifying. Crataegus acerifolia is usually considered to be a synonym of C. phaenopyrum, presumably on the basis of the name, and is therefore considered here for synonymy. However, Moench's herbarium no longer exists, and the protologue is poor and seemingly contradictory. Relevant detail, translated from the German (Moench, 1785), reads, “The trunk is 12 feet high and thornless. The leaves are alternate, round and of varying form, the (margins deeply sawtoothed,...dark green and not shiny above, below pale green, somewhat hairy on both sides, 5 inches long and 4 broad. The petioles are 1% inches long from which the leaves hang loosely. The 3-inch-long inflorescences. . .common flower stalks, which are hairy. . . . The five calyx lobes are lance-shaped, mostly with entire margins, stand upright. The five petals are white, concave, oval and as long as the calyx lobes. Stamens 16- 20, never more, white (e.g., filaments), as long as the petals. Anthers yellowish. Pistil is split into two diverging halves (— 2 styles?). Hypanthium is cam- panulate. Fruit red, ripening in October, the same size as round-leaved hawthorns, 2 pyrenes with compartments." About the only diagnostic points agreeing with C. phaenopyrum are leaf color, calyx margin, stamen number, and anther color. Leaf size, shape, inflorescence indumentum, and carpel num- ber are different. The subequal calyx lobe and petal length are implausible as is the leaf size unless for sucker shoots. The deficiencies of the protologue are typical of the period. In the case of C. populi- folia, the only Crataegus that could be found at n and brief, is pertinent. It reads, “With trilobed subcor- date leaves, the lobes incised-serrate, smooth on both sides, with long smooth petioles." phaenopyrum is so distinct a species that this di- agnosis probably applies. Neither of these names as any modern currency. The name Crataegus phaenopyrum has been reg- ularly in use on this continent for the Washington Thorn since Eggleston (1908) argued that Aiton's C. cordata (1789) represented another species, per- aps of series Tenuifoliae. Aiton's C. cordata is based on Philip Miller's Mespilus cordata, illustrat- ed (tab. 179) and described in his Figures of Plants... (1760). This illustration, however, does not compare very well with C. phaenopyrum, the flowers being too large, and having 10 instead of 20 stamens. In addition, the toothing of the leaf margins is stronger than is normal for C. phaeno- pyrum. It is clear that the illustration is somewhat inadequate for C. phaenopyrum. On the other hand, the deltate leaf of the illustrated specimen does in- deed resemble that of C. populnea Ashe or C. ira- cunda Beadle (both of ser. Silvicolae), and the fairly large flowers are in line with this. These taxa would have fallen within Eggeston's view of series Tenui- foliae. However, the illustrations are also somewhat Crataegus poor for the two last-mentioned species and also Miller explicitly stated that his Mespilus cordata flowered in late June, which is so late as to defi- nitely exclude series Tenuifoliae (sensu Eggleston) y a good month. It is clear to me that the identity of Philip Millers Mespilus cordata is unknowable with any certainty, on present knowledge of Amer- ican hawthorns, unless one were to embark on the arbitrary process of epitypification. Consequently, I have submitted a proposal to reject the name Cra- taegus cordata (Mill.) Aiton. Representative specimens examined. U.S.A. Arkansas: Washington: river banks, NW area, Harvey s.n. (UARK). Castle: hedgerows and fields, Canby 36 . Florida: Wakulla: Wakulla Springs along Wakulla R., Palmer 38573 (А, MO). Washington: rd. C-280 about 2 mi. from its a iie C-277, Godfrey et al. 81275 (FLAS). nett: from Buford to . Palmer 17648 (А). Maryland: College Park pus, Hayden 235 (MARY). Frederick: along stream 3. N of Wolfeville, Norris : . sr ‘Garfield Rd. 1 mi. from Rt. 15: HUH among the photostats of Walters herbarium 64 (CM). Worcester: roadside 5 mi. NE of Pocomoke City is C. marshallii, but the type description, though оп Rt. 113, Тапай 1822 (COV). Mississippi: Hinds: — Figure 5. Crataegus spathulata. —a, b. Infructescence and fruit parts from Phipps 5282 (UWO). -<-е. Inflores- cence, flower parts, an d magnified leaf from Phipps 5174 (UWO . —f-h. Range of short 1 leaves "s f. Phipps 5303 (UWO); g, two deeply lobed long-shoot leaves from Phipps 51 74 (UWO); h, Thomas e et al. 82031 (UWO). Volume 85, Number 3 1998 Phipps Crataegus 485 486 Annals of the Missouri Botanical Garden Jackson, Harbison 6051 (NCU). North Carolina: Beau- fort: 3 mi. SE of Aurora, Blair 665 (NCSC). Buncombe: low grounds, Biltmore, Anonymous 333b (NC D Catawba: Lyle Creek, between Newton & Statesville, o Of s.n. (NCU-77650). Chowan: 1.7 mi. ESE of кері in the vicinity of Middleton Creek, Ahles et al. 51082 (САМ). Cumberland: Methodist C college Campus, Crutchfield 5600 A, Р " MARY, NCU, NO, UNA. USL Н, WILLI, WVA). Davie: near foot of hill, W of Bear Creek on Statesville p near Mocksville, Totten (NCU). Durham: swamp of N y Hill- Raleigh dn Totten 1 (NC U), Guilford: near SW lim- its of Greensbo corner of intersection of 1.95 and E Lee Rd., ДІ 5120 (UWO). Henderson: NW of US. 6 5 М mi. NE of Hendersonville, Pittillo 115 (FSU, КҮ, NCU). Iredell: Statesville, Patterson s.n. (MSC-43895). Orange-Durham: Chapel Hill, swamp of Bowlins’ Creek near Handcock's Tyrrell: by i 5 of Sandy Point Landing: Fox et al. 4463 (NCSC). Wilson: near Contentnea Creek, 2 mi. E of Black Creek, Radford 35682 (NCU). Ohio: ton: near a park in Glendale, Adams 81 (KY). Bowling Green-southern, Price s.n. ennessee: Da- vidson: around Nashville, Gattinger BA Curtiss 803 (BM). IV. Crataegus series Microcarpae а Rehder, Man. cult. trees, Ed. 2: Crataegus sect. Microcarpae Lo d 2. frutic. brit. 825. 1835-1838. ТҮРЕ: Crataegus microcarpa Lindl. (= C. spathulata Michx.). Small tree to 7 m or bush, often with very tabulate branching when open grown; main trunk erect, bark smooth with thin flakes, cream to cinnamon-brown or light gray-brown in patches; thorns simple, abun- dant to few, straightish, of medium length, 3—4(—5) сш long, blackish. Leaves barely petiolate; blades small, coriaceous, long-persistent, lobed, somewhat glaucous, glabrous; those on short shoots elliptic to obovate, lobes 2-3, acute to rounded, LII about 5% to 30%, veins extending to sinuses, those on rapidly elongating shoots generally more diverse in shape, more deeply lobed and larger. Inflorescence mul- ti(20-30)-flowered, glabrous; anthesis late. Flowers small; calyx short-triangular; petals small, circular, white; stamens 20, anthers pale yellow; styles 3-5. Fruit small, 3-5 mm diam., globose, often numer- ous in a corymb, bright orange-red with 3-5 ру- renes obscurely dorsally grooved, their top quarters exposed. One species, common and widespread in the southeastern United States, west to Texas. 6. Crataegus spathulata Michx., Fl. bor.-amer. Ed. 1: 288. 1803. Mespilus spathulata (Michx.) Poir, in Lam., Encycl., suppl. 4: 68. 1816. PE: U.S.A. North Carolina: Michaux s.n. (P photostat). Crataegus microcarpa Lindl., Bot. Reg. 22: t. 1846. 1836. TYPE: U.S.A. Lindley s.n. (CGE not seen). Littlehip Hawthorn. Figure 5. Small tree to 7 m or bush, often with very tabulate branching when open grown. Main trunk usually erect; bark smooth with thin flakes, greenish when younger, then cream to cinnamon-brown or light gray-brown in patches; branchlets with extending shoots pubescent or glabrous, becoming dull red- dish; 1-year-old shoots dull gray; older gray; thorns simple, abundant to few, straightish, of medium length, 3—4(-5) ст long, blackish. Leaves decidu- ous; petioles + lacking; blades 1.5-3 cm long, + coriaceous, long-persistent, lobed, somewhat glau- cous; those of short shoots elliptic to obovate, nar- rowly cuneate below, lobes usually 0-2 per side, acute to rounded, LII about 596 to 30%, veins ex- tending to sinuses, if present, those on rapidly elon- gating shoots generally more diverse in shape, more deeply lobed and larger; glabrous abaxially; adax- ially somewhat hairy above especially near the mid- vein, becoming glabrous; rarely overall pubescent. Inflorescence multi(20-30)-flowered; branches gla- brous; anthesis medium-late. Flowers ca. 10 mm diam.; hypanthium glabrous; calyx lobes broad-tri- angular, 1.5-2.0 mm long, subentire, adaxially and abaxially glabrous; petals 4—5 mm long, + circular, white; stamens 20, anthers usually pale yellow, an- thers 0.5 mm long; styles 3-5. Fruit ca. 3-5 mm diam., globose, often numerous in a corymb, bright red or orange; calyx present, lobes reflexed; pyrenes 3-5, obscurely dorsally grooved. South-central Texas to northern Florida and north to Arkansas and Virginia (Fig. 6). A distinct and locally common species of the south, but hardly recorded from Mississippi. The fairly high frequen- cy of collection overall for C. spathulata suggests that this Mississippi gap is not a collection artifact but has natural causes in which case the hypothesis may be entertained that the separate eastern and western populations derive from different Pleisto- cene refugia. Woods, fence-lines, and brushy plac- es, on a variety of soils, apparently relatively he- liophilous. Crataegus spathulata exhibits considerable var- iation in leaf shape and fruit color. In addition, rare forms possess quite pubescent foliage. It appears to be most closely related to C. phaenopyrum (ser. Cordatae) and is possibly related to C. viridis (ser. Virides). It is very easily recognized. This is the only species of Crataegus known to the author reg- ularly possessing such strikingly thin, flaking outer bark, a characteristic by which it may be easily recognized even in winter. However, C. viridis L. Volume 85, Number 3 Phipps 487 1998 Crataegus T T poc 90? 809 |-- 369 ù + Ta + 36% = + a dut M + E +, ^ i + + + ++ x.» + + à. an + НЕ ЕД + + Ра + кк! Ұ ғу Т t+++ 4 Lt + " +\ + oe + + + TE + fn + ++ + + + FF pt + + + + + ++ + + + + + + + L30»** % ti +) + + 30% ++ + а CRATAEGUS SPATHULATA MICHX. o 90 0° Figu Distribution of Crataegus spathulata from collated herbarium records. Nearly all records seen for The eor Шш of the Southeastern United States area; inexact northward. + = exsiccatae seen by me; A = literature ords. also possesses this feature in a less marked degree. ПОЈ, Drew: Р.О. Wilmar, Demaree 24109 (FSU, СН). The chromosome number is not known, but the var- iation pattern suggests that C. spathulata is mainly MO). Hot Springs: Р.О. Magnet Cove, Demaree 124854 a sexual diploid. (GH). Howard: 5 mi. SE of Mineral Springs (ca. 35 mi. NNW of Texarkana), Лиз et al. 684 (UARK). Miller: Tex- 2 specia ns examined. A. Alabama: arkana, Pringle s.n. (GH). Montgomery: vicinity of Hous- Autau qiue attville, Smith 4 (UNA) ПШ Же. ley Point оп Lake Ouachita National Forest, ca. 10 mi. E T mi. NW of U 4. а is 7. “a ier" of Mount Ida, Thieret 18216 (FSU, USLH). Pike: Mur- es county сов ns d јип-олегокее — freesboro, Demaree 9963 (СІ, GH). Pulaski: P.O. Natural 4979 (TENN, UWO). Cullman: covering low banks, Mohr неке Hartford parvis 39328 (A, MO, US). Se- DeKalb: Al. 35, ca. 2 mi. N и of Little River, ~ prope des ЕЕ 2164710). Тейлор GEW under power line; Whetstone et al. 23122 (SU). E Lower road 10 mi. SW of Dardanelle, Henbest s.n. (UARK- rei МР Chur chi 7 2 gie (MSC). а. 36275). Florida: Gadsden: River Junction, Curtiss 5989 P rage FLAS, GAM, GH, MO, NCU, US). Jackson: ca. 9 Warrior River by Ala. 14, ca. 4.2 mi. E of Eutaw, Kral 46853 (VBD). е Birmingham, Harbison 25 (А). NW of Marianna, Godfrey et al. 54275 (FSU, DUKE, Lee Co.: Auburn, Earle et al. s.n. (MO-1917666, NCU, А à 2 dp ` S). Lowndes: US 80, 14 mi. г of Dallas ‘Gan line, Georgia: Barrow: Gay's Aeres, Morgan 38 (GAM). Kral 55038 (VDB). Macon: 2 mi. S of Hardaway, Grant Clarke: Athens, Agricultural Campus, Miller E3281 087 (AUA). Marshall: along dna Creek, Clark 12094 (UNA). Columbia: S of eastern summit of Rosemont Moun- UNC). Montgomery: 2 mi. S of Sellers, Uttal 10972 tain, Duncan 28969 (GAM). Decatur: 1 mi. N of Chatta- (NCU). Shelby: Oak Mountain State Park, towards the hoochee, Muenscher et al. 3077 (MO). Dougherty: along summ o of mountain, Kral 41513 (VDB). Talladega: on Flint River at "gui ye s.n. (A, US). Dade: on Sand Ala. 76, just E of Childerburg, location “Coosa Pines,” Mountain 4.1 mi. S of Trenton, Whetstone 96 (GAM). 25 2488 (UWO). Tuscaloosa: 11 mi. S of city of Tus- Early: between E and the Chattahoochee River, God- qe Cooley et al. 3595 (USF). T Ashley: Р.О. frey 79085 (FSU, MISSA). Floyd: Horseleg Mtn., Phipps . Demaree dp (СН). Bradley: Р.О. Warren, De- 5303 (UWO). Gwinnett: Yellow River Valley, near Mc- maree 24836 (СН). Calhoun: Р.О. Ham pias, Champag- Guire's Mill, паде g (A, GH, NCSC, US). Hall: nolle Bayou Bottoms, Demaree 16861 (TENN). Clark: be- E of Gainsville by 7 mi., W. H. Duncan 3232 (GAM). tween Gum Springs and Hollywood, Phipps et al. 5244 Harris: Blackman Place n near Cataula just S of Ossahatch- 488 Annals of the Missouri Botanical Garden ie Creek, Jones et al. 22322 (FLAS, GH, USF, VDB). Madison: 4 mi. E of Danielsville, Hume s.n. (DUKE, FLAS). Meriwether: near Durand, Jones 20941 (UNA, WVA). Richmond: Augusta, Harbison 6080 (A). Seminole: N side Ga. route 91 of Chattahoochee R., Phipps 5225 (UWO). Ao Curahee Mountain off Ga. route 23— U.S. route 123, ры et al. 17745 1! MO, e . NW of Plain Dealing, Allen et al. 8039 . FSCL, САМ, NCU, ХО, USLH, WVA). Caddo: Jack Price residence N of Blanchard-Shipp Road, W of Blanchard, R. D. Thomas et al. 41810 (NLU). Calcasieu: along the Calcasieu River, Nogle s.n. (USLH). Caldwell: hills N of Copenhagen, E of La. 849, R. D. Thomas et al. 35884 (NLU). Catahoula: woods poule La. 124, 3 mi. W of Duty Ferry, R. D. Thomas et al. 43282 (NLU). DeSoto: 3.2 mi. 5 of US 84, S of Logansport, R. D. Thomas et al. 68418 (NLU), D. Dixon 2088 (NLU). Grant: roadside of Bear Creek Rd., Parish road 521, Parker 88 (NLU). LaSalle: SW of US 84 and SE of US 165 SW of Tullos, R. D. Thomas 94592 (TENN, UWO). Morehouse: Spyker Estate, Crow et al. 551 (NLU). Natchitoches: Natchi- toches, Bush 5442 (CI, US). Ouachita: W of La. 557 be- tween Cypress Creek and Caldwell Parish line, R. D. Thomas et al. 93836 (UWO). Red River: E of Coushatta, R. D. Thomas 45635 (NLU). Richland: Alto, R. S. Cocks 22 (A). Winn: Kisatchie National Forest, Kessler 1213 (NLU). North Carolina: Macon: Clar Creek section, on Highlands-Wallkalla road, Wright s.n. (NCU-11425). Montgomery: Е of Narrows Dam on Yadkin River 2 mi. NW of town of Uwharrie, Wells 3048 (NCU). Бома 5тай et al. 47 (US). South Carolina: Aiken: along Co. Rt. vicinity of jet. of Co. Rt. 103 and с. Rt. 47, мене et al. 4943 (CLEMS, MO, NCU, NLU, VCU). Anderson: Anderson, Davis 198 (CI. NCU, US). Cherokee: SE of Gaffney, Ahles et al. 26959 (NCU, USF). Chester: where SC 72 crosses Broad River, Phipps 5124 (UWO). Henderson: Davis s.n. (СІ-245197). Lexington: vi- cinity of Batesburg, McGregor 107 (US). McCormick: dry pine-oak ridge 1 mi. NE of the Savannah River on S.C. 8, Bozeman et al. 8845 (AUA, CI, DHL, FLAS, FSU, GAM, KY, LYN, MARY, NCU, NO, TENN, 2. USCH, USLH, UWO, VDB, WILLI). Newberry: : f Chappels, Bell 9184 (NCU). Richland: B Salada Dam near Columbia, Coker s.n. (NCU-83574). Tennessee: Knox: Knoxville, Bright 13283 (CI, UWO). Texas: Braz- os: Bryan, Anonymous H8076 (US). Fort Bend: few mi. WSW of Rosenburg and 1 mi. E of 541 exit, Phipps et al. 6083 (UWO). Harris: ENE of Huffman on F.M. 1960, Phipps et al. 6067 (UWO). Hopkins: 7 mi. W of Sulphur Springs, Shinners 13045 (GAM, SMU). Newton: Saw Mill Road about 1.5 km W of Saw Mill Town, Phipps et al. 6078 (UWO). Shelby: 7 mi. S of Center, Shinners 18463 (SMU). Trinity: Chambers s.n. (TEX). Walker: 33 on Na- ture Trail, Huntsville State Park, Mahler 8926 (SMU, UWO). — - V. Crataegus series Brevispinae (Beadle) Reh- der, Man. cult. trees, Ed. 2: 366. 1940. Cra- taegus [subgroup] Brevispinae Beadle [without rank], in Small, Fl. s. e. U. S. Ed. 1: 532. 1903. nae Beadle TYPE: Crataegus brachyacantha Sarg. & Engelm Bush to medium-sized tree, commonly 6-10 m tall; trunk with plated grayish bark, when older with branched thorns; branches thornless to thorny, thorns short, recurved, usually « 1.5 cm long. Leaves deciduous, short-petiolate; blades coria- ceous and shiny, those of spur shoots small (ca. 2 cm) elliptic, crenate-margined, of elongation shoots often much larger (> б cm) and variously lobed, sometimes to the sinuses. /nflorescences with 15-25 flowers; branches glabrous; anthesis mid season. Flowers small, calyx lobes small; petals small, white until old; stamens 20, anthers cream; styles 4—5. Fruit + globose, mealy, black to bluish black, usu- ally with a heavy, waxy bloom at maturity; pyrenes 5, with slight dorsal grooving. One species, Louisiana and neighboring states, disjunct in Georgia. А very distinct series on account of its unusual, short, recurved thorns, very large plant size in some specimens, glossy leaves, among the smallest flow- ers in the genus, turning orange with age, and black fruit. The limited distribution range is remarkable and difficult to account for. 7. Crataegus 12. Sarg. & Engelm., Bot. Gaz. 7: 128. 1882. TYPE: U.S.A. Texas: таи io Aug. 1882, G. W. Lettermann s.n. (lectotype, here designated, A). Blueberry Haw, Hoghaw, Pomette Bleue. Fig- ure 7. Bush to medium-sized tree commonly 6-10(-15) m tall; trunk with plated gray bark; thornless to thorny, thorns short, recurved, usually < 1.5 ст long or branched on larger trunks. Leaves decidu- ous; petioles 3-8 mm long, glabrous; blades cori- aceous and shiny, those of spur shoots 2-3 cm long, elliptic, unlobed, crenate-margined, those of vege- tative shoots often much larger (> 6 cm long) and variously lobed, sometimes to the sinuses. Inflores- re 7. Crataegus brachyacantha. --а-с. — Fruiting branch and infructescence, fruit parts, and leaves from Duncan & Duncan 4128 (DUKE). —d, e. Inflorescence and flower section with calyx lobe from J. C. Chaffe s.n. (РОУ). Volume 85, Number 3 Phipps 1998 Crataegus 489 Tis. HG 4 Spi LV КА Ра «а д үз У ОРИ NS d ч ™ е” 19 ES a y 2 ФА) ў 556 p? A HAN à Жар» NUR UM " У 49% б ff А у À іст 490 Annals of the Missouri Botanical Garden 9 Pd + —340 34° + + +} н + + + -309 94 і | CRATAEGUS BRACHYACANTHA SARG. & ENGELM. 909 Figure 8. Distribution of и brachyacantha; main range essentially complete from collated herbarium re- PE pilas: incomplete for Ti cences with 15-25 flowers; branches glabrous; an- thesis mid season. Flowers 7-9(-11) mm diam.; hy- panthium glabrous externally; calyx lobes 1.5 mm long, triangular, margins entire, glabrous adaxially and abaxially; petals ca. 3 mm, circular, white until old; stamens 20, anthers cream, 0.5 mm; styles 4— 5. Fruit + oblate, 8-14 mm diam., flesh mealy, black to bluish black, with heavy, waxy bloom be- fore maturity; calyx remnants erecto-patent to ob- solete; pyrenes 4-5, with slight dorsal grooving. The distribution of Crataegus brachyacantha is centered in Louisiana, where this species is locally common. It is also found in southeastern Oklahoma, eastern Texas, southern Arkansas, and southwest- ern Mississippi. А disjunct population has been collected from Georgia (Fig. 8), but specimens from there cannot be found now. Although C. brachy- acantha has often been reported as a tree of wet prairies, alluvial flats, etc., I have frequently seen it, though generally as a smallish tree, on well- drained mesic sites in various parts of its natural range. This species is seemingly more shade-tol- erant than many species of hawthorn. It flowers dur- ing mid season. Crataegus brachyacantha is a very distinct spe- cies; the flowers turn orange with age or on drying, and are remarkable in this respect. The short, re- curved thorns and bitter, oblate, black fruits are also distinctive. А very heavy bloom on the some- what immature fruit, when the skin is purplish, gives rise to a bluish appearance and also to two of the common names. However, the fruit is nearly always full black at maturity. The fall foliage, bril- liantly colored lustrous orange, bronze, and red, suggests potential ornamental use. The frequently deeply lobed leaves with veins to their sinuses on elongating shoots constitute a presumptively ances- tral character that helps relate C. brachyacantha as well as such groups as series Virides to the other species treated in this paper. This species may be the tallest species of hawthorn in the United States. According to Sargent (1890), Crataegus brachy- acantha is not hardy in the north. The chromosome number is unknown. Crataegus brachyacantha f. leucocarpa Sarg. (J. Arnold Arbor 3: 10. 1922) is a white-fruited form that was collected at Nachitoches, Nachitoches Par- ish, Louisiana, in September 1915 (Palmer 8719, A). The holotype is the only known collection of this forma. Representative sj ns examined. U.S.A. Arkansa Ashley: P.O. € жне "160 ft., Demaree 22033 (А, МО). Little River: Ashdown, Palmer 8386 (A). Miller: Техаг- Капа, Palmer 22459 (^, MO, UARK, US). Georgia: Bak- Volume 85, Number 3 er: 4 mi. N of Newton, Duncan & Duncan 4128 (A, DHL, DUKE, FLAS, САМ, NCSC, NCU, US). Louisiana: Bien- ville: off Louisiana 4, ca. 2 mi. W of Friendship, Robinette 212 (NO). Caddo: ca. 4 mi. a of Keatchie, Thieret 22549 (USLH). DeSoto: ca. 1 mi. E of Naborton, ca. 8 mi. d of Mansfield, Thieret 27468 (DUKE. FSU, GAM, Franklin: Murphy Woods property 5 of Old Mixon Se bool S of La. 132 and NW of Archibald, R. D. Thomas et al. 88742 (UWO), 7032 (UWO). Nac рее. ere d а Cocks s.n. (A). Ouachita: cultivated, Purchase Gardens & Zoo, Monroe, R. D. Thomas et al. 30771 (NLU). Richland: E of La. 17 and S of La. 877 and Mitchiner, R. D. Thomas et al. 59400 (NO, NLU, LSU), 7864 (LSU). Sabine: Toro, posi 48200 (NCU, NO, UWO). St. Landry: Opelousas, n. (US-139229). St. Tammany: Glen Gordon. же itor. "Cocks s.n. (A, NO). 2. Lake Bistinean Stat Park campground « NW of Ringgold, Thieret uu pide FSU, GAM, "USL H). West gm N of La 134 at curve E of Little Colewa Cree W of Epps, R. D. Thomas 84381 (NLU). Texas: Fran Klin: ы km SE of Mt. V “arm Road 21, Phipps et al. 5260 (UWO). Liberty: « on FM 162, about 5 mi. E of Moss Hill : W of Phipps et al. 6069 (UWO). а th: С. С. Me- Donald тн along e Ка. in area “В,” a side of Lake Tyler East, ca of S sien Wilkin son 360 nun. Y Wood: near re NE Texas, ond man s.n. (А, MO, UWO). Literature Cited Aiton, W. 1789. Hortus Kewensis. George Nicol, London. Сћапеу, R. W. 1927. Geology and paleontology of the Crooked River Basin, with special reference to the Bridge Creek flora. Pp. 45-122 in R. pus et al. (editors), Contributions to Palaeontology from Basin Regions of North America IV. Carnegie Institute of Washington, Washington, D.C. Phipps 491 Crataegus Clark, R. C. 1971. + n pue of Alabama. Ann. Missouri Bot. Gard. 5 Cronquist, A. 1980. ы Flora of the Southeastern United States, Vol. 1. Univ. North Carolina Press, Chap- | Hill. Eggleston, W. W. О. The Crataegi of the northeastern United States к ye ү: Canada. Rhodora 10: 73— El-Gazzar, A. 1980. The taxonomic significance of lea morphology in Crataegus (Rosaceae) Bot. Jahrb. Syst. 101: 457-469. Miller, P- 1760. Figures of Plants Described in the T dener’s Dictionary, lst ed., Vol. 2. Rivington et al., aon. . Gardeners’ Dictionary, 8th ed. London. Miseni: с p^ Verzeichniss auslündicher Báume und Stauden. J. G. Fleischerischen Buchhandlung, Frank- urt. Phipps, J. B. 1983. res taxonomic and cla- distic relationships between Asiatic and North can Crataegus. Ann. Missouri Bot. С 1988. Crataegus (Maloideae, "Rossceae) of the soushagaienn United States, Introduction and series Aes- tivales. | Arnold Arbor. 69: 401—431. 297. Monograph ^t ез Mexican Cratae- gus. Sida | Miscell. 15: 1—94. Introduction 5 the red-fruited Crataegus (Ross bs p western North America. Canad. J. Bot. 56 (in press). & Muniyamma. 1980. A taxonomic revision of Crataegus (Rosaceae) in Ontario. Canad. J. Bot. 56: . J. Rohrer € P. С. Smith. 1991. Origins and evolution of Қаш cond Maloideae (Rosaceae). Syst. Bot puja Sarger y Б, . The Silva of e чаша Vol. 4. ШО o Boston and New Y: 922. Manual of the Trees of N North America, Vols. I, IT. Houghton, Mifflin, Boston and New 2nd x York. ADAPTIVE RADIATION OF Peter Goldblatt,? John C. Manning,? and BEE-POLLINATED Peter Bernhardt* GLADIOLUS SPECIES (IRIDACEAE) IN SOUTHERN AFRICA! ABSTRACT Observations on the flowers of 45 of 166 species of southern African Gladiolus (in sects. Blandus, Densiflorus, Hebea, Heterocolon, анарын сша Ophiolyza) show that 42 species are pollinated largely by polylectic bees in the family Apidae, 2 species by bees of the families Andrenidae or Halictidae, and 1 by a combination of hopliine beetles (Scarabaeidae) en Andrenidae The floral phenology, attractants, 4. of floral foragers, and sometimes the rewards, vary according to geography and are not correlated with taxonomy. Flowering in most Gladiolus species in the southern dig winter-rain fall zone coincides with the end of the ey ат August to Oc tober, but a few flower ге to April, from the mi o the end of the wet season, but a few species bloom from August to November, at the end of the dry season. Their n owers have no discernible fragrance and are colored mostly in shades of pink to mauve or white. Most floral foragers collected on these species were bees in the genus Amegilla, but other bee genera, as well as flies in the genera Prosoeca and Sten hin ipteron, were captured. Among the southern African species of Gladiolus pollinated by bees, there are in distinct пе nei strategies. The majority have bilabiate, “gullet” flowers or “flag” flowers that secrete sucrose-rich nectar at the base of an obliquely funnel- shaped floral tube 9-20 mm long with the r phora) contact dehiscent anthers m receptive stigmas while probing the tube for nectar with elongated к es. ate пл iantl offering little or no кесш at the dus of tubes Pen Шай 7 m m long. Andrena De les or oe Apis mellifera Ne both dehiscent anthers and receptiv e stigmas of G. stellatus or G. 4. respectively, while foraging for pollen. An additional 53 Gladiolus species also have bilabiate, gullet or flag flowers with obliquely funnel-shaped tubes 9— 20 mm long (the most common flower type in the genus), and are иза ве also to be adapted for pollination by long- tongued anthophorine and honey bees. The actinomorphic, rotate floral form is present in 2 more species. Thus, 60% of the Gladiolus species in southern Africa may be regarded as being pollinated by bees, and the overwhelming majority of these species (9596) have a or flag flowers and are visited ppm by long-tongued anthophorine bees that are foraging for nectar. The remaining species of Gladiolus in southern Africa have flowers with elongate perianth tubes and are adapted for pollination by sunbirds or insects other than bees, most importantly long-tongued flies (Nemestrin- idae, Tabanidae), moths, and the large satyrid butterfly, Aeropetes. Gladiolus, the е genus of Iridaceae subfam. ѕрр.), Hebea (31 spp.), Heterocolon (10 spp.), Hom- ioideae, consists of approximately 250 species — oglossum (51 spp.), Linearifolius (17 spp.), and (Goldblatt, 1996; ‘Goldblet & Manning, 1998) dis- — Ophiolyza (16 spp.)) based on vegetative, floral, and tributed throughout Africa, Madagascar, and west- d characteristics (Goldblatt & Manning, ern Eurasia. Southern Africa is the center of di- 1998). Species of the genus, especially those from versity of the genus, and some 166 species occur ш Africa, have been favorite horticultural in Africa south of the Zambezi River, over 95% of subjects in Europe since the beginning of the 19th them endemic (see Goldblatt, 1996). The southern — century, and cultivars based on interspecific hy- African species are currently divided into seven brids have been marketed since the 1820s (Gold- sections (sects. Blandus (21 spp.), Densiflorus (20 blatt, 1996). ' Support for this study by grant 5408-95 from the National Geographic Society is eo acknowledged. We thank R. W. Brooks and C. D. Michener, Snow Entomological Museum, xeu rsity of Kansas, and V. Whitehead, South African Museum, Cape Town, for identification of bees; B.-E. van nr Afrikaans University, tinea for the nectar analysés; and Yevonn Wilson-Ramsay, for Bie insect illust "i A. Krukoff Curator of African Botany, Missouri arena Garden, P.O. Box 299, St. Louis, Missouri 63166, 9 а on Herbarium, National Botanical Institute, P. Bag. X7, Claremont 7735, South Africa. * Biology Department, St. Louis University, St. Louis, Missouri 63103, U.S.A. ANN. MISSOURI Bor. GARD. 85: 492-517. 1998. Volume 85, Number 3 1998 Goldblatt et al. Bee-pollinated Gladiolus Species Despite a long history of cultivation, data on the reproductive biology of Gladiolus are meager. In- formation on intraspecific compatibility in the ge- nus has been derived primarily from experiments on cultivars and a few hardy species. Available lit- erature suggests that species are self-incompatible (Darwin, 1876; Goldblatt, 1971; Ohri & Khoshoo, 1981), while self-compatibility is frequently corre- lated with interspecific cultivars and/or high levels of ploidy (Knox et al, 1976; Ohri & Khoshoo, 1981). The self-incompatibility so far reported in Gladiolus is in marked contrast to the situation in Ixia, Romulea, Sparaxis, and Watsonia. Most spe- cies so far examined of these four genera, also members of subfamily Ixioideae, are self-compati- ble, but show reduced fertility (Horn, 1962; de Vos, 1972) when selfed as compared with xenogamous crosses. А few species of Sparaxis, however, are fully self-incompatible and others are autogamous (Goldblatt, 1992). The pollination ecology of Gladiolus species has until now been barely investigated. Cultivated specimens in India have been reported to be pol- linated by the Indian honey bee, Apis indica (Ohri & Khoshoo, 1981). Observations of wild species are few and scattered, and with hindsight, often unre- liable. Scott Elliot (1891) first commented that the hairy bodies of a hopliine (Scarabaeidae) beetle, Anisonyx ursus, carried the pollen of flowers of С. gracilis and G. hirsutus (as G. pilosus) in South Af- rica. Marloth (1908) reported a long-tongued fly (Philoliche rostrata, Tabanidae) foraging for nectar presumably on G. bonaspei Goldblatt & de Vos (as Antholyza merianella L.), an observation that is er- roneous since that fly is not on the wing when G. bonaspei blooms. Vogel (1954) suggested that the Gladiolus species of southern Africa could be sub- divided into four different pollination groups based on floral characters believed to be associated with either bee, sphinx moth, bird, or combined butterfly and long-tongued fly pollination. Unfortunately, Vo- gel's field observations on Gladiolus species were evidently limited to his sighting of a sunbird (Nec- tarinidae) foraging on the flowers of G. dalenii van More recent and reliable observations on the pol- lination of Gladiolus have been made in the course of studies of pollination guilds in southern Africa. Johnson and Bond (1994) have recorded the large butterfly, Aeropetes tulbaghia, as a visitor and likely pollinator of a few Gladiolus species with large, red flowers, including the well-known G. cardinalis Curt. Manning and Goldblatt (1995, 1997) recorded the long-tongued flies Moegistorhynchus longirostris (Nemestrinidae) and Philoliche rostrata (Tabanidae) foraging for nectar on G. angustus L. and G. un- dulatus L. and Prosoeca longipennis (Nemestrini- dae) foraging on G. bilineatus G. J. Lewis and G. engysiphon G. J. Lewis. The anthers and stigmas brush the dorsal surface of the thorax of these flies and deposit large quantities of pollen when the flies probe the floral tubes with their hyper-elongated mouthparts. Gladiolus meliusculus has been report- ed to be pollinated by a combination of andrenid bees (Andrena sp.) and hopliine beetles (Lepisia, Pachycnema; Goldblatt et al., 1998). Both bees and beetles carried the pollen of G. meliusculus and contacted the stigma lobes during foraging, while beetles also brushed against dehisced anthers and stigmas while mating or engaging in agonistic be- havior. Thus, existing observations are either un- reliable or involve unusual and highly specialized pollination systems based on a small sample of spe- cies, all from the southern African winter-rainfall zone. Available information does, however, indicate that pollination in Gladiolus is extremely diverse. This apparent diversity in pollination biology is matched by a broad range of colors and marking patterns in the flowers of southern African Gladi- olus species. The predominance of bilabiate, “gul- let" flowers (sensu Faegri & van der Pijl, 1979) with relatively short floral tubes, however, suggests that bees should be the dominant pollinators in the genus, and our aim has been to examine this hy- pothesis. Bees are known to contribute to the pol- lination of some species of Iridaceae in southern Africa, including Moraea (Goldblatt et al., 1989), Nivenia (Goldblatt & Bernhardt, 1990), and Lap- eirousia (Goldblatt et al., 1995). The bees appar- ently most important in the pollination of these taxa have long tongues and large bodies exceeding 10 mm in length (e.g., Apis, Amegilla, and Anthopho- ra). We present the following information as part of a broader survey of pollination in Gladiolus that aims to document the different pollination strate- gies that occur in the genus. MATERIALS AND METHODS Species examined. Хе present direct observations made during the years 1993 to 1998 in the field in southern Africa and in living collections at the Mis- souri Botanical Garden, St. Louis, and Kirsten- bosch Botanic Gardens, Cape Town, on aspects of the floral biology of the 80 species of southern Af- rican Gladiolus that have short-tubed, gullet, flag or rotate (stellate) flowers (Table 1). Some 19 more species have flowers closely resembling those ex- amined, but the remaining 67 species of the genus in southern Africa have flowers with elongate floral 494 Annals of the Missouri Botanical Garden Table 1. Species of southern African Gladiolus with flowers adapted for pollination by bees, with field sites and voucher information for species examined for pollinators and/or floral characters (sites marked * if no pollinators captured). Study sites are in South Africa unless otherwise indicated. Species are arranged taxonomically according to Goldblatt and Manning (1998). Vouchers are housed at MO (Goldblatt) or at NBG (other collectors). Study site Species Province, Locality Voucher Gladiolus sect. Densiflorus series Paludosus G. paludosus Baker Mpumalanga, near Belfast(*) Goldblatt & Manning 10133 G. papilio Herb. Mpumalanga, near Dullstroom Goldblatt & Manning 8941 series Densiflorus G. crassifolius Baker Free State, Witzieshoek Goldblatt & Manning 9861 Zimbabwe, Nyanga Goldblatt 9077 С. densiflorus С. J. Lewis Mpumalanga, Long Tom Pass(*) Goldblatt & Manning 9840 G. exiguus G. J. Lewis Mpumalanga, Long Tom Pass Goldblatt & Manning 9838 G. ferrugineus Goldblatt Mpumalanga, Long Tom Pass Manning 2109 & J. C. Manning Mpumalanga, Graskop Goldblatt & Manning 9826 G. hollandii L. Bolus Mpumalanga, near Barberton(*) Goldblatt & Manning 9845 G. serpenticola Goldblatt Mpumalanga, Barberton-Nelspruit(*) Goldblatt & Manning 9844 & J. C. Manning series Calcaratus G. appendiculatus G. J. Lewis Mpumalanga, Long Tom Pass Goldblatt & Manning 10644 series Scabrid G. brachyphyllus F. Bolus Mpumalanga, Tshokwane(*) Goldblatt & Manning s.n. G. ochroleucus Baker Eastern Cape, near East London(*) Goldblatt & Manning 9534 G. pavonia Goldblatt & Mpumalanga, Abel Erasmus Pass(*) Goldblatt & Manning 9831 C. Manning Gladiolus sect. Ophiolyza series Oppositiflorus G. dolomiticus Oberm. Northern Province, Makapansgat Goldblatt & Manning 10472 G. elliotii Baker Gauteng, Bronkhorstspruit(*) Goldblatt & Manning 10134 G. pole-evansii Oberm. Mpumalanga, Denilton(*) Goldblatt & Manning 9808 G. sericeovillosus Baker Mpumalanga, Amersfoort Goldblatt & Manning 9850 Zimbabwe, Nyanga Goldblatt 9083 series Ecklonii G. ecklonii Lehm. Mpumalanga, near Dullstroom Goldblatt & Manning 9803 G. vinosomaculatus Kies ;auteng, Pretoria hills(*) Goldblatt & Manning 9801 Gladiolus sect. Blandus series Phoenix G. crispulatus L. Bolus not studied G. phoenix Goldblatt & Western Cape, Bain's Kloof Goldblatt & Manning 10122 C. Manning G. oreocharis L. Bolus not studied series Sabulosus G. gueinzii Kuntze Western Cape, Great Brak River(*) Goldblatt & Manning 9522 & 9523 series Floribundus G. grandiflorus Mill. Western Cape, Burger’s Pass Goldblatt & Manning 10017 G. rudis Roem. & Schult. Western Cape, Die Galg(*) Lewis 6104 Volume 85, Number 3 1998 Goldblatt et a 495 „карик л Gladiolus Species Table 1. Continued. Species Study site Province, Locality Voucher Gladiolus sect. каш series Pubig G. G. G. G. G. G. dian Coldblan & J. Barres Goldbla & J Manning parvulus Schltr. pubigerus G. J. Lewis woodii Baker zimbabweensis Goldblatt series Linearifolius G. G. G. aure lus brevifolius Jacq. hirsutus Andr. Gladiolus sect. Heterocolon series Unguiculatus G. oatesii Rolfe series Heterocolon G. ze >. rubellus Goldblatt G. serie G. G. G. G. G. rufomarginatus G. J. Lewis G. pretoriensis Kunt s Vern hunctfotius Goldblatt kamiesbergensis G. J. Lewis marlothii G. J. Lewis mostertiae L. Bolus vernus G. J. Lewis Gladiolus sect. Hebea series Involutus G. G. involutus D. Delaroche loteniensis Hilliard series Permeabilis G 7. inandensis Baker G. E permeabilis D. Delaroche ul joa eed Coldblar wilsonii (Baker) Goldblatt & J. C. Manning series Deserticola G. arcuatus Klatt . deserticola Goldblatt . salteri L. Bolus . scullyi Baker z venustus Klatt >. viridiflorus G. J. Lewis Mpumalanga, Dullstroom(*) Mpumalanga, Stoffberg(*) KwaZulu-Natal, Underberg(*) Mpumalanga, Grasko KwaZulu-Natal, Underberg(*) Mpumalanga, Dullstroom(*) not studied Western Саре, Cape Peninsula Western Саре, Glencairn Near Riviersonderend Western Cape, Sir Lowry's Pass Mpumalanga, near Warmbad(*) Mpumalanga, Pretoria(*) not studied not studied Western Cape. Kamiesberg(*) Western Cape, Ganaga Pass Northern Cape, Bokkeveld Mts. Mpumalanga, Lydenburg not studied Western Cape, near Heidelberg KwaZulu-Natal. Loteni Valley(*) KwaZulu-Natal, Nchanga(*) Western Cape, Barrydale Gauteng, Bronkhorstspruit Western Cape, Heidelberg not studied not studied Western Cape, Vredendal Northern Cape, Richtersveld(*) Northern Cape, E of Springbok(*) Northern Cape, Calvinia Western Cape, S of Clanwilliam Northern Cape, Anenous flats(*) Goldblatt & Manning 10075 Goldblatt & Manning 10094 Goldblatt & Manning 10061 Goldblatt & Manning 10085 Goldblatt & Manning 10062A Goldblatt & Manning 10125 no voucher Goldblatt & Manning s.n. Goldblatt & Manning s.n. Goldblatt 2035 Goldblatt & Manning 10095 Goldblatt & Manning 9799 Goldblatt & Manning 9768 Goldblatt & Manning 10360 Goldblatt & Manning 10107 Manning 2115 Manning s.n. Goldblatt & Manning 10143 Goldblatt & Manning 10057 Goldblatt & Manning 10015 Goldblatt & Manning 9904 Goldblatt & Manning 10024 Goldblatt & Manning 9904, 10006 Goldblatt & Manning 9950 Goldblatt & Manning 9649 Goldblatt & Manning 9964 Steyn 525 Goldblatt & Manning 9288 496 Annals of the Missouri Botanical Garden Table 1. Continued. Study site Species Province, Locality Voucher series Hebea G. alatus L. G. ceresianus L. Bolus G. equitans Baker G. meliusculus (G. J. Lewis) Goldblatt & J. C. Manning G. orchidiflorus Andr. G. pulcherrimus (G. J. Lewis) Goldblatt & J. C. Manning G. speciosus (С. J. Lewis) Goldblatt & J. C. G. uysiae G. J. Lewis Manning G. virescens Thunb. С. watermeyeri L. Bolus Gladiolus sect. Homoglossum series Carinatus G. atropictus Goldblatt & J. nning G. жүннен Айоп G. comptonii G. J. Lewis G. s Goldblatt & J. €. Manning G. айа (D. Delaroche . T. Barnard G. 1.2. G. J. “емі series 4. б. ех x. J. Le G. hd С. “ Rt L. Bolus series Brevitubus G. vaginatus G. brevitubus G. J. Lewis series Gracilis G. bullatus G. J. Lewis G. blommesteinii L. Bolus G. caeruleus Goldblatt & J. C. Manning G. gracilis Jacq. G. 1. Goldblatt & G. qi Eckl. x. J. Lewis G. Gana Bolus G. ornatus Klatt G. rogersii Baker G. taubertianus Schltr. ) Western Cape, near Porterville Western Cape, Cape Town Western Cape, Klein Roggeveld Northern Cape, Spektakel Mts. Western Cape, Waylands, Darling Northern Cape, Spektakel Mts. Western Cape, Sandberg(*) Western Cape, Botterkloof Northern Cape, Bokkeveld Plateau Western Cape, Swellendam(*) Western Cape, Cold Bokkeveld Northern Cape, Bokkeveld Mts. Western Cape, Die Galg(*) Western Cape, Pakhuis Mts. Western Cape, Aurora Western Cape, Darling(*) Western Cape, Heerenlogement Mt.(* Western Cape, near Blouberg Strand Western Cape, Rondeberg(*) estern Cape, Edgemead — Western Cape, Pakhuis Pass(* Western Cape, Bain's Kloof Albertinia(*) Western Cape, Drayton(*) Western Cape, Western Cape, Vogelgat Western Cape, near Elim(*) Western Cape, Sir Lowry's Pass Western Cape, Langebaanweg(*) Western Cape, Aurora Western Саре, Darling Western Cape, Worcester Western Cape, near Glencairn Western Cape, Hermanus(*) Western Саре, Glencairn not studied Western Cape, Arniston(*) not studied — Goldblatt & Manning 9336 Goldblatt s.n. no voucher Goldblatt & Manning 10308A Goldblatt & Manning 10003 Goldblatt & Manning 10386A Goldblatt & Manning s.n. Goldblatt & Manning 10328 Goldblatt & Manning 10546 Goldblatt & Manning 10294A Goldblatt & Manning 9750A Goldblatt s.n. no voucher Lewis 5890 Manning 2010 Goldblatt & Manning 9922 Goldblatt s.n. no voucher rant 4649 Manning 2006 Manning s.n. Goldblatt & Manning 10542 Goldblatt & Manning 11024 Goldblatt & Manning 9921 Manning 1079 Manning 2018 Manning 1048 Manning s.n. Goldblatt & Manning 9916A Goldblatt 2505 Goldblatt & Manning 9928 Goldblatt & Manning 10227 Goldblatt s.n. no voucher Manning 2010 Goldblatt 10603A Goldblatt & Manning 9875 Goldblatt 10921 Manning 2034 Volume 85, Number 3 Goldblatt et al. Bee-pollinated Gladiolus Species 497 Table 1. Continued. Species Study site Province, Locality Voucher series Teretifolius G. delpierrei Goldblatt inflatus Thunb. patersoniae F. Bolus not studied Western Western Саре, Die Galg pritzelii Diels not studied subcaeruleus G. J. Lewis sufflavus (G. J. Lewis) trichonemifolius Jacq. Goldblatt & J. C. Manning series Tristis G. s DARAN symonsii F. Bolus not studied Western Cape, Jonaskop rn Cape, Worcester Western Cape, near Riviersonderend Northern Cape, Bokkeveld Mts Western Cape, Camphill road Goldblatt 10791 Goldblatt & Manning 10504 Manning s.n. Goldblatt & Manning 10186 Goldblatt & Manning 9783 Goldblatt 2175 tubes and are known or assumed to be pollinated by other organisms, including long-tongued flies, birds, a satyrid butterfly, or moths (Goldblatt & Manning, 1998), and will be the subject of separate papers. Plant vouchers (Table 1) are deposited at the Missouri Botanical Garden Herbarium, St. Lou- is (MO), and/or the Compton Herbarium, Cape Town (NBC). Seasonality, floral phenology, floral longevity, and reproductive success. Data on seasonality are taken from Goldblatt and Manning (1998) and are sum- marized below under Results. Observations of the mode and timing of anthesis (opening of individual buds), anther dehiscence, expansion of stigmatic lobes, and withering of the perianth were made on plants in cultivation or in the laboratory on cut stems placed in water. Reproductive success in terms of number of capsules developed per spike compared to the number of flowers produced was determined for one species, Gladiolus venustus. In- dividual fruiting spikes were sampled at 1-m inter- vals along a transect. Incompletely developed or malformed capsules were not scored. Fragrance. Floral scent was noted with the hu- man nose in the field and in cultivated plants. Pres- ence of scents too weak to be discerned in the open air was recorded after individual flowers were pick- ed and placed in clean, lidded glass jars and stored in a warm place. The contents of each jar were sniffed after a minimum of 60 minutes (Buchmann, 1983). Site of scent production was determined by immersing flowers in aqueous neutral red stain. Scent chemistry was examined by R. Kaiser, С1- vaudan-Roure Research Ltd., Switzerland, by gas chromatography using a DB-Wax Capillary column (Kaiser, 1993). Scents were captured in capsules though which air was drawn by a vacuum pump from a small, lidded chamber containing open flow- ers. Nectar-volume measurements were taken primarily from unbagged flowers in the field, reflecting both rates of secretion and deple- tion. In addition, nectar samples were removed from Gladiolus appendiculatus, G. brevifolius, G. crassifolius, and G. involutus using potted plants or cut stems placed in water in the laboratory for a minimum of 24 hours. Evidence that we have ac- cumulated on some Gladiolus species and other Ir- idaceae indicates that using flowers from cut stems Nectar analysis. does not affect nectar quality for the first one to two days after harvesting. To collect nectar whole flow- ers were picked and nectar was withdrawn from the base of the perianth tube with 3-4 capillary tubes after separating the ovary from the perianth base. The percentage of sucrose equivalents in fresh nec- tar was measured in the field or laboratory on a Bellingham & Stanley hand-held refractometer (0— 50%) from five or more individuals per population, unless fewer individuals were available. Additional nectar samples were dried on Whatman's filter pa- per no. 1 and sent to B.-E. van Wyk, Rand Afri- kaans University, Johannesburg, for HTLC sugar analysis. observation and pollen-load analys- es. Observations of insects on Gladiolus flowers of 45 species involved 4—20 hours per species at over 50 sites, and included the mode of foraging and whether insects contacted anthers and stigmas while foraging. Insects observed probing the floral tube or brushing the anthers or stigmas were cap- tured and killed in a jar using ethyl acetate fumes. Pollen was removed from insects after specimens were pinned. To prevent contamination of insect bodies with pollen on other insects in the same jar, Insect 498 Annals of the Missouri Botanical Garden the bodies of insect specimens were isolated by wrapping them in tissue prior to pinning. Capturing a single bee at any site appeared to reduce the local bee population significantly. We therefore captured only enough bees as necessary to obtain individuals for identification and pollen-load analysis. Removal of pollen from insect bodies involved either gentle scraping with a dissecting needle or gentle washing in drops of 9596 ethanol. The residue from needle probes or washes was collected on glass slides and mounted in 1-2 drops of Calberla's fluid (Ogden et al, 1974). Pollen grains were identified by com- parison with reference pollen-grain preparations made from plants flowering at study sites. Pollen of a species was recorded as present if more than 10 grains were counted. Gladiolus pollen grains are recognized by their large size, monosulcate aper- ture with prominent 2-banded operculum, and per- forate-scabrate exine. Body length and length of mouthparts of insects were measured from pinned specimens. Insect body length was measured from the base of the labrum to the tip of the abdomen. Mouthpart length was measured from the base of the labrum to the tip of the proboscis (flies) or the extended tongue (bees). Insect specimens were identified by R. W. Brooks (Apidae), C. D. Michener (Halicti- dae), both of Kansas State University, and V. White- head (Melittidae), South African Museum, Cape Town. Voucher specimens are deposited at the South African Museum, or the Snow Entomological Museum, Lawrence, Kansas. The bee classification used here is that of Roig-Alsina and Michener ) (1993 RESULTS Seasonality, floral phenology, апа longevi- ty. Species of Gladiolus are seasonal, corm-bear- ing geophytes of small to moderate size, typically 20—45 cm high, but in some species up to 1.5 m Individuals produce a single, simple or few- branched flowering stem annually, and flowering is closely synchronized in a population. Inflorescenc- es are spikes with the flowers usually secund, rarely distichous (only sect. Ophiolyza ser. Oppositiflorus) or spiral (species with actinomorphic flowers) (Figs. l, 2). Flowering in the Gladiolus species of southern Africa is correlated with their geographic ranges in one of two separate rainfall zones (Goldblatt & Manning, 1998). Species of the winter-rainfall zone mostly flower in spring (August to October), toward the end of the wet season, whereas most species of the summer-rainfall zone flower in early to late Figure 1, stigma maturation in Gladiolus alatus. The upper two flow- ers opened last and have the style reaching to the base of the anthers with the style branches Inflorescence form, flower shape, and style- 'es. Arrows indicate position of style branches. Scale approx. X24 Volume 85, Number 3 Goldblatt et al. 499 1998 Bee-pollinated Gladiolus Species re 2. Inflorescences and flowers of southern African Gladiolus, with half flowers for some species. —A. б. papilio (sect. Densiflorus). —B. G. rufomarginatus (sect. Heterocolon). —C. G. gracilis (sect. Homoglossum). —D. G. uysiae (sect. Hebea). —E. С. stellatus (sect. Hebea). —F. С. quadrangulus (sect. Homoglossum). Scale approx. Х%. 500 Annals of the Missouri Botanical Garden summer (December to April), again coinciding with the wet season (Table 2). In both cases this coin- cides with the period of optimal plant growth, dur- ing or soon after the main rainy period. Flowering near the end of the dry season is, how- ever, characteristic of several species of both cli- mate zones (Goldblatt & Manning, 1998). In the winter-rainfall zone, several species (Table 2) flower in the late summer and autumn (mainly March to May), and these species typically have reduced leaves on the flowering stem (foliage leaves for the accumulation of food reserves in the corm may be produced later, on separate shoots). In the summer- rainfall zone, flowering in the dry winter is largely precluded by cold temperatures, but several spe- cies flower in the spring (September to November), before or shortly after the first rains have fallen. These species also have reduced leaves on the flow- ering stem, but typically do not produce separate foliage leaves; instead, the leaves and stems remain green for months after flowering. The pattern of flower buds opening on an inflo- rescence is acropetal. In all Gladiolus species, a mature bud expands just before mid morning and the open flower typically lasts three or four days, and in some species up to five days. Flowers open one to two days apart, hence there are often two or three flowers open at any time on an inflorescence (Figs. 1, 2). The flowers of most species partly close at sunset, loosely enclosing the exserted anthers and stigmas. More precise flowering times occur in two species, Gladiolus quadrangulus and G. stel- latus. Flowers of G. quadrangulus close completely at 1500 hours and open at 1000, while those of G. stellatus close at 1230 and open at 0700--0800. Flowers of all species studied show mechanical protandry, a condition first described for southern African species of Gladiolus by Scott Elliot (1891). The anthers dehisce longitudinally within one to four hours after the flowers open. This depends to some extent on ambient temperature and humidity, and anthers dehisce later in wet, cool conditions. At this time the three style branches remain loosely clasped to each other and lie against the adaxial surfaces of the dehisced anthers, with the stylar lobes, the distal adaxial surfaces of which comprise the stigmas, folded together. The male phase of flowers of most species lasts at least two days, and a maximum of three days, as flowers wither after three or four (occasionally five) days. The male phase is then followed by a female phase that lasts either one or two days, during which the stylar lobes diverge and arch outward away from the dehisced anthers. At the same time, stigmatic surfaces at the distal ends of the stylar lobes unfold, and the visibly papillate and moist stigmatic areas are exposed and available for pollen deposition. Thus, species typically pass two days in an exclusively male phase, during which time pol- len is usually removed by insects. In undisturbed field sites it could readily be seen that no pollen remained in flowers by the time the stigma lobes unfolded, leaving the flowers in an exclusively fe- male phase for the last one or two days of anthesis. An exception to this pattern, G. trichonemifolius, has flowers that last just two days, the first in the male phase with dehisced anthers displaying pol- len, and the second in the female phase with the stigma lobes exposed (Fig. 1). As observed by Scott Elliot (1891), mechanical self-pollination cannot readily occur because of the physical separation of the pollen and stigmatic surfaces, even if pollen remains in the anthers by the time the stigmatic surfaces are exposed. We have no data on the bio- chemistry of stigmatic receptivity. А significant exception to the general pattern is Gladiolus gueinzii, a coastal strand species. The style divides between the base and middle of the anthers, and the stylar lobes remain entangled in the dehisced anthers. It has been found to be au- togamous in cultivation, with 80% of flowers setting capsules in the absence of hand manipulation or insect activity (Goldblatt & Manning, 1998) Population density. Our impression is that popu- lation density falls into two categories. Some Glad- iolus species typically occur in fairly dense popu- lations consisting of flowering individuals standing about 20-30 cm apart. These include species of specialized habitats, e.g., Gladiolus papilio and G. trichonemifolius, which favor marshy sites, and G. gueinzii, which grows on sandy beaches close to the high-tide mark. The majority of species, however, usually form populations in which flowering indi- viduals stand about 1-2 m apart (ignoring obviously disturbed habitats where density may be quite high due to recolonization). We have encountered sites with less than 10 individuals in flower (e.g., С. зиј- flavus, G. watermeyeri), and others with many hun- dreds of plants in flower (G. alatus, G. meliusculus), the latter usually in habitats disturbed by wild fires, partial clearing, or heavy grazing. Floral presentation and attractants. There are three more or less distinct floral forms in the Glad- iolus species visited by bees. The most common is the "gullet" flower, with a zygomorphic and bila- biate perianth and arcuate, unilateral stamens (Figs. ZA-C, 3A—C), present in 85 species of south- ern African Gladiolus. The dorsal (posterior) tepal (“tepal” is used throughout this paper for lobe of Volume 85, Number 3 1998 Goldblatt et al. Bee-pollinated Gladiolus Species the perianth) is always slightly larger than the oth- ers, and grades in orientation from arched to hood- ed and spooned. In flowers with a hooded and spooned dorsal tepal, the anthers are concealed and lie horizontally, close to or against the inner surface of the dorsal tepal. The upper lateral tepals are directed forward proximally and flare outward in the distal half. The lower three tepals are typically narrower than the upper three and are oriented hor- izontally or gently inclined. The distal parts of the lower tepals together form what resembles a landing platform, and have contrasting markings of various shapes and colors (nectar guides). The floral tube is typically obliquely funnel-shaped and 12-20 mm long (Figs. 2A-C, 3A-C). The proximal portion is hollow, suberect, and cylindric, mostly 5-15 mm long (Table 2) and 1-1.5 mm in diameter, thus too narrow to permit entry for the bodies of most in- sects. The distal part of the tube is flared and more or less horizontally oriented. Collectively the bases of the tepals and the distal part of the tube form a tapering throat leading to the narrow part of the tube. The throat readily accommodates the head and thorax of a large bee. In the sense of Faegri and van der Pijl (1979) these are gullet flowers. An unusual feature of the flowers of members of series Hebea (sect. Hebea) is ridges of unpigmented, iri- descent papillae in the throat. Pigmentation in these zygomorphic and bilabiate flowers is extremely variable (Table 2), particularly in species of the winter-rainfall zone. Flowers of species in section Homoglossum are typically shades of blue to mauve or purple, sometimes pink, or occasionally yellow. The lower tepals often have strongly contrasting nectar guides. Gladiolus suffla- vus is exceptional in having greenish flowers, and G. griseus has a gray-green perianth with yellow nectar guides. In some species of section Hebea, flowers are brilliant scarlet to orange or, in a few other species (including G. arcuatus, G. scullyi, and G. watermeyeri), dull greenish, brown, or muddy purple. Flowers with the latter range of colors tend to merge with the background colors of the sur- rounding vegetation and terrain and seem camou- flaged. Notably, these flowers always have a partic- ularly strong fragrance. In the summer-rainfall zone flowers are often pale pink, lilac, or mauve, and only occasionally brightly colored. In these species the tepals are poorly or not at all marked. Gullet flowers may be modified by atypical de- velopment of the tube or of the sexual organs in relation to the floral tube. First, Gladiolus aureus, of the winter-rainfall zone, has flowers with sube- qual tepals lacking contrasting markings. The lower part of the perianth is extremely narrow and tightly sheaths the style. In the summer-rainfall zone, С. pubigerus and G. parvulus also have subequal te- pals without markings and a short perianth tube. Perhaps better termed pseudo-gullet flowers, they produce only traces of nectar in quantities too small to measure. Second, the bases of the anthers of G. appendiculatus (sect. Densiflorus ser. Calcaratus) have elongated to form hard, green, sterile append- ages. These appendages extend downward across the floral throat, effectively closing off entry to the proximal part of the floral tube. As a large-bodied insect pushes its head and thorax into the flower, the appendages are elevated and the anthers pulled downward, swabbing the thorax with pollen. A second flower type conforms more closely to the definition of “flag” rather than to that of “gullet” sensu Faegri and van der Pijl (1979) and has been identified in eight species. The flag flowers of Glad- iolus differ from gullet flowers mainly in having an erect, enlarged dorsal tepal and widely flaring up- per lateral tepals (Figs. 2D, 3D). Like gullet flow- ers, they also have an obliquely funnel-shaped tube. Flag flowers are most common in series Hebea (sect. Hebea; Table 2), and in these species the sta- mens and style are strongly arched so that the an- ther apices are tilted downward and the style lobes face the ground. Large-bodied bees visiting these flowers typically contact the anthers or stigmas with the dorsal surface of their thorax or abdomen as they exit, but smaller bees do not contact the an- thers at all. The ridges of iridescent papillae in the throats of these flowers are especially prominent. Flag flowers often have bright coloring, usually pink, orange-red, or bright purple, and the lower tepals are marked or streaked with contrasting col- ors (Fig. 4). However, a few species of series Hebea with flag flowers, including G. ceresianus and G. uysiae, have dull-colored, purple-brown perianths always accompanied by particularly strong odors. Gladiolus meliusculus is remarkable in having un- usually conspicuous, dark purple markings, which may be “beetle marks” (Goldblatt et al., 1998). The third flower type is the rarest and has been identified in only five species, 596 (Table 2) of the short-tubed species. This is the rotate flower, in which the perianth is held more or less upright and the tepals spread horizontally, sometimes forming a shallow cup (Figs. 2E, F, 3E). This flower type is restricted to G. brevitubus, G. deserticola, G. quad- rangulus, and G. stellatus of the winter-rainfall zone, and G. parvulus in the southern African sum- mer-rainfall zone. Gladiolus brevitubus and G. de- serticola have a more or less actinomorphic peri- anth, but unilateral stamens. Visible contrasting markings (nectar guides) are lacking in G. quad- 502 Annals of the Missouri Botanical Garden Table 2. Floral and phenological data for southern African Gladiolus species with flowers adapted for pollination bees. Species are arranged according to the sectional classification of Goldblatt and Manning (1998). Note that only the length of the proximal, narrow and cylindric part of the perianth tube is listed to facilitate direct comparison with the mouth parts of visiting insects. Measurements of tube length are the range for the species. Scent is scored according to the typical condition of a species: ++ = strong scent; + = weak scent, — = no scent. Flowering times are from Goldblatt and Manning (1998), and an asterisk (*) indicates species flowering out of phase with the prevailing flowering peak for the rainfall zone. Flower Main Proximal Main flowering Rainfall Species Shape Color tube (mm) Scent reward time zone Gladiolus sect. Densiflorus appendiculatus gullet pale pink ca. 7 - nectar Apr—May summer G. brachyphyllus flat purple ca. 10 - nectar Nov.—Dec. | summer G. crassifolius gullet pink, mauve, 5-8 - nectar — Feb.-Mar. summer о G. densiflorus gullet pink, mauve, са. 6 - nectar — Feb.-Mar. summer or orange G. exiguus gullet pin ca. 7 = nectar Feb.—Mar. summer G. ferrugineus gullet white-pinkish 6-15 = nectar — Jan.-Apr summer G. hollandii gullet pink 9-12 = nectar ar—Apr. summer А ochroleucus gullet pink 7-8 - nectar Dec.—Feb. summer G. paludosus gullet ink ca. б = nectar ov.—Dec. summer G. papilio gullet pink, gray, 10-12 ES nectar Dec.—Feb. | summer or cream G. pavonia flag pink ca. 8 = nectar — Dec.-Jan. summer G. serpenticola gullet pink ca. 5.5 = nectar Feb.—Mar. summer Gladiolus sect. Ophiolyza G. dolomiticus gullet ink 9-12 = nectar — Feb.-Mar. summer G. ecklonii gullet pink, red, 8-10 = nectar — Dec.-Mar. summer or mauve G. еШош gullet pink-mauve ca. 8 = nectar Nov.—Dec. summer М 1 gullet ink ca. 8 = nectar Dec.—Jan. summer G. sericeovillosus gullet pink, cream, 5-8 = nectar Dec.—Feb. summer or purple G. vinosomaculatus gullet cream and 7-8 E nectar Dec.—Jan. | summer purple Gladiolus sect. Blandus G. crispulatus gullet pink 12-16 = nectar Dec Jan. winter G. grandiflorus gullet cream-pink 13-20(-40) — nectar — Sep.-Oct. winter G. gueinzii star mauve 10-12 = nectar Oct.—Nov. summer/winter G. oreocharis gullet pink 10-15(-20) — nectar Dec.-Jan. | winter G. phoenix gullet pink ca. 10 = nectar — Oct.-Nov. winter G. rudis gullet pin 8-10 = nectar — Sep.-Oct. winter Gladiolus sect. Linearifolius G. aureus pseudo- yellow 10-15 = pollen Aug.—Sep. winter gullet G. brevifolius star pink or occ. 6–7 =/+ nectar = Mar-Apr міпіег* стеат G. hirsutus gullet pink-mauve 8-12 - nectar — July-Sep. | winter G. malvinus gullet mauve ca. 5 = nectar Oct.—Nov. | summer* G. pardalinus gullet yellow or ca. 4 = nectar Oct.-Nov. summer* mauve G. parvulus star pink ca. 3 = ?pollen Oct-Dec. — summer* G. pubigerus gullet green 7-8 -/- ?nectar Oct-Nov. summer* G. woodii gullet purple or 4—6 = nectar Oct.-Nov. summer* yellow Volume 85, Number 3 1998 Goldblatt Я Gladiolus Species 503 Table 2. Continued. purple or pink Flower : ye А Proximal M flowering Rainfall Species Shape Color tube (mm) Scent reward ime zone Gladiolus sect. Heterocolon С. es gullet blue ca. 5 + nectar — Sep.-Oct. winter G. marlot gullet blue ca. 4 = nectar — Oct.-Nov. winter G. mostertiae gullet pink ca. 5 + nectar — Nov.-Dec. winter С. oatesii gullet mauve ca. 4 € nectar Oct.-Nov. ѕиттег* G. pretoriensis gullet pink ca. 6 = nectar — Dec.-Jan. summer G. rubellus gullet orange ca. 7 = nectar Nov.—Dec. summer G. rufomarginatus gullet pink ca. 4 = nectar Маг.-Арг ummer G. vernus gullet pink 4—6 = nectar — Aug.-Se summer* Gladiolus sect. Hebea G. alatus flat scarlet 7-9 + nectar — Aug.-Sep. winter G. arcuatus gullet brown 10-15 ++ nectar June-July winter ceresianus ag brown-purple 8-10 ++ nectar — Sep.-Oct winter с. deserticola star blue ca. 9.5 - nectar — Aug.-Sep. winter G. equitans gullet scarlet 9-11 "pep nectar Aug.—S winter G. inandensis gullet white 5-8 = nectar — July-Sep summer G. involutus gullet white-pink 16-18 - nectar — Sep.-Oct. winter G. loteniensis gullet mauve ca. 5 = nectar Nov.—Dec. summer G. meliusculus flag k 5-6 + nectar — Sep.-Oct. winter G. orchidiflorus gullet green or 7-10 FF nectar Aug.—Sep. winter brown G. permeabilis subsp. edulis gullet cream 7-10 ++ nectar — Sep.-Oct. | summer subsp. permeabilis gullet au 1-10 ++ nectar — Sep.-Oct. winter G. pulcherrimus flag scarlet 5-8 T nectar бер. winter G. eri gullet pink 15-18 ++ пес!аг ер. winter G. scullyi gullet brown-mauve 8-9 ++ nectar Aug.—Sep. winter G. speciosus gullet scarlet 5-7 + nectar Sep.—Oct winter G. stellatus a white-mauve 5-1 ++ pollen _ Sep.-Oct. winter С. uitenhagensis gullet mauve 15-20 ? ?nectar ct. winter G. uysiae flag brown-purple 5-1 ++ nectar — Sep.-Oct. winter G. venustus gullet mauve or 8—11 + nectar — Aug.-Sep. winter pin G. viresce flag yellow-brown 6-9 ++ nectar Sep.-Oct winter G. viridiflorus gullet green 9-12 ++ nectar May-July winter . watermeyeri gullet green ca. 10 ++ nectar — Aug.-Sep. winter G. wilsonii gullet white 3—6 = nectar — Oct.-Nov. winter Gladiolus sect. Homoglossum G. atropictus gullet blue ca. 8 ++ nectar Sep.—Oct. winter С. blommesteinii gullet pink 7-13 + nectar Sep.-Oct. winter G. brevitubus star range ca. 2 + pollen Oct.—Dec. winter G. bullatus gullet blue 5-8 + nectar "94 -Oct. winter С. caeruleus gullet blue 6-8 ++ пес!аг winter G. carinatus gullet blue or 5-8 ++ пес!аг E. -Sep. winter yellow G. comptonii gullet yellow 7-9 = nectar — July-Aug. winter G. delpierrei gullet yellow ca. 4 + nectar Dec winter* G. exilis gullet lue 7-10 oF nectar May-June winter* G. gracilis gullet ue 8-10 ++ nectar Aug.—Sep. winter G. griseus llet green-gray 3-5 = nectar June-July winter G. inflatus gullet 5-12 ~ nectar — Sep.-Oct. winter 504 Annals of the Missouri Botanical Garden Table 2. Continued. Flower Main Proximal a flowering Rainfall Species Shape Color tube (тт) Scent reward time zone G. inflexus gullet blue 6-8 ++ nectar — July-Aug. winter G. jonquilliodorus gullet yellow са. ! ++ nectar — Dec.-Feb. — winter* G. martleyi gullet pink ca. 7.5 ++ nectar — Mar-Apr. winter* G. mutabilis gullet blue or 8-12 ++ nectar — July-Aug. winter brown G. ornatus gullet pink ca. 12 = nectar — Sep.-Oct. winter G. patersoniae gullet blue 5-10 ++ nectar — Aug.-Sep. winter С. pritzelii gullet yellow ca. 6 ++ nectar — Sep.-Oct. winter G. quadrangulus star white or 4–6% - pollen — Sep.-Oct. winter purple ». TOgersii gullet blue 5-8 +/= nectar — Sep.-Oct. winter G. subcaeruleus gullet blue 8-10 =/%+ песі Mar-Apr. winter* G. sufflavus gullet green 7-8 ++ nectar — Aug.-Sep. winter С. зутопзи gullet pink 4—5 ? ? Dec.-Jan. | summer G. taubertianus gullet blue 5-7 + nectar — Aug.-Sep. winter G. trichonemifolius gullet yellow 7-10 ++ nectar Aug.—Sep. winter G. vaginatus gullet blue 3.5-12 ++ nectar — Apr-May winter* G. violaceolineatus gullet blue 7-9 ++ nectar — Sep.-Oct. winter rangulus and G. stellatus, the actinomorphic flowers of which are arranged spirally on the spikes. These two species also have prominently displayed an- thers and short perianth tubes. In the case of G. quadrangulus, the tube is extremely narrow and completely occluded by the tightly sheathed style. Flowers of all species with a rotate perianth pro- duce reduced amounts of nectar and are often odor- less, but flowers of G. stellatus are strongly scented. Pigmentation in species with rotate flowers is usu- ally pale pink, whitish, or lilac, but some popula- tions of G. quadrangulus have mauve flowers, and G. deserticola has dark blue flowers with white markings. Gladiolus brevitubus (sect. Homoglossum) has a bright orange perianth with yellow markings on the lower tepals. The flowers are lightly sweet- scented and have a filiform perianth tube, 2,5-4 mm long, and too narrow to permit entry to an in- sect's tongue. Discernible scent production in Gladiolus Scent. Figure 3. Front view and vertical section (or side view) of flowers of selected species of Gladiolus pollinated by bees, showing details of representative gullet (A-C), flag (D), and rotate (E) flowers. orus). —B. G. woodii (sect. Linearifolius). brevitubus (sect. Homoglossum). Scale approx. full size. —A. 6. crassifolius (sect. Densi- —C. G. permeabilis (sect. Hebea). —D. G. uysiae (sect. Hebea). — Volume 85, Number 3 Goldblatt et al. 1998 Bee-pollinated Gladiolus Species Figure 4. Floral foraging in southern African Gladiolus. —A. Amegilla obscuriceps оп G. brevifolius. —B. Amegilla spilostoma on G. rufomarginatus. —C. Anthophora diversipes on б. pom —D. Атерша она on 6. crassifolius. Arrows indicate position of anther 506 Annals of the Missouri Botanical Garden is correlated with phylogeny and geography (Table 2). Fragrance is released during the day and is sup- pressed at night. To the human nose, most species of the southern African winter-rainfall zone in sec- tions Hebea and Homoglossum have strong, sweet, but variable fragrances reminiscent of commercial cultivars of Rosa centifolia, Viola odorata, or Free- sia spp. The flowers of G. alatus (sect. Hebea) and a few other species in sections Hebea and Homo- glossum produce a distinctive, somewhat acrid odor, reminiscent of the terrestrial orchid genus Pterygodium. Within section Homoglossum, isolat- ed populations of G. carinatus, G. gracilis, and G. trichonemifolius were found to have flowers lacking discernible scent. Species of section Blandus pro- duce no apparent odor, and those of section Li- nearifolius in the winter-rainfall zone may have sweet, rose-like scents or no apparent odor. In contrast, most Gladiolus species of the sum- mer-rainfall zone, including those species belong- ing to sections Hebea and Homoglossum, produce no discernible scent. Gladiolus pubigerus (sect. Li- nearifolius) produces a scent similar to flowers of tuberose (Polianthes) in the southern populations in KwaZulu-Natal. We have recorded unscented flow- ers elsewhere in its range. Neutral red tests indicate a correlation between scent production in Gladiolus flowers and epider- mal vestiture of the tepals. Within 18 hours of soak- ing in aqueous neutral red, papillae on the lower tepals and ventral throat of Gladiolus flowers stain red. These papillate zones are restricted to regions of pale pigmentation. Flowers of species that lack these papillate zones appear unscente iochemical analysis of floral em in se- lected Gladiolus species in sections Hebea and Homoglossum identifies 16 major constituents (Ta- e 3). However, the scents of most of these species contain only four to seven major constituents. Glad- iolus virescens is unusual in producing only two scent constituents. Fragrances dominated by gera- niol, citronellol, and nerol, or their acetates, have a rose-like odor. Fragrances dominated by B-ionone and dihydro-B-ionone smell like a combination of violet and freesia (Table 3). The orchid-like odor of G. alatus appears to be produced primarily by lin- alool, balanced with a combination of four lesser constituents. Nectar. Nectar glands are septal in Gladiolus spe- cies, as in the entire subfamily Ixioideae (Goldblatt, , 1991; Manning & Goldblatt, unpublished). Nectar] is secreted from pores at the top of the ovary directly into the base of the perianth tube and is retained in the proximal part of the tube. The Scent characteristics of selected species of southern African Gladiolus with flowers adapted for bee pollination. Table 3. Scent composition (46 constituents above 1%) Dihy- Citro- Gera- B- Benzyl Neroli- Limo- Oci- Neryl niol Citro- nellyl acetate Gera- Scent description (-epixide) acetate ionone pinene/ lene/ ( | a-farn- acetate acetate Nerol alool топопе nene тепе esine a-) dol nellol niol Species Gladiolus sect. Hebea 12.5 floral orchid G. alatus rosy sweet 39.5 G. orchidiflorus G. scullyi violet-freesia G. virescens Gladiolus sect. Homoglossum 4.5 violet—freesia G. carinatus 30.5 rosy G. jonquilliodorus rosy G. patersoniae Volume 85, Number 3 Goldblatt e Bee- pomis a Gladiolus Species 507 Table 4. Nectar characteristics of selected species of southern African Gladiolus with flowers adapted for bee ек Sample number, from different individuals, (n) for nectar volume and concentration (conc.) is the same. Fru == fructose, Glu = glucose, Suc = sucrose Nectar . Range of sugars % Ratio of Volume Conc. 96 sucrose to Species pl (n) (+SD) Fru Glu Suc F + G (n) G. alatus 0.6-1.2 (6) 27.5 (3.6) 4 13-16 80-83 4.41 (2) G. appendiculatus 2.1–3.8 (2) 29.0–34.0 — — — G. arcuatus 2.1-3.1 (5) 43.2 (2.1) — — — — G. aureus quantity too small to measure 6 16 78 3.55 (1) G. blommesteinii 0.8-1.6 (3) 25.0 (3.1) — — — == G. brachyphyllus — — 2 10 88 7.33 (1) G. brevifolius 0.8-1.5 (5) 30.2 (3.2) — — — G. brevitubus no Use nectar m G. carinatus (Darling) 1.0-2.9 (5) 30.0 (0.9) 10 89 8.09 (1) (Aurora) 0.5-0.8 (8) 25.6 (1.5) e 4—9 9-84 41 (2) G. crassifolius 1.8-2.7 (3) 21.7 (1.2) 7-15 15-22 63-78 2.39 (2) G. ferrugineus 2.7–3.5 (3) 29.3 (2.1) 3-11 9-21 9 3.65 (2) G. gracilis (Darling) 0.8-1.8 (10) 36.6 (3.1) 1-6 7-12 82-93 7.51 (4) rora) 0.6-1.1 (10) 26.1 (2.6) 8-12 1-3 87-89 7.33 (2) G. grandiflorus 4.1—6.9 (5) 28.6 (2.7) — — — — G. hirsutus 1.2-1.9 (2) 25.0–27.0 — — — — G. inflatus 1.3-2.5 (6) 30.1 (2.3) = r = — G. inflexus 1.7-4.6 (6) 36.3 (7.5) — — -- С. involutus 3.1-3.8 (2) 9.0 18-21 22-24 57-58 1.35 (2) С. jonquilliodorus 1.6-2.7 (6) 33.7 (2.2) -— — — — G. malvinus 2.6-3.0 (2) 4 10-11 89-90 8.55 (2) G. marlothii 2.1-2.9 (4) 26.0 (2.2) 0—5 17-18 77-83 4.00 (2) G. mostertiae 2.1–3.8 (3) 29.0 (2.0) 0 10-18 82-90 6.14 (2) G. ochroleucus 2.4-4.9 (5) 27.1 (1.3) ? 3-13 87-97 11.50 (2) G. ornatus 1.8 (1) — 7 18 75 4.00 (1) G. patersoniae (Die Galg) 1.3-2.5 (5) 28.0 (2.5) — — — — (Worc este r) 0.9-1.7 (8) 33.4 (2.6) — -- — — G. pavon 1.5 (1) 36 26 66 1.94 (1) G. permeabilis subsp. permeabilis 2.8 (1) 29 — — — — G. pulcherrimus — — 1-8 22 70-71 2.39 (2) G. quadrangulus no measurable nectar produced С. rogersit 2.4—3.9 (2) 24-26 — -- -- — G. scullyi 2.6–3.9 (3) 28.7 (1.5) 9-15 31-36 49-60 1.20 (2) G. speciosus -- 6-8 16 76-78 3.35 (2) G. stellatus 0.4—0.5 (4) 45.8 (2.6) 10-15 20-27 63-70 1.88 (3) A 1.4—1.9 (2) 36.0 0- 17-19 78-81 3.88 (2) с. trichonemifolius 0.5-1.3 (10) a 9 (4.2) — — — — G. uysiae 0.4—0.8 (2) 10-33 18-22 49-68 1.41 (2) б. venustus 0.8-2.9 (10) id (2.1) 5-12 24–30 58–69 1.80 (3) G. NE 2.6—5.5 (2) 28.0 4—22 31-44 34—55 0.78 (3) С. watermey 0.7-1.3 (2) 25-32 10-28 17-39 33-73 0.97 (4) length of the perianth tube varies among the Glad- iolus species studied from 7 to 30 mm long, with the proximal, slender part mostly 4.5-15 mm (Table 2). All but 2 of 39 species examined for this char- acter were observed to secrete nectar. Nectar vol- umes are modest, mostly 1.5-3 pl, and rarely ex- ceed 5 ш (Table 4). Nectars are sucrose-rich to sucrose-dominant, with sugar solute making up 25-35% of the total volume of fluid (Table 4). The nectar of Gladiolus aureus, G. brevitubus, and G. quadrangulus is pro- duced in such small quantities that the volumes could not be measured. Nectar of С. stellatus, which has short-tubed flowers and is visited by short- 508 Annals of the Missouri Botanical Garden tongued andrenid bees, shows no difference as re- gards nectar sugars from species visited only by long-tongued bees, but it does produce nectar of the highest concentration of all Gladiolus species in this study, with a mean of 45% sucrose equiv- alents (Table 4). Nectar recorded here for Gladiolus species is similar in quality to that reported for other Iridaceae pollinated by active, long-tongued insects, including long-tongued flies and sphinx moths (Goldblatt et al., 1995; Manning & Goldblatt, 1995, 1997). Nectar volumes are, however, consid- erably larger in long-tongued fly flowers. Some bird-pollinated species of Gladiolus may also have sucrose-rich nectar, while others have hexose-dom- inant nectar with only trace amounts of sucrose (B.- E. van Wyk, pers. comm.). Pollinators, pollination mechanisms, and pollen- load analysis. Insect visitors collected on short- tubed species of Gladiolus were mostly bees, but in a few cases also flies, or hopliine beetles (Table 5). Bees included Andrena sp. (Andrenidae); Allodape (1 sp.), Amegilla (5 spp.), Anthophora (3 spp.), Apis (1 sp.), Pachymelus (1 sp.), Tetralonia (1 sp.), and Xylocopa (1 sp.) (Apidae); Lasioglossum (2 spp.), Lipotriche (2 spp.), and Patellapis (1 sp.) (Halicti- dae); and Rediviva (1 sp.) (Melittidae). Bees were predominantly female, but some males were also captured. Gullet and flag flowers are pollinated primarily by bees with large bodies and long tongues in the family Apidae (Figs. 4, 5), including Amegilla, An- thophora, and Pachymelus (Tables 5, 6). These bees appeared to be foraging solely for nectar, and no anthophorine bees were found visiting Gladiolus flowers to collect pollen. Nemestrinid and acrocerid flies, Prosoeca, Psilodera, and Stenobasipteron, with moderately long mouthparts (10-12 mm; Table 6), were also captured on flowers of some Gladiolus species visited by bees. They also appeared to pol- linate these species in their foraging for nectar, and, like the captured anthophorine bees, were found to carry Gladiolus pollen. Both bees and flies landed on the lower tepals and inserted their probosces into the floral tube (Fig. 4). As the insect pushed its head into the floral throat, its thorax made dorsal contact with the anthers and/or the stigma lobes. In . orchidiflorus and flag flowers of section Hebea ser. Hebea, the dorsal surface of the thorax and ab- domen contacted the anthers or stigma lobes as an insect exited a flower. The native and commercial strains of the honey bee, Apis mellifera, appear to be significant polli- nators of a few Gladiolus species of the winter-rain- fall zone. Individuals mostly behaved like antho- phorine bees, visiting flowers for nectar, but were also seen actively foraging for pollen. Our observations show that both anthophorine and honey bees display moderate floral constancy. Anthophorine bees consistently visited flowers of several plants of a particular Gladiolus species in succession before shifting their attention to the flowers of other plant species. Where we were able to track individual bees, the same bee was seen to revisit the same series of flowers several times dur- ing a morning. Honey bees exhibited stronger floral constancy, and numerous individuals could be ob- served visiting flowers of different individuals of a Gladiolus species for over an hour before shifting their foraging activity to other species. adiolus species of the southern African winter- rainfall zone receive a greater diversity of bee for- agers than those of the summer-rainfall zone. Sum- mer-rainfall species were visited primarily by Amegilla species (mostly A. aspergina, A. capensis, A. fallax), but also by Xylocopa rufitarsis and two species of long-tongued flies in the family Nemes- trinidae (Table 5). In the winter-rainfall zone, the most common pollinators were Anthophora diversi- pes and A. krugeri, both large bees (Table 5) active in the spring (August to October). Visits to spring- flowering Gladiolus species of the winter-rainfall zone by Pachymelus peringueyi (Apidae), the short- tongued Rediviva aurata (Melittidae), and Xylocopa rufitarsis (Apidae) have also been recorded. Apis mellifera appears to be an important pollinator only early in the season (June to August), when cool ambient temperatures often make it the only bee on the wing. Apis mellifera was the sole pollinator of the winter-flowering G. griseus and early spring- flowering G. aureus, G. quadrangulus, and G. te- nellus. Later in the season, this bee is rarely seen on Gladiolus species. In the summer and autumn, the most common pollinator in the winter-rainfall zone is Amegilla spilostoma. Gladiolus aureus is one of the few species stud- ied with trace amounts of nectar. Although the proximal part of the floral tube is 10-15 mm long, it is extremely narrow, forming a tight sheath around the style, and contains virtually no nectar, only traces of which are present at the mouth of the tube. Apis mellifera was the only insect visitor observed, and individuals were seen collecting pol- len from the anthers. Gladiolus aureus has bright yellow flowers that resemble blooms of yellow-flow- ered, nectariferous Oxalis pes-caprae L., a favorite plant resource of this bee (unpublished observa- tion). Active pollen collection, involving A. melli- fera, was noted in only two other Gladiolus species, G. gracilis and G. quadrangulus. Volume 85, Number 3 1998 Goldblatt et al ; 509 Bee-pollinated Gladiolus Species A striking exception to the pattern of bee polli- nation in short-tubed Gladiolus species is G. brev- ifolius, on which the acrocerid fly, Psilodera valida, was captured while foraging for nectar. This fly is a remarkable mimic of Amegilla obscuriceps and A. spilostoma, the most common insect visitors of G. brevifolius, in appearance, flight, and foraging be- havior (Goldblatt et al., 1997). Another exceptional species, С. meliusculus, is visited by both female andrenid bees and hopliine beetles (Goldblatt et al., 1998). The exceptionally large landing platform and unusually dark tepal markings are often hall- marks of hopliine beetle pollination, а common strategy in the southern African winter-rainfall zone. Although common in other genera of Irida- ceae, including /xia, Sparaxis, and Tritonia (Gold- blatt et al., 1998), hopliine beetle pollination is not known in other species of Gladiolus. Observations and bee collections on Gladiolus species with rotate flowers (G. brevitubus, G. quad- rangulus, and G. stellatus) indicated pollination by small-bodied Patelapis and Lasioglossum species (Halictidae), Apis mellifera (Apidae), and a large Andrena sp. (Andrenidae) (Table 5). Apis and the Andrena sp. landed directly on the anthers, while the halictids landed on the tepals and crawled onto the anthers. While both the Andrena sp. and the halictid bees were observed visiting the bases of the flowers for nectar, these bees spent most of their time during visits collecting pollen. Only Apis and the Andrena sp. were seen to contact the stigma lobes while foraging. The sweetly scented flowers of G. stellatus are unusual in opening at sunrise and closing completely soon after midday. Gladiolus stellatus would appear to optimize the number of insect visits by partitioning the diurnal timing of flowering with the seasonally coblooming Moraea polyanthos Thunb. The flowers of M. polyanthos open shortly after those of G. stellatus close and fade in the late afternoon. Both plants receive visits by the same bee species. The morphology of the bilabiate Gladiolus flower optimizes pollinator effectiveness because the in- sect is forced to push its upper body into the distal part of the floral tube as it inserts its mouthparts down the narrow proximal tube in search of nectar. The distal tube conforms closely to the shape and size of the pollinator's body, and the anthers, with their load of pollen, come into close contact with the dorsal surface of the thorax of the insect. Pollen deposition is thus passive, and confined to the dor- sal parts of the head and thorax. As flowers of Glad- iolus species are primarily sources of nectar, it is no surprise that both male and female anthophorine bees visit them, although female bees are recorded more often. In the case of flowers that offer pollen, only female bees were captured. The majority of bees captured on Gladiolus species were found to have pollen of Gladiolus on the dorsal surface of the thorax. Female anthophorines and halictids also carried Gladiolus pollen in their scopae, while Gladiolus pollen was usually present in the corbic- ulae of captured Apis mellifera. The hundreds of hours spent in the field, always at optimal times for pollinator activity, suggest low rates of visitation to Gladiolus flowers by antho- phorine bees. Visitation rates to species with gullet or flag flowers ranged from no visits to a maximum of three bees per flower per hour. These rates seemed highest in undisturbed sites, where flow- ering individuals are relatively widely dispersed over the landscape (usually 1-2 m apart). At sites where fire or clearing had occurred in the past one to two years, flowering individuals were often at rel- atively high density (ca. 20-30 cm apart) and visits seemed to be reduced, so that although bee indi- viduals were seen visiting some flowers, many flow- ers did not receive visits during the hours we spent at such sites. Typically, female anthophorines begin a foraging bout by visiting flowers of Gladiolus, but then forage on coblooming plants of other taxa of- fering either pollen or nectar, e.g., Asteraceae, Lachenalia, Lobostemon, Salvia. Female anthophor- ines returned to Gladiolus flowers repeatedly during the day, visiting three or more Gladiolus flowers and alternating these visits with bouts of foraging on flowers of nectariferous species in the families As- teraceae, Boraginaceae (Lobostemon), Fabaceae, Hyacinthaceae (especially Lachenalia), Tridaceae, Lamiaceae, and Sterculiaceae (Hermannia) (Table 5), as well as nectarless species in the families As- phodelaceae (Bulbine, Trachyandra) and Hyacin- thaceae (Tenicroa). Apis mellifera was observed to forage on Gladi- olus trichonemifolius more frequently and consis- tently than any female anthophorine. More than 20 bees were noted foraging on G. trichonemifolius be- tween 1100 and 1200 hours. However, after 1200, Apis mellifera abandoned flowers of this species and shifted foraging to Trachyandra and other cobloom- ing genera. Reproductive success. For one species where this was estimated, Gladiolus venustus, reproductive success was 88%. In a sample size of 61 plants that collectively produced 253 flowers, 202 well formed, full capsules were produced. Our general impression is that species with flowers adapted for pollination by bees almost always produce numer- ous capsules per plant. Failure to set capsules is 510 Annals of the Missouri Botanical Garden Table 5. Pollen-load analysis of bees and other insects captured on Gladiolus species with short-tubed flowers. Number of individuals of either sex indicated in parentheses. Gladiolus species are arranged by section according to Goldblatt and Manning (1998). Taxonomic associations of insects are as follows: Hymenoptera (bees): Allodape, Ame- gilla, Anthophora, Apis, Pachymelus, Tetralonia, Xylocopa (Apidae); Andrena (Andrenidae); Lasioglossum, Izoottichs, Patellapis (Halictidae); Rediviva (Mellittidae). Dio a (flies): Prosoeca, M (Nemestrinidae); Philoliche (Ta- banidae)i Psilodera (Acroceridae). Coleoptera (beetles): Lepisia, Pachycnema (Scarabaeidae: Rutelinae). Abbreviations: U nidentified monocot; UD = unide dicot. Families of 2. taxa: Aristea, Homeria, Ixia, Hesperantha, Lapeirousia, Moraea, Romulea (Iridaceae); Bulbine, Trachyandra sp. (Asphodelaceae); Cyphia (Campanulaceae); Erica (Ericaceae); Hermannia (Sterculiaceae); Lachenalia, Tenicroa (Hyacinthaceae); Lobostemon (Boraginaceae); Oxalis (Ox- alidaceae); Protea (Proteaceae); Salvia (Lamiaceae); Satyrium (Orchidaceae); Spiloxene (Hypoxidaceae); Wachendorfia (Haemodoraceae). umber of insects carrying pollen of Gladio- Other Pollen species carried Plant and insect taxon lus species (excl. Gladiolus) SUMMER-RAINFALL ZONE Gladiolus sect. Densiflorus G. appendiculatus Amegilla aspergina (2 9 | d) 3 0 n/a G. crassifolius Amegilla capensis (3 д) 3 2 Moraea brevistyla, UD (1) А. spilostoma (2 9) 2 2 UD X 2 (?Lamiaceae) Prosoeca T (1) 1 1 UD X 4 (?Gentianaceae, Fabaceae) G. den EI (2 9) 2 2 Lamiaceae, Fabaceae gos fallax (2 8) 2 2 1), Fabaceae A. spilostoma (3 9) 3 1 Fabaceae, UM (1) G. ferrugineus Amegilla capensis (1 9) l 1 Chlorophytum sp., а rupestris, Lamiaceae, ?Fabac Amegilla aspergina (1 9) 1 0 n/a G. papilio Атерша capensis (2 9) 2 0 n/a Amegilla aspergina (1 9) 1 0 Tetralonia sp. (1 l 1 Hibiscus sp. Gladiolus sect. Ophiolyza G. dolomiticus Amegilla spilostoma (2 9) 2 2 Fabaceae, Lamiaceae Xylocopa rufitarsis (2 9) 2 2 Fabaceae, Lamiaceae Allodape variegata (2 9) 0 1 Asteraceae Lasioglossum sp. (1 9) 1 0 n/a G. ecklonii ne So "on a (2 4) 2 2 Lamiaceae (?Hemizygia), UD G. sericeo A el la (2 9) 2 l Asteraceae, Ericaceae, UD (2), UM (1) Prosoeca sp. (1) 1 0 Gladiolus sect. Hebea G. permeabilis subsp. edulis Amegilla fallax (2 9) 2 2 UD (2) Gladiolus sect. Heterocolon G. rufomarginatus Amegilla langi (1 9 1 1 Asteraceae, UD (2) (?Lamiaceae) Amegilla spilostoma (1 9) l 1 ?Lamiaceae, UD (1) Stenobasipteron difficile (2) 2 1 Asteraceae Volume 85, Number 3 1998 Goldblatt e Bee- pae. Gladiolus Species Table 5. Continued. Number of insects carrying pollen of Gladio- = Pollen species carried Plant and insect taxon lus pecies (excl. Gladiolus) WINTER-RAINFALL ZONE Gladiolus sect. Blandus G. grandiflorus Anthophora diversipes (4 9) 4 4 Hermannia sp., Lobostemon, Pelargonium sp., Bombyliidae (1) 1 1 Hermannia sp., Lobostemon, UM (1) G. phoenix Amegilla spilostoma (2 9) 2 2 Lapeirousia neglecta, UD (?Fabaceae) Gladiolus sect. Linearifolius G. aureus Apis mellifera (4) 4 0 n/a G. brevifolius Allodape exoloma (1 9) 1 1 Asteraceae, Salvia sp., Fabaceae Al. pictifrons (1 9) 1 1 Asteraceae, Fabaceae Amegilla си A 9 1d) 2 2 Bulbine favosa, Salvia sp., Fabaceae Am. spilostoma (5 9 2 d) 7 1 Bulbine favosa, Pelargonium sp., Salvia sp., Fabaceae, Asteraceae Lipotriche (1 9) 0 0 a Psilodera valida (4) 4 4 Bulbine favosa, Salvia sp. G. hirsutus Apis mellifera (2) 2 2 Protea sp., Asteraceae Gladiolus sect. Heterocolon G. marlothii ү diversipes (3 9) 3 3 Tenicroa, Fabaceae (?Wiborgia sp.), Lamiaceae С. 2. ertia megilla и (29) 2 0 n/a Gladiolus sect. Hebea G. alatus Anthophora diversipes (2 9) 2 2 Aristea inaequalis (Iridaceae), Wachendorfia sp. (Haemodoraceae), UM (1) An. diversipes (2 9) 2 2 Babiana disticha, Asteraceae, UD e. EOM An. diversipes (1 9 1 d) 2 2 Lachenalia sp., Lobostemon sp., Оха Asteraceae Redivia aurata (1 9) 1 1 Lachenalia, Asteraceae G. ceresianus — diversipes (1 9) 1 1 Asteraceae, ?Lamiaceae G. equita Pac рани ЈЕ peringueyi (1 9) 1 1 Hermannia sp., UM (1) G. meliusculus Andrena 1 ? Homeria sp., Oxalis sp., Asteraceae Lepisia oh e 6 4 Ro ned ?eximia, Spiloxene capensis Pachycnema crassipes (6) 6 6 Drosera sp., Spiloxene capensis, Asteraceae G. orchidiflorus Anthophora diversipes (2 9) 2 2 Lobostemon, Asteraceae G. permeabilis subsp. permeabilis 2” diversipes (2 %) 2 2 Hermannia sp., Asteraceae G. s Anthophora diversipes (1 д 1 9) 2 1 Lobostemon sp., UM (1) G. speciosus Rediviva aurata (1 9) 1 1 Diascia sp., Ornithogalum (?thyrsoides) 512 Annals of the Missouri Botanical Garden Table 5. Continued. umber o insects carrying pollen of Gladio- Other Pollen species carried Plant and insect taxon lus species (excl. Gladiolus) G. stellat “Andrena sp. (2 9) 2 2 Moraea (?polyanthos), Lobostemon sp. 4 diversipes (1 9) 1 1 Lachenalia (?elegans), UD (1) (?Fabaceae) G. venustus Anthophora diversipes ( 9 1 d) 4 4 Lobostemon sp., Lachenalia sp. A. Mesi (19) 1 1 UD (2), Orchidaceae G. viresc Ри а diversipes (1 9) 1 1 Ixia (?latifolia), Satyrium sp. (Orchidaceae) G. watermeyeri jacens diversipes (2 9) 2 2 Lobostemon (Boraginaceae) A. krugeri (1 9) 1 1 Lachenalia (Hyacinthaceae), 1 (UD) Gladiolus sect. | У 6. blommestei Anthophora Geman = 1$ id) 2 2 Lachenalia sp., Lamiaceae (?Salvia sp.), Lobostemon sp.. UD (?Cyphia sp.) G. brevitubus Lasioglossum sp. (2 9) 2 2 Asteraceae пайиз Anthophora diversipes (2 d | 9) 3 3 Lobostemon sp., Hermannia sp., Homeria A. schulzei (2 9) 2 2 Hermannia sp., Homeria sp. A. krugeri (1 8) 1 1 Lobostemon sp., Hermannia sp., Homeria Apis mellifera (2) 2 2 Moraea sp. Lasioglossum sp. (1 9) 1 1 Asteraceae G. exilis Amegilla fallax (1 9) 1 1 Trachyandra sp. G. gracilis Anthophora diversipes (2 9) 2 2 Lobostemon sp., Fabaceae (?Aspalathus) Apis mellifera (5) 5 4 Lobostemon sp., Erica sp. Xylocopa rufitarsis (1 d) 1 1 Fabaceae (?Aspalathus) G. griseus Apis mellifera 4 4 Asteraceae (?Chrysanthemoides monilifera), Oxalis G. inflatus je aen obscuriceps (1 9 1 d) 2 1 Lobostemon sp. in Уа krugeri (1 d) 1 0 n/a Apis mellifera (3) 3 3 Oxalis зр., Romulea sp. Patellapis sp. (1 9) 1 1 Asteraceae G. jonquilliodorus megilla spilostoma (1 9) 1 1 Aristea Apis mellifera (1) 1 1 Aristea, UD (1) G. martleyi Amegilla spilostoma (2 9) 2 2 Rutaceae, UM (1) G. patersoniae Anthophora krugeri (3 9) 3 1 Lobostemon sp. Anthophora diversipes (1 9) 1 1 Lobostemon sp. Apis mellifera (3) 3 3 Lachenalia sp., Lobostemon sp., Oxalis sp., UD (1) G. quadrangulus Apis mellifera 5 5 Oxalis sp., Hermannia sp. avus Anthophora о (2 9) 2 2 Ixia (?brunneobracteata), Homeria sp., Lobostemon sp. G. trichonemifolius Apis mellifera (7) 7 7 Asteraceae, Oxalis sp., Romulea sp., Trachyandra sp. Volume 85, Number 3 Goldblatt et al. 513 1998 Bee- gane. Gladiolus Species умом (rigor палату ------------ 1597 Fig Representative long-tongued anthophorine bee species captured on Vip of Gladiolus, showing tongue beh. relative to body size. —A. Amegilla fallax. —B. Amegilla capensis. Scale bar 5 more likely to be the result of unfavorable weather conditions than ineffectiveness of pollinators. DISCUSSION Bee pollination in southern African Gladiolus comprises two different systems. A few species offer meager quantities of nectar or none at all, and have what appear to be derived, short-tubed, rotate flow- ers with actinomorphic perianths. These flowers are adapted for a relatively unspecialized pollination strategy of active pollen collection by female an- drenids, honey bees, and possibly halictid bees. In the second mode of bee pollination, flowers have zygomorphic perianths and obliquely funnel- shaped floral tubes containing appreciable amounts of sucrose-rich nectar. They are adapted for pollen dispersal by anthophorine and honey bees seeking nectar. In this mode, bees must navigate zygomor- phic perianths, and the transfer of pollen to the body of the bee is passive (sensu Bernhardt, 1996). In both these modes of bee pollination, bees are always polylectic foragers. Apis n and some arge-bodied anthophorines are know eralist foragers in southern Africa (Goldblatt et " 1989, 1995) and on other continents (Armstrong, 1979; Michener, 1974; Bernhardt & Weston, 1996). Another aspect of both modes of bee pollination is that the combination of floral dichogamy, herkoga- my, self-incompatibility (inferred), and fairly diffuse population patterns would appear to encourage out- breeding, with large-bodied bees acting as pollen vectors. The zygomorphic, gullet or flag flower offering nectar and pollinated by large-bodied, hairy, long- tongued bees is the most common floral type among the bee-pollinated Gladiolus species studied. This pollination system has now been established for 41 species of the genus in southern Africa. Ап addi- tional 53 species have flowers similar in size, shape, and coloring, and may be assumed to have the same pollination system. Thus 9546 of bee-pol- linated species and 94 of the 166 species (5696) o 514 Annals of the Missouri Botanical Garden Table 6. Body dimensions of bee and fly species cap- — have rotate perianths and are known or inferred to tured on flowers of Gladiolus species. Measurements are the range for each species that was captured, thus cov- ering different study sites. Mouth-part Body length length Taxon mm mm HYMENOPTERA ANDRENIDAE Andrena sp. 14—15 2—3 ANTHOPHORIDAE Amegilla capensis ca. 12.5 9-10 Am. fallax 10-12 4—6 Am. langi ca. 11 ca. 10 Am. obscuriceps 10-11 5-6 Am. cnn stoma 10-14 6-9 Am ca. 10 ca. 7.5 - diversipes 14-17 6.58 An. krugeri 11-14 6-7.5 An. schulzei 12-13 5—6 Pachymelus peringueyi 17 ca. 8 Xylocopa rufitarsus 14-15 ca. 4 APIDAE Apis mellifera 10-12 3-4 HALICTIDAE Patellapis sp. ca. 9 ca. 2 Lasioglossum sp. ca. 10 ca. 2 Lipotriche sp. ca. 11 ca. 2 MELLITIDAE Rediviva aurata ca. 12 ca. 2 DIPTERA ACROCERIDAE Psilodera valida 9-10 8-12 NEMESTRINIDAE Prosoeca sp. ca. 15 ca. 12 Stenobasipteron difficile 12-13 ca. 13 Gladiolus in southern Africa, including Zimbabwe, are known or assumed to be pollinated primarily by large, long-tongued bees foraging for nectar. This pollination strategy is also the most wide- spread within the genus, and predominates in six of the seven sections of Gladiolus in southern Af- rica (according to the infrageneric classification of the genus by Goldblatt & Manning, 1998). In the remaining section Ophiolyza, pollination by sun- birds is inferred as most common, given the elon- gated floral tubes and red or orange flower color of most species. The most parsimonious interpretation would suggest that bee pollination is the ancestral pollination syndrome in the genus ust five Gladiolus species in southern Africa be pollinated by bees, in this case short-tongued bees foraging for pollen. The gullet and flag flowers of many Gladiolus species show the same type of bee pollination as the bilabiate flowers of such dicot genera as Pedi- cularis (Scrophulariaceae; Macior, 1984), Aconitum (Ranunculaceae; Proctor et al., 1996), and the vast majority of papilionoid legumes (Kalin Arroyo, 1989). Both types of flowers exhibit passive collec- tion of pollen. More important, floral evolution in Gladiolus exhibits a pattern noted in the flowers of a few other plant genera (e.g., Pedicularis), in which passive pollen deposition appears labile. Within such taxa as Pedicularis (Macior, 1982), Ranun- culaceae tribe Cimicifugeae (Pellmyr, 1985), and the Lecythidaceae (Mori & P , 1990), nectar- iferous bilabiate flowers may be derived from nec- tarless actinomorphic flowers, or vice versa (Bern- hardt, 1996). In Gladiolus, as іп Pedicularis, nectarless species appear to have been derived rom ancestors with nectar-producing, bilabiate flowers, but nectarless or nectar-poor species re- main in the minority. In Gladiolus, this appears to have happened repeatedly, with examples in three separate sections (G. aureus being a striking tran- sitional example with its nectar-poor, but gullet-like flowers with an actinomorphic perianth). There are notable differences between the floral characteristics of summer- and winter-rainfall spe- cies of Gladiolus. In the summer-rainfall zone, flow- ers tend to be less variable and less strikingly col- ored, have obscure tepal markings, and lack scent. They are usually also somewhat smaller in size, and more numerous per inflorescence than in the win- ter-rainfall zone. The diversity of insect pollinators is lower, and pollinators have a smaller body size. In the winter-rainfall zone, the range of floral color is considerable, and floral markings usually show strong color contrasts. The flowers are often strongly scented, comparatively large, and few per inflores- cence. The diversity of pollinators is greater than in the summer-rainfall zone, and pollinator body size is often larger. Irrespective of geography, the selection of large anthophorines pollinating Gladi- olus flowers does not reflect the diversity of the large anthophorine bee fauna of southern Africa (Eardley, 1994; Eardley & Brooks, 1989). Only a fraction of the total number of Amegilla, Anthop ra, and Pachymelus species represented in the re- gion have been captured foraging for nectar on Gladiolus flowers. The same bee species have been captured on several different species of Gladiolus, rance and sometimes the same bee species has been cap- Volume 85, Number 3 1998 Goldblatt et al. Bee-pollinated Gladiolus Species 515 tured in both the summer- and winter-rainfall zones. Exclusive pollination by long-tongued bees, es- pecially anthophorines, has not often been docu- mented as a specialized mode of pollination in the southern African flora. Large-bodied anthophorines have been associated with Nivenia and Lapeirousia spp., but a large proportion of species in these gen- era are pollinated by long-tongued flies in the Ne- mestrinidae and/or Tabanidae (Goldblatt & Bern- hardt, 1990; Goldblatt et al., 1995). The bilabiate, nectariferous flowers of Gladiolus species appear to exploit long-tongued bees as pollinators in much the same manner as Nivenia and Lapeirousia ex- ploit long-tongued flies. The relationship between the bee-pollination system of the majority of Gladiolus species and the long-tongued fly pollination system of the minority of Gladiolus species (Manning & Goldblatt, 1997; Goldblatt & Manning, 1998) is of particular inter- est. In contrast to flowers of bee-pollinated Gladi- olus species, flowers adapted for pollination by long-tongued flies are scentless, have much longer floral tubes, are pink to white with red nectar guides, and are at best weakly bilabiate (Manning & Goldblatt, 1995, 1997). These floral differences, while minor, are entirely consistent within the ge- nus. Consequently, the adaptive radiation of polli- nation systems in Gladiolus appears to be the in- verse of that in Lapeirousia subg. Lapeirousia, where bee pollination is derived in some species. A specialized pollination system? Pollination by long-tongued anthophorine bees is not normally considered a specialized pollination strategy among rican plants. More often (e.g., in Lapeirousia, Goldblatt et al., 1995), anthophorine bees partici- pate in more generalized pollination systems that include honey bees, bee flies (Bombyliidae), but- terflies, and hopliine beetles. The vast majority of Gladiolus species with moderate-sized flowers and obliquely funnel-shaped tubes show a close fit be- tween the size of the flower and that of the antho- phorine pollinator—more specifically, between the shape, diameter, and length of the floral tube and the shape and size of the anthophorine’s head and thorax. This leads to direct contact between the dorsal surface of the bee’s thorax and the sexual organs of the flower. The fit is refined in G. appen- diculatus, where bees gain access to the floral tube by contact with the anther appendages. Evolution of these appendages in G. appendiculatus converges with the evolution of anther awns in the zygomor- phic flowers of Viola and its allies (Beattie, 1974) and of the sterile anther locule in Salvia. The pol- linators of Viola or Hybanthus flowers cannot feed on the nectar in the floral spur without tripping the anther awns that deposit pollen on the insects’ bod- ies (Bernhardt, 1996) Most other insects that we have occasionally cap- tured visiting Gladiolus flowers do not come close to matching this fit. Smaller bees, including Hal- ictidae, and hopliine beetles, are usually too small to be effective pollinators, and seldom carry Glad- iolus pollen, except in the case of G. meliusculus, which appears to be specifically adapted for polli- nation by hopliine beetles, evidently in combina- tion with andrenid bees (Goldblatt et al., 1998). Long-tongued flies, including Psilodera (Acroceridae), Prosoeca and Stenobasipteron (Nemestrinidae), with large bodies and tongues 10-15 mm long, behave much like anthophorine bees and appear to be po- tential pollinators, but are much less frequently ob- served than bees and thus seem to be of minor importance in the pollination ecology of Gladiolus. These flies are nectar feeders and are not known to consume pollen, nor do the females collect pollen to provision nests. We have only rarely seen anthophorine bees or Apis mellifera collecting pollen actively from Glad- iolus species with zygomorphic flowers. The accu- mulation of Gladiolus pollen in the scopae or cor- biculae of these bees thus must result from the grooming process. Analyses of pollen loads re- moved from the bodies of female bees, including the thorax and corbiculae or scopae, combined with field observation of bee activity, indicate that honey bees and female anthophorines depend on Gladio- lus for nectar but visit many other plants for a com- bination of pollen and nectar. In some cases, these bees visit nectarless flowers (Bulbine, Tenicroa, Trachyandra) exclusively for their pollen. Since Apis and anthophorines visit both nectar-rich and nectar-poor flowers in southern Africa, their forag- ing parallels the behavior of female anthophorines and halictids in southern Australia (Bernhardt, 1984, 1986, 1989, 1995). In southern Africa, Glad- iolus species produce nectar, satisfying the energy requirement of the adult bees, while other species offer pollen required by the bees' larvae. Thus Gladiolus species with short-tubed, zygomorphic flowers appear to be specifically adapted for a type of bee pollination in which nectar is the consistent reward, and which involves one or very few insect species. There is no indication that this strategy is part of a generalist system involving long-tongued bees and other unspecialized pollinators. Interestingly, the suppression of both zygomor- phy and nectar production in Gladiolus species does not make them more attractive to anthophorine 516 Annals of the Missouri Botanical Garden bees. Instead, these floral modifications appear to have resulted in a pollinator shift in which the usu- al pollinators (anthophorines) are replaced by bees belonging to different taxonomic associations (An- drenidae, Apis, and possibly Halictidae) and, in one case, hopliine beetles. А consequence of the limited selection of antho- phorine bee species that visit flowers of Gladiolus species is the constraint this imposes on the geo- graphic ranges of these plants. Species of Gladiolus sharing anthophorine bee pollinators rarely co-oc- cur, except when their flowering times do not over- lap. Otherwise, they would compete directly for the same pollinator(s) and be at risk not only of having the wrong pollen deposited on their stigmas, there- by reducing or preventing seed production, but also of having their gene pools disturbed by hybridiza- tion (many Gladiolus species are interfertile; Her- bert, 1847; Goldblatt, 1971, and unpublished). Dif- ferent species using the same pollination system may be isolated reproductively either by edaphic preferences or by flower shape, color, markings, and/or scent. We speculate that the reason floral variation is so great in Gladiolus species pollinated primarily by large-bodied bees in the winter-rain- fall zone is that flowering there is concentrated in the spring. The coincident flowering of so many species puts a premium on pollinators. By contrast, species of the genus that have shifted their flow- ering time significantly exhibit limited floral vari- ability. They have comparatively small flowers of unspecialized form, often virtually identical even in different sections, and more typical of species of the southern African summer-rainfall zone. In conclusion, pollination by anthophorines in Gladiolus shows both expected and unexpected trends. As expected, zygomorphic flowers coupled with high-sucrose nectars are often predicted in plants pollinated by bees with large bodies and long tongues (Baker & Baker, 1983, 1990). In con- trast, it is unusual to find two floral forms, flag and gullet, in the same plant genus and in association with the same pollinators. The floral forms of Glad- iolus species pollinated by bees alone are as vari- able as those of some neotropical orchid genera pollinated by euglossine bees (Dressler, 1981). The diversity of Gladiolus in southern Africa is account- ed for largely by species using long-tongued bees as specialist pollinators. Other pollination strate- gies occur in less than half the species in the re- gion. А final and unexpected feature of bee polli- nation in Gladiolus is the correlation between floral crypsis and strong odor. To the human eye, the flow- ers of several species of section Hebea (e.g., G. ar- cuatus, G. orchidiflorus, G. scullyi, and G. water- теуегі) appear well camouflaged, and while large, are dull colored and merge remarkably well with the terrain. One is more likely to be able to find such flowers by tracing the source of their rich odors than by sight. Could this be an adaptation to restrict the diversity of floral visitors, including those that might damage or consume flowers? Literature Cited E J. 1979. Biotic pollination mechanisms in the Australian dn review. N. Zealand J. Bot. 17: 467- 508. Baker, Н. С. & I. Baker. 1983. Floral nectar sugar con- stituents in relation to pollinator type. Pp. 117-141 in . E. Jones & R. J. Little (editors), Handbook of Ex- perimental Pollination Biology. Scientific and Academic шаш, New ork. . 1990. The predictive value of nectar uro to the rec о Ноћ of pollinator types. Israel . Bot. 39: 157-166. Beattie, A. J. 1974. Floral A in Viola. Ann. Mis- ri Bot. Gard. 61: 781-79 Bernhardt, P 1984. The lini in of Hibbertia stricta (illeniacsas’: РІ. 5 vol. 147: 267-271 1 Вее pollination in in M fasciculata (Dilleniacese), РІ. 5 Evol. 152: 231-241. 1989. The богы ecology of Australian Acacia. Ann. ‘Missouri Bot. Gard. 74: 42-50. 995. Buzz- Brod of Dianella caerulea var. assera v (юнде eae). Cun ninghamia 4: 9-20. 996. Anther adaptation in animal pollination. Pp. 192-220 i in W. G. D'Arcy & R. C. Keating (editors), The Anther: Form, Function, a Phylogeny. Cambridge Univ. Press, Cambridge, U. & P. H. Weston. 1996. The pollination ecology of Persoonia (Proteaceae) in eastern Australia. Telopea 6: 775-784. Buchmann, S. L. 1983. Buzz pollination in angiosperms. Pp. 73-113 in C. E. Jones & R. J. Little (editors), Handbook of Experimental Pollination. Van Nostrand Reinhold, e or arwi The Effect of Cross- and Self-Fer- tilization i in s Vegetable Kingdom. John Murray, Lon- resale R. L. 1981. The Orchids. Natural History and Classification. Harvard Univ. Press, Cambridge, Mas sachusetts. Eardley, C. D. 1994. ab ig Anthophoridae) in ern Afr mol. 1. Dept. Agric. S. Africa 91. к. W. Brooks. 1989 TS іп southern ја Entomol. Agric. Water Supply S. Africa 76. Faegri, L. van der Pijl. 1979. The Principles of Р‹ лабон Ecology, 3rd ed. Pergamon Press, New York Goldblatt, Р. 1971. Cytological and anatomical studies on the southern African Iridaceae. J. S. African Bot. 37 31 ал Ж Тће ol даи ЗА (Ну- uth ca. Ento- . The genus Anthophora Mem. Dept. ). Phylogeny 42: classification of Iridaceae. Ann. 2 Bot. Gar –627. 1. Ап overview i the systematics, phylogeny and ae p the southern African lridaceae. Contr. Bolus Bos : 1-74. ›. о in Tropical Africa. Press, коа Огероп. Timber Volume 85, Number 3 1998 Goldblatt e Bee- esti Gladiolus Species 517 —— а P. Bernhardt. 1990. Pollination biology of Niv- enia (Iridaceae) and the p of ер self- Masa qd Israel J. Bot. 39: ----- 1 nning. 1998. Gladiolus in Southern Africa. Fernwood yg Cape Tow ‚Р. йө hardt & Ј. 2 Mans 1989. Notes on the pollination mechanisms of Moraea inclinata and M. Heim Саа РІ. Sus Evol. 163: 201—209. 1997. Notes on the polli- nation of Gladiolus аи (Iridaceae) by bees (An- thophoridae) and bee mimicking flies pole Ас- 998. mun. of petaloid geophytes by monkey beetles Cds өз Rutelinae:Hopliini) in Southern Africa. Ann. Missouri Bot. Gard. 85: 215-230. J. C. Manning & P. Bernhardt. 1995. Pollination biology of Lapeirousia subgenus Lapeirousia (Iridaceae) in southern Africa: Floral divergence and adaptation for long-tongued fly-pollination. Ann. Missouri Bot. Gard 82: 517-534. Herbert, W. 1847. On hybridisation amongst vegetables. J. Hort. Soc. London 2: 81-107. Horn, W. 1962. Breeding research on South African plants: Ш. Intra- and interspecific compatibility in /xia L., Sparaxis Ker., Watsonia Mill. and Zantedeschia Spreng. J. S. African Bot. 28: 26 Johnson, S. D. & W. A. Bond. 1994. Re d flowers and butterfly pollination in ұм fynbos of South Africa. Pp. 137-148 : . The Scent of Orchids: Olfactory and Chemical Investigations. Elsevier, Amsterdam Kalin Arroyo, M. T. 1989. Breeding dd and polli- nation biology in the Leguminosae. Pp. 9 in R. Polhill & P. H. Raven (editors), 2. in аи нај - Royal Botanic Gardens, Kew. Knox, . Clarke, S. Harrison, P. Smith & J. J. vai eL 1976. Cell recognition in plants: Deter- minants of the stigma surface and their pollen interac- tions. Proc. Natl. Acad. Sci. U.S.A. 73: 2788-2792. Macior, L. W. 1982. Plant community and pollinator dy- namics in the evolution ol pollination mechanisms. Pp. J. A. Armstrong et al. (editors), Pollination ternational sur la Pollinisation. Colloq. I.N.R.A. 21: 7-261. Manning, a 4 & P. Goldblatt. 1995. Cupid comes in many guises: The not so humble fly and a pollination guild in be Оте Veld & Flora 8l: 50—52. & 1997. The Moegistorhynchus longi- rostris ЕЈ Nemestrinidae) pollination guild: Long- tubed flowers ecialized long-proboscid fly pol- тейен аы та к lias Africa. Pl. Syst. Evol. 206: 51 -69. шее». R. 1908. Some пер on entomophilous ers. S. ke Sci. 5: 113. Иа C. D. The E Behavior of the Bees. A quide. 7 s Belknap Press, Harvard Univ. Press, Cambridge, Massachusetts. Mori, S. A. & G. T. Prance. 1990. Flora Neotropica. Le- яа cues II. Zygomorphic-flowered New World gen- w York Botanical Garden, = York. Ogden, E. C. G. S. Raynor, J. V. Hayers & D. M. Lew 1974. Manual of Sampling pesada Pollen. Hafner Press, London. Ohri, D. & T. N. Khoshoo. 1981. Cytogenetics of garden Gladiolus. 1. Pollination mechanism and breedi tem. Proc. Indian Natl. Sci. Acad., B 47: М. in of salsa ний mutualism. Acta Univ. psa 2: 1-34 1996. The Baba History of Pollination. Timber hu Portland, Oregon Roig-Alsina, A . D. Michener. 1993. Studies of the phylogeny and classification of long-tongued bees (Hy- menoptera: Apoidea). Univ. Kansas Sci. Bull. 55: 123- 162 Scott Elliot, G. 1891. Notes on the fertilisation of South African and Madagascan flowering plants. Ann. Bot. (Oxford) 5: 333—405. Vogel, S. 1954. Blütenbiologische Types als Elemente der а Bot. Stud. 1: 1—338. Vos, M. P. de. 1972. The ag Romulea in South Africa. Ј.5 о ti Bot., Suppl. 9: 1–307. POME ANATOMY OF Juan José Aldasoro,? Carlos Aedo? and ROSACEAE SUBFAM. ај MALOIDEAE, WITH SPECIAL REFERENCE TO PYRUS' ABSTRACT Two anatomical features of the pome in Rosaceae subfam. Maloideae are investigated: sclereid type and epidermal structure. The large and irregular groups of sclereids in Pyrus are (ek rent from those in Sorbus subgenera Aria, Chamin "spilus, i Cormus, and similar to those in Cydonia. In additior шШиійуегей epidermis, from Pyrus, is documented in Pyrus sect. Pashia. Consequently, both the on of Pyrus and its current sec "tional classification are supported. The taxonomy of Rosaceae «Маш. Maloideae is — styles (free in Pyrus). This feature is consistent, but problematic in terms of generic delimitation. The шау be difficult to evaluate in practice. Thus, Bai- inconsistency of the main generic characters has ley (1949) reported the structure of the flower clus- generated a great deal of disagreement in the tax- (ег as the most obvious distinction between Pyrus onomic treatment of the group. A representative of and Malus: the Pyrus inflorescence has a rachis the more synthetic view was de Candolle (1825), — from which the pedicels emerge, while that of Mal- who included in Pyrus species now usually referred us has ап umbellate structure. Nevertheless, Rob- to Malus, Photinia, Eriolobus, and Sorbus. This ertson et al. (1991) showed that both Pyrus and classification was followed by Sax (1931) and Rob- Malus could have corymbs, panicles, or umbels. ertson (1974). Conversely, Decaisne (1874) and Finally, the supposed scarcity or absence of scler- Koehne (1890) used smaller generic concepts. They — eids in the pomes of Malus was contested by sev treated Pyrus іп a more restricted sense, and split eral authors, including Rehder (1940), Browicz off Photinia, Malus, and Sorbus. A comprehensive (1969), Terpó (1968), and Iketani and Ohashi review of taxonomic treatments applied to these — (1991). Robertson et al. (1991) reported that Malus genera was provided by Robertson et al. ( : may have abundant sclereids under the skin and Malus, Cydonia, Sorbus subg. Aria Pers., and around the core of the pomes. Hybridization and Sorbus subg. Chamaemespilus (Medik.) К. Koch grafting experiments provide additional data about have all been advanced as close relatives of Pyrus Pyrus relationships. According to Taylor (1983) Py- (Weber, 1964; Iketani & Ohashi, 1991; Campbell rus and Malus do not hybridize and cannot be graft- et al., 1995). According to Decaisne (1874), pomes ed one to the other. They also differ in flavonoid of both Sorbus subg. Aria and S. subg. Chamae- composition (Williams, 1982). However, Weber mespilus are characterized by their heterogeneous (1964) and Robertson (1974) reported that Pyrus, flesh. Flesh heterogeneity of pomes in subfamily Malus, and Cydonia can and do hybridize among Maloideae was studied by Kovanda (1961) and Ike- themselves. tani and Ohashi (1991), who showed that it was According to Rohrer et al. (1991: 78), the skin caused by groups of parenchyma cells filled with of the pomes of subfamily Maloideae “consists of a tannic substances. Cydonia, formerly included in single epidermal layer of tightly packed, anticlin- Pyrus by Linnaeus (1753), and closely related to it ally flattened, rectangular cells covered with а cu- according to Robertson et al. (1991), is easily dis- — ticle." Such an epidermal structure has been de- tinguishable by its solitary flowers and numerous scribed for Crataegus (Akhunova, 1986), Malus ovules per locule. Malus is separated by its connate (Clements, 1935), and Amelanchier (Olson & ! The authors thank M. Jerez for aiding with aiino iun microscopic Mi cipe Ginés López, N. Taylor, Spongberg, and the editors are thanked for helpful criticisms of the manuscript. are also indebted to 15 curators of the Royal Botanic Gardens, Kew, University of Live киет Botanic Gardens (NSS), Sir Harold Hillier Gardens and Arboretum, U.K., and Wakehurst Garden for their kind permission to collect living materials. ? Real Jardin Botánico, Consejo Superior de Investigaciones Científicas, Plaza de Murillo 2, 28014 Madrid, Spain (e-mail address: aedo@ma-rjb.csic.es). * Departamento de Biología Vege ма! ПІ, Facultad de Farmacia, Universidad Complutense, 28040 Madrid, Spain. ANN. Missouni Bor. GARD. 85: 518-527. 1998. Volume 85, Number 3 1998 Aldasoro et al. 519 Pome Anatomy of Rosaceae Steeves, 1982). On the other hand, Miller (1984) reported a multilayered epidermis in Mespilus ger- manica L. Our survey of anatomical characteristics of pomes of subfamily Maloideae has documented the occurrence of a multilayered epidermis in both Pyrus and Sorbus torminalis (Aldasoro et al., 1998). The supraspecific taxonomy of Pyrus is also con- troversial. Decaisne (1871-1872) recognized 23 species arranged in six informal groups. Koehne (1890) described two sections: Pashia and Achras. Fedorov (1954) recognized four sections: Pashia, Pyrus (= sect. Achras Koehne), Xeropyrenia Fed., and Argyromalon Fed. Tuz (1972) reduced these to two, Pashia and Pyrus, each with several subsec- tions. Terpó (1985) added his section Pontica, but the classification of Tuz (1972) was accepted by Browicz (1993), who pointed out that the two sec- tions could be distinguished by certain obscure characters. According to Browicz (1993) the more operative ones are: the sepal persistence on the pome, the presence or absence of whitish lenticels, and the thickness and flexibility of the pedicels in fruit. The character states of section Pyrus are: se- pals persistent, white lenticels absent, and thick, stiff pedicels; and of section Pashia: sepals decid- uous, white lenticels present, and thin, flexible pedicels. Nevertheless, these characters showed some inconsistency; for example, several species of section Pashia may have thick pedicels. The aim of the present work is to investigate some anatomical features of subfamily Maloideae pomes with special reference to Pyrus, and to dis- cuss their bearing on the taxonomic issues detailed above. The currently accepted concept of Pyrus 18 that of Decaisne (1874), and the sectional division of the genus that proposed by Tuz (1972), because they are better supported by morphological and an- atomical data (Robertson et al., 1991; Browicz, 1993; Aldasoro et al., 1996). MATERIAL AND METHODS Pomes were collected (see Table 1) and pre- served in Kew mixture (Forman & Bridson, 1989). They were cut with a razor blade both longitudi- nally and transversely in order to examine the in- ternal structure. Thin hand-cuts were taken in the proximal third of the pome and photographed by light microscopy. Other cuts were made with a SLEE-MAINZ-MTC microtome and stained with Fasga mixture (Tolivia & Tolivia, 1987). In some cuts, malachite green was used to stain the scler- eids. For scanning microscopy, dried pomes were cut, glued to aluminum stubs, coated with 40—50 nm gold and examined in а JEOL-TSM T330A scanning electron microscope at 20 kV. RESULTS Usually, sclereids are present in the flesh of pomes of subfamily Maloideae. They may occur un- der the skin, in the core or spread throughout the flesh, isolated or in groups, and vary considerably in shape and size. Four main sclereid types could be distinguished in the flesh (Table 1): isolated sclereids, as in Rha- phiolepis; small groups (less than 10), as in Ame- lanchier, Chaenomeles, Cotoneaster, Crataegus, Er- iobotrya, Malus, Photinia, and Sorbus subgenera Sorbus and Torminaria; large but irregular groups, as in Pyrus (Fig. 1A, B) and Cydonia; and large and rounded groups, as in Sorbus subgenera Aria, Chamaemespilus, and Cormus (Fig. 1C, D). The groups of sclereids in Pyrus and Cydonia are remarkably dense (over 50 sclereids can be counted in an equatorial section) and have an ir- regular outline, while in Sorbus subgenera Aria, Chamaemespilus, and Cormus they comprise less than 40 sclereids and have an elliptic outline (Fig. 1C, D). Some consistent differences in the size and shape of these sclereids were observed (Table 1). Pyrus and Cydonia sclereids are smaller and have a smaller lumen (40-80 шт long; lumen diameter 10-51 um; wall thickness 10-20 шт) than those of Sorbus subgenera Aria, Chamaemespilus, and Cormus (110-240 um long; lumen diameter 76- 180 um; wall thickness 6-32 jum) (Fig. 1). Scler- a in pomes of Malus were isolated or in small s, and were larger and with a greater lumen НА (75-360 шт long; lumen diameter 12- 310 um; 7 thickness 15-80 шт) than those of Pyrus pom We were ais to study the pomes of 16 of the 38 species of Pyrus accepted by Browicz (1993): 9 be- longing to section Pyrus, and 7 to section Pashia (Table 1). A multilayered epidermis was found only in Pyrus sect. Pashia, while species of section Py- rus had only a single layer of epidermal cells that produced a thick cuticle (Fig. 2C, D). The remain- ing species of subfamily Maloideae showed a sin- gle-layered epidermis, except for Mespilus german- ica and Sorbus torminalis (Table 1; Miller, 1984; Aldasoro et al., ( In Pyrus, the multilayered epidermis has 3-6 layers of cells, each layer with a cuticular mem- brane. These cells are tangentially compressed and filled with tannic substances (Fig. 2A, B). They de- velop from a tangential meristem layer that is some- what similar to the phellogen, a meristem that ap- Annals of the 520 Missouri Botanical Garden (VW) <69 o1ospp]y "ртәицос ‘виәріе>у тән пә Ápnis siu "IS ОР 001 081 запола ПЕШ "M “е) ("Ҷәиеі 4) sisuaupuuná W (УИ) £r9 o:ospp]y “y ipo Арга sip 16 8l 021 OSI sdnoi [jews ‘prouyog “Y 77) (CUIXPIA) 1ysouoyos, "Jy (УИ) S99 o1ospp]y “y по &pnis sip 16 sz 016 09€ sdnoi3 [peus — "prouyos ^y 7) (Тиде) озодој а ^y prouyos (VIN) 689 o4ospp]y *w “no Ápnis sip 15 SI S6 211 sdnoi2 [jews "M 70 хә әццәоҹ sisuaumppis “W (VIN) 829 o4ospp]y * [пә Ápnis sip 15 02 се 9L за пола [jews ләрчә прјодал ^y (УИ) 189 o4ospp]y “y по &pnis snp 15 08 el 001 за пола [peus ‘piouyog "у 72 (urpereg) sisuansuDy ‘W (УИ) 612 o4osop]y * po &pnis в 15 0c 021 есі sdnoi2 [jews чоп (poom) в15и201 "jy (VW) 909 o4osvpyy ‘suap гән "ano Ápnis sty} "IS 0c 0€ с; за пола [jews әицәоу DUDI]]DY W (VIN) 6 o4osopjy “WW по Артв snp 16 02 cc 08 за пола [jews "prougos “yD (yey) vasnf y (УИ) 269 оохррју “4 пә Ápnis в 15 vz оғ 06 sdnoi3 pews ‘prouyog “Y 70 (ооп) рипимој "jy (УИ) 169 o:osopjy “y ipo pnis si 15 0c 28 221 sdnox3 [jews "ppog (71) 0209209 snpopy (VW) 62r[ Nawojoyuvg ‘YSN :eruop[e?) Apnis в IS = = = juasqe sure1qy (18914 72) 2270/2105 y (VW) 621 0uD “мәд Apnis sip 15 == = = juasqe чоон озошттирј “y (VIA) 06892 svsv2au1mn:) *eiquio[o7) Ápnis stu] 1s — m == juasqe "[pur] ргизатлаој гојошолодкон (VIN) 912 оаозррју “y IA по &pnis в 15 ср ZI OLI за пола [jews "US M M sisuogoncua] 73] (УИ) FIZ o40sopjy “WW по &pnis sip 15 SZ 02 021 sdno13 [jews Мооң Dipjonad “4 (ҰЙ) SIZ огоѕорү “WW jo &pnis sip 16 Ра 02 001 за пола [jews "[pur] (quny p) poruodpf 73 (ҰЙ) 212 олозррју “WW no Артв sip 15 cc Ot 08 sdnoi2 pews "ооң ('qxoy) sisuo]pZuaq v11000143] (VIA) 195 ологррју ‘uredg Ápnis sty} "IS OI CZ 0S ва пола теүпдәли pue әде “ІШІ 22uo]qo viuop&?) (VIA) 222 49]og “uredg Арп sty} 1S SI OL 001 sdno13 [yews OUR Y поло SISUJUOUIISNLX 7) (VW) 624 12106 *uredg Аргус sin 116 OI 06 96 запола pews 71 57|01920 =п8ә07017) (VI) 08< ологррју "VIA “no Ápnis sty} 15 7 OI ОР запола jews "род snunla doy! 7) (VIN) 1696 орәү “WW "no Apms в 15 Il OF £9 sdnoi3 [pus — [pur] хә реду snrjofixnq 421$2u0]07) (WIN) PPS олозррју “VIA "ipo Ápms sip 16 01 се се запола [pus — [pur] Саипцј) oruodvf зојешоигруг) (VW) 195 o4osopjy “үр apo Ápms в 15 ZI се 19 за пола. [[ешѕ ЭПРӘЙ (7) чзигрриро лотузирјешу perpnis јеџаојвуј elep јо золпос (po1o&e[ (шт) (шті) (шті) sdno13 рталојов цвә|4 uoxe] -ә|ӛше 216 ssouxonp 19]2UBIP ціӛиә| “рәләХе|ц ye чәшп рталојос "mu IW) press ртәләрәс̧ 5їшләр -idə зшод "pai suoauitoads əy} шолу so[dures 9AU JO 6ИРӘШ зле вер 941, 'овортојвиј "urejqns IPIJESOYH ш $93.189] р!ә1ә[ә5 pue adÁ1 [guuopida эшо т әв, 521 Aldasoro et al. Volume 85, Number 3 1998 Pome Anatomy of Rosaceae (VIN) 202 олозррју “4 no Арпдѕ ви ТА -- -- -- ѕ1п018 1e[n33.111 pue 3318] '9поо(] DUDÁLA]]DI 4 (VW) 029 олокррју *y по &pmis snp "IW 02 0€ 9c sdno13 re[nZau pue әйтеү asung ођојутгд 4 DIysDg 1008 8744 (VIA) 899 олозррју “4 “no Ápnis siy} 16 61 Ie 0L sdnoi3 лејпволл pue 3318] "SSIOg ponds J (VIN) СОРТ 19 12 омтаруу *uredg Ápnis в 16 02 02 өс sdno13 теүпдәли pue әйте] "81104 nsouids 4 (6Z9862-V И) 10199][09 uMou»un *eruouLry (УИ) 069 o4ospp]y “VIA ano Ápnis sup "IS cI 01 0€ sdno13 лејтпдолт pue әйте] тва 2170/1210 y (VIN) [}9 олозррту *suaparee) зәң пә Ápnis sim 16 02. ОР 9L sdno13 лејпдоли pue 9318] "boef sparnu g (OZELTP-VW 92621%-УЙ) $10]93[[09 uMou»un '2181090) Ápnis siy} 16 91 Ра 9с sdno13 тејпдолл pue 9318] апу 02181098 J (VIN) PE9Z 42510) 15044 [erae[so3n] | :eruopooe]y (УИ) 069 0o40spnp]y “q "ino Apnys siy} 1s OI CZ cv sdnoi3 теүпдәли pue 93.18] еа »:0fru2v2avjo 4 (VI) 9911 ]? 1? 01427$»uojy “uredg Ápnis siy} "IS — m = ва пола теүпдәи pue әёте] “т $ununuo2 а (VW) ІСІ ол05ррју “VIA no Apnys siy} “IS eI 01 09 ва пола лејпвали pue 93.18] 'euoe(] рирәд тод J (ovetLY-VIN) 10]oo[[oO имоиҹ̧ип *eruouLry Арт sum 1S 9I ре 79 запола лејпвашт pue 3318] пи гпорлидшар Y епа 71298 87454 (VIN) 802 o1ospp]y “үр пә &pmis в IS 8l Ol ср ва пола [еш ueury[ey (389) 9070198 “Yd (VIN) POS 04080р|ү sddiyd ^g 1 Y *suapae«) зәң пә &pnis sty] 1 РІ Ob 04 sdno13 jjews uosuoqoy ^W ^W (шет) оуојиха ‘yd sddiyg ^g ^f Y uosua (УИ) 012 o4ospp]y “WW "imo Ápms вц "IS ZI се 001 sdnox3 pews -q0Y “Y ^N (xuory) »dapooupjout "uq (VW) 602 ооворту “үр no Ápms sp 15 el Ob v9 sdnoi2 [jews їорлтегу (‘auda(]) nupipiapp "uq (VW) 859 олозррју *w [no &pnis sty] qs CZ сє 001 вапола [jews ‘prouyog “y 79 DUDIPLIAMDA типоца Ápms si pue (VIN) 9911 оштату “шейс (4861) ти ПИ — — — juasqe "1 22100114192. правој perpnis јемојејј ғіер jo золпос (po1o&e] (шті) (шті) (шті) sdno13 ргалојов ysa]y uoxe[ -ә|4шв "Jg ssouxonp лојошегр qua] *рәләАвүп ем цошпј ртолајос -pw ту) — рголојос рталајос вїшләр -1Чә әшод трепициођ “әү, Annals of the 522 Missouri Botanical Garden (VIN) 6£9 o4ospp]y *suapaee) зәң пә Ápnis в 1$ 0c c6 PEI sdnois рорипол pue з8лвј preugos у 72 притргу "< (VIA) 629 o40sopjy “y [no Apms siu "IS CI cer со] sdnois рорипол pue әйір| uei[oiqee) DUDISDÍDY ^e (VIN) 112 олоўоруү “WW [no {pms sių} 1 CZ 92 121 5А пола рәрипог pue әйтер — ләрчә} (гргоицос ^N '5) игигјој `ç (8661) (VIN) 09822 орәү “uredg ЧЕ 19 оловерүү "IS FI 001 OTI sdnois рорипол pue 93.18] zuen (71) DUD % 4207 “Y (VIA) 229 o4o0spp]y “+ "по Ápnis sty} "IS Ic OLI OSI sdnoi3 рәрипол pue 3318] ('зопу Y рјодатс) vyofiujo палос Duy “3qNS snquog (VIN) 629 o1osop]y ‘SSN пә &pnis sty) 15 €I 001 vel sdno2 [[eurs PRUS A у пчиошји 5 (8661) | (VW) сер o40spp]y “uredg ТЕ 19 олозерју 716 Ol 0€ 09 за пола [jews "РРРФ 5 (VIN) OSS олозррју “VIA по Apnys stu "IS €I Ot OL за пола [геш preuqogs ^N 77 sisuayadny "$ (VIA) 209 o1ospp]y *y "ino Apis sup 15 Ol 001 ZI эшн Jis шец) Y ASHYO 1152440 6 (ұй) IFS 0108DP]Y “ұр спо &pnis siy} 16 “І ce 021 за пола [jews дицооч риртрәрәѕѕә “с (VIN) ESS олозррју “VIA пә Ápnis sių} qs ZI ce 09 за пода [jews ‘IPOH ррлшиоз << (8661) (VW) сәсе opoy “шғас ЧЕ 19 оловерүү "IS OI де cc ва пола [yews “| омратто «пдлос snq4og "qns 814106 (VW) сре o40spp]y “үр "no Арпа sip 15 Ol 0с 09 paje[os! ourqejy Cqunup) »7p]joquim ^M (УИ) 999 o40spp]y ‘Y "ipo Apnis sip 15 SI 09 06 penes ҙіриу тапозрјарх sidojonjdpiy (VIN) 929 o4ospnp]y *y по Ápnis siy} IW LI ос 09 5Чпол@ теүп#әли pue з8лвј эшіхвү{ SISUALINSSN qd (VIN) 299 040sDp]y *w “no Ápnis siy} ЛИ 11 ec 09 sdno13 лејпдолл pue 9318] тезе (у "uung) оуојиха а (YW) 699 олокррју *w "ino &pnis siy} TIA — -- -- 5Чпол@ лејпдаш pue oze[ Japyay pdun2ooanud а (WIN) 289 огоѕррү “Y по (VIN) [£9 010sDp y “SUSPIRO зәң “NO Арп stu] ТІ 01 ІС 92 sdnoi3 лејпдоли pue әйте] "ueH-'uong отуѕра а (VIN) 2272 орәү ‘ureds &pnis siy} ЛИ 0c OL 0€ за пола лејпвошт pue 9318] `^вә(] DIDPLO) а рәтрпіѕ јемајвуј elep |0 золпос (рәлә^е] (шті) (шті) (шті) за пола. ргәлә|ов uso[4 џохвј, -9[ÍUIS 216 ssouxongp 19]2UBIP y¡ Sua] *po1a&epn ем uauin ртолојос -пш ТА) рәләр props вишәр -Чә әшод "penunuo) 71 AQEL 523 Aldasoro et al. Volume 85, Number 3 Pome Anatomy of Rosaceae 1998 (8661) (VW) 0821 ошоарду *uredg ЧЕ 19 o10sep[V ТА 02 09 [072 за пола [[eurs тив”) (71) $1]ри1шло] snq4og Dununuo[ “Aqns snquog (8661) (VW) 095 олозррјју “uredg ЧЕ 19 оловврју 15 9 021 01 sdno4d popunoa pue 93.18] "1 роцвашор 84406 snui4o7) "qns 814406 (8661) (VW) OFTE opay "итедс ДЕ 19 o10sep[V 15 де 021 OPI sdnois рорипол pue әЗір| тш”) (71) впрФзәшәушоуо << snpidsawepupy’) "Запа 874106 (VIN) 202 олозррју “4 ano Ápnis siy} 16 02 06 егі sdnoid popunoa pue әйте] preugog ^N 77) 12u32n4qpqpz `Ç (VW) £02 олозррту ‘tuspa sny yey [пә Ápnis sun "IS 91 96 011 sdnois рәрипол pue әде 319q8guodg пита 6 (VIN) 859 огоѕортү *у [по Kpnis sių} 1s 1Z 091 COZ sdnoi3 рорипол pue ozue[ 'ppoT (под “5 хә сүү) 971824 "S (VIA) 089 o40spp]y * спо &pnis sip 15 ct да! 8/1 sdno13 рорипол pue әле] uerpouqeg nuvloryyn `$ (VIA) 289 o40spp]y “y спо Ápnis siy} "IS 0c OZI OVI вапола popunoi pue одлеј 'sstog ("'qape']) vosnfqns "< (VW) 002 o0sDp y Мо Kpnis sum "IS Ic 001 ОРТ sdno13 рорипол pue одлеј Japysy suaosa]¡vd 'g (VIA) Рад олозпрју ‘SSN по Ápnis stu] "IS ст 001 021 ва пола рорипол рив әле лопецос (uo(q '(]) vun} 5 (VIA) 259 o40sppjy "по Ápnis stu] "IS Ic 081 OFZ sdnoi3 рорипол pue әде _ Japyay ( preugogs “Y 52) 122/8822 "S (VI) РОД олозррју < no Ápnis sių} "IS 0c 06 OPI 5Чпол@ рорипол pue з8леј рән ('гизо() voruodof `ç (VW) 159 o40spp]y *y ino Apmis в 15 сс 001 wal sdno13 рорипол pue әйтер Japyay ('preuqos ^N 72) Majswoy `$ perpnis јеџоје ји elep jo золпос (ралолвј (шті) (шті) (шті) ѕ4по28 pro1o[os ysa]y похве], -ә|8шв 216 ssouxonup лојошетр y13ua] *pa1aAen ем иәшп] ртолајос mur IA) PPS РЕН StuLIp “до әшод PO qM Annals of the 524 Missouri Botanical Garden "um (o а “ur 002 :9 “ur! QOS :g :um 00S = :SIBQ IRIS 'spra1o[os jo sdno13 рәрипо 23.18] 3UIMOYS (002 010sDp]y) sua2sajjpd “с jo yder3or1onuojoyd WHS G— `5р!әләүәв Jo sdnoi3 рарипол 9318] 3UIMOUS (pzg 010SDP]y) Лны 1410$ jo цаелволопшојоца [pondo * )— 'spra19[os jo Апола лејпволи 9318, Эшмоцв (cgp[ 719 12 OLIDADN) nsouids ү] jo ude1do1otuiojoud WAS '4— 'spra19[s jo sdnoi2 лејпдашт oiv] SUIMOUS (999 ологррју) nonus sni&q- jo udeiso1oruojoud peond() *y— “sawuod snqu0g pue 87454 ш (8) spra1o[os jo sdno13 ayy jo sudeisod ormuojoud [eoudo pue WAS СІ 240314 Aldasoro et al. 525 Volume 85, Number 3 1998 Pome Anatomy of Rosaceae re 2. SEM and = val photomic rographs of the e ue in Pyrus pomes. pa ма (Aldasoro 641) s of P. pashia udis 641) кесене the а е piens (me) of P. spinosa —A. SEM еке rograph of Pyrus yered epidermis (me) and the hypodermis (h). — B. Optical photomicrograph and the парог rmis (h). i SE M photomicrograph Navarro et al. 1405) showing the one-layered epidermis (e), the cuticle (c) and the hypodermis (h). —D. Optical photomicrograph of P. spinosa жы et al. 1405 ipe owing the one- layered epidermis (e), the cuticle (c) and the hypodermis (h). Scale bars: A = 10 um; B: 25 um; C: m; D: 2 ~ = 526 Annals of the Missouri Botanical Garden pears in the subepidermal region of the incipient lenticel. Like the phellogen, the tangential meri- stem of the multilayered epidermis divides peri- clinally, producing layers of cells that undergo a progressive exfoliation. In some cases, it was ob- served that lenticel concrescence occurred prior to the development of a multilayered cuticle. DISCUSSION The hypothesis that Pyrus and Судота are sister taxa was advanced by Rohrer et al. (1994) on the basis of a single presumed synapomorphy: a pit in the floral cup surrounding the style group. The data contributed by Campbell et al. (1995) on ITS DNA sequences also support this view. Our studies show that these genera have sclereids similar in size, structure, and arrangement, which strengthens this idea. However, several other characters uphold the continued recognition of Cydonia and Pyrus as sep- arate genera: Cydonia has pluriovulate carpels, leaves with no adaxial glands, and solitary, pink flowers. In contrast, Pyrus has biovulate carpels, adaxial leaf glands, and corymbose, white flowers. Iketani and Ohashi (1991), Sterling (1966a, b), and Kalkman (1988) proposed that Pyrus may have ranched from the ancestor of Cydonia before the latter acquired the pluriovulate condition. Thus, the previously mentioned characters would support the monophyly of Pyrus sensu Decaisne (1874). This would be of remarkable interest in subfamily Ma- loideae, the genera of which have rather few apo- morphic character states. However, our data do not support a close relationship between Pyrus and Malus, since they have different types of sclereid oups. The distribution of the multilayered pome epi- dermis in Pyrus seems to support the infrageneric classification proposed by Tuz (1972) and Browicz 3), at least in terms of the sectional division. This is interesting because, as mentioned previous- ly, some of Browicz's sectional characters, such as pedicel thickness, are variable: the pedicels of P. pyrifolia and P. pashia (sect. 2. are thicker than those of some species in sectio Some other taxa of subfamily Maloideae (Mespi- lus, Sorbus) may also have a multilayered pome epi- ermis. According to Phipps et al. (1991) and Campbell et al. (1995), Mespilus, Pyrus, and Sorbus (subg. Torminaria) are not closely related. More- over, pomes with a multilayered epidermis were not present in any of the primitive genera of Maloideae studied (i.e., Cotoneaster, Eriobotrya, Heteromeles, Photinia, ind Rhaphiolepis; primitive according to Phipps et al., 1991; Campbell et al., 1995). Con- sequently, a multilayered epidermis is most parsi- moniously viewed as derived, and it seems an in- dependently acquired character state in these genera. The adaptative role of the multilayered epi- dermis is unknown, but it may be related to seed dispersal by mammals. All pomes of subfamily Ma- loideae studied with a multilayered epidermis pre- sent traits associated with mammalian zoochory syndromes: green or brown skin inconspicuous to birds, copious lenticels permitting scent to ema- nate, seeds protected against mammal-stomach gas- tric juices by many sclereids, tannins inhibiting bacterial or fungal damage in the ground, and high fiber content (Herrera, 1989). Literature Cited ви а, 5. 5. 1986. On systematic significance of peri- 'arp anatomical characters in some members of the ge- us Crataegus (Rosaceae). Bot. Zhurn. (Moscow & Len- ingrad) 72: к [In Russian.] Aldasoro, J. Ј., С. o & F. Muñoz Garmendia. 1996. The genus Pyrui ји 1. eae) іп south-west Europe and North Africa. Bot. | Linn. Soc. 121: arro & F. Muñoz байлар: 1998 . The mon зә (Maloideae, Rosaceae) in Eu- rope and in North Africa: Morphological analysis and systematics. Syst. Bot. (in press). b. 949. The Puras Иш puzzle. Gentes 5 `2, К. ee РЯ of woody Rosaceae іп W. Asi x Шы. Kórnickie 14: 5-19. 993. Conspect ii chorology of the genus Py- Ar Pres kie 38: 17— Campbell C. S., M. J. 4. | G Baldwin & M. F. Wojc 4. 1995. Phylogenetic relationships іп Maloideae (Rosaceae): Evidence 21 seq internal transcribed spacers of nu and its тшшш with iiie en Amer. J. Bot. 82: 903-91 Candolle, A P. de. 1825. Prodromus 1 naturalis i aris. 5. Morphology and physiology el Ji pome lentic {е of Pyrus malus. Bot. Gaz. 97: Decais "e J. 7. Ге jardin fruitier du Nx Vo e 1 Par 874. Mémoirs koe la famille des Pomacées. Nouv. e h. Mus. Hist. Nat. m. 7. pl. 8-15. Fedorov, A. 5. J. Sokolov ilc Dereva i Бај SSSR, 23. Academy of Scien ress, Moscow & . Tn Russ Forman, L. Bridson. 1989. The Berol Hand- Frugivory and seed dispersal by carnivorous прати ll and associated fruit character- istics, in undisturbed Mediterranean habitats. Oikos 55 250-202. Iketani, Н. & Н. Ohashi. 1991. Anatomical structure of пи and evolution of the tribe Sorbeae in the nm y Maloideae (Rosaceae). J. Jap. Bot. 66: 319-351. Kalkman, C. 1988. The phylogeny of the 2. Bot. J. Linn. Soc. 98: 31—59. Koehne, E. 1890. Die Gattungen der Pomaceen. R. Cy Volume 85, Number 3 Aldasoro et al. 527 Pome Anatomy of Rosaceae Gaertners Verlagsbuchhandlung Hermann Heyfelder, ‚ К. В. Robertson & J. B. Phipps. 1994. Floral Berlin. morphology of Maloideae (Rosaceae) ү its systematic Kovanda, On the generic concepts in the Ma- relevance. Amer. J. Bot. 81: 574-561. loideae. 2. sn 27-34 m: x, K. 1931 753. Species ias: Impensis Lauren- tii Salvii, боска. Miller, К. Н. 1984. The multiple epidermis—cuticle com- plex of medlar fruit 4. germanica L. (Rosaceae). inn Bot. (Oxford) 53: 779-792. Olson, A. R. & T. A. Steeves. 1982. Structural changes in the етапе fruit wall of Amelanchier alnifolia. Canad. J. Bot. 60: 1880-1887. sr B., K. R. Robertson, J. R. Rohrer & P. G. Smith. 99]. Origins and evolution E subfam. Maloideae (Ro- saceae). Syst. Bot. 16: 303-3 Rehder, A. 1940. Manual of са Trees and Shrubs Hardy in North America Exclusive of the Subtropical and Warmer Temperature Regions, 2nd ed. Macmillan, New York. Robertson, K. R. 1974. The genera of Rosaceae in the southeastern United States. J. Arnold Arbor. 55: 303— 332, 344-401, 611-662. ——-, J. B. Phipps, J. В. Rohrer & P. С. Smith. 1991. A synopsis of genera in Maloideae (Rosaceae). Syst. Bot. 16: 376-394 Rohrer, J. R., K. R. Robertson & J. B. Phipps. 1991. Variation in structure among fruits of Maloideae (Ro- saceae). Amer. J. Bot. 78: 1617—1635. oideae. J. Arnold Arbor. 12 Sterling, C. 1966a. Comparative O of the carpel in the Rosaceae. VII Pom Е M Cydo- nia, md Amer. J. Bot. 53: 2 —231. ————. 1966Ь. Comparative 2. of the carpel in the 4. IX. Spiraeoideae: Quillajeae. Sorbarieae. Amer. J. Bot. 53: 951—960 Taylor, N. P. е Malus sikkimensis Rockii* Bot. Mag. 184: 168-17 Terpó, A. 2! “Маш Miller. Pp. 2. іп ^ de ird V. W^ D. M. Val- . Nh pe UE Europaea, a 2. Univ. Press, emite —— . Studies on taxonomy pu кош of Py- rus Species ‘Feddes pert. 96: 73-8 Tolivia, D. & J. Tolivia. "1987. Fasga: 1 new polychro- matic method for simultaneous and differential staining of plant tissues. J. Microscopy 148: 113-117 Tuz, А. 5. K voprosu klassifikacii poda: Pyrus L. Trudy Prikl. Bot. 46: 70-91. [In Russian.] Weber, C. 1964. The genus Chaenomeles (Rosaceae). J. A 05. . The eden дай. relationships of the Po- —22. . Chemical evidence from flavonoids relevant to the clasica of Malus species. Bot. J. Linn. Soc. 84: 31-39, Volume 85, Number 3, PP. 367-530 of the ANNALS OF я Т BOTANICAL GARDEN as published on November 23, Flora of the Venezuelan Guayana Located in the southeastern half of Venezuela, the Venezuelan Guayana is the core area of what has been called “Тће Lost World." The area is dominated by massive table mountains eds as tepuis and includes many endemic species and genera, with much of the area still n pristine condition. There are nearly 10,000 species in the flora area, and over half will be illustinied by line drawings. Volumes 3 and 4 of the Flora of the Venezuelan Guayana are now available from Missouri Botanical Garden Press: Berry, P. E., B. K. Holst, and K. Yatskievych, editors. Flora of the Venezuelan Guayana. Volume 3, Araliaceae-Cactaceae. 1997. ISBN 0-915279-46-0. 774 pp. 1113 species treated. 628 line drawings. $67.95 Volume 4, Caesalpiniaceae-Ericaceae. 1998. ISBN 0-915279-52-5. 799 pp. 1329 species treat- ed. 621 line drawings. $67.95 Also still available: Volume 1, Introduction (includes Vegetation Map and Topographical Map). 1995. ISBN 0- x 313-3. 320 pp. of text, plus 44 pp. of color plates, 10 b/w photos, 51 line drawings. $52.95 Volume 2, Pteridophytes, Spermatophytes (Acanthaceae-Araceae). 1995. ISBN 0-88192-326- 5. 706 pp. 1285 species treated. 618 line drawings. $67.95. Vegetation Map and Topographical Map, 2-map set: rolled and shipped in tube $17.00; or professionally folded $15.00. Volume 1 (& 2-map set) DIGO: > чая Make checks pavable to: Volume 2 ? LES рата Missouri Botanical Garden Press Volume 3 SOLOS oL cede 4344 Shaw Blvd. Volume 4 piv А см Аы St. Louis, MO 63110-2991 2-map set, rolled BLU 2 ле -2-map set, folded $15.00 ааа Or order using Visa or MasterCard: Add for we and handling: | Check one: Visa MasterCard Within the U.S. Card No i Exp. date for the first book PO o Сы а Signature х for each additional book 15s CN о њи phone, 314-577-9534 Canada and Mexico fax, 314-577-9591 e-mail, mbgpress@mobot.org i web site, www.mobot.org. International for each additional book аа dts for the first book $H00 E. Ship to: for each additional book 62:00 .— Street Air rates available on s erac City Total enclosed г .; State Zip CONTENTS A Taxonomic Revision of Grimmia subgenus Orthogrimmia (Musci, Соает жеше леч ез eda Jesús Muñoz Revisión Sistemática y Análisis Cladístico del Género Chaetium (Poaceae: Panicoideae: Paniceae) |... svaldo Morrone, Fernando O. Zuloaga, Mirta O. Arriaga, Raúl Pozner y Sandra S. Aliscioni Revisión del Género Cucurbitella (Cucurbitaceae) — Raúl Pozner A Taxonomic Revision of Dicoma (Asteraceae: Cichorioideae: Mutisieae) for the Dont а ARCA лас O cela rencontre те 5 Santiago Ortiz, Juan Rodríguez-Oubiña & Mesfin Tadesse A Review of the Genus Paragonia (Bignoniaceae) ________-- Warren D. Hauk Synopsis of Crataegus Series Apiifoliae, Cordatae, Microcarpae, and Brevispinae (Rosaceae subfam. Maloideae) ____- J. B. Phipps Adaptive Radiation of dp eer Gladiolus Species (Iridaceae) in Southern A S s Peter Goldblatt, John C. Manning & Peter Bernhardt Pome Anatomy of Rosaceae Subfam. Maloideae, with Special Reference to Pyrus Juan José Aldasoro, Carlos Aedo & Carmen Navarro 440 ин m RR Cover illustration. Drawing by John Myers vectes from cover м кедей; for 43rd Анын cas PARUM. Missouri Botanical Garden 222% ый O олен тін еі ней O PUT TRES. SIT ofthe _ Missouri D otanic al Volume 85, Number 4 Fall 1998 Annals of the Missouri Botanical Garden The Annals, published quarterly, contains papers, primarily in systematic botany, contributed from the Missouri Botanical Garden, St. Louis. Papers originating out- side the Garden will also be accepted. All manuscripts are reviewed by qualified, independent reviewers. Authors should write the Managing Editor for information concerning arrangements for publishing in the ANNALS. Instructions to Authors are printed in the back of the last issue of each volume. Editorial Committee Michael H. Grayum Editor, Missouri Botanical Garden Amy Scheuler McPherson Managing Editor, Missouri Botanical Garden Diana Gunter Editorial Assistant, Missouri Botanical Garden Vicki Couture Secretary Ihsan A. Al-Shehbaz Missouri Botanical Garden Gerrit Davidse Missouri Botanical Garden Roy E. Gereau Missouri Botanical Garden Peter Goldblatt Missouri Botanical Garden Gordon McPherson Missouri Botanical Garden P. Mick Richardson Missouri Botanical Garden Henk van der Werft | Missouri Botanical Garden | For subscription information contact ANNALS OF THE MISSOURI GARDEN, % Allen Marketing & Management, Р.О. Box 1897, Lawrence, KS 66044-8897. Subscription price is $125 per volume U.S., $135 Canada & Mexico, $160 all other countries. Four issues per volume. The journal Novon is included in the Subscription price of the ANNALS. L Ст. A ls _ amy.me g (editorial queries) http://www.mobot. org | © Missouri Botanical Garden 1998 . The mission of the Missouri Botanical Garden is to discover and кз knowledge boni t plants nd ; The ANNALS OF THE MISSOURI BOTANICAL GARDEN (ISSN 0026-6493) is published quar- terly by the Missouri Botanical Garden, 2345 Tower Grove Avenue, St. Louis, MO 63110. Pe- riodicals postage paid at St. Louis, MO and ad- ditional mailing offices. POSTMASTER: Send ad- | dress changes to ANNALS OF THE MISSOURI BorANICAL GARDEN, % Allen Marketing & - Management, Р.О. Вох 1897, Lawrence, KS 66044-8897. Ки their environment, in order to preserve and enrich life. 5h Ө This paper meets the requirements of ANSI/NISO 239.48-1992 (Permanence of Paper). | 2 s CP X4 da" vU, Ww) Ue yo ciis iat om ~ Volume 85 Number 4 1998 Annals of the Missouri Botanical Garden NZ AN ORDINAL CLASSIFICATION FOR THE FAMILIES OF FLOWERING PLANTS The Angiosperm Phylogeny Group! ABSTRACT Recent cladistic analyses are revealing the phylogeny of flowering plants in increasing eee and — is suppor for the monophyly of many major groups above the family level. With many elements of the major bran of phylogeny established, a revised suprafamilial classification of flowering plants becomes ud feasible 2 deal ble. Here we present a classification of 462 flowering plant families in 40 putatively monophyletic orders and a small number of кинен, informal higher groups. The latter are the monocots, commelinoids, eudicots, core eudicots, rosids including eurosids I and II, and asterids including euasterids I and II. Under these informal groups there аге also listed a number of families without assignment to order. At the end of the system is an additional list of families of uncertain position for which no firm data exist regarding placement anywhere within the system. Why rearrange families, still less formalize or- ders? Higher-level classifications, the grouping of species into families, orders, etc., are needed as reference tools not only in systematics but also in many other branches of biology. Knowledge of phy- logenetic relationships of major groups of organ- isms, that is, a phylogenetic perspective, is becom- ing increasingly important, and hence the need for a phylogenetic classification as a reference tool is also becoming imperative. Our primary focus is on orders with a secondary emphasis on families of flowering plants. The family is central in flowering plant systematics. For ex- ample, in studying an unknown plant we usually first identify it to family. The orders, on the other and, have until quite recently been of little im- portance, either being morphologically unrecogniz- able or in most cases lacking any evolutionary co- herence (Heywood, 1977; Merxmiiller, 1977). However, orders are useful in teaching, for studying ! Recommended citation, abbreviated as “APG, 1998.” This paper was compiled by Káre Bremer, Mark W. Chase, and Peter F. Stevens, equally responsible and listed here in alphabetical order only, with contributions from Arne A. Anderberg, Anders Backlund, Birgitta Bremer, Barbara G. Briggs, Peter K. Endress, Michael F. Fay, Peter Goldblatt, Mats H. С. Gustafsson, Sara В. Hoot, Walter S. Judd, Mari Kallersjé, Elizabeth A. Kellogg, Kathleen A. Kron, Donald H. Les, Cynthia M. Morton, Daniel L. Nickrent, Richard G. Olmstead, Robert A. Price, Christopher J. Quinn, James E. Rodman, Paula J. Rudall, Vincent Savolainen, br e E. Soltis, Pamela S. Soltis, Kenneth J. Sytsma, and Mats Thulin (in alphabetical order). Addresses: K. Bremer, Department of Systematic Botany, Uppsala University, Villavügen 6, 5-752 36 Uppsala, Sweden; M. W. Chase, Jodrell D. Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3DS, U.K.; P. F. Stevens, Harvard University Herbaria, 22 Divinity Avenue, Cambridge, Massachusetts 02138, U.S.A. ANN. Missouni Bor. GARD. 85: 531—553. 1998. 532 Annals of the Missouri Botanical Garden family relationships, and in positioning genera of doubtful affinity. The didactic value of suprafami- lial duas has been emphasized by various au- thors (e.g., Dahlgren, 1975; Thorne, 1976; Davis, 1978; ты, 1997). Т evident now that the phylogeny of flowering plants is being disclosed in increasing detail. Many of the orders recognized by earlier authors are not mono- phyletic, yet there is a pressing need for names to communicate the knowledge of monophyletic groupings of families that are becoming evident. With the major branching sequence of flowering plant phylogeny becoming clearer, a revised famil- ial and ordinal classification is feasible. Flowering plant classification systems from the late 1970s seemed to be stable and show substan- tial agreement, but this stability has been rudely shattered as new kinds of data and new methods of analyzing conventional data have become firmly es- tablished (Stevens, 1986). Classifications such a those by Cronquist (1981) and Takhtajan (1980). although still in frequent use, have become outdat- ed. Of more recent classifications, that by Goldberg (1986) of the dicotyledons predates the advent of molecular studies at higher levels, as does that by Dahlgren et al. (1985) of the monocotyledons. How- ever, the latter incorporated much new data and provided synapomorphy schemes for many groups. The recent system of Takhtajan (1997), although extremely elaborate, is made less useful because his value is even more his propensity for splitting often results in well- known families being dismembered, then reassem- bled as orders. Furthermore, the findings of recent molecular studies, despite being cited, have hardly influenced his classification. We conclude that there is a great need for a new, phylogenetic classification of flowering plants, pro viding names for major monophyletic groups of families. Obviously, it is not possible, nor is it de- sirable, to name all clades in the entire phylogeny. Any such complete classification would be so cum- bersome that it would be useless for general com- munication. Systematists need to come to some kind of agreement concerning which clades to rec- ognize and name, so that a reference tool of broad utility can be formulated and used to discuss di- versity. An ordinal classification of flowering plant families is here proposed for that purpose (pp. 538— 542). It recognizes a selected number of monophy- Јенс suprafamilial groups, that is, clades in the phylogeny of flowering plants that are supported by at least one, and often several, lines of evidence. These are clades to which we find it useful to refer when we communicate information about higher- level interrelationships of the flowering plants. We note that the selection of clades to be rep- resented in a formal classification is different from the procedure of naming these clades. The latter issue of biological nomenclature in phylogenetics is currently much debated (e.g., Cantino et al., 1997; de Queiroz, 1997; Lidén et al., 1997), but we have not adopted any "phylogenetic naming" sensu de Queiroz and Gauthier (1994). We operate under the current International Code of Botanical Nomencla- ture (Greuter et al., 1994) and choose to emphasize the ranks of family and order. The Linnaean cate- gories serve as a convenient mnemonic device for remembering hierarchical relationships, but it should of course be realized that groups of the same rank are evolutionarily non-comparable units un- less they are sister groups. There are noteworthy problems when establish- ing the names for taxa at ordinal and other higher taxonomic levels. Until recently, little attention has been paid to the nomenclature at these levels, and our knowledge of the early literature in which such names were used is imperfect. This situation has in considerable part been rectified by Reveal's (1998) Herculean labors. The principle of ie is not mily, al- though authors are exhorted "generally" to follow this principle (Greuter et al., 1994). We have tried to balance priority with general usage when assign- mandatory for taxa above the rank of ing names to orders, but even if future bibliograph- ic work discloses earlier ordinal names, changes are not mandate Which clades dodi be recognized in classifi- cation, or in our case, how should the orders be circumscribed? Given the primary principle of monophyly, that of recognizing clades and not grades in classification, there are nevertheless many considerations to be taken into account when circumscribing taxa at ordinal as well as all other hierarchical levels above that of species. Classifi- cation is not only a matter of grouping according to the principle of monophyly, but it is also a matter of communication (note that whatever philosophy of naming is adopted, there has to be some consensus as to the clades we are going to use in general botanical communication). For us, this raises the question of ranking, that is, after having selected clades in the phylogeny to be named, they have to be assigned an appropriate place in the hierarchy, in our case, family and order (e.g., Backlund & K. Bremer, 1998; Stevens, 1998). In choosing between alternative circumscriptions it is desirable to rec- ognize groups that are well supported. It is also useful to select groups that have some kind of eas- ily observed morphological synapomorphies, al- though this may be difficult at the ordinal level and Volume 85, Number 4 1998 Angiosperm Phylogeny Group Ordinal Classification even sometimes at the family level. Synapomor- phies also often include (sometimes exclusively) anatomical, biochemical, and developmental char- acters. Many of our ordinal names are already well es- tablished and used in earlier classifications and systematic treatments. So far as they represent monophyletic groups, we retain well-known orders in the interest of preserving stability. In other cases, the size of the orders comes into consideration. However, what is reasonably broad circumscrip- tion? From the point of view of memorization of names, groups of 2–6 or a few more would seem 10 be ideal, and there is evidence that systematists in the past have commonly recognized groups of this size (Stevens, 1997). However, with the discoveries of new species, genera, and families, the sizes of genera, families, and orders have increased, and many orders now comprise 10—20 families, or even more. Other orders contain a few families only, and if there are only two or three families in an order, "one is not far from leaving the families unplaced" (Copeland, 1957). Concerns about the doubtful val- ue of recognizing similarly small groups have also been expressed by others (e.g., Burtt, 1977). Nev- ertheless, we have chosen to recognize a number of small orders because these represent clades for which monophyly and relationships are well sup- ported, and this better conveys the interrelation- ships of the families included rather than leaving them unclassified to order. In general, we adopt a broad circumscription of the orders. We recognize 462 families and 40 or- ders of flowering plants. Cronquist (1981) recog- nized 321 families and 64 orders, Thorne (1992) 440 families and 69 orders, and Takhtajan (1997) no less than 589 families in 232 orders. Our wider ordinal circumscription is not because finer details of the phylogeny within the orders are as yet un- clear, but because we think the classification will be more useful with a limited number of larger or- ders. As we develop more firmly supported phylog- enies within and among orders, groups at the in- fraordinal and supraordinal levels can be recognized. Hence we anticipate that there will be little need to change the circumscription of the or- ders recognized here, except for inclusion of yet unassigned families of unknown systematic position and the transfer of misplaced families. Additional orders may have to be recognized as the phyloge- netic relationships of families that are not yet placed are clarified. Discussion as to whether a widely accepted monophyletic group should be a superorder, order, suborder, or family is largely vac- uous because this will always be an arbitrary de- e - sion. Takhtajan (1997) opted in favor of “smaller, more natural families and orders, which are more coher- ent and better-defined, where characters are easily grasped, and which are more suitable for informa- tion retrieval and phylogenetic studies, including cladistic analyses (e.g., because it reduces poly- morphic codings)." However, the size of a group has nothing to do with its “naturalness.” For a smaller group, one will often be able to say more about all of its constituent members, and so the characters may be more easily grasped. However, segregates of well established monophyletic families like Ru- biaceae (Gentianales) or Asteraceae (Asterales) would by Takhtajan's generalization also be more natural; by this criterion, the smaller the group, the more natural it will necessarily be, so there is no ranking criterion to be derived from “naturalness.” If by “more natural” is meant “has more synapo- morphies” then this, too, is incorrect; the number of synapomorphies is not connected to the size of the group or the hierarchical level at which it is > n our classification, these considerations have had is impact. The principle of monophyly in combination with the desirability of maintaining al- ready well established and familiar entities has largely formed the ordinal classification. Monofam- ilial orders (and monogeneric families) are avoided as much as possible, minimizing redundancy in classification. In a few cases we have, however, rec- ognized some monofamilial orders (Ceratophyllales, Acorales, Arecales) because these are sister groups of more than one other order. Hence, the families of these monofamilial orders cannot be included in any other order without violating monophyly. The principle of monophyly in combination with the mandatory usage of the family category (Greuter et al., 1994) may lead to the recognition of many small families. For example, in Dipsacales, if Dip- sacaceae and Valerianaceae are to be retained as families separate from Caprifoliaceae, the principle of monophyly requires the recognition also of Dier- villaceae, Linnaeaceae, and Morinaceae (Backlund & K. Bremer, 1998; Backlund & Pyck, 1998). This is because each of these latter families is the sister group of more than one family so they cannot be merged with any other family without violating monophyly. Similar considerations apply at the or- dinal level. Unfortunately, no absolute guidelines as to reasonable practice can be offered, but we simply observe that caution is always in order. In other cases there are small families that may be reduced to synonymy of their sister group if the Annals of the Missouri Botanical Garden latter consists of a single family. Examples are Ca- bombaceae, which may be merged with Nymphae- aceae, and Kingdoniaceae, which may be merged with Circaeasteraceae (Ranunculales). Such com- monly recognized families that nevertheless may be merged with their sister family are in our classifi- cation placed within square brackets below the family with which they may be merged (in Ran- unculales either Fumariaceae or both Fumariaceae and Pteridophyllaceae may be merged with Papav- eraceae; alternatively, either Pteridophyllaceae or both Fumariaceae and Pteridophyllaceae may be retained as distinct). We do not attempt to thoroughly revise family circumscriptions. In general we follow recent au- thors and attempt to recognize as many monophy- letic families as possible. It should be emphasized, however, that following additional investigation some families listed below may be shown to be non- monophyletic; revised circumscriptions, either by merging or splitting, into monophyletic taxa are not yet possible given our current knowledge. Exam- ples are Euphorbiaceae and Flacourtiaceae of Mal- pighiales (КаПегзјб et al., 1998) and several fami- lies of Myrtales (Conti et al., 1996; Gadek et al., 1996) and core Caryophyllales (which comprise Achatocarpaceae, Aizoaceae, Amaranthaceae, Bas- ellaceae, Cactaceae, Caryophyllaceae, Didierea- ceae, Molluginaceae, Nyctaginaceae, Phytolacca- ceae, Portulacaceae, Sarcobataceae, апа Stegnospermataceae; Hershkovitz € Zimmer, 1997). Other probably non-monophyletic families that cannot yet be recircumscribed are Boragina- ceae (euasterids I; Chase et al., 1993), Scrophular- iaceae (Lamiales; Olmstead & Reeves, 1995), and Santalaceae (Santalales; Nickrent & Duff, 1996; Nickrent et al., 1998). Brassicaceae (Brassicales) include also the former, paraphyletic Capparaceae (Brassicaceae sensu stricto being nested inside Capparaceae; Judd et al., 1994; Rodman et al., 1996). A supposedly parallel case comprises Api- aceae and Araliaceae (Apiales), since the former have been assumed to be nested inside the latter (Plunkett et al., 1996). However, with a transfer of Hydrocotyloideae from Apiaceae to Araliaceae, it seems that two monophyletic families can be rec- ognized, only a few genera remaining unplaced (Plunkett et al., 1997). Delimitation of Bombaca- ceae, Malvaceae, Sterculiaceae, and Tiliaceae (Malvales) is problematical, and only Malvaceae are monophyletic (Alverson et al., 1998; Bayer et al., . Here all four are treated together as a single monophyletic family, Malvaceae sensu lato (Judd & Manchester, 1997). Our proposed classification is a modification of that conceived by Bremer et al. (1995, 1996, 1997) and since 1996 available on the Internet (Bremer et al., 1998). This classification is based on various recently published mostly molecular phylogenetic analyses (e.g., Chase et al., 1993; Chase et al., 1995; Bremer et al., 1994; Struwe et al., 1994; Na- dot et al., 1995; Nickrent & Soltis, 1995; Soltis et al., 1995; Gadek et al., 1996; Gustafsson et al., 1996; Morton et al., 1996; Soltis & Soltis, 1997; Soltis et al., 1997; Anderberg et al., 1998; Back- lund & B. Bremer, 1998; Bakker et al., 1998; Käl- lersjó et al., 1998; Soltis et al., 1998; Thulin et al., 1998; further references above). The major differ- ences are in the expansion of Alismatales (includ- ing also Araceae), Caryophyllales (including Dro- seraceae, Nepenthaceae, Polygonaceae, Plumbagi- naceae, and several other families outside the tra- ditional, core Caryophyllales), the recognition of a comparatively widely circumscribed Rosales (in- cluding Rhamnaceae, Urticaceae, Moraceae, and their allies), in the addition of a number of smaller orders (Ceratophyllales, Acorales, Arecales, Prote- ales, Garryales, Aquifoliales), and in the deletion of a few others (Aristolochiales, Nymphaeales, Bro- meliales, Trochodendrales, Zygophyllales). Mono- cots and eudicots are not formally ranked and named because it is not yet clear at which level they should be recognized. The same problems oc- cur with commelinoids (a phylogenetically derived subgroup of monocots) and with rosids and asterids (subgroups of eudicots), although these are com- monly known as subclasses чш Rosi- dae, and Asteridae, respectiv Well supported ordinal к кышы аге shown in Figure 1. Interrelationships among the basal branches of the tree and the position of the root of the flowering plant phylogeny remain elu- sive. Within the eudicots there is increasing sup- port for a large subgroup with predominantly pen- tamerous and isomerous flowers, the core eudicots, mainly comprising Caryophyllales, Santalales, Sax- ifragales, rosids, and asterids. Rosids and asterids each comprise two large subgroups, eurosids I and II and euasterids I and П, also receiving increasing support as monophyletic. These correspond to the similarly numbered rosid and asterid clades of Chase et al. (1993) Under each of the supraordinal groups of mono- cots, commelinoids, core eudicots, rosids, etc., there are a number of families listed without as- signment to order. These families are known to be- long within the major group under which they are listed, but their ordinal position is still uncertain. Similarly, Amborellaceae, Austrobaileyaceae, Ca- nellaceae, etc., are listed at the beginning because Volume 85, Number 4 Angiosperm Phylogeny Group 535 1998 Ordinal Classification Ceratophyllales Lau 100/75 rales 100/54 ,, sat Piperales Acorales Alismatales Asparagales 99/08 96/100 99/68 ne s}ovouow |& commelinoids мын area suuedsoibue E Ranunculales Proteales Caryophyllales Santalales Saxifragales -— sjooipne|§ ае Geraniales » | мо. Malpighiales 77/55 ^ Oxalidales Fabales sisi 68- 10077 eurosids | 4/- 1 60/- ми жива 5 е SpISOJ | Brassicales 94 г T cn Malvales “и Sapindales eurosids 11 Sjooipne 3109 8 3 Cornales Garryales Gentianales = L euasterids | Spuojse | euasterids 11 = ewe Figure 1. пе pibe s а of the orders of flowering plants, compiled from recent cladistic analyses cited in dE text. Jackknife support is given on the branches (a dash for values < 50%), first jackknife values from о of 545 $e eas ‘of the diy atpB, and 185 rDNA genes (D. E. Soltis, M. W. Chase, P. S. Soltis, D. Albach, M. Sav en, M. Zanis & J. S. Farris, unpublished, in prep.) and second jackknife values from analysis of 05 rbcL sequences s (Kallersjó B al., 1998). 536 Annals of the . Botanical Garden they belong neither in any of the phylogenetically “basal” orders at the beginning nor in the monocots or eudicots. Furthermore, families listed directly under monocots without an order are monocots but not commelinoids, and families similarly listed di- rectly under eudicots and core eudicots are eudi- cots or core eudicots, respectively, but neither ros- ids nor asterids. At the end of the system is an additional list of families of uncertain position. Most of these are probably eudicots (including core eudicots, rosids, and asterids), but so far there are no firm data supporting their placement anywhere within the eudicots. Literature Cited d W. S., K. G. Pai D. A. жоне. M. W. Chase, 5. M. Swensen, R. McC 1 sma. 1998. Cir- cumscription of the Malvales and «тат to other Rosidae: Evidence from rbcL sequence data. Amer. J. Bot. 85: 876-877. Anderberg, A hl & M. Killersjó. 1998. Phylo- genetic interrelationships in the Primulales inferred from rbcL sequence data. Pl. Syst. Evol. 211: 93-102. & B. Bremer 1998. Phylogeny of the Aster- idae s. str. based on rbcL sequences, with particular reference to the Dipsacales. Pl. Syst. Evol. 207: 225- 254. mer. 1998. To be or not to be—Prin ples of c lif 'ation and monotypic plant families. Tax- on 47: 391—400. & N. Pyck. 1998. Diervillaceae and MA two new families of caprifolioids. Taxon 47: € Bakker, F. = D. D. Vassiliades, C. Morton & V. Savola Phylogenetic relationships of Biebersteinia Soka А eae) inferred 8 аг and atpB se- quence comparisons. . 127: 1 49- 158. Bayer, C., M. F. Fay, A. Y. de 54. E Savolainen, C. M. Morton, K. Kubitzki & M. W. Chase. 1999, Support for an expanded concept of Malvaceae within a recir- cumscribed order Malvales: A combined analysis of plastid atpB апа rbcL DNA sequences. Bot. J. Linn. Soc. [in press]. Bremer, н К. С. Olmstead, L. Struwe & J. A. Sweere. cL sequences support exclusion of Retzia, Des- fontainia, and Nicode 7 from the Gentianales. РІ. iind ie €— 213-23( Breme mer & “4 . 1995, 1996, 1997. In tin г Phylogeny Ee с Ра of Flowering Plants. lst Les 3rd eds. Compendium, Uppsala Uni- veis Uppea - = —— Classification of flow- ering plants: Internet http: - systbot.uu.se/classifi- cation/overview.htm 971. Classification above the . às ex- ith pa de тот sed groups. /n K. Kubitzki (editor), Flowering plants: а lution and classification of higher categories. Pl. Evol. Suppl. 1: 97-109. Cantino, P. D., R. G. Olmstead & S. J. Wagstaff. 1997. A comparison of phylogenetic nomenclature with the cur- rent system: А botanical case study. Syst. Biol. 46: 313— Chase, M. W., D. E. Soltis, R. G. Olmstead, D. Morgan, D. H. Les, в р. Mishler, M. R. Duvall, R. A. Price, . Karol, W. D. Clark, M. R. K. Jansen, K.-J. Kim, C. F. Wimpee, J. F. Sith. у. R. Furnier, S. H. Strauss, Q.- Y. Xiang, G. M. Plunkett, P. S. Soltis, 5. M. Swensen, S. E. Williams, P. A. к C. J. Quinn, L. E. Eguiarte, E. rara erg, G. H arn, Jr., S. W. Graham, S. C. H. Barrett, i die та & V. A. Albert. 1993. Phylo- genetics a seed plants: An analysis of nucleotide se- quences es the plastid gene rbcL. Ann. Missouri Bot Gard. 80: 580. -----, D. 22. P. Wilkin & P. J. Rudall. 1995. Monoc ‘ot systematics: A combined analysis. Pp. 68 730 in P. J. Rudall, P. J. Cribb, D. F. Cutler & C. J. Humphries (editors), Monocotyledons: Systematics and Evolution. Royal Botanic Gardens, Kew. Conti, E., A. Lit & K. J. Sytsma. 1996. Circumscription of Myrtales and their relationships to other ow Ev- idence from rbcL sequence data. Amer. J. Bot. 83: 221— Сарыал; H. F. 1957. 11 cast of a system of the dicot- yledons. Madroño 14: 1- Cronquist, A. 1981. An а System of Classification of dad A 7 Columbia Univ. Press, Ne Dahlgren, R. A system of classifies sation of the angiosperms to ү used to demonstrate the distribution of characters. Bot. Not. 128: 119-147. . H. T. Clifford & P. F. Yeo. 1985. The Families of the Monocotyledons. а E lin Davis, P. H. 8. The moving staircase: An analy: sis of taxonomic sis and affinity. Notes Roy. Bot. Gard. Ed- inburgh 36: ж ES De Queiroz, K. . The Linnaean hierarchy and the 2. “of taxonomy, with emphasis on the problem of аря Aliso 15: 125-144. . Gauthier. 1994. Toward a phylogenetic sys- tem of biological nomenclature. Trends Ecol. Evol. 9: 2 2 = Cadok, P. A., E. 5. Fernando, C. J. Quinn, S. B. Hoot, T. Terrazas, C. Sheahan & M. W. Chase. 1996. Sap- indales: Molec ‘ular delimitation and infraordinal groups. Amer. Д ue 83: 802-811. P. G. Wilson & C. J. Quinn. 1996. Phylogenetic rec ойнор in Myrtaceae using matK, with particular reference to the position of 2 and Heteropyxis. Austral. Syst. Bot. 9: 283 Goldberg. A. 1986. Cla i il evolution, and phylog- eny pA the families of dicotyledons. Smithsonian Contr. Bot. 1-314. ;reute . R. Barrie, H. M. Burdet, W. С. Chaloner, V. Demoulin, D. L. Hawksworth, P. M. 7. D. Н. Nicolson, P. C. Silva, P. Trehane & J. McNeill. 1994. rM Code "i Botanical Nomenclature. Regnum . 131. Қ. М. ‚ А. Backlund & B. Bremer. 1996. Picloros of the dumis sensu lato based on rbcL ш with particular reference to the Goodeni- e. Pl. Syst. Evol. 199: 217-242. 2 M. A. & E. A. Zimmer. 1997. On the evo- lutionary origins of the cacti. Taxon 46: 217-232. Heywood, V. H. 1977. Principles and concepts in the clas- sification of higher taxa. /n K. Kubitzki (editor) Flow- ering plants: Evolution and classification of higher cat- egories. Р]. Syst . Evol. Sup : 1-12. Judd, W. S. & S. R. Manchester. 1997. Circumscription Volume 85, Number 4 Angiosperm Phylogeny Group Ordinal Classification of Malvaceae (Malvales) as determined by a preliminary cladistic analysis of morphological, anatomical, paly- MN and chemical characters. Brittonia 49: 384— 405 W. Sanders & M. J. Donoghue. 1994. An сы а pairs: 2” phylogenetic Mp Harvard Pap. Bot. 5: 1- sr M., Ј. S. Faris. M. W. Chase, B. Bremer, M. F. C. J. Humphries, G. Petersen, O. Seberg & K. 1. 1998. Simultaneous parsimony jackknife anal- ysis of 2538 rbcL DNA sequences reveals support for major clades of green plants, land plants, seed plants, and flowering plants. Pl. Syst. Evol. [in press]. Lidén, M., B. Oxelman, A. Backlund, L. Andersson, B. ‚ Persson, Merxmüller, H. PE Summary lecture. /n К. Kubitzki (editor), Flowering plants: Evolution and classification of higher categories. E Syst. Evol. iud 1: 2. Могіоп, С ‚М. е, К. А. Кго S. M. Swensen. 1996. A ba de Burch of de м! of the order rn based upon rbcL sequence data. Syst. Bot. 567—586 Nadot, a Є Bitt ck Carter, R. Lacroix & B. 1995. À 1 analysis of 0 oe on the chloroplast gene rps4 using parsimony and a new numerical phenetics method. Molec. Phylog. Evol. 4: 57-282. Nickrent, D. L. & J. R. Duff. 1996. Molecular studies of parasitic plants using ribosomal RNA. Pp. 28-52 in M. Moreno, J. E Cubero, selman & C. Parker bans ces in Paras Plant Кош | dac i Ando lieti: iB General de Investigac ión a usd in. = & D. E 1995. небы of angiosperm 2 = nuclear 188 "DNA Sed A sequenc- s. ri Bot. Gard. 82: D jJ. R Duff, A. E. Colwell, "Walle, | ung, K. E. Steiner & C. W. dePamphilis pu та те аг К апі е опагу — es of para- sitic plants. 11-241 in D. E. Soltis, P. S. Soltis & J. J. Doyle Aul is Мак: ene 5 7 Plants II: Pappe Bosto P. A. Reeves. 1995. Evidence for the polyphyly of he Scrophulariaceae based on chloroplast rbcL and ndhF sequences. Ann. Missouri Bot. Gard. 82: dece G. M., D. E. m & P. S. Soltis. 1996. Higher relationships of Apiales (Apiaceae and Araliaceae) d “ phylogenetic analysis of rbcL sequences. Amer. J. B : 399—415. Classification of the relstioniblp between Apiaceae and Araliaceae based on matK and rbcL sequence data. Amer. J. Bot. 84: 565- 580 Reveal, J. L. 1998. Indices nominum supragenericorum plantarum vascularium. Internet. http://www.inform.umd. edu/PBIO/. dp man html. м... Е., . Karol, R. А. Price & K. J. Sytsma. 996. Senis mbrphology, and Dahlgren’s expanded СЕ Capparales. Syst. Bot. 21: 289—307. Soltis, D. E. & P. S. Soltis. 1997. пе relation- ships in Sa кызыш. gn sensu lato: А comparison of to- pologies based оп 185 rDNA апа rbcL деді Беј Amer. J. Bot. 84: 5 2 , D. R. Morgan, S. M. Swensen, B. С. Mullin, J. M. Dowd & P. G. Martin. 1995. Chloroplast gene sequence data suggest a single origin of the pre- sperms. Proc. Nat y M S. B. Ноо! & С phylogenies using parsimony: An using three large DNA data sets for angiosperms. Syst. Biol. 47: 32-42 , D. L. rs L. A. Johnson, W. J. . B. Hoot, J. A. Sweere, R. K. Kuzoff, K. A. Kron. P W. Chase, S. M. Swensen, E. A. Zimmer, C. Shu-Miaw, L. J. Cilliespie, W. J. Kress & K. J. Sytsma. 1997. Angiosperm phylogeny inferred from 185 ribo- somal DNA sequences. Ann. Missouri Bot. Gard. 84: 1—49. Stevens, Р. К. 1986. Evolutionary classification in botany, 1960-1985. J. Arnold Arbor. 67: 313-339. . 1997. How to interpret botanical classifications: та from history. Bioscience 47: 250—250. ----- 1998. What kind of classification should the "—€— taxonomist use to be saved? Pp. 295-319 in J. Dransfield, M. J. E. Coode & D. A. Simpson (editors), Plant Diversity in Malesia III. Royal Botanic Gardens, Ke ew. Struwe, L., V. A. Albert & B. Bremer. 1994. Cladistics and c level classification of the Gentianales. Cla- distics 10: 175-206 Takhtajan, A. 1980. Outline of the PM 'ation of flow- ering ee jones: нун. Bot. Rev. 46: 225-359. ersity and а Ё Flowering Plaati Река Univ. Press, New Y Thorne, R. F. 1976. A Bc classification of the Ana Evol. Biol. 9 ——. An updated po ган“ classification of flowering plants. Aliso 13: 365-389. РЕН M., remer, J. Richardson, J. Niklasson, M. F. Fay & M. W. Chase. 1998. Family relationships of the enigmatic rosid genera Barbeya and Dirachma from the Horn of Africa region. Pl. Syst. Evol. 213: 103-119. 538 Annals of the Missouri Botanical Garden CLASSIFICATION OF FLOWERING PLANTS Amborellaceae Limnocharitaceae Austrobaileyaceae Posidoniaceae Canellaceae Potamogetonaceae Chloranthaceae Ruppiaceae Hydnoraceae Scheuchzeriaceae Illiciaceae Tofieldiaceae Nymphaeaceae Zosteraceae [+ Cabombaceae] Rafflesiaceae Asparagales Bromhead Schisandraceae Agapanthaceae Trimeniaceae Agavaceae Winteraceae Alliaceae Amaryllidaceae Ceratophyllales Bisch. Anemarrhenaceae Ceratophyllaceae Anthericaceae Aphyllanthaceae Laurales Perleb Asparagaceae Atherospermataceae Asphodelaceae Calycanthacea Asteliaceae Gomortegaceae Behniaceae Hernandiaceae Blandfordiaceae Lauraceae Boryaceae Monimiaceae Convallariaceae Siparunaceae Doryanthaceae Hemerocallidaceae Magnoliales Bromhead Herreriaceae nnonaceae Hesperocallidaceae Degeneriaceae Hyacinthaceae Eupomatiaceae Hypoxidaceae Himantandraceae Iridaceae Magnoliaceae Ixioliriaceae Myristicaceae anariaceae Laxmanniaceae Piperales Dumort. Orchidaceae Aristolochiaceae Tecophilaeaceae Lactoridaceae Themidaceae Piperaceae Xanthorrhoeaceae Saururaceae Xeronemataceae MONOCOTS Dioscoreales Hook. f. Corsiaceae urmanniaceae Japonoliriaceae Dioscoreaceae Nartheciaceae Taccaceae Petrosaviaceae Thismiaceae Triuridaceae Trichopodaceae Acorales Reveal Liliales Perleb Acoraceae Alstroemeriaceae ampynemataceae Alismatales Dumort. Colchicaceae Alismataceae Liliaceae Aponogetonaceae Luzuriagaceae Araceae Melanthiaceae Butomaceae Philesiaceae Cymodoceaceae Ripogonaceae Hydrocharitaceae Smilacaceae Juncaginaceae Volume 85, Number 4 1998 Angiosperm Phylogeny Group Ordinal Classification 539 CLASSIFICATION OF FLOWERING PLANTS (cont'd Pandanales Lindl. Cyclanthaceae Velloziaceae COMMELINOIDS Abolbodaceae Bromeliaceae Dasypogonaceae Rapateaceae Arecales Bromhead Arecaceae Commelinales Dumort. Philydraceae Pontederiaceae Poales Small Anarthriaceae Centrolepidaceae Cyperaceae Ecdeiocoleaceae Eriocaulaceae Flagellariaceae Xyridaceae Zingiberales Griseb. Cannaceae Costaceae — Lowi Ке ај Musaceae Strelitziaceae Zingiberaceae EUDICOTS Buxaceae Didymelaceae Trochodendraceae [+ Tetracentraceae] Proteales Dumort. Nelumbonaceae Platanaceae Proteaceae Ranunculales Dumort. eae [+ Kingdoniaceae] Eupteleaceae Lardizabalaceae Menispermaceae Papaveraceae t Fumariace ea [+ Pteridophyllaceae] R anunculaceae CORE EUDICOTS Aextoxicaceae Berberidopsidaceae Dilleniaceae Gunneraceae 1... Vitac Caryophyllales Perleb rpaceae Caryophyllaceae Didiereaceae Dioncophyllaceae Droseraceae Drosophyllaceae Frankeniaceae Molluginaceae Rhabdodendraceae apris и Simmondsiace Stegnospermatacese eae Tamaricac Santalales Dumort. Olacaceae Opiliaceae 540 Annals of the Missouri Botanical Garden CLASSIFICATION OF FLOWERING PLANTS Loranthaceae Fabales Bromhead Misodendraceae Fabaceae Santalaceae Polygalaceae Quillajaceae Saxifragales Dumort. Surianaceae Altingiaceae Cercidiphyllaceae Fagales Engl. Crassulaceae Betulaceae Daphniphyllaceae Casuarinaceae Grossulariaceae Fagaceae Haloragaceae ш Hamamelidaceae Myric Iteaceae Nothofagaceae Paeoniaceae Rhoipteleaceae Penthoraceae Ticodendraceae Pterostemonaceae Saxifragaceae Malpighiales Mart. Tetracarpaeaceae chariaceae Balanopaceae ROSIDS Caryocaraceae Aphloiac 2 Crossosomataceae Clusiace Ixerbacea Doe 78 кен Erythroxylaceae Picramniaceae Euphorbiaceae Podostemaceae Euphroniaceae Stachyuraceae Flacourtiaceae Staphyleaceae Goupiaceae Tristichaceae Hugoniaceae Zygophyllaceae Humiriaceae Irvingiaceae Geraniales Dumort. Ixonanthaceae Francoaceae Lacistemataceae Geraniaceae Linaceae E ылады Malesherbiaceae Greyiac Malpighiaceae aceae Medusagynaceae Melianthaceae Ochnaceae Vivianiaceae Pandaceae Passifloraceae EUROSIDS I Putranjivaceae Celastraceae Quiinaceae Huaceae Rhizophoraceae Parnassiaceae Salicaceae [+ Lepuropetalaceae] Sc cs eae Stackhousiaceae Trigoniac urneraceae Violaceae Cucurbitales Dumort. ale AE | | Begoniac Oxalidales Heintze ies Cephalotaceae Corynocarpaceae 1. Cucurbitacea unoniaceae Datiscac Elaeocarpaceae Tet ко а Oxalidaceae Tremandraceae Volume 85, Number 4 1998 Angiosperm Phylogeny Group Ordinal Classification CLASSIFICATION OF FLOWERING PLANTS э Rosales Perleb Elaeagnaceae Moraceae Rhamnaceae Urticaceae EUROSIDS II Tapisciaceae Brassicales Bromhead Akaniaceae [ + Bretschneideraceae] їасеае Brassicaceae Caricaceae Emblingiaceae Gyrostemonaceae Koeberliniaceae Limnanthaceae Moringaceae Pentadiplandraceae Resedaceae Salvadoraceae Setchellanthaceae Tovariaceae Tropaeolaceae Malvales Dumort. ixaceae [+ Diegodendraceae] Cistaceae Cochlospermaceae Dipterocarpaceae aceae Muntingiaceae Neuradaceae Sarcolaenaceae Sphaerosepalaceae Thymelaeaceae Myrtales Rchb. rypteroniaceae Oliniaceae Onagraceae Penaeaceae Psiloxylaceae Rhynchocalycaceae Vochysiaceae Sapindales Dumort. Anacardiaceae Biebersteiniaceae e Nitrariaceae [+ Peganaceae] Rutaceae Sapindaceae Simaroubaceae ASTERIDS Cornales Dumort. Hydrangeaceae Hydrostachyaceae Loasaceae Ericales Dumort. Diapensiaceae ceae Ebena Lecythidaceae Marcgraviaceae Ternstroemiaceae Tetrameristaceae Theaceae Theophrastaceae EUASTERIDS I Vahliaceae 542 Annals of the Missouri Botanical Garden CLASSIFICATION OF FLOWERING PLANTS ont'd.) Garryales Lindl. Apiales Nakai Aucubaceae Apiaceae Eucommiaceae Araliaceae Garryaceae Aralidiaceae Oncothecaceae Griseliniaceae Melanophyllaceae Gentianales Lindl. Pittosporaceae Apocyna Torricelliaceae Gelsemiaceae Gentianaceae Aquifoliales Senft Loganiaceae Aquifoliaceae Rubiaceae Helwingiaceae Phyllonomaceae Lamiales eme Acanthac Asterales Lindl. 22-2 Iseuosmiaceae Bignoniaceae Argophyllaceae Buddlejaceae Asteraceae Byblidaceae Calyceraceae Cyclocheilaceae Campanulaceae Gesneriaceae [+Lobeliaceae] Lamiaceae Carpodetaceae Lentibulariaceae Donatiaceae Myoporaceae Goodeniaceae Oleace Menyanthaceae Orobanchaceae Pentaphragmataceae Paulowniaceae Phellinaceae Pedaliaceae Rousseaceae [+ Martyniaceae] Stylidiaceae Phrymaceae | Plantaginaceae Dipsacales Dumort. Schlegeliaceae Caprifoliaceae Scrophulariaceae Di ill Stilba Dipsacaceae Tetrachondraceae Linnaeaceae Verbenaceae Mo nacea Valerianaceae Solanales Dumort FAMILIES OF UNCERTAIN POSITION Convolvulaceae Hydroleaceae Balanophoraceae Montiniaceae Bonnetiaceae Solanaceae Cardiopteridaceae Sphenocleaceae Ctenolophonaceae Cynomoriaceae EUASTERIDS II Cytinaceae Adoxaceae Di oe Bruniaceae Elatina Carlemanniaceae aa cee Columelliaceae Hoplesti 4. [+ Desfontainiaceae] Kaliphoraceae Eremosynaceae Lepidobotryaceae Escalloniaceae Lissocarpacea Icacinaceae Lophopyxidaceae Polyosmaceae Medusandraceae Sphenostemonaceae Metteniusaceae Tribelaceae Mitrastemonaceae Paracryphiaceae Volume 85, Number 4 1998 Angiosperm Phylogeny Group Ordinal Classification 543 CLASSIFICATION OF FLOWERING PLANTS (cont'd.) Pentaphylacaceae Peridiscaceae Plagiopteraceae Pottingeriaceae Sladeniaceae Strasburgeriaceae Tepuianthaceae ORDINAL SYNONYMS Acanthales Lindl. — Lamiales Acerales Lindl. 5 Actinidiales Takht. ex Reveal — Ericales Adoxales Nakai - not accepted, family under euasterids II Aesculales Bromhead — Sapindales Agavales Hutch. — Asparagales Alliales Traub sparagale Alstroemeriales Hutch. Amaranthales Dumort. — Caryophyllales Amaryllidales Bromhead — Asparagales Ambrosiales Dumort. — Asterales Ammiales Small Amomales Lindl. — Zingiberales Ancistrocladales Takht. — Caryophyllales Annonales Lindl. — Magnoliales Anthobolales Dumort. — Santalales Apocynales Bromhead — Gentianales Aponogetonales Hutch. — Alismatales Arales Dumort — Alismatales Araliales Reveal piales Aralidiales Takht. ex Reveal piales Aristolochiales Dumort. Piperales Asarales Horan. — Piperales Asclepiadales Dumort. Asteliales Dumort. — Asparagales Atriplicales Horan. — Caryophyllales Aucubales Takht — Garryales Austrobaileyales Takht. ex Reveal - not accepted, family at beginning of system Avenales Bromhead — Poales Balanopales Engl. — Malpighiales Balanophorales Dumort. - not accepted, family unplaced Balsaminales Lindl. ricales ние Takht. & Reveal — Rosales Batales Engl. — Brassicales Begoniales Dumort. — Cucurbitales Berberidales Dumort. — Ranunculales pi od Biebersteiniales Takht. Bignoniales Lindl. = Lamiales Bixales Lindl. = Malvales Boraginales Dum - not accepted, family under euasterids I Brexiales Lindl. - not accepted, family under eurosids I Bromeliales Dum - not и. family under commelino Bruniales Dus - not ый кыли family under euasterids II Brunoniales Lindl. — Asterales Burmanniales Heintze — Dioscoreales Burserales Baskerville — Sapindale Butomales Hutch. — Alismatales 544 Annals of the Missouri Botanical Garden CLA SSIFICATION OF FLOWERING PLANTS cont Buxales Takht. ex Reveal - not accepted, family under eudicots Byblidales Nakai ex Reveal 1 = Lamiales rt. — Caryophyllales Callitrichales Dumort. — Lamiales Calycanthales Mart. — Laurales —— еи ех Reveal ети Rchb. — Asterales Canellales Cronquist - not accepted, family at beginning of system быб Dumort. — Zingiberales ар oeil Hutch. — Brassicales Caprifoliales Lindl. ipsacales бшшш ы Takht. - not accepted, family under euasterids II Carduales Small — Asterales Caricales L. D. Benson — Brassicales Cassiales Horan. abales Casuarinales Lindl. agales Celastrales Baskerville - not accepted, family under eurosids I iex c ina Takht. ales Cephalotales n — Oxa Cercidiphyllales Hu ex Reveal = Saxifragales Chenopodiales Dumort. = Caryophyllales Chironiales Griseb. = Gentianales Chloranthales A. C. Sm. ex J. -F. Leroy not ac Een family at beginning of sy T dd Lindl. — Gentianales Circaeasterales Takht. = Ranunculales Cistales Rchb. — Malvales Citrales Dumort. — Sapindales Cocosales Nakai = Arecales Colchicales Dumort. = Liliales Combretales Baskerville — Myrtales Connarales Takht. ex Reveal — Cunoniales Convolvulales Dumort. И Бы - Fag Corynocarpales Takht. rbitales Crassulales Lindl. = Saxifragales Crossosomatales Takht. ex Reveal - not accepted, family under rosids Cunoniales im ch xalidal Cyclanthales T H. Schaffn. = Pandanales Cymodoceales Nakai i ales Cynomoriales Burnett - not accepted, family unplaced Cyperales Hutch. Cytinales Dumort - not жү family unplaced стоте р indl. 21: Pulle ex Cronquist ifragales Datiscales Dumort. — Cucurbitales Desfontainiales Takht - not e family under euasteri Diapensiale Engl. & Gilg — Ericales Didymelales Takht. - not accepted, family under eudicots Dilleniales Hutch. ot accepted, family under core eudic Dioncophyllales Takht. ex Reveal — Caryophyllales Diospyrales Prantl = Е 1 Droserales Griseb. — Caryophyllales Angiosperm Phylogeny Group Volume 85, Number 4 1998 Ordinal Classification CLASSIFICATION OF FLOWERING PLANTS (cont'd.) Ebenales Engl. ricales Elaeagnales Bromhead — Rosales Elaeocarpales Takht. = lid Elatinales Nakai - not accepted, family unplaced Elodeales Nakai ales Did Nakai — Poales Eucommiales Nemejc ex Cronquist arryales Euphorbiales Lindl. — Malpighiales (€ 2 ex Reveal — Magnoli pir on Hu. ex Reveal — Ranunculales Ea H.L.Li - not accepted, family at beginning of system Ficales Dumort. Rosales Flacourtiales Heintze = Malpighiales Fouquieriales Takht. ex Reveal = Ericales Francoales Takht. = Geraniales F rangulales Wirtg. = Rosales Galiales 22. — Gen Geisslomatles Takht. ex Reveal ted, family unplaced без jns — Lamiales Glaucidiales Takht. ex Reveal — Ranunculales те Dumort. — Lamia байа Lindl. — Asterales eis, uh 2. = Сет Grossulariales Lindl. = Saxifragales unae Takht. ex Reveal - not accepted, family under core eudicots Gyrocarpales Dumort. Gyrostemonales Takht. Haemodorales Hutch. = Commelinales Haloragales Bromhead = Saxifragales Hamamelidales Griseb. = Saxifrag Hanguanales R. Dahlgren ex Reveal = not accepted, family under melinoids Helleborales Nakai = Ranunculales Helwingiales | II — Aquifoli Himantandrales Baci & Shevyryova Magnolia Hippuridales Palle ex Reveal AA Troad — Malpighiales ча FS Hydatellales йен oales Hora Takht. ex Reveal un. family at beginning of sy ы Nakai ornales Hydrastidales Takht. anunculales Hydropeltidales (Bartl.) Spenn. - not accepted, family Nymphaeaceae at beginning of system Hydrostachyales Diels ex Reveal — Cornales Hypericales Dumort. — Malpighiales Hypoxidales Takht. = aragales Icacinales Tiegh. ex Reveal - not accepted, family under euasterids I Illiciales Hu ex Cronquist - not accepted, family at beginning Jasminales Dumort. — Lamiales Juglandales Dumort. — Fagales 546 Annals of the Missouri Botanical Garden CLASSIFICATION OF FLOWERING PLANTS Julianiales Engl. = indales Juncaginales Hutch. — Alismatales Juncales ua oale Laima Baskerville = Malpi Lactoridales Takht. ex Reveal Piperales Lardizabalales Loconte s Lecythidales Cronquist = Ericales Leitneriales Engl. = Sapindales Lentibulariales Lindl. = Lamiales Ligustrales Bartl. ex Bisch. Limnanthales Nakai Loasales 2 y — Cornale Loganiales Lindl. = Gentianales Lonicerales T. Liebe = Dipsacales Loranthales Dumort. = Santalales к ~ Marathrales. Dauid - not accepted, fam Podostemaceae under rosids Mayacales Nakai - not и family under commelino Кас Такы. = Malpighiales МИРТА Вгепап accepted, family unplaced мали К. Dahlgren ex Reveal iale Melastomatales Oliv. Meliales Lindl. = Sapin йш ошл Bromhead — Ranunculales Menyanthales T. Yamaz. ex Takht. — Asterales Metteniusales Takht. - not accepted, family unplaced Mitrastemonales Makino - not accepted, family unplaced Monimiales Dumort. — Laurales Moringales Nakai icales Myricales Engl. — Fagales Myristicales Thomé = Magnoliales Myrothamnales Nakai ex Reveal - not accepted, family under core eudicots Myrsinales Spenn. — Ericales Najadales Dumort. — Alismatales Narcissales Dumort. paragales Nartheciales Reveal & Zomlefer - not accepted, family under monocots Nelumbonales Reveal — Proteales Nepenthales Dumort. — Caryophyllales Nolanales а = Solan Nyctaginales Duk aid Nymphacales D = not accepted, family at beginning a Hutch. ex Reveal = Malpighiales 5. Bromhead ales Olacales Benth. = Lam Onagrales Rchb. yrta Opuntiales Willk. — Caryophyllales Orchidales Raf. — Asparagales Paeoniales Heintze — Saxifragales Pandales Engl. & Gilg — Malpighiales Papaverales Dumort. unculales sisi Takht. - not си dms unplaced Paridales Dum — Liliales Volume 85, Number 4 Angiosperm Phylogeny Group 1998 Ordinal Classification CLASSIFICATION OF FLOWERING PLANTS (cont'd.) Parnassiales Nakai көне Dumort. - not accepted, family under — Lam eurosids I Rhizophorales Tiegh. ex Reveal Passiflorales Dumort. — Malpighiales — Malpighiales Rhodorales A Penaeales Lindl. = = Myrtales Rhoiptleae Novák ex Reveal Petiveriales Lindl. — Caryophyllales Roridulales Nakai Petrosaviales Takht = Ericales - not accepted, family under 4. Dumort. monocots — Gentianales Philydrales Dumort. Ruppiales Nakai = Commelinales = Alismatales Physenales Takht. Rutales Perleb = Caryophyllales = Sapindales Pinguiculales Dumort. Sabiales Takht. = Lamiales = not accepted, family under т Lindl. eudicots Salicales Lindl. Plantaginales Lindl. = Malpighiales Salvadorales R. Dahlgren ex Reveal dais І. n Schaffn. — Brassicales — Proteales Samolales Dumort. Plumbaginales Lindl. — Ericales — Caryophyllales Samydales Dumort. Podophyllales Dumort. — Malpighiales = nculales Sanguisorbales Dumort. Podostemales Lindl. — Rosales — not accepted, family under rosids ериш Hook. f. Polemoniales Bromhead ricales — Ericales Saraceniales Bromhead Polygalales Dumort. cale = Fabales аы B. Boivin Polygonales Dumort. — Alismatales — Caryophyllales да Dumort. Pontederiales Hook. f. = Caryophyllales melinales Serophulariales Lindl. Беса) Dumort. — Caryophyllales Seyphostegiales Croizat Posidoniales Nakai Malpighiales — Alismatales Sedalás Rchb. Potamogetonales Dumort. — Saxifragales — Alismatales Silenales Lindl. Primulales Dumort. — Caryophyllales — Ericales Simmondsiales Reveal Quercales Burnett — Caryophyllales — Fagales Smilacales Lindl. Rafflesiales Oliv. = Liliales - not Accepted, family at beginning Stellariales Dumort. of system = Caryophyllales Resedales Dumort. Stylidiales Takht. ex Reveal rassicales = Asterales Restionales J. H. Schaffn. Siyracales Bisch. = Poales = Ericales Rhamnales Dumort. Taccales Dumort. = Rosales = Dioscoreales 548 Annals of the Missouri Botanical Garden CLA: SSIFICATION OF FLOWERING PLANTS (cont'd.) Tamales Dumort. hyllales Tecophilaeales Traub ex Reveal — Asparagales ue Lindl. ricales Tied: Nakai — Gentianale Thymelaeales Willk. = Malvales Tiliales Caruel vales Tofieldiales Reveal & Zomlefer = Alismatales Torricelliales Takht. ex Reveal = Apiales Tovariales Nakai = Brassicales Trilliales Takht. iliales Triuridales Hook. f. - not accepted, family under monocots Trochodendrales Takht. ex Cronquist - not accepted, family under eudicots Tropaeolales Takht. ex Reveal Turnerales Dumort. = Malpighiales Typhales Dumort. = Poales Ulmales Lindl. се lid = Ros Vacciniales Dumort rica Vallisneriales Nakai = Alismatales Velloziales R. Dahlgren ex Reveal Pandanales Veratrales Dumort. iliales Verbenales Horan. Viburnales Dumort. - not accepted, family under euasterids II Violales Perleb = Malpighiales Vitales Reveal - not accepted, family under core eudicots Vochysiales Dumort. Winterales A. C. Sm. ex Reveal - not accepted, family at beginning of system Xyridales Lindl. = Poales Zosterales Nakai = Alismatales Zygophyllales Takht. - not accepted, family under rosids SELECTED FAMILIAL SYNONYMS Abrophyllaceae = Carpodetaceae SE на Velloz Ming — Sapindaceae Achradaceae — Sapotaceae Aegiceratacess = Myrsinaceae AgnesiMaceae hytolaccaceae Гали = Meliaceae Alangiaceae — Cornaceae Aloaceae — Asphodelaceae Amygdalaceae — Rosaceae Androstachyaceae — Euphorbiaceae Aristoteliaceae — Elaeocarpaceae Asclepiadaceae — Dioscoreaceae Volume 85, Number 4 Angiosperm Phylogeny Group 1998 Ordinal Classification CLASSIFICATION OF FLOWERING PLANTS (cont'd.) Balanitaceae Capparaceae = 2 жын = Brassicaceae Barbeiuacea Carduaceae eig eek NN Asteraceae Barclayaceae Cassythaceae — Nymphaeaceae La Barringtoniaceae Chailletiaceae — Lecythidaceae — Dichapetalaceae Baueraceae сеу = Сипошасеае = Ат же 94 Baxteriaceae Соло = Das O = Melanthiaceae Bembiciacea Chloanthaceae = Ку шне = Lamiaceae Berzeliaceae Cichoriaceae runiaceae — Asteraceae Dias Cleomaceae = Euphorbiaceae = Brassicaceae Blepharocaryace Cneoraceae nacardiaceae = Rutaceae Boerlagellaceae Cobaeacea Sapotaceae = Polemoniaceae Bombacaceae Compositae = Malvaceae = Asteraceae Boopidaceae Conostylidaceae = Calyceraceae = Haemodoraceae Bretschneideraceae Cordiaceae = Akaniaceae = Boraginaceae Brexiaceae Coridaceae = Celastraceae = O Brunelliaceae Corokiacea = Cunoniaceae = A төріуінеег Brunoniaceae Corylacea — Goodeniaceae = аи Витећасеае Croomiaceae — Sapotaceae — Stemonaceae Burchardiaceae Cruciferae — Colchicaceae = Brassicaceae Byttneriaceae Curtisiaceae — Malvace = сеае Cabombaceae онн: ае = Nymphaeaceae = Convolvulaceae Саеѕа]ріпіасеае Cyananthaceae — Fabaceae — Campanulaceae Calectasiaceae Cyanastrac — Dasypogonaceae — Tecophilaeaceae Callitrichaceae Cynocrambaceae nom. illeg. — Plantaginaceae — Rubiac Calochortaceae Cyphiacea = Lili — Campanulaceae Camelliaceae Cyphocarpaceae — Theaceae — Campanulaceae Canotiaceae Cypripediaceae — Celastraceae — Orchidaceae Cansjeraceae Dactylanthaceae — Opiliaceae — Balanophoraceae 550 Annals of the Missouri Botanical Garden CLASSIFICATION OF FLOWERING PLANTS Davidiaceae Frangulaceae — Cornaceae — Rhamnaceae Davidsoniaceae Fumariaceae — Cunoniaceae — Papaveraceae Decaisneaceae Funkiaceae — Lardizabalaceae — Agavaceae Desfontainiaceae Galacaceae — Columelliaceae = iid ое Dialypetalanthaceae Geitonoplesiac — Rubiaceae — He т РРА Dianellaceae Geniostomaceae — Hemerocallidaceae — Loganiaceae Dichondrace Сао = Convolvulaceae = Iridaceae Diclidantheraceae Gisekiaceae = асеае = Phytolaccaceae Diegodendraceae Glaucidiaceae — Bixaceae — Ranunculaceae Dionaeaceae Globulariaceae — Droseraceae — Plantaginaceae Dracaenaceae Goetzeaceae — Convallariaceae — Solanaceae Duabangaceae Gonystylaceae - = Thymelaeaceae 2... Gouaniaceae — Solanaceae — Rhamnaceae Dulongiaccae nom. illeg. Gramineae — Phyllonomaceae — Po Dysphaniaceae Gronoviaceae = A nthaceae — Loasaceae Ehretiaceae Gustaviaceae — Boraginaceae — Lecythidaceae Ellisiophyllaceae Guttiferae = Scrophulariaceae — Clusiaceae йрке дә Gyrocarpaceae — Eri — Hernandiaceae Epacridaceae pag irat — Ericaceae — Hydrocharitaceae Eremolepidaceae Halophytaceae — Santalaceae — Amaranthaceae Eriospermacea Hectorellaceae е = Portulacaceae Erycibacea Heliotropiaceae = 5 йы — Boraginaceae Erythropalacea Heloniadaceae — Olacaceae o Eucryphiaceae ие pk = Cunoniaceae 224 Euryalaceae нне унии = Nymphaeaceae = Rubiaceae Ехосаграсеае Hippocastanaceae Santalaceae — Sapindaceae Flindersiaceae Hippocrateaceae — Rutace — Celastraceae Foetidiaceae Hippuridacea — Lecythidaceae — Plantaginaceae Volume 85, Number 4 1998 Angiosperm Phylogeny Group Ordinal Classification CLASSIFICATION OF FLOWERING PLANTS (cont'd.) Hortoniaceae — Monimiaceae Hostaceae = Agavaceae Humbertiaceae — Convolvulaceae Hydrastidaceae — Ranunculaceae Hydrocotylaceae — Araliaceae Hydropeltidaceae — Nymphaeaceae Hydrophyllaceae — Boraginaceae Hymenocardiaceae ы аа Illecebraceae — Caryophyllaceae Jasionaceae — Campanulaceae Jasminiaceae Johnsoniaceae — Hemerocallidaceae Julianiaceae nacardiaceae = Circaeasteraceae Kirengeshomace eae = Hydrangeaceae Labiatae Lamiaceae Langsdorffiaceae = Balanophoraceae Leeaceae = Vitaceae Leguminosae = Fabaceae Leitneriaceae = Simaroubaceae = Boraginaceae Leoniaceae = Violaceae Lepuropetalaceae = Parnassiaceae Lilaeaceae = Juncaginaceae Limoniaceae = Plumbaginaceae Liriodendraceae = Campanulaceae Lomandraceae = Laxmanniaceae = Pedaliaceae Mastixiaceae = Cornaceae Medeolaceae iliaceae Meliosmaceae abiaceae Mendonciaceae ceae Mesembryanthemaceae — Aizoaceae Mimosaceae = Dipterocarpaceae Monotropaceae = Ericace: Mouririaceae = Memecylaceae Moutabeaceae = Polygalaceae Myriophyllaceae — Haloragaceae Mystropetalaceae — Balanophoraceae Najadaceae — Hydrocharitaceae Nandinaceae — Berberidaceae Napoleonaceae — Lecythidaceae Naucleaceae — Rubiaceae Nectaropetalaceae — Erythroxylaceae 552 Annals of the Missouri Botanical Garden CLASSIFICATION OF FLOWERING PLANTS Nelsoniaceae Pistiaceae — Acanthaceae — Áraceae Nemacladaceae Platystemonaceae — Campanulaceae — Papaveraceae Nesogenaceae Plumeriaceae — Cyclocheilaceae = Аросупасеае Nolanaceae Podoaceae — Solanaceae — Anacardiaceae Nolinaceae Podophyllaceae — Convallariaceae — Berberidaceae Nupharaceae Polygonanthaceae — Nymphaeaceae — Anisophylleaceae Nyctanthaceae Potaliaceae — Oleaceae — Gentianaceae Nyssaceae Ptaeroxylaceae — Cornacea = Octoknemaceae Pteridophyllaceae — Olacacea — Papaveraceae Oftiaceae Punicaceae — Scrophulariaceae Lythraceae Ophiopogonaceae Pyrolaceae — Convallariaceae — Ericaceae Osyridaceae Ranzaniaceae — Santalaceae = Berberidaceae Pachysandraceae Reaumuriaceae é = Tamaricaceae Palmae Retziaceae = Arecaceae = Stilbaceae Papilionaceae Rhinanthaceae = eae = Orobanchaceae Peganaceae Rhodoleiaceae = Nitrariaceae = Hamamelidaceae Pentastemonaceae Rhopalocarpace eae = Stemonaceae = Sphaerosepalaceae Peperomiaceae Rhynchothecaceae = Piperaceae = Ledocarpaceae Periplocaceae Roxburghiaceae =A е = Stemonaceae Feripterygiao Ruscaceae — Ca is = Convallariaceae Petermanniaceae Saccifoliaceae = Colchicaceae = Gentianaceae Petiveriaceae Salaciaceae = Phytolaccaceae = Celastraceae Philadelphaceae Salicorniaceae — Hydrangeaceae maranthaceae Phormiac is = Hemerocallidaceae = Solanaceae Phylicaceae Sambucaceae = Rhamnaceae = Adoxaceae Picrodendraceae Samolaceae = Euphorbiaceae = Primulaceae Pinguiculaceae secan = Lentibulariaceae ae Pistaciaceae Sarcophytacea = Anacardiaceae Balanophoraceae Volume 85, Number 4 1998 Angiosperm Phylogeny Group Ordinal Classification 553 CLASSIFICATION OF FLOWERING PLANTS (cont'd.) Sarcospermataceae — Sapotaceae Sargentodoxaceae = Lardizabalaceae Saurauiaceae = Actinidiaceae Sauvagesiaceae = Ochnaceae Scaevolaceae = Goodeniaceae Scepaceae = Euphorbiaceae Schoepfiaceae welerophiacdcene olanaceae Scoliopaceae Selaginacea = Sc нети ПА Зезатасеае = Pedaliaceae Sesuviaceae = 27 Simethidace — Hem eocalidacese Siphonodontac = Ce - Sonneratiaceae vac ьалаа = Euphorbiaceae Surychnaceae ganiaceae Stylobasiaceae = Surianaceae Stylocerataceae = Buxaceae Symphoremataceae = Lamiaceae Syringaceae = Oleace Tetracentraceae = Trochodendraceae Tetradiclidaceae Tetrastylidiaceae = Olacaceae Thalassiaceae = Hydrocharitaceae Ineligunacene ubiace лика acon = Acanthaceae Tiliaceae — Malvaceae ola Tone ын еае = Liliaceae Trilliaceae — Melanthiaceae Таріоевневвя = папасеае Uapacacea = Euphorbiaceae = Colchicaceae Vacciniaceae Viticaceae — Lamiaceae leche cophilacaceae Wellstediacea = Boraginaceae Xanthophyllaceae = Polygalaceae Xerophyllaceae = Melanthiaceae Zannichelliaceae = Potamogetonaceae TAXONOMY OF THE PAEONIA DELAVAYI COMPLEX (PAEONIACEAE)! Hong De-yuan?, Pan Kat-yu?, and Yu Hon ABSTRACT The Paeonia delavayi complex, a group of 2 7 endemic to China, is problematic because of the lack of agreement among the taxonomic treatment 5 м the axonomists recognize one species with three infraspecific nd ye E others accept three species with two е taxa. The present analysis of certain characters. The results show that the complex i is extremely variable both within and between йш: in the nua length, ud width of leaf segments and in number, size, and color of all the floral parts. This variation correlation between the charac ‘ters. Only one species, P. күүнү is recognized, without infraspecific tax is continuous, and the re 18 no P. delavayi. Vegetative vincire by roots are thickened fusiformly. These two features probably n and somewhat d D and allow for rapi id colonization whenever plants become established. Although P. delavayi has been listed as endangered, it is unlikely to become extinct if wanton digging is d. controlle The Paeonia delavayi (Paeoniaceae) complex comprises a group of woody peonies endemic to southwestern China. It and other shrubs of the ge- nus Paeonia belong to section Moutan DC., while the herbaceous members belong to section Paeonia and section Опаера Lindl. Members of section Moutan are readily distinguished by the following key: la. Flowers solitary and terminal, erec 2a. Carpels nearly always 5, вене entirely enveloped by floral disk at ant : suffruticosa ra 2b. Carpels 2—4(or 5), glabrous, enveloped b floral disk only in the lower half P. dec composita Hand.-Mazz. . Flowers сте 2 or 3, both terminal and axil- lary, = pe 3a. бил ou always solitary; petals, fila- ments, and stigmas always pure yellow; plants 1.5-3.5 m tall; follicles 4.7-7 x 3.3 cr 5 — — - TN P. ludlowii igi & С. Taylor) D. Y. Hong 3b. Carpels mostly vetals, filaments, and stigmas 4. іп color; plants < 2 m tall; follicles < 4 X 1.5 ст . P. delavayi complex Four species and numerous infraspecific taxa have been allied with Paeonia delavayi. The dar red-flowered Р. delavayi Franch. and the yellow- flowered Р. lutea Delavay ex Franch. were both based on collections from Northwest Yunnan [Li- jiang (Likiang) and Eryuan, respectively] and were described on the same page (Franchet, 1886: 382). A third species, P. potaninii Kom., was described in 1921 based on a specimen from the Yalong Val- ley in West Sichuan. It has deep maroon-red flowers and was said to differ from P. delavayi in having narrower leaf segments, smaller flowers, paler sta- men color, and in the absence of the conspicuous involucre, and from P. lutea in having much nar- rower leaf segments, smaller flowers, and deep ma- roon-red flowers. Stern (1931) added the ped flowered Р. trollioides Stapf ex Stern, based on a specimen from Baima Shan, Deqen County, North- west Yunnan. In addition, a number of infraspecific taxa have been described, including P. delavayi var. atropurpurea Schipez., P. delavayi var. angustiloba Rehder & E. H. Wilson, P. delavayi var. alba Bean, and P. delavayi var. lutea f. superba Lemoine. aeonia lutea var. ludlowii Stern & G. Taylor, described in 1953 from Southeast Xizang (Tibet), has recently been shown by Hong 7) to be quite different from the remaining taxa of Paeonia lutea in a number of characters. It is now recog- deeds was inane 'ially supported by the National Geographic Society (Grant 5515-95), to which the authors are very grateful. The authors are indebted to Luo Yi-bo a hua (C hengdu itte of 2. iao-ling for the measurements ng gratitude i is due to Іһвап AL Shehbaz for his encouragement and help utionary Botany, Institute of ee Chinese Academy of Sciences, 20 Nanxincun, ? Laboratory of Systematic and Evolut Xiangshan, Beijing 100093, People’s Republic of China. лу of Sciences) for calculations in statistical study, and for typing the manuscript. Our sincere Shu-ren (Institute of Botany, Beijing) and He Yong- their assistance during fieldwork. We thank Li with the manus script. ANN. Missouni Bor. GARD. 85: 554—564. 1998. Volume 85, Number 4 1998 Hong et al. 555 Paeonia delavayi Complex nized as a distinct species, P. ludlowii (Stern & G. Taylor) D. Y. Hong, and has been excluded from the P. delavayi complex under study. Finet and Gagnepain (1904) treated Paeonia lu- tea as P. delavayi var. lutea (Delavay ex Franch.) Finet & Gagnep. In his monograph of Paeonia, Stern (1946) recognized in this complex three spe- cies (P. delavayi, P. lutea, and P. potaninii) and reduced P. trollioides to a variety of P. potaninii. Fang (1958) followed Stern (1946) in the treatment of this complex, but ignored Р. lutea var. ludlowii. Wu (1984) recognized two species, P. delavayi, P. delavayi var. angustiloba (= P. potaninii) and P. lutea. Gong (1990) also followed Stern (1946) and recognized three with two infraspecific taxa (P. de- lavayi, P. lutea, P. potaninii, P. potaninii var. trol- lioides, and P. potaninii f. alba). However, Рап (1979, 1993) recognized only Р. delavayi, with two varieties, var. lutea and var. angustiloba and ig- nored (Pan, 1979) P. lutea var. ludlowii. Obviously, there is a lack of agreement among the taxonomists who have studied this complex. They have various- ly emphasized the presence of a conspicuous in- volucre (P. delavayi), petal color (dark red or deep maroon-red in P. delavayi and P. potaninii, yellow in P. lutea and P. potaninii var. trollioides, white in P. potaninii f. alba), and the width of leaf segments (narrower in Р. potaninii than in P. delavayi and Р. lutea). The aim of the present paper is to present a revision of this fascinating group of woody peo- nies based on extensive field observations, popu- lation sampling, and analysis of certain characters. MATERIALS AND METHODS In order to fully understand the variation of cer- tain characters, especially flower color, presence of an involucre, and width of leaf segments, the first author (alone or with the other two) made five trips to the distribution range of this group. The first (in 1988) and second (1993) were to the West Hill of Kunming; the third to West Sichuan (1995), in- cluding Yajiang County, the type locality of both P. potaninii and P. delavayi var. angustiloba; the fourth to Southeast Xizang (1996), including the type locality of P. lutea var. ludlowii; and the fifth to Northwest Yunnan (1997), including the type lo- calities of P. delavayi, P. lutea, and P. trollioides. In addition to the critical examination of all mor- phological characters and their variation, biological features (e.g., seed set, presence or absence of clon- ing) were also recorded. Eighteen populations were studied in the field by the first author (alone or with the other two), and vouchers are deposited in PE, A, CAS, K, MO, and — Figure 1. Vegetative reproduction by stolon in the Paeonia pom complex (population H97112). US. The populations studied are vouchered by the following herbarium specimens (collection numbers of Hong) H97077, H97078, H97087, H97095, H97102, H97103, H97108, H97112, H97119, and H97128 (from Yunnan); H95063, H95070, H95074, and H97110 (from Sichuan); and H96003, H96019, H96024, and H96028. In addition, hundreds of oth- er herbarium specimens from CPB, E, KUN, PE, WU, and XZ were studied morphologically. OBSERVATIONS AND DISCUSSION Plants of the Paeonia delavayi complex are al- ways dwarf shrubs. The tallest plants (ca. 1.8 m) were found in Yunshanping in Lijiang, Northwest Yunnan (population H97103), where they grew in Picea likiangensis forest at an altitude of ca. 3200 m. By contrast, the shortest plants (rarely reaching 1 m) were found in Ganghaizi, Lijiang (H97095), а population only about 20 km southeast of the first population, and grew in dry, sparse Pinus densata— Quercus gilliana forest. Many short individuals had underground woody parts, and only current-season shoots were aboveground. Plants of the other pop- ulations that we studied were intermediate between these two. Vegetative reproduction is probably predominant in the Paeonia delavayi complex, and seedlings are very rarely found in the field. It is even more pre- dominant at the northwestern and northern bound- aries of the distribution range. In Yajiang County, West Sichuan, a population (H95070) was found near a village, where it was represented by some individuals growing by fences and on newly stabi- lized debris. In this population, only about 50% of the follicles were developed and, because of insect damage, only 2096 had seeds. Cloning by stolons, however, was common. Based on an examination of 556 Annals of the Missouri Botanical Garden Figure 2. by stolon in the Paeonia delavayi complex (population 2). Fusiform roots and vegetative reproduction all follicles in the spring of 1996, the five popula- tions observed in Tibet in 1995 produced no seeds. No follicles were observed in a population (temple ruins in Xituan village of the Gengzhanglungba val- ley, Nyingchi County, Tibet) that consisted of nu- merous individuals and occupied an area of about 250 m?. This “population” probably consisted of individuals produced through cloning. Cloning was found in every population visited (Figs. 1, 2). The production of fusiform, thickened roots is charac- teristic of the Paeonia delavayi complex (Fig. 2). Although leaves in the Paeonia delavayi com- 20 100 plex are always biternate, the leaf segments are quite variable in number, length, and width. number of segments ranged widely from 17 to 312 (Fig. 3), and seemed to differ between populations. For example, the number varied from 17 to 49 in population H96024, and from 68 to 312 in popu- lation H95063. However, the standard deviation of the variation in leaf segment number (Fig. 3) shows that these two populations were just the extremes of a wide variation that was also observed within given populations. Taking all the populations into consideration, the number of segments varied con- tinuously in the complex. Раеота potaninii was described as new because it was considered to have narrower leaf segments than P. delavayi and P. lutea. However, we observed that the width of leaf segments varies greatly, and ranges from 0.4 to 2.86 cm within the complex, and from 0.76 to 1.83 cm in population H95070, the type locality of P. potaninii, which falls in the mid- dle of the overall variation range in the complex Fig. 4 length of leaf segments. Therefore, it is evident that P. potaninii is rather similar to P. delavayi and P. lutea in this regard. Stern (1946) distinguished Paeonia delavayi within the complex by the presence of a conspic- uous involucre immediately below the calyx. How- ever, in the whole genus it is difficult to distinguish . The same situation was found for the ~ clearly between the bracts and sepals, and there is 200 250 + H97095 -amj H97110 — = H97078 — A — H97077 H97087 а а, H97119 -—am — H97112 — — ——À — 312 H95063 тр AMENS H95070 — —— Há— Pm— — H96024 — d——— Figure 3. Variation of leaf-segment number both within and between populations in the Paeonia delavayi gimp ertical line is mean, and thickened horizontal line is standa rd deviation. Localities of populations are: H97 (Zhongdian, Yunnan), H97095 (Lijiang, Yunnan), H97110 (Yanyuan, Sichuan), H97078 (Chenggong, Yunnan), mos (Kunming, Yunnan), H97087 (Cang Shan, Yunnan), H97103 Lijiang, Yunnan), H97119 (Deqen, Yunnan), H97112 (Zhongdian, Yunnan), H95063 (Dawu, Sichuan), H95070 (Yajiang, Sichuan), and H96024 (Bomi, Tibet). Volume 85, Number 4 1998 Hong et al. Paeonia delavayi Complex 2.0 2.5cm width H97077 H97078 H97087 H97095 H97103 H97128 H97112 H97119 H97110 H95070 Е H95063 H96024 e 4. Variation of leaf-segment width both within and between populations in the Paeonia delavayi complex. Figur Vertical line is mean, and thic ened horizontal line is stan ard deviation. Localities of populations are: H97077 (Kunming, Yunnan), H97078 (Chenggong, Yunnan), H97087 (Cang Shan, Yunnan), H97095 (Lijiang, Yunnan), H97103 (Lijiang, Yunnan), H97128 (Zhongdian, Yunnan), H97112 (Zhongdian, Yunnan), H97119 (Deqen, Yunnan), H97110 (Yanyuan, Sichuan), H95070 (Yajiang, Sichuan), H95063 (Dawu, Sichuan), and H96024 (Bomi, Tibet). a gradation between leaves and bracts as well. We designate the laminae that are borne some distance below the flowers as leaves, while those at the top of shoots and immediately below the calyx are bracts. Bracts, so designated, have various forms, ranging from segmented and leaflike to entire and sepal-like. The sepals have a much broader proxi- mal part and a dark green, smaller, and narrower distal part with a mucronate or rounded apex. The total number of bracts and sepals varied greatly both within and between populations of the Paeonia delavayi complex. Population H97103 (with dark red flowers) at Yunshanping, Lijiang, the type lo- cality of P. delavayi (Stern, 1946), indeed had the highest number (10 or 11) of bracts and sepals forming the so-called "conspicuous involucre,” while the other populations observed had fewer bracts and sepals. er, the difference was not distinct (Table 1). Population H97095, only about 20 km from H97103, was very variable in the num- ber of both bracts and sepals, and some flowers had 10 or 11 bracts and sepals (Plate 1, vii), just as those in population H97103. Population H97087 ad the same total number of bracts and sepals, and its petals were pure yellow or yellow with a dark red spot at the base. Table 1 and the remarks above show clearly that there is continuous varia- tion in the number of bracts and sepals, and no correlation between petal color and the total num- ber of bracts and sepals. Therefore, P. delavayi is not distinct from the other putative taxa in the com- plex under study. Variation in the number and color of floral parts involves the sepals, which vary in color both within a flower "9 Ме populations (Table 2). They аге usually green, but sometimes dark red or purple ate 1, vii, viii; Table 2). In addition, they vary greatly in size both within and between popula- tions, and the variation is continuous (Fig. 5). Al- though the population H97103 had larger sepals than the other populations, the formation of a “con- spicuous involucre" (Stern, 1946) was not unique in the complex, as alleged. Petal color has been much emphasized in the taxonomy of the Paeonia delavayi complex, and was used by various authors in distinguishing P. lutea (yellow) from P. delavayi (dark red). As shown in Table 2, however, petal color is extremely vari- able between and within populations (Plate 1, i—vi). In populations H97112 and H97128, various petal colors could be found, and a few individuals in the latter population had white petals. On the basis of the literature and our own observations, red, dark red, or dark purple-red petals occurred in the northeastern part of the distribution range, while yellow petals or yellow petals with a dark red spot at the base were found in the northwestern, west- ern, and southern parts. For example, populations in Southwest Sichuan (Dawu, Yajiang, Muli, Yan- yuan) and Northwest Yunnan (Lijiang and Ning- — Annals of the Missouri Botanical Garden Variation of bract and sepal number in the Paeonia delavayi complex. Table 1. Total bract Note and sepal # Bract # Sepal + 7 or 8 Petal color Population and locality H97077, West Hill, Kunming 3 or 4 no bracts segmented yellow H97078, Mt. Liangwang, Cheng-Gong Co. H97087. Huadianba, Cang Shan, Dali H97095, Ganghaizhi, Lijiang 10 or 11 4 or 5 yellow some flowers with 2-segmented 6 or 8—11 4 to 6 red to dark red most flowers with one 3- or 4- 10 or 11 7 109 2 to 4 dark red H97103, Yunshanping, Lijiang segmented bract no bracts segmented 6, or 8 2105 1 to 3 yellow, orange, to dark H97112, Hala village. Zhongdian red only one flower with one 3-seg- 5 2105 5108 roo 1,3 white to red H97128, Gezha Township, Zhongdian mented bract lang), here designated as the “red region,” had red petals, while those in Southeast Tibet (Nyingchi, Bomi, Zayu), Northwest Yunnan (Deqen and Gong- shan), and Yunnan (Weixi, Eryuan, Dali, Cheng- gong, and Kunming), here designated as the “yel- low region,” had yellow petals with a red spot at base, pure yellow petals, and petals either yellow or yellow with a dark red spot at the base, respec- tively. Therefore, petal color shows a weak geo- graphical differentiation, but it would be unwise to accord formal recognition to the populations in these two “regions” because petal color is uncor- related with other = and is very variable within a given “region” or population. The most pronounced petal-color variation occurred in pop- ulations H97119 and H97078 of the “yellow re- gion,” and some specimens with yellow flowers (e.g., Rock 16110, Rock 161576, Kingdon-Ward 4043, Forrest 12565, and McLaren 89, all at E) were also found within the “red region.” Further- more, petal color was extremely variable in popu- lations such as H97112, H971 28, both from North- west Yunnan. Petal number in the Paeonia delavayi complex varied enormously, and ranged from 4 to 13, but it also varied greatly within populations. For example, petals were 4—7 or 10 in population H97119, and 8, 10, 11, or 13 in H97087. Stamen number able. In population H97112 the number varied from 25 to 128. The filaments were yellow, pale red, red, or dark red, and the anthers were yellow, orange, red, or purple in populations H97077, H97078, H97087, H97095, H97103, H97110, and H97112 (Plate 1, viii). But in the populations H97119 and H97128, both filaments and anthers were yellow even in flowers with red petals. The color of the filaments and anthers was not always correlated with that of the petals, although flowers with red petals usually had red or purple filaments and an- thers. We conclude that the characters of the an- and color were also very vari- droecium cannot be used for dividing this complex. Unlike the nectar disk of the other woody peo- nies of section Moutan, the disk in the Paeonia delavayi complex is generally short and fleshy. In populations H97112 and H97119, the disk secreted abundant nectar in some flowers, and it seems like- ly that the secretion made these flowers more scent- ed. The disk (including teeth) varied in height from 1 to 3 mm, and in color from pale yellow or yellow red, even within a single population (e.g., poe 12). Therefore, the characters of the disk are of little, if any, importance taxonomically in this complex. The gynoecium also varied greatly in the Paeon- 559 Hong et al. Volume 85, Number 4 1998 Paeonia delavayi Complex "(6026H uone[ndod) вүәйтеә jo 10[oo pue лодшпи ш иопеиед “(14311 ‘MOI шоцод) хі- ‘(Z [26H uoue[ndod) сешаце pue “sip “ѕлэщив *sjuaure ry *spedas jo UOMBLIBA 10009 (әүрріш ‘мол woyoq) utA— (660/61 иоце[падод) s[edos pue sjoe1q уо 10109 pue 1oquinu ш попемед “(1191 ‘MOI шопод) пл- *(977764] чопејпдод) лојо2 зәмор ш uoneneg (1udu “мол әүррїш) 1à— ‘(ZI 126H uone[ndod) JO[09 домој ur поцемед *(o[pprur “мол әррріш) A— ‘(0126H uoue[ndod) ләмор pay (1191 “мол әррріш) ^1— '(21126Н uone[ndod) домој SUBIC) "Qua ‘мол Чоу) ti— “(1/026H иоце[паод) 1240] морәќ-чәәле) “(әррріш ‘мол doy) п— (9601/6Н uone[ndod) ләмор) әні “(ey “мол doy) т— :хојашоо 1€napjap ютиоәр әчү ш syed үелор jo uoreugA ‘I eld a ТҮЗ, E os JN Missouri Botanical Garden Annals of the 560 uen рал рал-оја paz -Y91S “UBNÁUPA paz -o[dund *цпәә# 2107 -ind әз MO[[94 pa1 QOL 91 08 pax-a[dind ‘pai 2 018$ -a[dind “ә ‘yeg пап јо 4 0II26H pal ‘әлоде 10 2seq чеџипх pax *^o[[94 ye wed pal е цим ‘иетр8иоц7 Јо N “пәә18-мо(|әХ цә Coz MO[[94 мојј94 per 'мојј94 (01 0108 МОЦ(әХ “морәќ ‘эим дос pei-ajdind шу gp *inusduow — 901/6Н ueuuny ‘uəbəq *ue[ u9913 -izuag “әЗеүүгл -MO[[9Á *u2248 иә poz MO[[94 MO[[94 MO[[94 011 вә мо(әА-ә?ивло *^o[[Á 01' оф 19918 Виовпсјо M ux € 6IIZ6H ивиипд MO[[9Á pal гер “paz ‘Zue ‘umo, uerpguouZ pol “рәл ‘мо pax ‘par pal »xrep -10 *jods [eseq pol pau JO A wy *u3213-Mo[[9 19218 21027, -[94-19918 o[ed “мо(әХ ‘por ‘моәХ 821 01 CZ т un^ морәќ “морәќ IL 01. ¢ -o[dund *uo218 ес ‘әйел рең 21126Н apis yorq 10 191 әрізші aseq 091 -u39 ш 1045 «o[[o4 pax oed ueuunj ‘Buel pal 9228 00146 par yep _ рол-ојфпа p34-o[dund о (су вцим “pal ер ‘pas ст 91 OI цим useis -r] “Suidueysunx £01/6H pa21-o[d -ind жвр *pa1-o[d pal par -and ‘pos “Аләцйи ueuuny Guell рал-оја па — o[dund “иәә83 род para¡dind — -o[dund ‘pau -e[dund ‘pel OZI o1 08 -əd мојјах ҷим рад 198 зјфапа зала -r[ чугтеудиво) — сбог6Н рал-оја 1046 [eseq por -and ‘pal -a¡dund ло pau зјед СТ 10 ивиипд мо[әА әүеа 19918 сос MO[[9Á MO[[94 әрра *^o[[oÁ (01-001 2 Чим мојјо *моцәА "TT “OT “8 19918 ‘eq иес 3ue) 790761] 1045 ¡eseq pa1-o[dund e ивиипх Чим мојј94 ХПвош *duod3uau7) MO[[9Á o[ed изала 2107 MO[[94 MOT[9Á pax-a]dind стр во “мо(әХ 'мо|[94-изола 6198 19918 'SuewzZuer] “UN 22026Н 1046 ¡eseq рал-оја па е цим морәќ ивиипд “suru морәА o[ed пәә# 6107 MO[[94 MO[[94 рал-ојф та 021 99 “мо(|әХ 'мојјоА-изала 1108 ицәәл2 -uny ‘JH мод 22026H 10ds [е зд, ‘wog MO[[94 o[ed изал8 6107 Mo[[94 мо[әќ рел-ајф та -Seq pol в цим морәќ ua391d dIysumo] gueixn? (02096H apis [ uo 10ds [e 19q мо[әќ ayed aj¡dind ‘иәәз pz MO[[94 мо|ј94 рал-ојфпа OIL "Вә -seq pal e цим мојј94 OL 19918 -1] ‘имо ови 61096H 10|02 euridtjg — лојоо Ал) # [1814 10[09 osiq J10[09 19ujuy лојоо шәш у = uaurels лојоо [8194 # 1194 лојоо [2496 Áy[eoo'T uon -e[ndoq "хојашоо 1íDaD]2p гтиогра ay) ut syed [e1og jo uorgueA “Z 9[qe], Volume 85, Number 4 1998 Hong et al. Paeonia delavayi Complex E 9 Е ЕЭ at 5 © жж x* = L * O o * а М 9 0 * 5 O O O O ü Ж H97103 e 2 "T О H97087 ы yo о H97095 a 0 20 30 40 mm length Figure 5. Variation of sepal size both within and between populations in the Paeonia delavayi complex. Populations H97103 and H97095 from Lijiang (Yunnan), and H97087 from Cang Shan (Yunnan). ia delavayi complex. Pistil (carpel) number was usually 2—4 in most of the populations studied, but in population H97103 it was mostly 5 and very rarely 3, 4, or even 6, and in one collection, Dai, Li & Tang 64—4 (CPB), it ranged from 4 to 8. This variation in pistil number was found in every pop- ulation observed (Plate 1, ix; Table 2). The pistils were usually green, with a pale green, yellow, or red stigma, but some individuals in populations H97095, H97103, H97112, H97110, and H97128 had entirely purple pistils (Plate 1, viii, ix). Al- though characters of the gynoecium are generally regarded as significant in the taxonomy of angio- sperms, our observations show that they are insig- nificant taxonomically in this complex. In summary, the Paeonia delavayi complex ex- hibits tremendous and continuous variation in char- acters of the leaves (number, length, and width of segments), bracts (number), and floral parts (color and number). Except for the very weak correlation between petal color and geography, the variation in these characters is clearly insignificant taxonomi- cally. Therefore, none of these characters can be considered as justifying the subdivision of this complex, which we thus consider as comprising only one species, without infraspecific taxa. TAXONOMIC TREATMENT Paeonia delavayi Franch., Bull. Soc. Bot. France 33: 382. 1886. TYPE: China. NW Yunnan: Li- jiang (Likiang), Delavay 1142 (holotype, P; isotype, Paeonia lutea Delavay ex Franch., Bull. Soc. Bot. France 33: 382. 1886. Paeonia noe var. lutea (Delavay ex Pow Finet & Gagnep., Bull. Soc. Bot. France 51: 524. 1904. TYPE: China. NW ину Eryuan ais Mt. Hea Chan Men, 25 May 1883, Delavay n. (lectotype, here designated, P; isolectotype, K). Paconi = Kom., Bot. Mater. Gerb. Glavn. Bot. RSFSR 2: 7. 1921. TYPE: China. W Sichuan: ШІ 2. lu district in Yalung valley), Potan- ‚п. (holotyp Pisonia purae puma ex Stern, J. Roy. Hort. Soc. 56 77. 1931. Paeonia 2. var. trollioides (Stapf ex Stern) P. Stud. Paeonia 50. 1946. TYPE: China. nan: па iom Shan, Mekong- Yangtse divide, “11.000 ft. n stony pastures, Forrest 13195 (holotype, E). Paeonia delavayi var. lutea f. superba Lemoine, Rev. Hort. 1 m tab., 1906. Paeonia lutea var. superba (Le- moine) hort. ex Gard. Chron., Ser. 3. 44: 50, cum rd 1908. TYPE: [*pl. 14 in Lemoine, Rev. Hort s vds var. angustiloba Rehder & E. H. Wilson, in Sarg., Pl. Wilson. 1: 318. 1913. TYPE: China. W Si chus Yajiang (W of Tachien-lu, descent to Ya- lung river), 3000-3600 m, Oct. 1908, Wilson 1333 (holotype, A). 562 Annals of the Missouri Botanical Garden Bot. Mater. TYPE: Paeonia delavayi var. A A ere Schipez., ;erb. Glavn. Bot. Sada 2: 47. 1921. China. No locality given een ?LE). Paeonia 2. var. alba Bean, Trees Shrubs 3: 265. 19 aeonia potaninii f. alba Aged ia Stud. . 49. 1946. TYPE: "t. іп Stern, Stud. Paeonia, 1946" (holotype). Shrubs 0.2-1.8 m tall, glabrous throughout. Pet- ioles 10-15 cm long; lowermost 2 (or 3) leaves bi- ternate, these best developed and with the most leaflets and segments, other leaves becoming small- er upward and with fewer leaflets and segments. Leaflets first divided into 3-11 primary segments, halfway into 2-11 secondary segments or lobes, thus each leaf with (1725-1006 312) segments or lobes; leaves ovate in outline, 15-30 cm long (excl. petiole), 10-22 cm wide; petiolules of middle pri- mary divisions 5-9 cm long, petiolules of lateral primary divisions 1-3 cm long, petiolules of sec- ondary divisions much shorter; segments or lobes linear, linear-lanceolate, entire or only occasionally dentate, 0. x0 shoot, terminal and axillary, = pendulous, maturing basipetally. Bracts 1-5, gradually differentiated from the upper leaves and hardly distinguished from the sepals, the outer ones 2—4-segmented, green. Sepals 2-9,-ртееп outside, green with a pink base inside, or entirely purple or purple-red, round- ed or triangular-rounded, acuminate or mucronate to rounded at apex, 1.3-3.7 X 0.6-2.3 cm. Petals (4—)7-11(-13), yellow, yellow with a red or purple- red basal spot, red, dark red, or dark purple-red, sometimes white, orange, green-yellow, or yellow with red margin. Stamens 25—160; filaments yellow, pink, red, or dark purple-red; anthers yellow, pink, red, or dark purple-red. Disks short, annular or short-cylindric, 1-3 mm high, dentate, green, yel- lowish, yellow, red, or dark red. Carpels 2-4, very rarely 6—8; ovary usually green, sometimes purple; stigma yellow-green, yellow, red, purple-red; ovules 7-17 per carpel. Follicles fub 2- 3.5 X 1–1.5 ст, brown at maturity. Seeds 1–6 in each follicle, brown-blac ст. Flowers 1-3 on a Distribution. Endemic to China and restricted to Sichuan, Tibet (Xizang), and Yunnan. Paeonia delavayi grows at 2000-3600 m, primarily in sparse thickets or dry Pinus and Quercus woods, or rarely in grassy slopes or glades of virgin Picea forests (Fig. 6). Phenology. Flowering from mid May to mid June; fruit maturing August. 2n — 10 (Gong et al., 1991) Additional specimens examined (all from CHI- NA). SICHUAN. Batang: Zongza Township, Anonymous 1547 (PE), by Jinsan River, Li & Xu eee 2. East District, Yanrigong Township, Li & Xu 64—38 (CPB); Za- ngqenrong, Li & Xu 64—47 (CPB); узду Township, Li & Xu 64—48 (CPB). Daocheng: Dongnyi, Gawa Town ship, Kasishe, Daocheng Division 2397 (PE). Dawu: Mazi 12% Жар village. Hor ko wi & He H95063 (А, ?, US); Dai, Li & Tan (CPB); Qiangning, S nd 'un, Dai Li we Tang Huius e Huidong: Bais- anpo, Paomaping, Li, Li & Хи 64-9 (CPB). Huili: Xiacun, Heilaoling, Qiu. Zhu, Deng & Shi 66-023 (CPB). Mian- ning: Jinping Shan, 2. M. . Luning Dis- trict, Xiamatou Shan, Z. M. Xu 64—30 (CPB). Muli: moun- tains of Kopati, Diago & Muli, 16157 (E); кн between Muli and Pew Rock 24123 (E); W of Muli, Mt. Mitzuga, Rock 0 (E); no precise locality, Ды; ngdon- Ward a (E); Yalientsa, m ciis T. T. Yü 14147 (PE); i 5560 . T. T. Yü 6003 (PE); Luobo Township, near the ferry. S. Tana diao Township, Shangjiamiwan, Li, , Xu & Tang 64-19 (CPB), 64-20 (CPB); Zhuao ее. чы Aoxia village, Li & Хи 64—26 (CPB). cheng: Qingda, L. D. Sheng 64-113 (CPB). Yajiang: Niri A 22. Hong, Luo & Не Н95070 (А, K, MO, ы . 5 km E of Yajiang Town, Hong, Luo & Не 0 (PE у; no precise locality, Li & Хи 64—64 (CPB). i Yanshi 7. to Jiudao Zhuling, Qinghai- ud NE 11458 (PE). Yanyuan: Zuosuo Township, by Lugu Lake, Eu Pan, Yu & pis H97110 (А, CAS, К, MO. PE. US); CM Township, near Shanjiacun, Dai, : Zuosuo District, Dai, Li & Tang Tang 64—7 (CPB); Shangjiagou, Dai, Li & Tang (СРВ); E of Yongning (Yunnan), 27°50'N, 100%56'E, For- rest 20458 (E); Татари, 27°37'N, Handel-Mazzetti 2067 (E, WU). TIBET (Xizang). Bomi: Guxiang Township, Ying & Hong 39 (PE), 214 (PE), Hong, Luo & Zhang H96024 (A, K, MO, PE, US); between Zamu & Guxiang, Zhang & Lang 379 (PE); near Zamu, behind Army Sta- tion, Xiao, Xia & Mi 2233 (KUN, PE); Sumzom Township, Sumzom, S of River, Hong, Luo & Zhang H96028 (А, К, MO, PE, US). Markan 66335 (PE); Cawarong, Dula, C Cawalong, Songta Snow Mountain, Qinghai- Xizang Exped. 7633 (KUN, PE); E Himalayas, no precise loc ality, don- Ward 5691 (E). . Xituan Village, 2 Luo & Zhar H96003 (A, К, MO, PE, US); Bayi Town, ы Hong, Luo & Zhang Н96004 (А, К, MO, PE, US). Zayu: Gujing District, Xizang Biol. Inst. Pl. Resources Ехреа. 3895 (XZ), near Guyu to Cinong, Qinghai- -Xizang км 23-294 (РЕ); Gujing District, Ni, Wang, Cidou & Cidan no precise locality, Ludlow & Sherriff Mt. Cangshan, Huadianba, Hong, Pan, Yu & Dai H97087 (A, CAS, K, MO, PE, US), R. C. Ching 22954 (KUN, PE), Sino-British Cangshan Exped. 0684 (E, KUN), Xiaohua- dianba, Q. Ling 7708 (KUN); no precise locality, T. N Liou 16149 (KUN, PE), . Liou 2 4352 (E), 6787 (E), 30998 (E); Mt. Pi-iou-se, above Tapin- tze, 11 June 1883, red s.n. (P); Tali (Dali), Mt. Che- tcho-tze, 9 May 1883, Delavay s.n. (P); Tali (Dali), Mt. Che-tcho-tze, 10 Па 883, Delavay s.n. (P). Deqen: Benzilan, 3 km W of Susong village, Hong, Pan, Yu & Volume 85, Number 4 Hon al. ng e 563 Баји delavayi Complex Figure 6. Distribution of Paeonia delavayi. Dai H97119 (A, CAS, K, MO, PE, US); Cizhong to Yongzi, by the Lancang River, К. M. Feng 5765 (KUN, PE); m ling, Nanshui Beidiao Exped. 9194 (К UN ‚ РЕ); E flank o i Shan, jon zilan от ghai-Xizang | Exped. me (KUN, PE); ber Yonglobu (Forest Farm), Xizang Exped. 1878 (KUN, PE); Benzilan, Dongzuling, Qinghai-Xizang Exped. 2209 (KUN, PE); no precise lo- Mekong-Salwin divide, 28°12'N, Forrest 16339 (E), Forres 16527 (E). H Laugkong-Hoching divide, 26^16'N, Forrest 10062 (E): Baiya, Sanxi, R. C. Ching AM PE); Songgui, Mt. Maer, R. C. Ching 24191 (KUN, PE). Jianchuan: Xin- sheng Township, Huajiaping, P Y. Mao 236 (KUN). Kunming: Shanqing temple to Shitoushan, Liou 20677 (PE); West Hill, Hong, Pan, Yu & Dai H97077 (A, CAS, К, MO, PE, US) ‚В. ү Qiu 51019 (KUN), B. Y. Qiu 57101 ~ of Xiaoyanjing, Hengduan Mt. Ехреа. 908 (PE). Lijiang Yulong Snow Range, Ganghaizi, Hong, Pan, Yu & Dai H97095 (A, CAS, K, MO, PE, US), Qinghai-Xizang Ex- ped. 201 (PE), Edinburgh Bot. Gard. Exped. 85-6 (E, KUN), R. C. Ching 20438 (KUN, PE), Z. W. Lu pa (KUN), Kingdon- кеш 238 (E); Yulong Snow Range, zhugou to Ganghaizi, X. Zhou 1030 (KUN), 5. to Canghaizi Lijiang Bot. Gard. 100486 (KUN); Yulong ~. Snow Range, Yunnan Univ. Biol. Dept. Vegetation Ехреа. 233 22. К. C. Ching 30987 (KUN, PE); Yulong Sn now Range, Yuhu, Xuesong aros Mt. Beibazi, R. с 30175 (KUN, PE), K. 2 village, К. C. Ching diruo, K. M. бы 22229 (KUN): . K. M. Ee 23029 (KUN, PE); Y ange, Baishui, Hong, Pan, Yu & Dai H97102 (A, CAS, К, MO, PE, US); Yulong Snow Range. Yunshanping, Hong, Pan, Yu & Dai H97103 (A, CAS, К, MI, PE, US); Yulong Snow Range, Mahuangba to Wutoudi, Z. H. Yang & Y. C. Cai 101777 (PE); Yulong Snow Range, Jiuzihai to Yulong Snow Range, Heishui, J. 5. Yang 4157 (KUN): Heishui River, left slope, Lijiang Bot. Gard. 100026 (KUN); Yulong Snow Range, Mt. Gyina Loko, Rock 24984 (E); E flank of Yulong Snow Range, Forrest 5716 (E), T. T. Yü 15016 (E, PE); Mt. Wenbi, T. T. Yú 8107 (KUN, PE); northern Part, Hongshiyan, R. C. Ching 20595 (KUN, PE); Shapingze, T. T. Yü 5160 (KUN, PE), T. T. Yü 5163 (КИМ, PE); Sixth District, X. L. Wang s.n. (СРВ); Zili, by the river, R. C. Ching 22192 (KUN, PE); S of Lijiang, Sungkwe Pass between Lijiang & Heqing, Rock 251 79 (E): Mao 01565 (KUN). Ninglang: Gouzuandong, J. 5. Yang 4123 (KUN); Yongning, Geao Pass, S. (KUN); hills around Yongning, Forrest 12503 (E), 12565 (E); Yongning, Shize Shan, Т. T. Үй (PE); Yongning, Wen- quan Township, near Wenquan, Dai, Li & Tang 64–2 564 Annals of the Missouri Botanical Garden (CPB), Liujiashe, Dai, Li & Tang 64—3 (CPB); р 5. 4 Tang 64—4 (CPB); E of ш Yangtze bend, 100°45'E, Kingdon-Ward 3981 (E); no precise аы а. McLaren 5 (E), 89 (E), Kingdon-Ward 5055 (E). Weixi: Yezhi, by 2.” (Mekong) River, К. M. Feng 4220 (KUN, PE); Yezi, C. W. Wang 68199 (PE); Yangtze- Ме- kong divide, 29°45' М, Handel-Mazzetti 8868 (E, Pantiange District, Wucun village қ A Yongsheng: Shunzhou District Shicanauo, J. S. Yang 4402 (CPB). Zhongdian: 23 km NW of Zhongdian Town, Hala village, Hong, Pan, Yu & Dai H97112 (A, CAS, K, MO, PE, US); 45 km N of рі а Томп, Gezan Town- ship, Hong, Pan, Yu, Dai H97128 CAS, К, MO, РЕ, U , Gongbi, К. е Feng ee (KUN, PE); Zhondisn to Annachang, K. M. Feng 938 (KUN, PE); Tuguancun to Haba di A Zhongdian Ехреа. 63— 2389 (KUN, PE); Haba, Longwang Binsanba, Zhongdian Exped. 63-2738 (KUN, PE); Huangdong, Reshuitang, Zhongdian Exped. 63-2597 (KUN, PE); Qiaotou to Xiao- Feng 885 (КОМ, PE); — 2: = Baidi, 7. Т. Үй 11389 (KUN, РЕ); T. Т. Үй 14915 (РЕ); Tuguancun, Zhongdian Exped. 63-2375 (KUN, РЕ); Haba, Annazai, Zhongdian Exped. 63-2647 (KUN, PE); Haba, near Longwanbin, Zhongdian Exped. 1634 (KUN, PE); SE of Zhongdian Town, Jiuluo, gr Exped. 929 (KUN, PE) no precise locality, 50'N, Forrest 15162 (E); Haba Shan, N of the Yangtze n. Rock 24758 (E); no precise locality, 27°30'N, Forrest 1256 ЦЕ); Chih- ren, T. T. Yü 11247 (E, PE); N of Zhongdian Town, Tonwa Territory, Rock 24717 (E). Paeonia delavayi has the widest geographical range of any member of section Moutan (Fig. 6). The plants reproduce predominantly vegetatively, and cloning by stolons (Fig. 1) was commonly seen in every population visited except for population H97103, in Picea likiangensis forest in Yunshan- ping Lijiang, Yunnan. Vegetative reproduction is particularly prominent in West Sichuan (Dawu, Ya- jiang) and Southeast Tibet, where seedlings have never been found and clones often covered large areas. The roots of P. delavayi are always fusiformly thickened (Fig. 2). Such roots and stolons probably make the species more adapted to open, somewhat dry and disturbed habitats, and enable the species to become established very rapidly in a given pop- ulation. It may also account for the scattered dis- tribution of the species and the large number of "individuals" lavayi has been listed as an endangered species in the China Plant Red Data Book (Feng, 1992). On the basis of its vegetative reproduction and rela- tively wide distribution, however, it is reasonable in a given population. Paeonia de- to conclude that it will not become extinct if wanton oue is controlled. aeonia delavayi is most closely related to P. 4. from which it is readily distinguished by having predominantly vegetative reproduction by stolons and fusiform roots, a non-caespitose habit, stems 0.2-1.8 m tall, segmented leaves with narrow and acute segments, variously colored floral parts (petals, stamens, disk, and stigmas), a 2-8-pistillate gynoecium, and small (2-3.5 X 1-1.5 cm) follicles rarely producing seeds. In contrast, P. ludlowii is obligately sexual, without stolons, and has slender, terete roots, a caespitose base forming a large clump with dozens of stems, stems 2-3.5 m tall, lobed leaves with short and acuminate lobes, uni- formly yellow floral parts (petals, stamens, disk, and stigmas), a l(or very rarely 2)-pistillate gynoecium, and larger (4.7-7 X 2-3.3 cm) follicles always bearing seeds. In fact, P. delavayi and P. ludlowii are better considered as members of two separate species complexes Literature Cited Fang, W.-P. 1958. Notes on Chinese paeonies. Acta Phy- totax. Sin. 7: 297-323. Feng, K.-M. 1992. Paeonia delavayi Franch. var. lutea (Delavayi ex Franch. ) Finet & Gagnep. Pp. 530-531 in ,-K. Fu & J aig Plant Red Data gnepain. 1904. Contributions 5 | Flore de % TUM ам Bull. Soc. Bot. Fra 524—525. Franchet, A. 1886. Plantas yunnanensis. Bull. Soc. Bot. France 33: 358-467. Gong, X. 1990. A Taxonomical Study of the Раеота de- "pud complex. Master's Thesis, Kunming Institute of (one X. d J. Gu & Q. A. Wu. 1991. A cytological ed of seven populations in Paeonia delavayi var. lut Acta Bot. Yunnan. 13: 402-410. Hong, D.-Y. 1997. Paeonia (Paeoniaceae) in Xizang (Ti- bet). Novon 7: 156-161. Pan, K.-Y. 1979. Paeonia. In: W.-T. Wang (editor), Fl. Reipubl. Popularis Sin. 27: 37—59. Science Press, Bei- jing. . 1993. Paeoniaceae. Pp. —546 in W.-T. Wang & S. ©. Wu (editors), Vascular 1. of in Hengduan Mountains. Science Press, Beijing Stern, F. C. 1931. Paeony species. ji Roy. Hort. Soc. 56: 71-71 946. A Study of 7 Genus Paeonia. Royal Hor- ticultural T Londo & G. он . 1953. Драва lutea var. ludlowii. . 209. ) 1984. Index florae yunnanensis. 1: 122- -123. The People's Publishing House, Kunming, China. A CLADISTIC ANALYSIS OF Jason C. Bradford? SPECIES GROUPS IN WEINMANNIA (CUNONIACEAE) BASED ON MORPHOLOGY AND INFLORESCENCE ARCHITECTURE! ABSTRACT кч 2... is а woody p of about 150 d widely distributed in the tropics and the southern temperate zone. Herbarium and living "es ns were examined to determine characters for a cladistic analysis to test the жене чем of the genus puc its sectio d matrix of - vun ie a morphological characters was analyzed to find the most РА trees. The strict consensus cladog su s the monophyly of the genus Weinmannia and sections Leiospe , Weinmannia (including sect. Simplicifoia), e and Spicata, but section Fasciculata is paraphyletic 4. iii ct to a highly derived section Weinmannia. Section Leiospermum from the South Pacific is the sister taxon to the rest of the genus. Some of the most parsimonious trees support a monophyletic clade from Madagascar of sections Inspersa and Spicata, but this node is unresolved in the strict consensus tree. Although the deep nodes of the phylogeny are not well supported, and few evolutionary interpretations are ventured, it appears that dioecy has arisen independently at least three times in the genus. The method n: theory used to analyze variation in inflorescence architecture, which emerges from the metameric construction of plan а positional homology, are emphasized. The cladistic coding of positional characters and the traci ng of their 1. on the cladogram аге a study іп heterotopy, 5$ pd change in the position of development. ic evolution i i on in eterotop ence 15 common mannia, Puggesting a role for an evolutionary- 2 process that has typically been overlooked in favor of oikee rony SYSTEMATIC BACKGROUND genera that have been treated as separate but re- . 2, lated families are included (Bauera, Brunellia, Eu- Weinmannia is- a. genus of м Meee апа cryphia). However, Orozco (1997) advocated split- shrubs common to montane tropical and southern temperate floras. With approximately 150 species, it is the largest genus in the Cunoniaceae, a mor- phologically diverse family that includes 27 genera and ca. 370 species. The family has been consid- ered a distinct, isolated lineage within the Rosidae (Dickison & Rutishauser, 1990) that is morpholog- nonia and Pancheria, which was nested among a ically recognized by its woody habit, usually inter- large, monophyletic group of genera characterized petiolar stipules, toothed leaves, and decussate leaf by the presence of a stylar canal and a fused, bi- arrangement (Hufford & Dickison, 1992). A cladis- —carpellate ovary. tic analysis using morphological and anatomical The Cunoniaceae are austral in distribution: 15 characters (Hufford € Dickison, 1992) supports of the genera occur in Australia and Tasmania, 9 the monophyly of the Cunoniaceae only when a few іп New Guinea, 11 in the South Pacific, and ting the family to exclude Brunellia and its putative Cunoniaceae relatives. Davidsonia may also belong in a monophyletic Cunoniaceae (Doweld, 1998). Hufford and Dickison's (1992) consensus clado- gram placed Weinmannia in a small clade with Cu- ! My doctoral dissertation research at Washington University has been supported by the Missouri Botanical Garden through the Mellon Foundation, U.S. National Science Foundation training grant BIR-9256779 to Washington University, U.S. National Science Foundation Dissertation Improvement Grant #57479, and a travel grant from the American Society of Plant Taxonomists. I especially thank Helen Fortune Ho pkins for detailed discussions of characters, and for showing inflorescence architecture of Malesian-Pacific species and R. Hoogland's unpublished work in the family. I gratefully acknowledge the hospitality of scientists and support staff from institutions around the world who have helped with information, transportation, communication, food, and lodging. The Field Museum of Natural History provided useful material on loan. The suggestions of Paul Berry, Bill 2. Mike Grayum, Helen Fortune Hopkins, Neil Snow, Steve Wagstaff, and an anonymous reviewer improved this paper. ? Missouri Botanical Garden, P.O. Box 299, St. Louis, Missouri 63166. U.S.A. ANN. Missouni Bor. GARD. 85: 565—593. 1998. 566 Annals of th Missouri Botanical Garden only 5 (3 endemic) in South America. Like the Cu- Table 1. А comparison of previous sectional classifi- noniaceae in general, the present-day distribution of Weinmannia suggests a Gondwanaland origin (Raven & Axelrod, 1974). Weinmannia is wide- spread relative to other genera in the family, and its distribution includes the following geographic regions: Central and South America and the Carib- bean islands; Madagascar, the Comores, and the Mascarenes; Malesia (the Malay Peninsula, Greater and Lesser Sundas, the Philippines, Celebes, Mo- luccas, New Guinea, and the Bismarck Archipela- go); and the South Pacific high-elevation islands, from the Solomons and Vanuatu in the west to the Societies and Marquesas in the east, including New Caledonia and New Zealand. About half of the spe- cies occur in tropical America, a quarter are con- centrated in Madagascar, and another quarter are distributed in the vast Malesian-Pacific region. Sympatric species of Weinmannia are common in middle- to upper-elevation montane forests of the Andes and Madagascar (Humbert & Darne, 1965; Ríos, 1986; Kelly et al., 1994; Gentry, 1995). Spe- cies can have radically distinct life forms that cor- respond to local habitats. For example, at upper elevations, especially in exposed, windswept con- itions, Weinmannia species are typically small, sometimes prostrate shrubs, whereas in eastern Madagascar some species are among the largest trees in the lowlands. Weinmannia is not as locally species-rich in Malesia and the Pacific, but species are often locally abundant in this region, E te in the Society Islands (Florence, 1982; Fosberg, 1992) and New Zealand (Wardle & MacRae, 1966). The genus typically has small, fragrant flowers that attract bees (Renner, 1989; Bradford, pers. obs.), and small, hairy seeds easily dispersed by wind (J. Bradford & H. C. F. Hopkins, pers. obs.). Engler's (1928) revision of Weinmannia divided the genus into two sections, section Leiospermum and section Euweinmannia, based upon variation in the persistence of the calyx in fruit. Bernardi (1961, 1963b, 1964) incorporated variation in the form of the floral nectary, raceme structure, and leaf complexity to recognize six sections: section Ra- cemosa, which corresponds to Englers section Leiospermum and has caducous sepals, and five other sections corresponding to Engler's section Eu- weinmannia. 5” is an illegal name (see Greuter et al., 1994, Article 21.3) and is no longer used.] In general, the al of Ber- nardi is followed, although it is recognized that En- glers section Leiospermum has priority over Ber- nardi's section Racemosae. Also, Bernardi's section Simplicifoliae has been sunk into section Weinman- nia. Section Simplicifoliae separates species with cations of Weinmannia vs. the one usec Bernardi (1961, 1963b, 1964) Engler (1928) Bradford (present paper <~ Leiospermum Racemosae Leiospermum Weinmannia Weinmanniae vuweinmannia Weinmannia Simplic ifoliae Euweinmannia Fasciculata Fasciculatae Euweinmannia Spicata Spicatae Euweinmannia Inspersa Inspersae Euweinmannia simple leaves from the compound-leaved species in section. Weinmannia and appears to have been erected by Bernardi for taxonomic convenience. However, the number of leaflets often varies within individuals, is extremely homoplastic among spe- cies, and, therefore, has little taxonomic value at the sectional level. Previous and current classifi- cations are compared in Table 1 PURPOSE Recent systematic studies of Weinmannia have identified more characters than were used by Ber- nardi (1961, ‚ 1964) to circumscribe sec- tions. Many new characters, especially from inflo- architecture, are Thirty-one qualitative characters are used in a cla- distic analysis to resolve these systematic ques- tions: (1) Is Weinmannia monophyletic? (2) Are the sections monophyletic? (3) How are sections related to each other? and (4) What е. сһаг- acter states support each clade? Answers to these questions may help interpret biogeographic and character-distribution patterns. For example, a dioecious to polygamodioecious breeding system is nearly ubiquitous in the Male- sian section Fasciculata, among most species of the South Pacific section Leiospermum, and in the Mas- rescence escribed here. carene island species of section Weinmannia. An understanding of whether dioecy is derived or basal within the genus and the number of times it has evolved depends upon the relationships inferred among taxa. Because most sectional delimitations (е.в., Madagascar and the Comores), the cladistic rela- tionships among species groups of Weinmannia may inform us about area relationships among southern continents. To understand the cladistic coding and the evo- lution of the inflorescence in Weinmannia, it is nec- essary to describe inflorescence characters in de- tail, and to discuss the methods and theory behind Volume 85, Number 4 1998 Bradford 567 Species Groups in Weinmannia determinations of homology. Three points are em- phasized: (1) the inflorescence is formed by the spatial arrangement of nested sets of parts, (2) the coding of inflorescence characters usually requires an assessment of positional homology (at least for this study), and (3) evolutionary change in the po- sition of inflorescence units is a form of heterotopy. OVERVIEW OF INFLORESCENCE ARCHITECTURE IN WEINMANNIA An attempt to apply general inflorescence ter- minology, such as that of Briggs and Johnson (1979) and Weberling (1989), was not satisfactory. Struc- tural definitions only approximate many inflores- cence features of Weinmannia, or terms cannot be applied consistently among species. In a later sec- up I discuss why standard terminology does not ork in Weinmannia, especially when the goal is identifying cladistic characters. But first, the ter- minology used here to describe Weinmannia inflo- rescences is introduced. Following the system of Briggs and Johnson (1979), flower-bearing axes in nonia can precisely be called “anauxotelic botrya,” which includes forms of “spikes” and “pseudora- To avoid these cumbersome terms in all Weinmannia and Cu- cemes." subsequent discussion, the term *raceme" is used broadly to include any unbranched, flower-bearing axis. The terms Inflorescence Module (IM) and To- tal Gee (TI) will refer to two other distinct levels of organization. Architectural variation in the Weinmannia inflorescence occurs at up to three hi- erarchical levels: (1) the organization of flowers along a “raceme,” (2) the development of racemes in conjunction with supporting stems and buds (IM), and (3) the position these raceme-stem units occupy in relation to the main stem axis (TI). Flowers vary in their organization along the ra- ways: (1) how they are initiated and (0) whether their relative positions change durin axis elongation (Fig. 1). Flowers are initiated soli- tarily or grouped, in the ах! of a bract. During elongation of the raceme axis, flowers may remain in the axil of the bract or move away from it and each other. Differences in the positional relation- ships among flowers give distinct forms to the ra- eme in two w ceme In most sections, racemes usually develop as parts of developmentally integrated units (modules) composed of internodes, nodes, meristems, and ra- cemes. Within are organized predictably and may be repeated along the main stem axis in predictable patterns (Fig. 2). Each of these raceme-stem units will be a species-group, raceme-stem units ны to as an Inflorescence Module (IM). The term “module” has had more or less precise usage (see White, 1979; Grimes, 1992; Barlow, 1994). Here it refers to a particular level of organization, or integration of a group of metamers, to form a natural structural unit (Wagner, 1996). Of course higher and lower levels of organization, such as the raceme, are modules, but may already have useful names. The IM is at an intermediate level of or- ganization, above the raceme but below the entire inflorescence. Within a locally dominant stem sys- tem, the structure of the racemes, the organization of racemes into an IM, and the arrangement of IMs along the main axis constitute the Total Inflores- cence (TI). DESCRIPTION OF INFLORESCENCE ARCHITECTURES IN WEINMANNIA Inflorescence architecture is taxonomically infor- mative and may help identify an entire section, or a portion of a section, or distinguish a single spe- cies from other members of its section. Variation occurs in: (1) whether racemes develop as part of an IM or directly along the main stem axis, (2) the form of the IM, (3) the number of main stem meta- mers that bear racemes or IMs, and (4) the position of IMs relative to the main stem axis. There are 17 terminal taxa of Weinmannia in the data matrix Appendix 4). When variation within the raceme ie. that shown in Fig. 1) is ignored, there a about 11 distinct inflorescence architectures among am taxa, which are described below (A) Twelve of 18 species in section Fasciculata have supernumerary lateral buds that develop into lateral IMs in series at a node (e.g., W. fraxinea). These IMs each consist of one metamer with an opposite pair of racemes and a vegetative bud be- tween them. The apical meristem of the main stem axis usually remains vegetative, although vegetative growth is usually suspended during reproduction. The TI is acrotonic in that the distal node of the TI usually has more IMs per series than occur at subdistal nodes (Fig. ( ost other members of section Fasciculata have similar IM morphology, but do not develop Ms in a series at a node. Also, while it is rare for IMs to form in medial positions in the W. fraxinea group, this is common in some other species in the section (Fig. 2b) (C) Weinmannia descombesiana has a highly var- iable inflorescence architecture that may appear most similar to that of the W. fraxinea group (Fig. 3a) or other members of section Fasciculata (Fig. 568 Annals of the Missouri Botanical Garden 5 Е E E Early Middle Developmental Stage INS-ven FAC- fraxinea (b. form also) FAC-pullei FAC-clemensiae Cunonia pulchella SPI-icacifoli 522-445 FAC-fraxinea (a. form also) Cunonia-macrophylla Cunonia-purpurea 22... Cunonia-bal Cunonia аа Cunonia-alticola d. SPI-bojeriana INS-louveliana Late Figure l. The distribution of flowers along a "raceme" is determined by me ied of flowers initiated i in the axil of a bract and whether the flowers remain in the ax К as the axis elongates: Solitary inception + ber аан —d. Solitary arily to scale and only a portion of the а analysis with its r (SPI). See discussion of characters 19 and 20 (Appendix 3) and the character matrix (Appendix 4). --Һ. Fasciculate inception + floral migration. —c. migration. Parts not neces Weinmannia and Сипота OTU in the cladistic (D) Nearly half the species in the genus are in section Weinmannia, which can be recognized by its unique inflorescences. In this section, the inflo- rescence is limited to a pair of racemes developing from axillary buds at the most distal node of the main stem (Fig. 2c). There is no IM in this group and much variation in the suppression of leaf de- velopment at the node bearing the racemes. Within individual plants leaves may be fully developed at the raceme-bearing node or extremely reduced. (E) About 20 species in section Leiospermum have a characteristic inflorescence in which the apical meristem within an IM either develops into a raceme or aborts. In many species this is fixed one way or the other [e.g., the bud almost always axis is shown. The adjoining table lists each raceme form. Flowers are sessile in section Spicata aborts in W. dichotoma (Fig. ЗЬ)], but in several species this character varies within individual plants. Also typical of this section is that the IM may consist of sequentially arranged metamers with long internodes. The position of the IMs along the main stem is both lateral and terminal. The TI is acrotonic, with the terminal IM often developing more metamers than the laterals (Fig. 2a). Leaf de- velopment is suppressed within the IM, but gen- erally less so at nodes proximal to the main stem when the IM consists of sequential metamers. Two species endemic to New Zealand have distinctive inflorescences. Weinmannia sylvicola is similar to W. dichotoma in that the IMs terminate with an aborted meristem, but differs in that the Volume 85, Number 4 1998 adford 569 Species Groups in Weinmannia medial vegetative bud Inflorescence Module b. LE = "Raceme" n f 1 т о г 9 е t S а с 1 e n c e (TI) leaf ca. 5 cm later vegetative bud Figure 2. Enn with IMs potentially 2 of se and medially. —b. Parts of sections /nspersa, „а pns laterally at vegetative nodes. See text roups, as not all patterns are shown. жаш, the actual arrangement is decussate. medial IM often produces two metamers. Also, there is usually only one lateral IM (Fig. 3c). (G) The other New Zealand species, Weinmannia racemosa, is unique within section Leiospermum in that the IM does not terminate in a raceme or abort, but produces a vegetative bud (Fig. 9b). This is the only species within the section in which vegetative growth continues beyond the inflorescence. Fur- thermore, the IM is only borne in a medial position. А more detailed discussion of the inflorescence in sections Fasciculata and Leiospermum is given in Hopkins and Bradford (in Hopkins, 19982). (H) Most species in the two Madagascan sec- tions, Inspersa and Spicata, have similar inflores- cences. About 22 species in section Spicata and 5 species in section /nspersa have IMs consisting of Diagram of inflorescence terms and examples of inflorescence architecture in Weinmannia. quential metamers, terminating ма а raceme, and ES 4. аға, and Fas vegetative bud, and borne medially and (мае at distal and subdistal nodes. —c. oa ossi miss or more complete descriptions of architectural variation within species Actual leaves may be t —а. Section ciculata, with IM one metamer, term simple or compound. Although the figures are in two one metamer with an opposite pair of racemes and a vegetative bud between them. These IMs may be positioned laterally at both distal and subdistal nodes along the main stem, as well as medially at the distal node. Racemes can also develop directly from axillary buds, but only at the most distal node of the TI (Fig. 2b). (Т) Five species in section Spicata (two of which, W. comorensis and baehniana, were placed in sect. Leiospermum by Bernardi, 1964) develop ra- cemes from axillary a along the main stem at sequential nodes (Fig. 3 (J) Weinmannia m (not illustrated) in section Inspersa has extremely plastic inflorescence development. IMs may be of one or more metamers, sometimes branched but usually not, and may ter- 570 Annals of the Missouri Botanical Garden e. Figure З. Examples of inflorescence architecture among Weinmannia species groups. —a. W. fraxinea (sect. Fas- ciculata). ie W. dichotoma (sect. Leiospermum sylvicola (sect. buc —d. ІМ. comorensis (sect. Spicata). —e. W. venusta (sect. Inspersa). Note that b кесе с jue aborted terminal meristems in the IM. See text for full к т minate in a bud ог a raceme. When reproduction is prolific, IMs develop at several nodes and may vary in form within a TI. In general, larger IMs develop from lower nodes. Weinmannia hepatica- rum, known only from the type collection, appears to be closely related to W. rutenbergii, but its inflo- rescence Жалы) is unknown. Bernardi (1964, 1965) placed W. rutenbergii in section Weinmannia. (K) Two species in section /nspersa, W. venusta and W. sp. nov. 1, have branched IMs. that bear racemes. The length of these racemes is variable within an IM. The IM may terminate either in a raceme or in a vegetative bud. The position of IMs along the main stem may be lateral and medial. The TI is basitonic, with the largest IMs developing at lower nodes (one of the variants is shown in Fig. 3e and includes only a subdistal node of the TI). Bernardi (1964, 1965) placed W. venusta in section Weinmannia. Most species in the closely related genus Cu- nonia have an inflorescence architecture similar to that depicted in Figure 9c, but often with IMs at two successive nodes of the TI. However, several species diverge ae this pattern (see illustrations in Hoogland et al., 7, and coding in character matrix, Appendix e photos in Plate 1 show some common forms f the inflorescence among Weinmannia sections. Additional color plates will be available from the World Wide Web by searching the species name in W*TROPICOS at http://www.mobot.org. THE USE OF INFLORESCENCE ARCHITECTURE IN CLADISTICS The hierarchy of inflorescence architecture and its iid ap ee have been recognized in diverse groups (Venkata jn 1965; Mabberley, 1975; Weberling, 1977; s & Johnson, 1979; Kaul & Abbe, 1984; Tucker. 1987; 2. 1988; Weberling, 1988; Schlessman et al., 1990; Grimes, 92; Liede & Weberling, 1995; rs et al., 1996; Timonen, 1998). While many studies have proposed pathways of inflorescence evolution based on comparative morphology and development, few have used cladistic methodology to do so. Volume 85, Number 4 Bradford 1998 Species Groups in Weinmannia Plate 1. Photos doc 'umenting. Weinmannia inflorescence arc "иес чиге illustrate di in Figure 8 2, 3, and 9. —i, Section Weinmannia; W. dryadifolia (J. C. Bra ын m ч ccn with Figure 2c. ii, iii, Section Fasciculata. —ii. W. fraxinea d .. Brad da compare with F За. —iii. W. clemensiae (Н. Е — 5011); сотраге with Figure 9c. — Section “Spica ; W. stenostachya Ti С. Ва. 650); eye with Figure 2b. v, vi, Section Leiospermum. — syl icola (J. C. pube 732); compare with Figure Зе. —vi. W. parviflora m C. Bradford 919); compare with Figure 572 Annals of the Missouri Botanical Garden dl. d2. d4. a. ure 4. Morphological evolution occurs through temporal and spatial modification of development. As shown here, Fi the ancestor us as racem or both. —dl. which the racemes develop from the terminal bud and heteroto shifts that are probably heterochronic (d Although inflorescence architecture is diverse in many taxa, cladists may have avoided inflorescence characters because of difficulties in assessing ho- mology. Progress came when Grimes (1992) ana- lyzed the inflorescence of the Pithecellobium-com- plex by breaking it into nested, repeated units. Similarly, Weinmannia inflorescences have nested components and, in order to identify cladistic char- acters, the relative positions of parts within each component can be compared. Recognizing position- al homology is the key to coding inflorescence char- acters. Ав a consequence of recognizing positional homologies, evolutionary changes in the positions at which parts develop, i.e., heterotopy (Haeckel, 1905; Gould, 1977; Sattler, 1988), can be inferred by tracing character transformations on the phylog- eny. Therefore, this work advances a view of evo- lutionary development that recognizes both tempo- ral and spatial transformation (see also Zelditch & Fink, 1996) HETEROTOPY: DOES ІТ HAPPEN? While it is well accepted that heterochrony caus- es the modification of form (Jong & Burtt, 1975; Gould, 1977; Raff & Wray, 1989; Kellogg, 1990; Boughton et al., 1991; McKinney & McNamara, 1991), positional changes in development have of- ten been either ignored or considered byproducts of underlying heterochrony (see discussions in Raff, terochronic, paedomorphic form in which the raceme axes are constricted. —d2. and not the axillary on racemes develop in in their plesiomorphie, axillary position, and from a new position Hete opic form. Heterotopy is recognized by changes in the relative position of parts (d2, d3), not just quantitative 1). ies developing from the leaf axils. The descendants (41-44) are heterochronic, heterotopic, Hete erotopic form in Heterotopic form in which the , at the terminus rochronic es. —d3. 1996; Zelditch & Fink, 1996). Heterotopic patterns are recognized by qualitative shifts in the positional relationships among parts, not just quantitative changes in relative distances among parts, whic may often be due to heterochrony (Fig. 4). For example, reproductive structures can shift from lateral to terminal positions, or from distal branches to the trunk. A possible outcome may be the evolution of monocarpy from polycarpy, or changes in how flowers are pollinated and fruit is dispersed. Distinct inflorescence modules that de- velop either male or female flowers are common in monoecious breeding systems [e.g., pistillate vs. staminate spikes in Quercus (Kaul & Abbe, and in corn]. Once male and female modules are established, they can be expressed at separate po- sitions within the plant, which may have functional enetits. Even though heterotopy has been mostly over- looked by evolutionary biologists, some obvious (and bizarre) examples of morphological changes in plants appear to be purely heterotopic. These in- clude: the switch in position of leaves and lateral buds in Utricularia (Sattler, 1994); the positional reversal of stamens and carpels in Lacandonia schismatica (Martínez & Ramos, 1989); and epi- phyllous inflorescences (Dickinson & Sattler, 1974, — © Plant morphologists have long accepted that spa- Volume 85, Number 4 1998 Bradford 573 Species Groups in Weinmannia tial and temporal changes in development contrib- 1 "translocation" ÍT сеуіп. 1905) and "phylogenetic shifting" (includ- ing homeosis, Zimmermann, n discussing the “morphogenetic” basis of plant form, Sachs (1982) suggested that mutations affecting gene reg- ulation could change the spatial and temporal ex- pression of developmental processes to account for morphological evolution. This notion is supported by developmental-genetic studies that show how sets of genes, especially ones encoding transcrip- tion factors (Sommer et al., 1990; Yanofsky et al., ), act to establish the position in which an organ will form (Bradley et al., 1997; Meyerowitz, 1997). Sattler (1988) discussed how both hetero- chrony and heterotopy are related to homeosis. The potential for heterotopy is also embodied in Sattler's term “homotopy,” referring to homology of position (Sattler, 1994). Taken together, the data from systematists, plant morphologists, and plant molecular biologists show that heterotopy is widespread among plants. This provides the conceptual and empirical framework on which positional homologies of inflorescence ar- chitecture are here coded. MATERIALS AND METHODS SAMPLING METHODS Except for recent work by Н. C. Е Hopkins (1998а-с; Hopkins & Florence, 1998) in the Ma- lesian-Pacific region, Weinmannia has not been re- vised at the species-level since Bernardi (1961, 963b, 1964, 1965). Numerous recent collections show that many currently recognized taxa are poor- ly circumscribed. Sampling of specimens was done to cover the distribution of phylogenetically infor- mative characters independently of previous spe- cies circumscriptions or determinations. For this reason, some taxa recognized as varieties of the same species by previous authors are members of different terminal taxa in this analysis. Because there will be changes in nomenclature and new species descriptions, both specific names and a set of specimens examined are listed in Appendix 1. My goal was to examine at least one specimen of each species of Weinmannia. The lack of a recent treatment of Malagasy and American species made sampling more uncertain in these areas. Data come from field studies in tropical America, Madagascar, Malaysia, and the South Pacific, and examination of herbarium specimens primarily at the Labora- toire de Phanérogamie in Paris (P) and the Missouri Botanical Garden (MO). Type material was avail- able for nearly all species from Madagascar and the Malesian-Pacific region. Several American species are known only from type specimens that were not available, but most of the characters used in this analysis were at least seen in illustrations or photos. Furthermore, the Neotropics are well represented by recent collections at MO and F, and I have col- lected extensively in the region. Species sampling of Cunonia relied on the taxonomy of R. Hoogland and his unpublished monograph of the genus at P. This study has revealed several undescribed spe- cies, each of which can be placed in an Operational Taxonomic Unit (OTU) with at least one described species. Many inflorescence characters are coded based on developmental potential, which is fixed in some taxa but variable in others (e.g., see discussion of character 23, Appendix 3; description of W. ruten- bergii, inflorescence J). To accurately code these taxa, groups with greater complexity and variety of inflorescence architecture, such as section Leio- spermum, were sampled more intensively than those with less, such as section Weinmannia. Poorly rep- resented species that are not known for all of their characteristics (e.g., of which fruits have never been collected) could be provisionally placed into an OTU, since an intact inflorescence with either flowers or fruits is sufficient to discriminate among the OTUs The observed morphological characteristics of hundreds of specimens were managed in two ways. Sketches and notes of specimens were made, and descriptive information on specimens was entered into a computerized database. The database fields were used to search for unique combinations of character states among species, and characters were reconfirmed by checking sketches and by re- peated specimen examination. The OTUs in the cladistic analysis were circumscribed according to the possession of unique combinations of character OTU was polymorphic for a character, it was due to poly- morphism within some Lii of that OTU. Identifying OTUs by ue combinations of characters could yield E iie OTUs. fore, each ingroup OTU was studied for potential autapomorphies, which are listed in Appendix 2. Most OTUs were arguably monophyletic, but a few were not. А separate cladistic analysis was run that removed the following OTUs to see if their absence affected the tree topology: FAC-descombesiana (possible hybrid taxon), FAC-clemensiae, SPI-ica- cifolia, INS-madagascariensis, LEI-serrata, Cunon- la-purpurea. states for the current character matrix. If an here- Annals of the Missouri Botanical Garden CHOICE OF OUTGROUPS The character matrix and cladograms of Hufford and Dickison (1992) were used to choose out- groups. All outgroups were part of the large clade that includes Weinmannia. Two of them, Caldcluvia paniculata and Spiraeopsis celebica, were consid- ered useful because they apparently lack many de- rived floral, would make comparison with Weinmannia difficult. Within the Cunoniaceae, Spiraeopsis (with 6 spe- inflorescence, or fruit characters that cies) has the autapomorphy of stellate pubescence on the leaves, but no obvious autapomorphies exist for the monotypic Caldcluvia s. str. (Hufford & Dickison, 1992). Within this data set, Caldcluvia paniculata 18 unique in possessing four stipules per node, as opposed to two stipules per node in other taxa (Dickison & Rutishauser, 1990). In Hufford and Dickison's (1992) analysis, the position of the monotypic Vesselowskya was poorly resolved, but in some trees this was the sister taxon of the Cunonia(Pancheria-Weinmannia) clade and, therefore, was included here. However, their matrix ad some incorrect scoring of characters in Vesse- lowskya, which has palmately compound leaves, flowers borne in racemes, and valvate sepals. These traits suggest a close relationship between Vesse- lowskya and the Lamanonia/Geissois-Pseudowein- mannia clade, rather than the Cunonia(Pancheria— Weinmannia) clade. The Lamanonia/Geissois—Pseu- doweinmannia clade has two features shared with Cunonia and Weinmannnia: a racemose floral axis and a fused vascular bundle between the fruit car- pels, the latter not included by Hufford and Dick- ison (1992). Possible autapomorphies for Vesselow- skya are its 3-merous perianth and androecium, and its decurrent stigmas. Ackama rosifolia was examined because this ge- nus (with 3 species) is the only one in the Cunon- iaceae other than Weinmannia with hairy seeds. However, for these hairs to be homologous with those of Weinmannia would require that Ackama be the sister taxon to Weinmannia, adding five steps to the shortest trees of Hufford and Dickison (1992). In fact, Ackama shares some possibly de- rived leaf trichome and domatium features with Spi- raeopsis. The monophyly of Weinmannia with respect to Cunonia was tested by treating Cunonia as part of the ingroup. Pancheria was excluded because its many derived floral and inflorescence features (Hufford & Dickison, 1992) make it difficult to rec- ognize corresponding structures in Weinmannia. For example, Pancheria has whorled phyllotaxy, flowers arranged in spherical heads, unisexual flow- ers with a reduced number of parts, and an apo- carpous gynoecium. Їп contrast, Weinmannia and Cunonia have similar floral structures and inflores- cences composed of racemes. Including Pancheria in an analysis with Weinmannia and Cunonia would require that many characters be scored as inappli- cable in Pancheria, and vice versa. Therefore, de- spite the possibility that Pancheria is the sister ge- nus to Weinmannia, it was not useful to include it in a morphological analysis at this level. CODING OF CHARACTERS Because phylogenetic studies attempt to discover relationships at different hierarchical levels, char- acters that vary within one set of taxa may be in- applicable in other taxa (what Pleijel, 1995, termed *hierarchical dependence"). Other coding problems may arise when there is non-independence of char- acters due to developmental or functional coupling. For example, stem architecture may be coded sep- arately from inflorescence architecture, although one may influence the othe plex characters” has generated some recent dis- cussion (Pleijel, 1995; Wilkinson, 1995), which has recognized the different assumptions and tradeoffs er. The scoring of “com- in different coding methods. For example, multi- state coding may reduce the problem of non-inde- pendence among characters, but may compound problems of inapplicable characters among taxa. Reductive coding (such as presence vs. absence) eliminates the problem of inapplicable characters among taxa, but may increase the frequency of non- independence among characters. Which tradeoffs or assumptions are most acceptable for a particular analysis may be difficult or impossible to know a riorl. Within this data set it was decided to adopt a moderate, consistent approach, as recommended by Hawkins et al. (1997). For features present and var- iable among some taxa and absent in others, the information was coded into two characters: a pres- ence/absence character and a character denoting character-state variation for those taxa for which the character was present. Taxa for which the complex character was absent were denoted as inapplicable for variation within the character by a dash (“—”) in the data matrix, and taxa for which character ta were missing were denoted by a question mark ("2"). For features that may be developmentally re- lated, the strength of this relationship was consid- ered before establishing a character (e.g., see dis- cussion about W. dichotoma for characters 3 and 27, Appendix 3). However, developmental studies Volume 85, Number 4 1998 Bradford 575 Species Groups in Weinmannia have not been done, nor are the pleiotropic effects of trait evolution known. he characters used in the cladistic analysis are discussed in Appendix 3. Several conspicuous characters that were not used also merit some dis- cussion. Leaf characters, such as imparipinnate vs. unifoliate leaves, winged vs. unwinged leaf rachis, and entire vs. toothed margins, varied within most species-rich OTUs used here and, therefore, could not test the monophyly of sections or their relation- ships. Variation in leaf complexity was considered to divide the species-rich OTU of American section Weinmannia, but these characters were difficult to score qualitatively (see discussion about Bernardi’s sect. Simplicifoliae) and would require the breakup of nearly all other multi-species OTUs. Another po- tential character that differed within the American OTU was caducous vs. persistent petals. However, this appeared related to the degree of opening of the sepals, a character more difficult to define, and not to the absence of petal abscission. Other char- acters that could not be defined with precision or were difficult to score for the majority of species included: variation in the thickness of the pericarp, whether the endocarp detaches in fruit, stipule shape, and leaf anatomical characters, such as whether areolation is complete, presence or ab- sence of a hypodermis, and the degree of lignifi- cation of tertiary vascular bundles. Such characters may be found useful with more detailed anatomical, morphological, or developmental information. CLADISTIC METHODS Data were entered into MacClade (Maddison & D. ‚ 1992), and analyses used PAUP version 1 бхон 1993) run on an Apple PowerMac E /100. All characters were scored as unordered (nonadditive) except for “Morphology of branches” (see discussion of character 2, Appendix 3), which was treated as an ordered (“Wagner”) character. No data were missing, but inapplicable entries ac- counted for ca. 7% of the data and were treated as “unknown” in the analysis. Polymorphisms were treated as “uncertainties,” because of the way char- acter data were partitioned in the data set. In PAUP, polymorphism is intended to represent variation among monophyletic clades (Swoffor Begle, 1993), such as between species, whereas the coding in this data set represents variation at the popula- tion level (i.e., variation within species). The heu- ristic-search option was implemented with 100 rep- licates of random-taxon addition using the TBR branch-swapping algorithm. Zero-length branches were collapsed. All shortest trees were saved, and was used to make consensus trees A bootstrap analysis (Felsenstein, 1985) was done with 1000 replicates, with one round of ran- dom-taxon addition for each replicate, using the heuristic-search option and TBR branch swapping. To prevent some searches from never swapping to completion, the “maxtrees” setting was held at 500 trees. Decay analyses were done by saving all trees one, and then two steps longer than the most par- simonious trees. RESULTS The first round of taxon addition found 192 most- parsimonious trees of 90 steps [Consistency Index С.І.) = 0.456, Retention Index (R.I.) = 0.745], and all subsequent searches located the same is- and of trees. Among the outgroups, Caldcluvia is arbitrarily used to root the cladogram. The strict consensus tree is shown in Figure 5 and includes bootstrap values of 50% or greater and decay val- ues. The search for trees two steps longer than par- simony could not be completed because the number of possible trees filled available memory. One of the most parsimonious trees is shown in Figures 6 and 7, and was chosen to represent the Malagasy OTUs sects. Spicata + Inspera) as forming a monophy- letic group. hen the six possibly non-monophyletic OTUs were removed, the strict consensus cladogram was less resolved than the one shown in Figure 5. Un- resolved nodes included the monophyly of Wein- mannia and the monophylies of sections Spicata and /nspersa. The node uniting sections Fasciculata and Weinmannia collapsed. The poorer resolution probably resulted from an inability to polarize char- acter states when plesiomorphic taxa were removed, ut the results were congruent with the complete — ж matrix tree. The monophyly of Weinmannia is supported with Cunonia as the sister taxon. One character syna- pomorphic for the Weinmannia-Cunonia clade, im- bricate calyx aestivation, was expected, but other synapomorphies are homoplastic within the clade, including fused lateral stipules and medial meri- stem abortion. An unbranched floral axis (e.g., ra- ceme), a fused fruit column, and IM development are synapomorphies for a Vesselowskya (Weinman- nia—Cunonia) clade. nonia has three distinctive apomorphies: a nectary whorl adnate to the carpel wall, circum- basal dehiscence of fruits, and a double internode on branches Monophyly of Weinmannia is supported by the 576 Annals of the Missouri Botanical Garden Mos H PE WEI-pinnata(73) AM WEI-tinctoria(2) MC FAC-fraxinea(13) ML* FAC-descombesiana ML | FAC-pullei(3) ML FAC-clemensiae(2) ML SPI-bojeriana(16) MD — PE SPI-icacifolia(6) MD SPI-comorensis(5) MD/CO INS-rutenbergii(2) MD ё 62{__ INS-venusta(2) MD | NS-louveliana(3) MD INS-madagascariensis(2) MD LEI-serrata(5 NC во] _[— ші-әуізісота NZ — LEI-racemosa NZ LEI-affinis(14) SP нека! Cunonia-macrophylla(3)NC/SA Cunonia-purpurea(13) Cunonia-aoupiniensis(2) NC х— 7 8 Cunonia-balansae(2) NC pe Cunonia-alticola(2) NC Cunonia pulchella NC Cunonia bullata NC + Vesselowskya rubifolia AS Ackama rosifolia NZ Spiraeopsis celebica ML Caldcluvia paniculata АМ ure 5. Strict consensus of 192 most parsimonious trees, and taxon-area cladogram. OTUs of species groups in Weinmannia are shown with the first three letters of their section capitalized followed by the name of one species. WEI = Weinmannia, FAC = Fasciculata, SPI = Spicata, INS = Inspersa, LEI = Leiospermum. The total number of species in each OTU is given in parentheses. For areas: AM = Americas, MC = Mascarenes, ML = Malesia (*with a few species in Melanesia), MD = Madagascar, CO = Comores, NC = New Caledonia, NZ = New Zealand, SP = South Pacific (Melanesia and Polynesia), AS — Australia, SA — South Africa. Numbers on branches are bootstrap values. Decay values are 1, unless indicated by a star at the Ше with * — d ** — decay value prob ably structed on the cladogram, unisexual flowers have evolved from bisexual ones at least three one in Weinmannia (solid bars), and the reverse has happened twice (open ba absence of seed wings and the presence of seed а perigynous perianth found only in New Caledonia hairs, even though these characters are also present and New Zealand. The two New Zealand species in Ackama. Another supporting character is the de- — are further distinguished by their inflorescences (as velopment of IMs in lateral and medial positions, previously discussed). but this character is relatively homoplastic overall. Four unambiguous synapomorphies unite all oth- The first split within the Weinmannia clade is er Weinmannia species: calyx persistent in fruit, between the South Pacific section Leiospermum and — fasciculate flowers, persistent floral bracts, and the rest of the genus. Section Leiospermum appears short IM shoots. Except for the character of fascic- to be monophyletic based upon two characters of — ulate/solitary floral inception (which changes 6 the inflorescence architecture: sequential arrange- — times in all most parsimonious trees), these char- ment of metamers within the IMs, and acrotonic acters are not very homoplastic, as each changes development of IMs in the medial position. Within only 2-3 times in all most parsimonious trees (Fig. this section is a group of 14 species with unisexual 8). However, this clade is not supported by boot- flowers widespread in Melanesia and Polynesia, strap or decay analyses (Fig. 5). and a clade of 7 species with bisexual flowers and Relationships within this larger Weinmannia Volume 85, Number 4 Bradford 577 1998 Species Groups in Weinmannia WEI-pinnata(73) AM WEI-tinctoria(2) MC Ш FAC-fraxinea(13) ML Ш FAC-descombesiana ML zum FAC-pullei(3 ML TI dominance mE EH P pn EHE] FAC-clemensiae(2) ML [ — ]basitonic EH SPI-bojeriana(16) MD ШШ isotonic на SPI-icacifolia(6) MD ШЕШ acrotonic Р ^ SPI-comorensis(5) MD/CO ES equivocal T ОШОО П INS-rutenbergii(2) MD [] INS-venusta(2) MD = EE INS-louveliana(3) MD E INS-madagascariensis(2) MD g LEI-serrata(5) NC B LEI-sylvicola NZ LEI-racemosa NZ lg LEI-affinis(14) SP Cunonia-macrophylla(3)NC/SA Cunonia-balansae(2) NC Cunonia-aoupiniensis(2) NC B Cunonia-purpurea(13) NC EE] Cunonia-alticola(2) NC БЫШ Cunonia pulchella NC B Cunonia bullata NC Vesselowskya rubifolia AS HB Ackama rosifolia NZ ЈСО Spiraeopsis celebica ML 30 Caldcluvia paniculata AM Figure 6. evolution of character 31: “TI dominance. ” clade are not fully resolved. Although the mono- phyly of sections Inspera and Spicata is supported, the relationship of these clades to one another and to the Weinmannia—Fasciculata clade is uncertain. Assuming that the Malagasy sections are most closely related to one another can resolve this po- One of the most-parsimonious trees showing a monophyletic group from Madagascar and tracing the lytomy as shown in Figures 6 and 7. However, there are no unambiguous synapomorphies for the Mal- agasy clade. Section Spicata has only one recog- nized apomorphy, sessile flowers. Section /nspera has two apomorphies, ribbed nectaries (also present in sect. Weinmannia and erratically in sect. Fas- 578 Annals of the Missouri Botanical Garden Floral inception Ш wEI-pinnata(73) С] solitary ШЕШ fasciculate uncertain B WEI-tinctoria(2) B FAC-fraxinea(13) [] FAC-descombesiana B РАС-ри11е1 (3) B FAC-clemensiae(2) [] SPI-bojeriana(16) B sPI-icacifolia(6) B sPi-comorensis(5) [] INS-rutenbergii(2) B INS-venusta(2) О INS-louveliana(3) Ш INS-madagascariensis(2) MD О LEI-serrata(5) O LEI-sylvicola [] LEI-racemosa MD NC NZ NZ SP =j] [] Cunonia-macrophylla(3)NC/SA О Cunonia-balansae(2) NC О Cunonia-aoupiniensis(2) NC [] Cunonia-purpurea(13) NC [ES Cunonia-alticola(2) NC JÜ LEI-affinis(14) - Ш Cunonia pulchella NC 7] JO Cunonia bullata NC _1П Vesselowskya rubifolia AS [] Ackama rosifolia NZ [] Spiraeopsis celebica ML JO Caldcluvia paniculata AM Figure 7. Floral inception (character 19) traced on one of the most-parsimonious cladograms. ciculata) and seeds covered with dense hairs (some- times found in sect. Fasciculata). In general, spe- cies-groups in Madagascar show a great diversity of characters relative to other taxa. No most-parsimonious reconstruction supports the monophyly of section Fasciculata, which in this analysis is paraphyletic with respect to a highly differentiated section Weinmannia. These two sec- tions are united by the absence of persistent, fused lateral stipules and the absence of a bud at the base of the axillary shoots. The node uniting the species of section Wein- mannia, which are disjunct between the Americas and the Mascarene Islands, is the best supported, Volume 85, Number 4 1998 Bradford Species Groups in Weinmannia 579 T alll гора о > N OQ ~ л о ~ LL. 1 2 _ 1 а D) ГГ 1 1011 шиши Character Figure 8. The number of steps (y axis) per character (x axis) calculated over all most-parsimonious trees, showin the upper bound and minimum values. Characters а represent variation in . These characters comprise 2646 of the characters in the matrix (8/31), and about 29% of the total number of steps 21 When all inflorescence cha values are 4296 (13/31) and 4496 (39—40/90), а demonstrating that they do not have the level of the raceme axis (as shown in Figs. 2, 3, € acters 19—31), these the inflorescence architecture above racters are included (char- levels of homoplasy significantly different from traditionally used charact with a bootstrap support of 9396 and a decay index of >2 (Fig. 5). changes: ribbed nectaries, sparsely distributed seed hairs, reniform seed shape, absence of IM devel- opment, presence of development of racemes along the main stem axis, and only one TI metamer. With- in section Weinmannia, the Mascarene species have his node has six unambiguous two distinguishing characters: unisexual flowers formed by late suppression, and 5-merous flowers (in contrast to bisexual, 4-merous flowers in Amer- ican species). Аз noted previously, most American species have caducous petals and may be a mono- phyletic sister taxon to the Mascarene clade. The American species appear monomorphic in this analysis, which relied on qualitative character var- iation, but are highly diversified in terms of leaf shape, distribution and form of pubescence, and size of racemes. DISCUSSION SYSTEMATIC IMPLICATIONS At the generic level, this analysis supports the continued recognition of Weinmannia and Cunonia as separate genera, a distinction that has been questioned by some systematists (Cronquist, 1981: ). However, more systematic studies within the family are needed to be more certain of cladistic relationships among genera. ithin Weinmannia, the sections proposed by Bernardi (1961, 1963b, 1964) appear monophylet- ic, with the exception of section Fasciculata, with some rearrangement of species indicated. Specifi- cally, Weinmannia descombesiana on the one hand, and W. comorensis and W. baehniana, on the other, do not belong in section Leiospermum, but in sec- tions Fasciculata and Spicata, respectively. Wein- mannia venusta and W. rutenbergii do not belong in section Weinmannia, but in section Inspersa (see Appendix 1). With these rearrangements, the mono- phyly of sections Leiospermum, Spicata, Inspersa, and Weinmannia (including sect. Simplicifolia) is supported. Section Fasciculata is paraphyletic with respect to section Weinmannia, but several char- acters were scored as inapplicable in section Wein- тапта (since this section lacks IM development) and may have skewed patterns within section Fas- ciculata. Furthermore, preliminary results of mo- lecular-systematic studies (Bradford, unpublished) strongly support the monophyly of section Fasci- The poor resolution in Cunonia may have been due to the selection of characters in Weinmannia. An analysis focusing on Cunonia might be more informative. Interestingly, the South African Cu- nonia capensis, the only Cunonia species outside of New Caledonia, shares many characters with two other species, C. macrophylla and C. schinziana. Furthermore, these three species appear very dis- tinctive when compared to other Cunonia species. UTILITY OF ARCHITECTURAL CHARACTERS FOR WEINMANNIA SYSTEMATICS As judged by relative levels of homoplasy, inflo- rescence architecture is as informative as other sources of variation, such as vegetative, floral, and fruit characters (Fig. 580 Annals of the Missouri Botanical Garden Parts of the cladogram rely heavily on inflores- cence characters. For example, the monophyly of section Leiospermum is supported by inflorescence architecture alone, and section Weinmannia is sup- ported by three inflorescence characters. The inclusion of inflorescence characters has permitted the resolution of relationships within sec- tions for the first time. This is especially true in the species-rich section Spicata, which can be di- vided into three groups largely on the basis of in- i Inflorescence characters are also useful at the species level in section /n- florescence differences. spersa. It is often easier to recognize species using qualitative differences in the inflorescence than us- ing differences in leaves, which may be fairly sim- ilar among species and highly plastic within spe- es. The fact that inflorescence characters have been overlooked is illustrated by Smith (1985), who con- sidered W. richii and W. vitiensis, both from Fiji, as fairly indistinct based upon leaves and flowers. However, at a glance one can recognize that ra- cemes terminate the IMs of W. vitiensis (sect. Leio- spermum), whereas the IMs terminate in a bud and are borne serially in ЈУ. пећи (sect. Fasciculata). Although these characters have gone unnoticed, they are macroscopic, with clear patterns discern- ible by the naked eye and not requiring rehydration and dissection of parts, as is often necessary for flowers and fruits. THE CONCEPT OF POSITIONAL HOMOLOGY: IMPLICATIONS AND BENEFITS Heterotopy is the development of an organ in a different position in a descendant than where it had developed in the ancestor. This definition views the organ as primary and the position as secondary. An- other perspective of heterotopy is: at a given posi- tion, a different organ develops in the descendant than that which had developed in the ancestor. This view gives primacy to positional homology, and with this perspective it may be easier to understand that an organ can develop in its plesiomorphic position and in an apomorphic position within a plant. Con- sideration of the homology of position has two im- portant implications: (1) it broader set of characters, and (2) it may lead to raws attention to a more precise formulation of characters. The coding of the inflorescence using positional homology ex- emplifies these points. Often, the use of inflorescence characters in sys- tematics is vague (e.g., not”) because the level of organization is not spec- ified. By contrast, consider character 26, “Fate of “inflorescence terminal or ” IM terminus,” which accounts for the organ (ra- ceme, bud, or aborted meristem) that forms at a particular position, the terminus of an IM. This cod- ing specifies topographic information in the proper hierarchic context. Because such characters are po- sitional, patterns of heterotopy can be studied from cladograms. The mapping of character 31, “TI dom- inance," on the cladogram (Fig. 6) shows that TIs are generally isotonic within Cunonia and Wein- mannia, but that acrotonic TIs have evolved twice Figs. 2a, 3a) and basitonic TIs once (Fig. 3e). At the TI level, this is a positional change in domi- nance, but it is expressed by differences among IMs possibly caused by unequal timing of development. In other words, timing differences are among seri- ~ ally homologous IMs within a TI, whereas positional differences are among clades within Weinmannia. It should be stressed that nothing is being im- plied about the underlying molecular development of these characters. Heterotopy is only being used to describe patterns observed at the morphological level in a phylogenetic context. However, because the deep nodes of the cladograms are not well sup- ported (Fig. 5), extensive discussion of character evolution is not warranted. ORIGIN OF RACEMES In the discussion of character 22 (*Flower-bear- ing axis: branched/unbranched") it was suggested that if fascicled flowers (character 19) were ple- siomorphic in the cladogram, then the hypothesis that floral fascicles are homologous to flower-bear- ing short shoots would be supported. Mapping of character 19 on the cladograms does not support this hypothesis (Fig. 7). Fascicled flowers appear derived within the sister clade to section Leiosper- mum, but within this clade there are four reversals. The hypothesis is not completely rejected, however, due to the high level of homoplasy of character 19 and the lack of support for deep nodes. INFLORESCENCE CLASSIFICATION AND CLADISTICS: SOME OBSERVATIONS AND CAUTIONS Standard inflorescence terminology may inade- quately describe cladistic characters because tra- ditional terms frequently confound distinct char- acters and levels of organization. For instance, previous decriptions of inflorescence variation in Weinmannia | 1963а) горе three orms: “racemes,” “pseudoracemes,” and "spikes." The first pair of terms refers to the final stage of floral position along the axis but mixes the char- acters “Floral inception" caracte 19) and “Bract fidelity” (character 20; Fig. 1). The term “spike” че Volume 85, Number 4 1998 adford Species Groups in Weinmannia Homothetic compound raceme Figure 9. “homothetic compound racemes,” anc characters coded using the methodology in ix this paper, standard terminology a IM position IM metamers IM terminus Homothetic compound raceme с. lateral one raceme Inflorescence typology and rw 8. Three species and their inflorescences are shown, with two labeled ] one ^heterothetic compound raceme," follow wing Weberling (1989). Of three accounts for only one, IM terminus. That is why standard terminology classifies b and c the same, whereas this analysis suggests that a and b are most closely related. specifies flowers borne on an unbranched axis with- out pedicels and is therefore unrelated to variation in the distribution of flowers along the axis that is dealt with (poorly) by the other two terms. A cla- distic analysis that codified traditional terminology would therefore be comparing non-homologous parts. As mentioned in the description of inflorescence architecture, general systems of inflorescence clas- sification do not apply to many structures in Wein- T The various forms of the IM may loosely "compound racemes" following Weber- e (1989), although they do not fit the definition exactly. Standard terminology cannot be expected to describe in detail the unique aspects of a group, but details are crucial for cladistics. For example, compare the inflorescences shown in Figure 9, which are common in W. marquesana (Fig. 9a) and W. racemosa (Fig. 9b; sect. Leiospermum) and W. clemensiae (Fig. 9c; sect. Fasciculata). The inflo- rescence of W. marquesana can be called a *het- erothetic compound raceme," and the others can be called *homothetic compound racemes." The stan- dard terminology addresses whether the terminus of the IM produces a bud or a raceme. This fails to convey other characters important in Weinman- nia, such as the number of IM metamers and the position the IM occupies within the It is also difficult to apply standard names to the inflorescences of some highly plastic species, such as W. rutenbergii, especially since variation is ram- pant at different levels of organization. However, by 582 Annals of the Missouri Botanical Garden breaking down a complex structure into its parts, variation relevant to different hierarchical d can be distinguished. EVOLUTION OF DIOECY The cladogram (Fig. 5) shows that bisexual flow- ers are plesiomorphic in Weinmannia. Two clearly monophyletic groups (corresponding to sects. Wein- mannia and Leiospermum) have some members with bisexual and others with unisexual flowers (and which are more or less dioecious). Unisexual flowers are also most common in section Fascicu- lata, but because relationships in this portion of the tree are not well resolved, the pattern of char- acter evolution is unclear. Minimally, unisexual flowers have arisen three times, with a possible re- versal in W. descombesiana and in the branch of the section Weinmannia clade. Weinmannia flowers are small, simply structured, and mature more or less simultaneously within an inflorescence. The pollinators of Weinmannia prob- ably conform to the "generalist" category of small insects, especially bees, that Bawa (1994) and Beach (1981) suggested drove the evolution of di- duction of honey in New Zealand (Matheson, 1991; Walsh, 1978) and Madagascar (Ralimanana, 1994) It is noteworthy that dioecy is prevalent on is- lands, which has been suggested as a general trend (Baker, 1967; Baker & Cox, 1984). The only di- oecious species in section Weinmannia occur on the Mascarene Islands, and the dioecious members of section Leiospermum are distributed among the smaller South Pacific islands, with bisexual species on the relatively larger islands of New Caledonia and New Zealand. In the South Pacific, dioecy is usually imperfect (e.g., “leaky dioecy” or polyga- modioecy), which may promote the colonization of small, ephemeral islands (Baker € Cox, 1984). BIOGEOGRAPHY Weinmannia has attracted the interest of bioge- ographers because its distribution suggests that the genus was widespread on Gondwanaland, and its current disjunctions may have much to do with plate-tectonic processes and resultant continental vicariance (Good, 1950; Bernardi, 1963a; van Bal- gooy, 1971; Raven & Axelrod, 1974). However, ex- tensive distribution among South Pacific volcanic islands shows that long-distance dispersal can also be important in this genus. Because deep phylogeny of Weinmannia is not well supported, no firm conclusions about Gond- wanan-area relationships can be made. However, the well-supported sister-group relationship be- tween Атепсап and Mascarene taxa is impossible to explain without invoking dispersal, because the Mascarene Islands are of relatively recent, hot-spot volcanic origin (Patriat & Seqoufin, 1988). This dis- junction can be explained by one of two general classes of hypotheses: (1) long-distance dispersal between the Americas and the Mascarenes, or (2) short-distance dispersal from a non-extant source area from Africa, Madagascar, or other, older is- lands in the Indian Ocean. The second hypothesis implies an historically more widespread occurrence of section Weinmannia, with intervening extinction. If section Weinmannia is in fact a relatively young, highly derived lineage, as its current posi- tion in the cladogram suggests, then the long-dis- tance dispersal hypothesis may be favored. How- ever, the hypothesis of formerly widespread occurrence and short-distance dispersal gains some support from fossils. Oligocene whole-leaf fossils of either Weinmannia or Cunonia from Tasmania (Car- penter & Buchanan, 1993) look very similar to W. trichosperma from southern Chile. CONCLUSIONS AND FUTURE DIRECTIONS This cladistic study of morphological characters has identified congruent patterns of variation both within and among previous taxonomic groupings of Weinmannia. This helps to focus research into the origin and transformation of characters within Wein- mannia and related genera. While good support ex- ists for the monophyly of some sections, cladistic relationships and patterns of character evolution among sections are still poorly understood. How- ever, molecular-systematic work will be aided by a better understanding of the disparity in morpholog- ical diversity and species richness among clades, and the overall congruence of character variation with geography. It is likely that the methods used here to identify and cladistically code inflorescence architecture in Weinmannia and Cunonia will be useful in other taxa. This is certainly true within and among other genera of Cunoniaceae. An understanding of the nested, modular construction of plants and the po- tential for heterotopy in evolution may often be re- quired to code inflorescence architecture for cla- distic analysis. It is suggested that heterotopic patterns are prevalent in inflorescence evolution, and that coding of positional homologies is an ef- fective way to infer character-state polarities of in- florescence features. Volume 85, Number 4 Bradford 583 Species Groups in Weinmannia Literature Cited Baker, H. G. 1967. pon for Bakers law—As a rule. Evolution 21: 853- — —— & Р. A. Cox. pom Further thoughts on p 'ism and idands. Ann. Missouri Bot. Gard. 71: 244—253. Balgooy, M. M. J. van. x а -geography of the Pa- cific. Blumea Suppl. 6: Barlow, P. W. 1994. From coll * system: Repetitive units of growth in the development of roots and shoots. Pp. 19—58 in M. Iqbal (editor), Growth Patterns in Vascular Plants. Dioscorides Press, Portland, Oregon. Bawa, K. S. 1994. Pollinators of tropical dioecious ico Ке А reassessment? No, not yet. Amer. J. Bot. 81 bra J. H. 1981. Pollinator foraging po the evolution of 2 Amer. Naturalist 118: 572 Bernardi, Revisio generis Де did Pars I: Sectio Le Candollea 17: 123-189. ———. 1963а. 2. phytogéographiques et morphogénétiques sur le acées). por nia 3: 404-412. Ў Revisio generis Weinmanniae. Pars 11: Sectio simple Candollea 18: 285-334. — evisio i anniae. Par Ш: 22 Ш- IV-V-VI (veteris orbis). Bot. Jahrb. [om e Weinmannia (Cunoni- . 1965. Cnipa, V. 93 in H. Humbert (editor), Flore de Madagascar et des Comores. Muséum National d'Histoire Naturelle, Laboratoire de Phanérogamie, Par- IS. Boughton, D. A., B. B. _ & A. R. McCune. 1991. Heterochrony in jaw morphology 5. a (Te- leostei: Belonidae). Syst. Zool. 40: 3 Bradley, D., O. Ratcliffe, C. Vincent, Е 14 ter & E Coen. 1997. Inflorescence со and architecture in Arabidopsis. Science 275 Briggs, B. G. . Johnson. n up шы in the Myr- аселе «Епа фецсе from inflorescence structure. Proc. Li inn. Soc. New South Wales 102: 157-2 Carpenter, R. J. & A. M. Buchanan. 1993. Oligocene каи. fruit and flowers of the Cunoniaceae from Ceth- , Tasmania. Austral. Syst. Bot. 6: 91-109 «алшы А. 1981. Ап Integrated System of M esed epiphyllous inflorescence of P. (Turez.) Loes.: еи for comparative morpholog Bot. J. Linn. Soc. 69: 1-13 & -----. 1975. Developman of the epiphyllous аам of pilin japonica (Helwingiaceae). r. J. Bot. 62: 3. Dickison. W. С. Ts Studies on the floral anatomy of the Cunoniaceae. Amer. J. Bot. 62: 433-447 . 1984. Fruits and m of the Cunas, J. Arnold Arbor. 65: 149— R. пите жабын е Developmental mor- phology of stipules and systematics of the Cunoniaceae and presumed allies. II. Taxa without interpetiolar stip- ules and € Bot. Helv. 1 -95. Doweld, А. 998. Тһе carpology and taxonomic rela- tionships а parte (Davidsoniaceae). Edinburgh J. Bot. 55: 13-25. Engler, A. 1928. Cunoniaceae. Pp. 229-262 in A. Engler & K. Prantl (editors), Die natürlichen Pflanzenfamilien a. Engelmann, Leipzig ски, Ј. 1985. бын limits оп phylogenies: T approach using the bootstrap. Evolution 39: 783- Да Ј. 1982. Recherches botaniques en Polynésie aise. ORSTROM-Tahiti 992. ноя of the Society Islands. Pacific Sei. 46: 232-250. Gentry, A. H. 1995. Patterns of diversity and floristic com- position in neotro pi S. P. Churchill, H. Balslev, E. Forero & J. Luteyn (ed- ы Biodiversity апа Conservation of Neotropical ne Forests. New York Botanical Garden, Bronx. eae "E J. 1983. The fruit in Ackama, Caldcluvia, and 21: Weinmannia (Cunoniaceae). New Zealand J. Bot 55—456. Good, R. 1950. ag: da and New > ашы A prob- lem in geography. Blumea 6: 470—4 Gould, S. J. 1977. Ts ria and зена The Belknap Press of Harvard Univ. Press, Cambridge Greuter, Е arrie, H. M. Burdet, W. С. Chaloner. V. Demoulin, D. L. Hawksworth, P. M. Jorgensen, D. Н. Nicol P. ilva, P. Trehan J. McNeill. 1994. n C. a, ane International Code of Botanical Nomenclature (Tokyo Code). Eres Veg. Grimes, J. 1 ‚ Metiiperism, heterochrony and inflores- cence ШТ dog gy of the Pithecellobium-complex (Le- osae: Mimosoideae: Ingeae). Brittonia 44: 140— Haeckel, E. Pec The Evolution of cx Watts, London. Hawkins, J. A., Hughes & R. W. Scotland. 1997. Primary ho ie assessment, characters and character states. Cladistics E ipo Hoogland, R. D., mie & H. C. F. Hopkins. 1997. Le genre uina (там) еп Nonsaliss Calédon- ie. Description E cinq евресев nouvelles. Bull. Mus. Natl. Hist. Nat., B, Adansonia 19: 7-20. Hopkins, Н. C. Е 19983. A revision of Weinmannia (Cu- noniaceae) in Malesia and the Pacific 1. Introduction florescence and W. hooglandii with J. C. Bradford). Ad- ansonia series 3, vol. 20: 5—41. . 1998b. A revision of Weinmannia (Cunoniaceae) in Malesia and the Pacific 2. pee and the Philip- pines. Adansonia series 3, vol. 2 . 1998c. A revision of сан) (Cunoniaceae) in 55. sid the Pacific 3. New Guinea, Solom Islands, Vanuatu and Fiji, vidi notes on the «ресін of Samoa, Rarotonga, New Caledonia and New Zealand. dri а; and New Caledonia with R. D. тое 2 with J. C. Bradford). Adansonia series 3, vol. 106. J. Florence. 1998. A revision of Weinmannia (Cunoniaceae) in Malesia and the Pacific 4. The Soci- ety, Marquesas and Austral Islands. Adansonia series 3, vol. 20: 107-130. Hufford, L. & W. C. Dickison. 1992. A col a anal- ysis of Cunoniaceae. Syst. Bot. 17: 181– 1965. Description des ies de вани. р. 46—78 in Н. Humbert urs Darne (editors), Notice de la carte de Madagascar E es Sci. Techn. Inst. Frang. Pondichéry. Vol. 6 Jong, K. & B. L. Burtt. 1975. The piles of NUN logical novelty exemplified in the growth Dm of some Gesneriaceae. New Phytol. 75: 297-3 Kaul, R. B. & E. C. Abbe. 1984. ud oes азоне ес- ture and avolhidon in the Fagaceae. J. Arnold Arbor. 65: 375-401. 584 Annals of the Missouri Botanical Garden Kellogg, E. A. 1990. Ontogenetic studies of florets in Poa мете Allometry and heterochrony. Evolution 44: 1978-1989. Kelly, D. L., E. V. J. Tanner, E. M. NicLughadha & V. Kapos. 1994. Floristi 's zi biogeography of a rain E est in the с. Andes. J. Biogeogr. 21: 421-4 Leavitt, R. G. one ral ‘ation of characters in p M 5 13—: Liede, S. 12. 71995. On the "us 'ence structure of Asclepiadaceae. Pl. Syst. Evol. : 99 Mabberley, D. J. 1975. The giant Lobelias: Toxicity, inflo- ence a and tree-building in the Campanulaceae. New 295 1992. MacClade: r Evolution V 3.0 Maddison. Analysis of Phylogeny p аи у 22. a ads Massachuse Martínez, E. . Ramos. 1989. Lacandoniaceae (Triuridales): Tie M 5” de México. Ann. Mis- souri Bot. Ga urd. 76: 12 199]. 5. ee agricultural change in New Zealand. Bee ids ier +73. г К гі, ete m n Press, New York. Meyerowitz, E. M. 1997. Genetic control ul cell division atterns in developing plants. Cell (Cambridge) 8 299-308. Orozco, C. I. 1997. Sobre la posición sistemática de Bru- m Ruiz & Pavon. Caldasia 19: 145-164 Patriat, P. & J. Seqoufin. 1988. Rec oniiraction of ihe Tan: жүз idian Ocean. Tectonophysics 155: 211-234. Pleijel, К. 1995. On charac el apr for phylogeny re- construction. Cla distics 11: 309-3 Raff, R. A. 1996. The Shape of ae Genes, 2. and the Evolution of Animal Form. Univ. Chicago Press Chicago. & С. A. Wray. 1989. Heterochrony: Developmen- tal mechanisms and evolutionary results. J. Evol. Biol —434. Helindnane, H. 1994. Contribution à la connaissance de 'apiculture et à la melissopalynologie dans le parc na- tional de Ranomafana. Unpublished Masters thesis, Université d'Antananarivo, Antananarivo. Raven, P. H. & D. I. Axelrod. 1974. Angiosperm bioge- ography and ipe ont movements. Ann. Missouri Bot. Gard. 61: 539-673. Renner, 5, S. С Floral и ‘al observations on He- liamphora tatei (Sarraceniaceae) and other plants s | erro de la Neblina in ala РІ. Syst. Evol. 29. -— “а 0. V. 1986. Estudios ecológicos en un relicto de bosque de Weinmannia tomentosa у Drimys granaden- 5. En la región de Monserrate. Perez-Arbelaezia 1: —356 56. Sachs, T. 1982. A Vid m basis for plant mor- phology. Pp. 118-131 in R. Sattler (editor), Axioms and Principles of Plant d. onstruction. Martinus Nijhoff/Dr W. Junk Aged ue Sattler, R . ^r ag . Homeosis in йө Amer. J. Bot. 75: 1994. Homology. homeosis, and process mor- бартер in plants. Рр. 424-475 іп B. Hall (editor), Ho- mology: The Hierarc ical о of Comparative Biology. Academic dara San Die Schlessman, M. A., D. С. Lloyd & P. P. Lowry П. 19€ Evolution of Mud systems in New Caledonian пи асеае. Mem. New York Bot. Gard. 55: 105-117. Smith, A. C. 1985. Flora Vitiensis Nova: A New Flora of in. 3. н Tropical Botanical Garden, Lawai, Kauai, Hawa "Ве ltrán, P. Huijser, Н. Pape, W. Lénnig, H. Suedler & Z. Schwarz-Sommer. 1990 Deficiens, a homeotic gene involved in the control of flower mor- phogenesis in Antirrhinum majus: The pug shows 22 to transcription factors. E. M. B. O. J. 9: 605- Swofford, D. L. 1993. PAUP: Phylogenetic Analysis саш d deni Ll V 3.1. Illinois Natural History Survey P. Begle. 1993. PAUP: Phylogenetic Anal- ysis и И version 3.1, User's Manual. Lab- oratory of Molec r Systematics, Smithsonian Institu- tion, Washington, D.C Timonen, T. 1998. лет 'ence structure in the sedge tribe Cariceae (Cyperaceae). Publ. Bot. Univ. Helsinki 26: 1-35. Tortosa, R. D., L. Aagesen & G. M. Tourn. 1996. Mor- phological studies i in the tribe Colleticas Ше eae): Analysis of architecture and inflorescences. Bot. Lon. Soc. 122: 353-367 Tucker, S. C. 1987. Pseudoracemes in papilionoid le- gumes: Their nature, oom and variation. Bot. J. Linn. Soc. 95: 181-206. Venkata Rao, C. 1965. Studies in the Proteaceae V. Evo- lution of the inflorescence. J. Indian Bot. Soc. 44: 244— 211. Wagner, G. P. 1996. Homologues, natural kinds and the evolution of modularity. Amer. Zoologist 36: 36—43 Valsh, R. S. 1978. Nectar and Pollen Sources of New Zealand. 2. Beekeepers Association of New Zea- land, Welling Wardle, P. & A. H Mac Rae. 1966. Biological. flora of New Zealand. Vol. ік BE = = 8 = = =. 8 т с ^5 ER о P4 с 3 u- 4-1 31. . А. Simpson. 1991. Seed morpbilogy in ‘elation to DNO in New Zealand species of Wein- mannia, rama, and the related South American 2. luvia Rin Ма (Cunoniaceae). New Zealand J. Bot. 29: 451—453. Weber, A. 1988. Development, interpretation, and phylog- eny of the в гепсе of Epithema. Beitr. Biol. Pflan- zen 63: 431—4 Weberling, F. on : . Beitrüge zur Morphologie der Rubi- aceen-Infloreszenzen. Pp. 191-209 in Y. Sell, F. We- berling & H. Lorenzen (editors), Morphology, Anatomy and Systematics of Higher Plants. Gustav Fischer Ver- lag, Stuttgart. 1988. The architecture of inflorescences in the Myrtales Ann. Missouri Bot. Gard. 75: 226— 1989. Morphology of Flowers and ba. ences. Cam bride Univ. Press, Cambridge. е J. 1979. The plant as a S MIN Annual „ Ecol. Syst. 10: 109-145. Wilkinson. M. 1995. A comparison of two methods of character construction. Cladisties 11: 297-308. Yanofsk y, Ma, Bowman, 52 hs A. Feldinann « E rowitz. 1990, The р encoded by the hoger se на gene agamous sembles I ription od — ture 346: 35-39. Zelditch, M. L. & . Heterochrony and 5 Stability ae innovation in the evolution of orm. Paleobiology 22: 241-254. Zimmermann, W. 1961. Phylogenetic shifting of organs, Volume 85, Number 4 1998 Bradford 585 Species Groups in Weinmannia tissues, and phases in pteridophytes. Canad. J. Bot. 39: 1547-1553. APPENDIX 1 List of species assigned to each OTU and a selected Only the name s the е of spe in eac +h OTU includes у к эне of Қа денің ог isotypes unless indicated otherwise, Herbaria holding specimens are given in parentheses Section Weinmannia WEI-pinnata 73 species (in addition to the species cited below there are ca. 10 undescribed species) Weinmannia antopiytla Standl. ч L. О. Williams, Smith 262: type), Costa Ric Weinmannia pun D. Don, Bradford 392 (MO), Ven- ezuela; Jorgensen 1260 (MO), Ecuador Weinmannia auriculifera Hieron., Cuatrecasas 6313 (V), Colombia Vemm balbisiana Kunth, Bradford 459 (MO), Ven- zue Weinmannia bangii gere Bradford 525 (MO), Bolivia; Solomon 10683 (MO), Bolivia pu boliviensis R. E. Fr., Boliv E bogotensis Cuatrec., Bradford 746 (MO), Co- lombia; Cuatrecasas 8005 (COL, F; type), Colombia pu а Standl., Bello и dd (MO), Costa а; Bradford 97 (MO), Costa ue Weinmannia ei grs Hieron., Bradfo 3 (МО), Ecuador; ford 339 (MO), Ecuador; ba 762 (MO), Colombia Weinmannia corocoroensis P. E. Berry & J. C. Bradford. Huber 12296 (MO), Venezuela Weinmannia costulata Cuatrec., ype), Ecuador "> crassifolia Ruiz 4 Pav., аи 510 (МО), : Solomon 17344 (МО), Bol Weinmann cundinamarcensis Cua ка у 3 (COL, Е; type). Colombia bend cymbifolia Diels, ind 2654 (MO). Llantas Quir 2625 (МО), Per Weinmannia discolor Gardner, poc 5020 (MO), Bo- livia; Hatschbach 43003 (MO), Brazil г ее атуайуоНа Мопс., Bradford 150 (МО), Ec- : Gentry 61436 (MO), Peru; Smith 5011 (Е о ен Solomon 10591 (MO), Steyermark 53545 (F; Cuatrecasas Peru; [д мег» ntha Diels, vie id 538 (МО), Boliv- (M Weinmannia 2. Kunth, Bradford 295 ne uador Weinmannia glabra L.f., Breedlove 19958 (F), Mexico Weinmannia glomerata C. Presl, Gentry 19302 (F). Peru; Gentry 44877 (F, MO), Peru; Young 593 (F, MO), Peru Weinmannia guyanensis Klotzsch ех Engl., Delascio 11861 (MO), Venezuela; ms 9085 (US), Venezuela Weinmannia haenkeana Engl., Llatas Quiroz 1320 (F), Peru; Smith 5039 (MO), Peru Weinmannia humilis Engl., Hatschbach 26323 (MO), Bra- zil Weinmannia ilutepuiensis J. C. Bradford & P. E. Berry, Liesner 23338 (MO), Venezuela; Liesner 23413 (MO; ype), Venezuela; Maguire 3.3508 Weinmannia ane Schltr. & Cham., ), Venezuela Ventura 1089 Weinmannia elski Szyszyl., Gentry 80472 (MO), Ecuador; mith 7789 (MO), Peru Weinmannia karsteniana Szyszyl., Venezuela Weinmannia kunthiana D. Don, Jaramillo Mejia 167 (F. olombia Weinmannia latifolia C. Presl, d aded 23723 (F). Co- lombia; Smith 7937 (MO), Weinmannia laurina Kunth, P 7686 MON Panama Weinmannia laxiramea КІШ ‚ Steyermark 128321 (MO), Venezuela; Шер. 129882 (US). Bradford 383 (МО), Venezuela Weinmannia lentisifolia C. Presl, Palacios 4128 (MO), Ec- uador Weinmannia macrophylla Kunth, Јагрепзеп 92729 (MO). Ecuador ш magnifolia Cuatrec., Cuatrecasas 8590 (COL, F; type), Colombia; Palacios 13415 (MO), Ecuador Weinmanvia mariquitae Szyszyl., Bradford 208 (MO), Ec- uador; Romoleroux 253 (MO), Ecuado Weinmannia microphylla Ruiz & Pav., Bradford 492 (MO). Bolivia Weinmannia жаке d Killip & A. C. Sm., Bruijn 1296 MO), Venezuela; Penland 1198 (F), Ecuador Weinmannia 4. ia Cuatrec., Bradford 745 (MO), Co- Іот Ма; Cuatrecasas 9451 (COL; type), Colombia Weinmannia organensis Gardner, Irwin 8601 (MO), Brazil Weinmannia ovata Cav., Bradford 541 (MO), D'Arcy 13770 (MO), Peru Weinmannia port ifoliolata те COL, Bolivia; Cuatrecasas 8486-A type), Colom Weinmannia paulliniifolia Pohl ex Ser., Pohl s.n. (P; type), Brazil; Sucre 6824 (MO), Brazil Weinmannia pentaphylla Ruiz & Pav., Gentry 43227 (MO), Peru Weinmannia pinnata L., Wilbur 7952 (MO), Dominica Weinmannia ко Diels, Smith 4459 (Е, MO), Peru Weinmannia polyphy Ша Мо oric. ex Ser., 2.” 2 (МО). Ес ие produc Sagástegui 7766 (MO Weinmannia pubescens Kunth, Bradford 13 (MO), Ecua- dor; Palacios 9648 (MO), Ecuador Weinmannia ие Ruiz & Pav., Bradford 296 (MO). Ecua йрй rhoifolia Rusby, Lectae 1989 (МО; type). Bo- livia Weinmannia rollottii Killip, eee 357 (MO), Ecuador; Bradford 747 (MO), Col 2. sibundoya une - aio 11624 (F . Colombia; Cuatrecasas бет 7 (F), Colombia: “hate recasas 19308 (MO), Colon dai von sorbifolia Kunth, 4: 15997 (МО), Во- livi пе subsessiliflora Ruiz & Pav. COL), Colombia Weinmannia ternata Р: fe deg 2114 (MO), Peru; Woyt- kowski 8314 Ше XN Weinmannia tolimensis Cuatrec., Cuatrecasas 20424 (МО), olombia; Wilson Devia 663 (MO), Colombia е tomentosa L.f., Berry 175 (MO), Venezuela; Bradford 751 (MO), Colombia; Gentry 34725 (MO). Culombía E E Mu 7146 (MO). Schultes 5329 586 Annals of the Missouri Botanical Garden Weinmannia trianaea Wedd., Cuatrecasas 11635 (COL E t (Ма; Cuatrecasas 8431 (COL, F), Colombia; 438 (MO), Ecuador Weinmannia trichocarpa Ратр., Weinmannia tric 5” Сау., tina; West 4679 (MO Chile Weinmannia velutina O. cM 'hmidt, Steyermark 104504 (F, US), Venezuela; Steyermark 92437 (F), Venezuela Haber 2438 (MO), Costa + 1 704 (MO), Peru m 4537 (MO), Argen- Weinmannia wercklei Standl., ica WEI-tinctoria 2 species Badré 932 (P), Mascarenes; carenes; Cadet 1704 (P), Mas- carenes; uiid 395 (P), Mascarenes; Friedmann Mascarenes Weinmannia tinctoria Sm., Barthe s.n. (P), Mascarenes; Desvaux s.n. (P), Mascare Ven Frappier 397 (P), carenes; Kramer 9299 Mascarenes; Lorence 2427 (MO), Mascarenes; 1. 2676 (MO), Mas- carenes Weinmannia pena Lud 5 Section Fasciculata FAC-fraxinea 13 species Weinmannia аралаша Airy Shaw, Brunig 5 8785 (1), ysia, Sarawak; Chew Wee-Lek 380 (L), Malaysia, Sarawak; Clemens 29476 (L), Malaysia, Sabah; Clemens 33076 (1), Malaysia, Sabah; Endert 4125 (L), Indonesia, Kalimantan; Paie S 26531 (L), Ma- laysia, Sarawak TM - Н. C. Hopkins, Coode 6197 (L; type). ndor 27. Weinmann С Н. Hopkins, Balgooy 3809 (L Indonesia, Sulawesi; V Vogel 5682 (L), blond: Sulawesi; de Vogel 5959 (L), AN Sulawesi; de Vogel 6122 (L; type) Indonesia, Sulawesi; Meijer 11147 (L), Indonesia, Sulawesi; Schmid 5512 (L, P), Indonesia, Sulawesi ке. dulitensis Airy Shaw, Hopkins 5014 (МО), Malaysia, Sabah . Howard 89 (BISH), Fiji: more BSIP 995 uii. a), Solomon Islands iine е Sm. ). Don, Bradford 830 (MO), n Islands; "ok 5004 (MO), Malaysia, Sa- ei Hopkins 30 О), Malaysia, Sarawak; Sar- gent s.n. Indonesia “Journey Round the World”; Takeuchi 71 35 (MO), Papua New Guinea Lignes furfuracea H. C. Hopkins, Balgooy 3255 (L; ratype), Indonesia, CM MEN Balgooy 3464 быы ы. Indonesia, deuil bb 20787 (BO; par: atype), Indonesia, Sulawesi; Rutten 2231 (BO, L paratype), Indonesia, 4. Татға 1595 (1; Indonesia, Sulawesi Шын ене" hutchinsonii Merr., Elmer 14228 (МО), m ippines; Elmer 14918 (MO), Philippines; Ramos 23494 (мо) Philippines; Wenzel 1088 (MO), Phil- ірріп Weinmannia luzoniensis Vidal, Elmer 18024 ippines; Vanoverbergh 1253 (MO), Weinmannia exigua hit т type). MO), Phil- ~ б 322 (P), Vanu- (P), Vanuatu; Schmid 5083 (NOU) Vanuatu Weinmannia negrosensis Elmer, Ramo. 287 'hilippines; Wenzel 1057 (MO), Philip ines М. richii А. Gray, Hopkins 5023 (MO), Fiji; De- r 14379 (F, MO), Fiji; Smith 6813 (L, P), Fiji 7 (MO), , P), Vanuatu; Schmid 3557 ~ ВО, 1; (уре), m NS Perry, Brass 3215 Solon FAC-descombesiana | species Weinmannia descombesiana Ad bb 20870 (L), In- donesi: >Р; p ia ulaw 2» (L), In- donesia, асыны Kjellberg 1618 (S: is Indone- sia, Sula FAC-pullei 3 species Weinmannia eymaeana H. L; type), Indonesia, Su Каштан pullei 5 s Schltr., Hoogland 6979 (BISH, BO, L), nd 7685 (L), Papua New nea; Kalkman 4859 (BO, L), Papu an pr ew Guine Veldkam Guinea; “Vink 17098 (BO, L, P. Papua New Guinea; Frodin NGF 26964 (BISH, L), Papua New Guinea Weinmannia urdanetensis Elmer, Bowers 401 о w Guinea; Hoogland 5463 (BISH, 1), P oe Eyma 3578 (BO, Papua a New u New fora , E), Indonesia, Глап Jaya; 5 (L), Papua New Guinea; Sterly a 40 (L), Pa E Guinea; BW 5588 (L), Indonesia, Irian ps5 Robbin 191 (L), Papua New gis Tak- euchi 6340 (MO), Papua New Guinea; ANU 7663 (L), a New > Womersley NGF 15240 (L), Papua New Guin FAC-clemensiae 2 species Weinmannia clemensiae Steenis, Beaman 9132 (1), Ma- laysia, Sabah; Beaman 9837 (L), Malaysia, Sabah; Clemens 50793 (L), Malaysia, Sabah; Clemens 50877 (L). Malaysia, Sabah; Hopkins 5011 (MO), Malaysia, Sabah; Chew RSNB 4508 (L, SAN), Malaysia, Sabah; Chew RSNB 4755 (L, SAN), HUE Sabah Weinmannia Due H. C. Hopkins & J. C. Bradford, on 32246 (КЕР), Peninsular Malaysia; Whit- more UL 2582 (KEP, L, SAN; type), Peninsular Malay Section na £ — SPI-bojeriana 16 species (in addition to ui species cit- ed below there are 7 undescribed species) Weinmannia о Tul., Bojer s.n. (Р; type), Madagas- r 13321 (P), Madagascar; Bradford 639 (MO), Маа Decary 5010 (Р), Машааны: Decary 5366 Madagascar; Humbert 4849 (P), Madagascar; Malcomber 2 (MO), Madagascar; Perrier 6433 (P), Madag Weinmannia decora Tul., Bernardi 11961 (P), rr ar; rnier 324 (P; тре), 2-2. scar; R.N. 3452 (P), 7 (P), Madagascar; SF 13228 Weinmannia eriocarpa Tul., Baillon s.n. (P), Madagascar; Bojer s.n. (P; Madagascar; Decary 15104 (P), Madag e 42. 27809 (P). Madagascar; Hil- дбн 3562 (P) Madagascar; Baron 1674 (P), adagas Weinmannia hildebrandtii Baill. var. yi Bernardi, Br дың а ө42 (МО Ы Mx лава. S.F, 8848 (P; type ariety), Madaga ШО humbertiana е Bradford 703 (MO), dagascar; Humbert 23532 (P), Madagascar; Hum- Volume 85, Number 4 1998 Bradford Species Groups in Weinmannia 587 bert 23814 (P; type), Madagascar; Humbert 22505 adagascar Кйтап humblotii Baill. var. humblotii, Baron 4434 (Р; type of W. leptostachya), Madagascar; Humbert nih (P), Madagascar; 2. 613 (P; type), Mad- gascar; Weinmannia humblotii var. anceps Bernardi, Bradford 705 (MO), Madagascar: Perrier de la Bâthie 421 (P), Madagascar Жы таттеа Bernardi, Louvel 191 (P; type), adagascar; Morat 2803 (P), Madagascar; Humbert 24496 (P), Madagascar; Humbert 24781 (P), Mada- gascar Weinmannia sanguisugarum Bernardi, Bernardi 11999 (Р; 2), Madagascar; Humbert 3823 (P), Madagascar; Malcomber 2363 (MO), Madagascar Weinmannia stenostachya Baker, Baron 3148 (P; type of P), Madagascar; Ташы. ‚ Madagascar; Perrier 6422 (P), Madagascar; S.F. 12547 (Р), Madagascar; S.F. 3 (P), Madagascar SPl-icacifolia 6 species (in addition to = species cited below there are 4 undescribed species) Weinman ae hildebrandtii Baill., ipe hn 3695 (P; Cours 4792 (P), Mada- 5809 (P). M po icacifolia Bernardi, Perrier de la Вале 6 (P), Madagascar; Perrier de la Báthie 16122 (P), Madagascar; Perrier de la Báthie 16464 (P; type), Madagascar SPI-comorensis 5 species (in addition to the species cit- below there is 1 undescribed species) Weinmannia baehniana Bernardi, Perrier de la Báthie 6431 (P), Madagascar; R.N. 1398 (P), Madagascar; R.N. 1837 (P), Madagascar; S.F. 10198 (P), Mada- Boivin s.n. (P; type), Mada- gascar; Bosser 17997 (P), 27 Soda 68 (TEF), M 2. S.F. 16591 (P), Madagasc Weinmannia lucens Baker, Decary 13241 (P), Мә Болыс; Humbert 28744 (P), Madagascar Weinmannia minutiflora Baker, Baron 2542 (P; type). ea Jardin oin bd 4710 (P), Madagas- ; S.F. 9044 (P), Мадар, gascar Weinmannia на dog Section /nspersa INS-rutenbergii Weinmannia hepaticarum Bernardi, Humbert 23545 (P; type), Madagascar pons 4. Engl., Chauvet 429 (P), Madagas- 7. Madagascar; SF 13518 (P), Mad- Бине x 2 (P), Madagascar; SF 21468 (P), Madagascar; a 227 (P), Madagascar INS-venusta 1, R.N. 111 (MO, P), 2. Bradford 655 (MO), Madagascar; R.N. 3 (P), Madagascar; S.F. 5464 (P), Madagascar; 5 P), Madagascar; ауда 295 (Р), ғасы ponia venusta Bernardi, Humbert 21944 (P; type). agascar; a S.F. 27632 (P), Madagascar; Cheistaphe S.F. 17610 (TEF), Madagascar; S.F. 17231 (TEF), Madagascar Weinmannia sp. nov. d INS-louveliana 3 species (in addition to the species cit below there is 1 undescribed species) Weinmannia commersonii Bernardi, S.F. 16718 (TEF), type), Madagascar; S.F. 10164 (P), Madagascar Weinmannia louvelia ana Bernardi, S.F. 9669 (P), Madagas- car; S.F 5182 (P), Madagascar; R.N. 6211 (P; type), Madarat S.F. 10402 (P), Madagascar INS-madagascariensis 2 species шее henricorum Bernardi, Humbert 7018 (P; ype), Madagascar Weinmennia madagas sis DC. ex Ser, S.F 17814 . Madagascar; Bradford 660 (MO), Madagascar; Dus 535 (P), Madagascar; 5. 17814 (P), Mad- Weinmannia madagascariensis var. aniba Ветий, R.N. 9765 (P; type of variety aniba), Mad- agasc Section Leiospermum LElI-serrata 5 species Weinmannia dichotoma Brongn. & Gris, MacKee 15000 P), New Caledonia; MacKee 18867 (P), New Cale- donia; MacKee 31635 (P), New Caledonia; Schmid 4135 (P), New Caledonia; Veillon 1914 (P), New Cal- edonia; Veillon 3839 (P), New Caledonia Weinmannia monticola Daniker, Aubréville 230 (P), New Caledonia; Hürlimann 1875 (P), New Caledonia; Jaf- fré 2707 (P), New Caledonia; MacKee 44406 (P), w Caledonia; McPherson 5809 (MO), New Cale- a ill. & Virot) Hoogland, МасКее 18670 (P), New Caledonia; MacKee 34088 (P). New Caledonia; MacKee ec >), New Cale- donia; MacKee 3600 w Caledonia; Schmid 3370 (P), New Caledonia; Veillon 2273 ud New Cal- edonia; Virot 731 (P; type), New Caledon Weinmannia paitensis Schltr., Bernardi 9881 (P). Naw Cal- edonia; McPherson 3403 (MO), New Caledonia; Schlechter 14941 (P; рр, New Caledonia; Thorne 28733 (P), New po Weinmannia serrata Bron M" Gris, Balansa 2298 (P), ew Caledonia; Ed 12816 (P), New Caledonia; 2. 627 (MO), New Caledonia; 2. 746 (Р), ew Caledonia; MacKee 35545 (P), New Caledonia; Pide 16608 (P), New Caledonia; Thishaus 22 (P), w Caledonia be eee T LEI-sylvicola 1 species Weinmannia sylvicola Sol. ex A. Cunn., Bradford 912 MO), New Zealand; Gardner 1621 (MO), New Zea- land; Gardner 2659 (MO), New Zealand; Gardner 5384 (MO), New Zealand; Orchard 4048 (MO), New Zealand; Walker 5243 (MO), New Zealand LEI-racemosa | species Weinmannia racemosa L.f., Bradford 910 (MO), New Zea- land; Chapman CHR 258594 (MO), New Zealand; Loh CHR 359035 (MO), New Zealand; Gardner 169 (MO), New Zealand; Gardner 5350 (MO), New Zea- land; Thompson 526 (MO), New Zealand; Wood 31653 (MO), New Zealand LEI-affinis 15 species (in addition to the species cited below there are 2 undescribed species Weinmannia affinis А. pi Bradford 597 (MO), Fiji; Gibbs 642 (BISH), Fiji; Hopkins 5022 (MO), Fiji; Seemann 200 (P), Fiji; Smith 4905 (BISH, L, P), Fiji; 588 Annals of the Missouri Botanical Garden Smith 7608 (BISH, L, ped. US 48070 (US; type), Fij н У croftii Н. C. Hopkins, спе ІДЕ 63017 (К, 4 paratype), Papua New Guinea; Ridsdale МСЕ 33981 (BISH, L; type) Papua New Guinea; Stevens E 51 ; paratype) Papua New Guinea; Vinas LAE 59724 (BISH, L; paratype), Papua New Guinea Weinmannia denhamii Seem., Aubert de la Rüe s.n. (P). anuatu e) rnardi 13258 (L, P), Vanuatu; Morat 5897 6 Р), -— Kajewski 317 (BISH, P). Vani 1. Gillivray dis isolectotype), Vanuatu Weinmannia. marquesana e . Hallé 2057 (Р), Mar- quesas; Perlman 10259 mE MO, P), Mar Weinmannia marquesana var. myrsinites, 9667 (MO), Marquesas . P, 4. Fijiz U.S. Expl. Ех- uesas; Florence ddp 920 (MO), So- € Tahiti; Florence 3 0 (P), Society Is- ; Florence 7935 e. Society Islands; md (BISH, P), Society Islands, Ta hiti sands dimit Perry, Kajew ski 1738 (L; type), Pap- u uine a; Kajews 4 s.n. (L). Solomon Islands; Mauriasi ВР. 12092 (ВУЗЈЕ, L), Solomo Weinmannia raiateensis J. W. Moore, Bradford 929 (MO). y ds, Raiatea; Florence 3554 (P), Society асе Raiatea: Florence 3746 (P), Society Islands, i 3754 (P), Society Islands, Кашек Нер и ПИ ra G. For - < Gagné 1457 bis (P), Soc iety Islands, ан io Se F. Br, Florence 6395 (MO). Rapa: 6514 (P), Rapa; Hallé 7517 (P). Кара; N. Hallé 7700 (P). Вая St. John 15304 (Р). Кара; St. John 15305 (Р), Кара Weinmannia ма Hemsl. ex Cheeseman, Gardne 2503 (MO), е МасКее 44191 (Р), ин ga; MacKee 44309 (MO, P), Rarotonga — samoensis А. Gray, Bradford 800 (MO), Upu- lo, Western Samoa; Bradford 2. Savall, West- en ¿Samos Christophersen 1943 (BISH, MO), Samoa; Expl. Exped. US 4 2. (05; чүн.) Samoa eds diee des H. C. Hopkins & Florence, Flor- ence 9581 (P: type). M o. Islands Weinmannia vescoi Drake, Bradford 932 (MO), Society Is- ands, Rait 2 699] (P), Society Islands, Raiatea; Morat 6990 (P), Society Islands, Raiatea; St. John 17255 (P), Soc |“ UN Raiatea; Vescoi s.n. (Р; type), Society Islar Weinmannia vitiensis Seem., Brya 7 (BISH), da Hop- kins 5041 (MO), Fiji; ini us (Р; type), Fiji Cunonia Cunonia-macrophylla 3 species Cunonia на Brongn. & Gris, Bradford 607 (МО), бшш zaledonia; McPherson P (MO), Med Cale- a; cx жей 2277 (MO), New Caledon бле capensis 4, Bradford 2. (MO), United. States Cultivated); pb 1387 (MO), South Africa; Rouske 734 (MO), South Africa; Werdermann 2441 South Africa Cunonia sc - Diniker, Düniker 506 (P; type), New edonia; MacKee 17799 (Р), New Caledonia; MacKee 27546 (P), New Caledonia; MacKee 38224 (P), New Caledonia Cunonia-purpurea 13 species Cunonia atrorubens Schltr., Bradford 614 (MO), New Cal- edonia; McPherson 2004 (MO), New Caledonia; Mc- Pherson 2227 (MO), MN Caledonia; McPherson 2 (MO), New Caledon Cunonia austrocaledonica BERE ex Guill., Mi uad 12904 (MO), New Caledonia; McPherson 1876 (MO ie Caledonia; McPherson 1944 (MO), New Cale- McPherson 3434 (MO), New Caledonia бар «тїш 2. McPherson 4450 (P: paraty- pe). New Caledor Cunonia ta ae Ei & Gris, McPherson 4364 ( New Caledonia; McPherson 6429 (MO), New E па Сипота lenormandii Vieill. ex NE & Gris, McPher- son 6418 (MO), New Caledoni Ши ава e n € ) Benard McPherson 2878 ‚ New pam Caledc n nervosa dud МаскКее 15715 (P: type), New Caledor Cunonia pi erticillata Guill., Blanchon 196 (P), New Caledon Cunonia ШО Schltr., Mc 2. 2119 (МО), New ;:aledor McPherson 4445 (MO), New Caledonia Me Pa 2163 (МС ), Hin Caledonia Cunonia ЛЕ л 217 ns (MO), E donia; Mc la 2637 w Cale- a: McPherson 4396 (MO), New 1. Cunonia "piel Hoogland, MacKee 19129 (Р; P» New Caledon as ШЇ Hoogland, МасКее 22886 (P; type), New Caledoni Cunonia vieillardii Brongn. & Gris, McPherson 3998 (P), New Caledonia; McPherson 4634 (МО), New Cale- donia; McPherson 4535 (MO), New Caledonia Cunonia-aoupiniensis 2 species сы here мерес Hoogland, Morat 7977 (Р; paratype). Caledonia Cunonia montana Schltr., Bradford 609 (MO), New Cal- edonia; McPherson 2930 (MO), New Caledonia; Ber- eeg 12729 (MO), New Caledonia Cunonia-balansae 2 species шш = wee rus & Gris, Bradford 617 (MO). : McPherson 3313 (MO), New Cale- ~ ei 4127 (MO), New Caledonia; Hoog- M 12731 (MO), New Caledonia Cunonia rotundifolia Düniker, Jaffré 1954 (P), New Cal- edonia Cunonia-alticola 2 species Сипота alticola Guill., Bradford 611 (MO), New Cale- donia Cunonia bernieri Guill., Morat 7660 (P), New Caledonia Cunonia pulchella Cunonia pulchella Brongn. & Gris, Bradford 635 (MO), N aledonia; McPherson 4037 (MO), New Cale- donia; McPherson 6034 (MO), New Caledonia Cunonia bullata Cunonia bullata Brongn. & Gris, McPherson 2241 (MO), New Caledonia Outgroups Ackama Ackama rosifolia А. Cunn., Bradford 909, New Zealand Volume 85, Number 4 1998 Bradford Species Groups in Weinmannia 589 (MO); Gardner 358, New Zealand (MO); Jessup s.n., New Zealand (МОЯ1616100); Orchard 4040, New Zealand (MO) Caldcluvia Caldcluvia paniculata D. Don, Landrum 4476 (MO), Chile; Morrison 17552 (MO), Chile; Werdermann 681 а. ), Chile; Werdermann 1855 (MO), Chile; Zóllner 9639 (MO), Chile Spiraeopsis I celebica Blume, Bradford 834 (MO), B Islands; Bradford 840 (MO), Solomon Islands: 11402 (MO), Philippines; Elmer 14157 (MO), Phil- ippines; Elmer 15184 (MO). Philippines; Wenzel 1087 (MO), Philippines Vesselowskya Vesselowskya rubifolia Pamp., Boorman NSW 104729 NS Australia; Bradford 879 (MO), Australia: rade 882 (MO). Australia; Coveny 5676 (MO NSW), Australia; Coreny 10877 (MO, NSW), Aus- tralia С Эи s.n. NSW 104749 (MO, NSW), Austra- lia; Webb 11474 (MO), Australia APPENDIX 2 List of possible autapomorphies for each ingroup OTU. Some assumptions of relationships were needed to hy- pothesize derived states. For example, comparisons of character states were made within sections, and the most generalized character state was usually considered prim- itive. WEI-pinnata: caducous petals (not present in all species, see discussion) WEI-tinctoria: unisexual flow SPI-bojeriana: solitary нен double internode of branches SPI-icacifolia: no clear autapomorphies SPI-comorensis: no I INS-rutenbergii: double internode of branches INS-louveliana: solitary flowers INS-madagasc ariensis: no clear autapomorphies INS-venusta: IM branchin FAC-fraxinea: fixed s superimieraty ca y еса. solitary flowers (has hybrid- id cteristics ы sects. Fasc iculata and Le perm FAC- pullei СХ of medial meristem FAC-c _ clear aimapomorphies morphies LEl-serrata: no cles ear Го LEI- dinis ан нанио LEI-sylvicola: IM with two ЕТЕ and terminal abor- tion LEI-racemosa: no lateral IMs, IM terminus а bud Cunonia-macrophylla: large flowers, no medial meristem abortion, racemes borne directly along main stem Cunonia-purpurea: no clear autapomorphies Cunonia-aoupiniensis: lateral n UN IMs Cunonia-balansae: medial IM Cunonia bullata: IM with medial: raceme Cunonia-alticola: sessile flowers Cunonia pulchella: fascicled flowers APPENDIX 3 CHARACTERS USED IN THE CLADISTIC ANALYSIS. Numbers after character states refer to the coding in the data matrix as shown in Appendix 4. The plesiomor- phic condition implied by the outgroups is listed first state 0). Illustrations for many of these characters can be ound in Bernardi (1961, 1963b, 1964, 1965) and Hop- kins (19984). ~ Vegetative Characters (1) Persistent, fused lateral stipules: present (0), absent (1). Stipule pairs that enclose lateral ci may be fused at their base. Even when most of t the fused tissue may leave a the young branches. The lateral stipules may rarely be barely fused in section Fasciculata, but they do not persist as "collars." (2) ake ues of branches: basal bud absent (0), basal bud present (1), double internode present baal о leaves at the first node. In another ме а fent е this iati that are oriented in a pla the second metamer of this type of branch s development of internode and leaves. This is treated as an ordered character, because the basal bud form is con- sidered intermediate between the other conditions. Taxa with basal buds often produce stems from them. This is especially true for section Leiospermum, in which the dominant medial and distal- enm meristems are com- mitted to reproduction. The double internode form is most pronounced in Cunonia. (3) Abortion of medial meristem: absent (0), present (1). ort the medial meristem. This branching ree produces two new dim- inan occurs as ecies it is idis м А peo ma h this wa every node. (4) Lateral bud formation: additional pair(s) (0), one pair only Among OTUs, the number of 5 _ formed at a single, usually ы = 3 © = = =. ix] e ~ a 52 ш - © А шег | develop and may fall off quickly or persist during stem thickening. In section Fasciculata, many species also pro- duce additional axillary buds at a node, but with no clear dominance among buds so that the extra buds are more easily visible. The additional buds in many species of sec- tion i Рабин often develo op into lateral IMs to form a s of IMs at a node. This is scored by character 31. Flower and Fruit Characters (5) Pedicels: present (0), absent (1). In section Spicata t e pedicel is so short that the re- ceptacle lies against dE axis of the raceme, although it is 590 Annals of the Missouri Botanical Garden not fused to the axis. This condition is scored as “pedicel absent." However, this character must be scored in flower, as some species develop a slender pedicel in fruit. A few species (e.g., W. comorensis) have a narrow, elongate re- ceptacle that resembles a pedicel but is here mn non- oo See eus (1964: tabs. 1—9; 1965 figs r illustration (6) Calyx sestvtion Lu (0). —€— Imbricate calyx aestivation is considered a sy e hy for the Cunonia(Pancheria— Weinmannia) clade. ford and Dickison (1992) i a pia vation in Vesselowskya, but the this study show valvate aestiva a (7) Calyx persistence in fruit: present (0), absent (1). This character is s i ate calyx ped "mens examined for means that the calyx falls off maturation. (8) ) Number of perianth parts: 5 3-merous ( In most OTUs the релелі) tends to be either 4-merous o -merous (0), 4-merous (1), The floral receptacle bears tissue between the staminal whorl and the gynoecium, or as ribbed outgrowths at the ase of the outer carpel wall. Traditionally, these tissues have been called nectaries. Whether nectar is produc 'ed has not been _ Ne carefully, but it is likely since commercial honey is produced from Weinmannia flowers (Walsh, 1978; Mallissen. 1991; Ralimanana, 1994). Most Cunoniaceae have free nectaries, developing unattached to the ipd e wall. The nectaries of Cunonia are adnate to the carpe (10) Form of the nectary: segmented (0), ribbed (1), mem- branous (2 ). Free nectaries are found in different shapes. Ribbed u a rate, often slender parts usually positioned between the stamens. ü 1) Floral sexuality: bisexual only (0), early ena (1), ate unisexual (2), сот is unisexual (: па de bre little stud- , it appears that nis owers are mostly be. on different in- dividuals (i.e., dioecious), but strict dioecy probably imes both sexes, or or polysamodioec y). velop unisexual flowers was scored, even oed bec Py were known from a species. When unisexual flowers were unknown this was coded as 2. ук See illustrations in Bernardi (1964) ere seem to be two distinct kinds of unisexual flowers in Weinmannia. The most common has relatively early 1 е ve ale flowers and the stamens wea el- oped in female flowers. Two species of section Weinman- nia from the Mascarene islands have flowers that ss conspicuously unisexual because sexual dimorphism is not very great. These are coded as a distinct kind of uni- sexual flower that is presumably caused by late suppres- sion of the opposite sex. A third type of development is found in Vesselowskya rubifolia, which has extreme sup- pression of parts of he opposite sex so that stamens do not develop in female flowers and carpels do not develop male flowers. (12) Position of the perianth: hypogynous (0), perigynous -5 = Most OTUs have hypogynous flowers with a fairly nene flat receptacle upon which the carpel is borne at the sa level as the perianth. In several species of section E permum the carpel is sunken slightly into a narrowly con- e perianth e edid ча base of scored as being perigy (13) Fruit debis 'ence: idi (0), circ Mc il a ). Circumbasal fruit dehiscence occurs wh > bahe af the "bere ke Eom the receptacle. Usually this ehiscence is non-directional along the longitudinal suture between the carpels although circumbasal dehiscence is present. When the capsule splits open from the top, dehiscence is basipetal, as in Weinmannia and most other capsular-fruit- ed Cunoniaceae. (14) Fruit pum split pair a fused (1). All OTUs sed of two united carpels that dehisce pane their Mite exposing two locules full of seeds. The seeds are borne in two rows in each locule, along axile placentae (Dickison, 1975, 1984). The degree each bundle acies of the opposing edges of carpels. i aracter is easily visible as the carpels split in the matu e fruit. Weinmann, Cunonia, and oe have a s single central column. In many spec ‘ies this central column is prominent е it remains intact and u pright 2. the separate carpels after fruit ен În M the column is less conspicuous because it re cp r- tially attached to one edge of a ари] and p fruit i a single, short stub remains near the re- air of columns that adheres to the carpels during dehiscence. Only by examining the fruit near the receptacle can the pair of Sun be seen. In contrast, Caldcluvia and Spiraeopsis have conspicuous pairs of columns that are detached from the ca patterns of fruit column morphology were illustrated by Godley (1983), although he did not notice the subtle vas- cular pair in Ackama and its underlying similarity to the columns of Caldcluvia. (15) Seed wings: present (0), absent (1). a 1 tissue а around the re Ma 2 = = eed then wine! i present" because they are very small. Because mi- nute wings occur, wings and hairs appear to be ontoge- beme unrelated, and the two features were treated as arate characters, eak ii they may serve the same Sade in seed dispersal. (16) Seed hairs: к (0. present (1). Volume 85, Number 4 1998 radford 591 Species Groups in Weinmannia Long, slender, unicellular vy iis from the seed coat are found in all species of Weinmannia and Ackama i ; Webb & Simpson, оо). (17) Distribution of hairs on seeds: sparse and widely dis- i )), comose at both ends (1), dense and a s evenly distributed around the seed and may be either so dense see surface is obscured, or sparse, with the seed tice vis- f. c = ny spec les are fixed for one of these ранен, but a must be scored ; phic. For e mbe comose hairs and sparsely distributed on the same seed. [n a few species of section Fasciculata, some spec- imens have comose quads a other тре s have seeds with dense hairs all o surfac (18) Seed gene ree s же (1). Most seeds are straight or only slightly asymmetrica Asymmetry is most Ана in section Weinmannia, in which the seeds have a curved shape so that a line from one pole to the aer SAY pass far from the center. Floral Axis/Raceme Characters (19) Floral inception: solitary (0 d fasciculate (1) Floral meristems develop in the axils of small bracts in all OTUs (Fig In some g Ше ? o several floral meristems de each bract axil, and in others only a sin loral е develops velop Sy ы multiple жаш meristems at a bract often results in ulate flowers (but character 20 *bract fidelity"), while a “alien же fle icis from a single meristem Taxa with fasciculate Soci gd of flowers have a large bract subtending the entire fascicle and smaller “bracte- oles" subtending ilis individual flowers. This character may vary within a raceme, but this var- iation is usually minor and occurs predictably, with soli- t the very tip of the raceme and fasciculate e highly reduced racemes with relativ s the flowers "d arise solitary Because this feature is correlated with li section Weinmannia are pi be to ere not scored in the matrix. 20) Bract fidelity: present 0. absent (1). Although flowe ers are initi iated from bract axils, a. ju) not all to be “present.” due to a visible zone of elongation among the flowers, then bract fidelity is "absent." (21) Persistence of floral bract: absent (0), present (1). This character is scored around the time of floral ma- turity, such as when the anthers dehisce. Precise timing is difficult, because specimens are not collected at uni- form р stages. If the bracts are consistently or firmly attached to the axis, not just ir жылдын on 1 loosely after abscision, persistence is coded as "prese ~ 2b. 3a, b). Species (22) Flower-bearing axis: branched (0), unbranched (1). a g in Cunoniaceae is almost u t es, and branched axes, which A lateral UP EARN branches and floral me The term "raceme" is used here to denote any un- branched, bearing axis. Family-level cladograms > роону of fascicled-flowered racemes (character 19) sug- gests that they may be derived from branched the suppression of lateral sho Jane es кыд aris fascicle. If fascicled-flowered rac branched flower- bearing axes Пик is че n 19) is expec sted to be ке расни" 1 | 23) Ds arising as split pairs: present (0), absent 1 = In some taxa, гасете meristems have the tendency (1.е., this character is not always expressed) to split and form a pair of racemes from a single initial. Often, the pair of racemes is united by common rachis tissue near their )ase Inflorescence-Module (IM) Characters (24) а Ne of or panne axes as part of IMs: ent (0), absent Species of section E oasis (Fig. 2c) and a few of section Spicata (Fig. 3d) have flower-bearing axes only developing directly from axillary buds along the main such characteristics as usually shortened internodes, com- plete or partial ө, ыы of leaf development, and dis- у. ry short 4-1 to vegetative internodes in most Weinmannia species. In sec- tions Leiospermum and o a IM кшй а аге к in length to vegetative o (26) Fate of IM terminus: iN axis (0), vegeta- tive bud (1), aborted (2). character accounts for what happens to the domi- nant 5 meristem within ап IM. There аге three istem fates: M ug e bud (Figs. 2b, 3a). axis (Figs. 2a, 3e), or aborted (Fig. 3b, c). fepe шш, terminating the IM often develop vegeta- e branche r fruiting on IM меб, эн da branched (0), sequential (1), uninodal (2). IMs may be composed of one to several metamers. Іп many species of Weinmannia and Cunonia, the IM is lim- ited to a single raceme-bearing metamer (uninodal; 2e in section — and sor members of section /nspersa have IMs with the pote а о develop more than one metamer in a sequence, without ranching, and with racemes at each node (sequential; — 592 Annals of the Missouri Botanical Garden Figs. 2a, 3c). A pair of species in section /nspersa have racemes borne at the ends of a decussate-branched mod- ule (branched; Fig. 3e). As mentioned previously, Weinmannia Nee iud and W. monticola have a fixed vegetative arc 'hit mers. in а sequence. Her re, this Is re дие. as a ee a = = = © 2 © 3 E & fo iz z= = a = ES = Z ò = = = ~ о © ~ S & ЕЕ = zE = % Ф о = =| “ч = ы б о Ф б. = m E 3 m = ~ - = = т O OTU alec ла species (W. oua mensis, W. paitensis, and W rrata). Total Inflorescence (TI) Characters (28) Du cn axis borne along the main stem: ab- sent (0), present (1). This character distinguishes whether nodes along the main stem axis bear racemes ee from axillary buds. owever, direct development of racemes from axillary = of the main stem M. not prec 11: the dev elopment emes as part of an 1 oa Position p qiue aule (О), lateral and medial (1). medial only This refers to i ig ation of IMs relative to the main LE axis of stem growth. Lateral IMs develop from axillary buds. A medial IM develops from medial, apical buds. This character is inapplicable when racemes only develop directly from meristems along the main stem, e.g., in sec- tion А (Fi morensis group (Ве section Spicata, for which characters 24 and are e d as state 1. E TI formed by more than one main stem metamer: present (0), absent is character scores whether the main stem produces racemes or IMs at successive nodes. For example, this character г sc кісі "absent" for Figure 2c and “present ~ Figure 31) TI jc basitonic (0), isotonic (1), acrotonic 2) б. ~ T The size or оа of IMs ог flower-bearing ах may depend on their position within the TI. Isotonic Ts have IMs or flower-bearing axes of the same size at all nodes. Basitonic Tls have pea basal, subdistal meta- mers, while acrotonic TIs have larger distal metamers. opment is manifested in two distinct umber of metamers к 1 an IM (e. ретип, "ig. 2a). ariatio the р ан of IMs at a node in response t б ds eU within the TI (e.g.. n 2.1. Fig. 3 593 e еттојтвол eweyoy T/0| 0 ci м, eo er[ojgtqua А L 4. o = | ~ eorqereo езетпотива erAn[opT?o (ds 1) ertteuornd etuouns (ds 7) етоотзте-Ртџоџп2 a "<< O| Aj 4 O о о о о 0; о о (ds т) езеттпа etuouny inmannia (ds z) eesuerqeq-eruoun? (ds z) dnoe-eruoun;j (ds єт) eeaindand-eruoun? e ~ ~ ~ ci ~ (ds с) еттАчдолорш-Ртиоџпо сч ~ (ds т) esousoe1-137 (ds Т) етоотлтАв-тят (ds рт) STUTIJP-1371 Species Groups in We Bradford сч ~ (ds с) e3e1198-137 (ds z) зетвџешето-2ма ~ | ~ 2/Т (ds с) тәттпа-очя (ds І) eue 244 E 4 ы a ~ ті ~ a ~ (ds ZI) eeurxez3-5v4 [m] ~ = ~ (ds z) easnueA-SNI CMT DNL = Qu u e ~ тай қаймана ч ОО ооо ojl ojl ojl ojl ojoje (dsg) euerqeAnoT-SNI O| H| A O NI NI e e, | N| N O| яя Ala) Ala) A] Aj A] Aj =“ Чо о ач o olo ON NIO, NI NINI (QNI | AA e| | ON | NI NINI N| N e мч O| e| e| O| e| e ala) O| O| N mj e| e| Al A Ol A] ~ “| =ч) 4] Ol «| Aj ч Oj О) а О О оо oo =ч ~ т (ds z) тт que3 SNI = ~ о (ds с) вт IdS ~“ (ds 9) еттојтоеот-145 e| гч | CN | CN і e| =ч e ~ (ds 91) eueraze(oq-rias (ds 2) еттозоитз-їям| (ds £4) еземита-там чаа Оч ч оо ооо оо оо чо ооо оо ч ч.о ч со N| A| ч оо ча Оу О ооо ОО ооо/о/ооо о О ООО = чоч со N| e| 4 Al A] о A] Aj 4] Al Al A] Aj A] Aj Aj A] Al Al A] | A) AlAs ASO}; О ч о Чоо Al ч ооч чоо со о о" о)"о о о/о ооо оо! о о|оо о |і ооо іі Hio я ао осо осо осо ос осо чоо оо со чо << о о о о о оооо ооо о ооо о о о о о о о ос оо ч чч) Al aA A] Al чоч ч чуч чо Al чуч) Aj HA O о) OF; о оу о оо о о + [79] "m| e| e| e| e, mj e| Aj по Aj A] Aj mA] чече е dolo O 9)! ооо о О ©! ч < 4-4 4 іі Al Al A] Al Ala) 1.14.4 Al aA Al Aj | = |! | Al A oO] О nt oO со et Оо ој оу оу ој о OF] OF О ОО Оо о О о O| O| m| Al aA | 4) ч о ОП 01! о a) Oo} О о О о Оо Оо о Оо о о Ооо ч о ч ч о ol olol о о о ооо О айел О! О) О ©] ооо aA] O| аке ч| оо ojoj; ojoj ojl o| olol olim; o моо оо оо оо O| O| O| O| O| =ч, Al «ч A es A| A m| A m| m| “| О Ft о N| O| O| ну H| tt чч ч ч ч O| O| me, о чч! чч тч Al Aj H| H| A e| nri! n! nd сч 0 0 0 0 0 0 0 0 I I I I 0 [4 [4 I I I AO} O| O| A| A| N| N| N| N; mj mj СО rn, n! n! n ч 92| sz + "ч сч © о ~ сч “ч 0 о о оо ооо ооо о ОО о Оо о о O| ~ чо al m| === оооо о чоо ооо O| O| O| m| H| m| <= == ча со ООО ооо чо сч о 0 0 0 0 I I I I I I I I I I I I I I I I I I I I I I I I 9 ТІ 01 010 0000 0} 0; 0| 0 отог отте I| T] Tj € O T| TZ OLT TZ O| T| Tj 2 0; T| Tj 2 от оё OJ T| T/T O T| TIT оттт от тт отоо оо T| 0 0| 0j T| 0 0; 0 0; 0 0; TITT от I| T 0; Tj) TIT O TITZ T T| TIT тттт TITT 0,000 0710.00 S v" eE iz e|ece[cdjoojojolojojo іі оо о| ооо оо O CO] OC] m| m| | ч 19308 1PYI/SNLO Volume 85, Number 4 1998 “SIIJIBIBYO JO U0ISSNISIP 10] € xipuadd y pue “SOLO jo uornrsoduioo 10) | x1puadd y ээ ‘хеш Iere) p XIANAAd Y TAXONOMIC REVISION OF GERANIUM SECTIONS BATRACHIOIDEA AND DIVARICATA (GERANIACEAE)! Carlos Aedo?, Juan José Aldasoro,? and armen Navarro ABSTRACT Geranium и Robertium (Geraniac ien comprises eight sections, of ibo h section аы "hioidea and section Di- here varic ata are sed Divaricata comprises two s the other endemic to the Caucasus. | relationship between 'hioidea consists о We also accept G. aequale to шшде plants similar to G. molle but ih smooth mericarps. Geranium ecies centered in Eurasia, between the eses. роп and the Hin DNA rerum ain oth sections e = = Ф = .= = = = = Ф st > E < = = = a Y 2 = = ~ a ^ a 5 ( os stic aspects of morphology. Nomenclature for all species is reviewed, 32 lectotypes are designated, and descriptions, m tribution maps, and illustrations are provided. The genus Geranium L. (Geraniaceae) is distrib- uted throughout most of the world. A brief history of generic delimitation and infrageneric classifica- tion, as well as a Е of the genus, can be found in Aedo (1996). O species and is divided, according to the cur- rently accepted classification. (Yeo, 1984), three subgenera: subg. Geranium, subg. Erodioidea (Picard) Yeo, and subg. Robertium (Picard) Rouy. Only subgenus Erodioidea has been monographed recently (Aedo, 1996). Geranium subg. Geranium comprises over 380 species, grouped in at least 10 Geranium comprises about into sections. Some of these sections have been revised (Davis, 1970; Carlquist & Bissing, 1976), but much more work is necessary to attain a satisfactory knowledge of subgenus Geranium. ranium sects. Batrachioidea W. D. J. Koch and Divaricata Rouy, the taxa studied here, belong to subgenus Robertium, which is firmly supported by both morphological and chloroplast-DNA data (Yeo, 1984; Price & Palmer, 1993). According to Yeo's (1984) sectional classification, шры Rob- ertium comprises eight sections and 30 s Section Polyantha Reiche (8 species) is endemic ecies. to the eastern Himalayas and southern China. Sec- tion Anemonifolia R. Knuth (2 species) also has a limited distribution, being endemic to Madeira Is- land. Section Trilopha Yeo (5 species) is restricted to mountains in tropical Africa, western Asia, and the eastern Himalayas. The d of the re- maining five sections, Lucida , Ruberta Dumort., Divaricata, Batrachioidea, and [; lata (Boiss) Reiche, is centered in the Mediterra- nean region and western Asia, though section Rub- erta extends in the east to Japan, and in the south to mountains of tropical Africa. Sections Anemonifolia and Ruberta were revised by Yeo (1973). The same author also studied most id the species of section Polyantha (Тео, 92). Following upon the recent revision of Gera- ы subg. Erodioidea (Aedo, 1996), and in pursuit of our aim to prepare a comprehensive monograph of the genus, we here present a revision of two sec- tions of subgenus Robertium: sect. Batrachioidea and sect. Divaricata. The taxonomic problems of section Batrachioi- dea were confined to two species, G. pyrenaicum and G. molle. The variability of G. pyrenaicum had ! The authors thank R. Morales, J. Mufioz, P. Perret, L. е апа М ,arcía García for help with some ope C. Jarvis, F. Mufioz, and M. Laínz for nomenclatural advi of be; manuscript; and 5. C о for uncompromising penu We herbaria for kind assistance during our visits anc Dirección General de Inve sstigaci ión Científic ? Real Jardín Botánico, Conse (e-mail address: aedo@ma-rjb.csic.es). * Departamento de Biología Vegetal II, Facultad de ANN. Missouni Bor. for specimen loar а y Técnica (DGICYT) through the researc Н project PB91—0070-CO03-X o Superior de Investigaciones Científicas, Plaza de Murillo 2, 28014 Madrid, ен Rico, А. Pur cis and R. Vogt for help with literature; 1. ill and P. Buek for help with some localities; D and an anonymous reviewer for accurate are also de to the curators of the cited . This work was partly financed by the Xen dea . Mil e reviews Farmacia, Universidad Complutense, 28040 Madrid, Spain. GARD. 85: 594—630. 1998. Volume 85, Number 4 1998 Aedo et al. 595 Geranium not been well studied except for Ortiz's (1989) work, which included mainly Iberian material. Ge- ranium molle is a highly variable species from which the most robust forms had been segregated and named G. brutium. Another problem to address in this revision was the taxonomic status of some plants similar to G. molle but with smooth meri- carps. Section Divaricata, comprising two поп- problematic species, had not been revised since Knuth's (1912: 57, 154) monograph. Both sections could constitute a monophyletic group, as suggest- ed by the presence of a derived character state (blue pollen). However, at present, there are no oth- er data to confirm this hypothesis. This revision of sections Batrachioidea and Di- varicata is a first attempt to explore the phyloge- netic relationships within Geranium subg. Rober- tium. Future work may be focused only on sections Trilopha and Unguiculata, because section Lucida is monotypic and non-problematic. MATERIALS AND METHODS This revision is based on more than 2000 her- barium specimens from the following herbaria: AK, B, BAF, BC, BISH, BM, BR, C, CAN, CAS, CHR, G, JE, H, K, L, LE, LISI, LOU, LY, M, MA, MAF, MO, MPU, MUB, NY, OXF, PAL, PH, PO, RO, and W. Furthermore, microfiche, photographs, and other data have been examined from the following addi- tional herbaria: BREG, DS, E, GB, GFW, HAL, LD, LINN, LISU, MANCH, NAP, S, SGO, SZU, TBI, U, UPS, US, and WRSL. Unfortunately, we have had difficulties in obtaining some types on loan. The most relevant cases are those of F. Schur, A. Ter- racciano, and N. Terracciano. Schur's original ma- terial is spread through several herbaria. B, C, E, L, PH, and W have none of Schur's original mate- rial, while BP, BRNU, GOET, LW, MW, NA, P, and WU did not respond to our requests. Terracciano's herbarium constitutes a separate collection in МАР. However, this collection has not been available for study since World War II (fide A. Santangelo, in litt.). The dispositions of names for which no type material could be located or obtained are based on the opinions of previous authors (as indicated). Where no reliable opinion was found, these names are included in a *Dubious Names" section. Cladistic analyses were carried out using the PAUP software package (Swofford, 1993). АП char- acters were unweighted and unordered. Data were analyzed using the exhaustive option. Polarization of characters into plesiomorphic and apomorphic states was assessed using the standard procedure of outgroup comparison (Watrous & Wheeler, 1981). MacClade version 3.04 was used to edit the data set analyzed with PAUP (Maddison & Mad- dison, 1992). It was also used to map the distri- bution of particular character-state changes. A bootstrap analysis (Felsenstein, 1985) with 1000 replicates was conducte Descriptions of leaf venation in this work follow the terminology of Hickey (1973). Seeds were cut with a razor blade both longitudinally and trans- versely in order to reveal their internal structure. Thin hand-cut sections were taken in the micro- pylar third and photographed under optical micros- copy. Other sections were made with a SLEE- MAINZ-MTC microtome and stained with Казга mixture (Tolivia & Tolivia, 1987) or with Sudan red and Malachite green. For scanning electron mi- croscopy (SEM), samples were glued to aluminum stubs, coated with 40—50 nm gold, and examined with a JEOL-TSM T330A scanning electron micro- scope at 20 kV. Species-distribution maps were based primarily on exsiccatae, though for G. albanum literature rec- ords were also used. RESULTS ANIMA AAA УДК, АД, CHARACTERS Duration and habit. are herbaceous perennial plants with horizontal rhi- Most Geranium species zomes; however, some species in subgenera Rob- ertium and Geranium are annuals. The different an- nual species in the genus do not resemble one another, and they share characters with different groups of perennials. Consequently, we consider that annual species have probably been derived from perennials several times in the genus. One species of section Divaricata is perennial and the other is annual, while section Batrachioidea has one perennial and three annual species. Geranium albanum (sect. Divaricata) has a horizontal rhizome that has been codified, according to the outgroup, as plesiomorphic. Geranium pyrenaicum (sect. Ba- trachioidea) is also perennial, but it has a vertical, napiform rhizome. All remaining species from both sections are annuals. We consider that this could be interpreted as a linear transformation series, with horizontal to vertical rhizomes leading to an- nuals without rhizomes. Among the perennials, a vertical rhizome should arise from a horizontal one, and thus ought to be considered as derived. Finally, in Geranium annuals are usually considered de- rived while perennials are primitive (Yeo, 1984). Moreover, Sanderson (1991) proposed a similar multistate series for Astragalus (Fabaceae: Papi- lionoideae): 0 = perennials with well-developed 596 Annals of the Missouri Botanical Garden Figure 1. folding, —A. Seeds of Geraniu with conduplicate cotyledons s.n. (MAF-118241)), s, each lyin section Divaricata, with = half of cotyledon C,; P, = petiole of cotyledon C: ком short-lived developed rhizomes, and 2 rhizomes, 1 perennials with poorly — annuals without rhi- zomes. The aerial portion of the stem is usually erect in all species of sections Batrachioidea and Divari- cata; however, the annual species of section Batra- chioidea can also have decumbent stems. Cotyledons. The cotyledons usually have entire margins in Geranium, but those of a few species are incised. Geranium aculeolatum Oliv., with cot- yledons having two notches on each side, is the only species in subgenus Erodioidea without entire cotyledons (Yeo, 1990: 13). In subgenus Geranium, only G. bohemicum L. has cotyledons with a single incision on each side (Dahlgren, 1943: 137 fig. 5). Finally, in subgenus Robertium, the two species in- cluded in section Divaricata have cotyledons with a single incision on each margin. According to the outgroup, in all these cases incised cotyledons are considered as a derived condition. The cotyledons in Geranium are always condu- plicate, one half of each cotyledon lying in the pri- Optical photomicrographs of transverse sections of Geranium seeds um pyrenaic um subsp. lusitanicum pe arreira s.n. (МА- 47: ng in the primary fold of the the proximal part o longer than in section Batrachioidea; thus, transverse sections at the n micropylar third of the seed show the cotyle in um о pats (the basal auricles of the cotyledons) and their petioles. С Р, g diff t patterns of cotyledon 3325)), section Batrachioidea, Seeds of G. divaricatum (Sánchez Mat ol colado ns 2. cordate and cotyledon petioles dons + С, = cotyledons; CH, = half of cotyledon С; D of cotyledon С; К = nde le. Scale bar — 300 mary fold of the opposite cotyledon (Yeo, 1990: 14). Moreover, seeds of section Batrachioidea show some differences from those of section Divaricata. In section Divaricata, the proximal part of the cot- yledons is deeply cordate and the cotyledon peti- oles are longer than in section Batrachioidea. Con- sequently, transverse sections at the micropylar third of seeds of section Divaricata showed both the petiole and the cotyledons, the latter divided into two parts (the basal auricles). Conversely, in section Batrachioidea, as in most of Geranium, the coty- ledon base is truncate and the petioles are very short (Fig. 1). These differences were not consid- ered by Tokarski (1972), who showed a similar pat- tern of simply conduplicate cotyledons in these sec- tions АП species in sections Divaricata and Batrachioidea have more or less deeply palmatifid leaves. Leaf outline is pentagonal in section Divar- icata, whereas it is usually orbicular to reniform in section Batrachioidea. The segments can be rhom- bic, as in section Divaricata, or obdeltate, as in Volume 85, Number 4 Aedo et al. Geranium section Batrachioidea. Obdeltate segments seem to be derived, according to the outgroup (Nieto Feli- ner & Aedo, 1995). The number of lobes per seg- ment varies between 7 and 15 in section Divari- cata, and between 3 and 12 in section Batrachioidea. The lower cauline leaves can be ei- ther opposite or alternate in both sections. Accord- ing to Davis (1970), alternate leaves should be the primitive state in Geranium, as well as the out- group. Inflorescence and branching. The inflorescence in both sections is cymose, composed of axillary, two-flowered cymules. All cymules arise along ae- rial stems. The most significant inflorescence fea- ture taxonomically is the indumentum of the peduncles and pedicels. Almost all species have two types of hairs. One type comprises patent, eglandular hairs 0.7–1.8 mm long, as in G. alban- um, G. divaricatum, G. pyrenaicum subsp. lusitan- icum, G. molle, and G. aequale. This type of hair is lacking in G. pusillum, and usually also in G. py- renaicum subsp. pyrenaicum. The other type of in- dumentum, composed of glandular or eglandular patent hairs less than 0.5 mm long, is present in all species. Sepals. The mucro of the sepals is very short (less than 0.6 mm) in all species of both sections, except for 6. divaricatum, in which it is ca. 1 mm long. The last-mentioned species seems to be un- usual in this regard, in subgenus Robertium, ac- cording to Yeo's (1992) description. Long, eglan- dular sepal hairs are common in most species, but lacking in G. divaricatum and G. pyrenaicum. Petals. Petals in both sections have emarginate apices, with the notch usually ca. 1 mm deep. Ge- ranium pyrenaicum has more deeply emarginate (2-3 mm) petals, while its closest relative, G. pus- illum, has shallowly emarginate petals (0.2—0.5 mm). The longest are those of the perennial species, G. albanum and G. pyrenaicum. However, the an- nual С. molle occasionally also has long petals, as discussed under that species. In section Divaricata, as in most other Geranium taxa, the petals are ta- pered uniformly toward the base, without any claw; however, in section Batrachioidea, a very short claw is evident. Stamens and pollen. Іп both sections, as in the entire genus, the ten stamens are arranged in two whorls. In G. pusillum, the anthers of the external whorl are missing. The filaments are usually hairy on the abaxial side, as in many species of the ge- nus. However, those of G. molle and G. aequale are glabrous abaxially, which should be interpreted as a derived character state, according to the out- group. Another character is the presence or ab- sence of cilia along the filament margins. All spe- cies studied have ciliate filament margins, except for G. albanum. According to Stafford and Blackmore (1991: 51), pollen of Geranium divaricatum, G. pyrenaicum, G. pusillum, and G. molle belongs to the G. molle type, which includes most of the Geranium species stud- ied by them. This type is characterized by reticu- late exine ornamentation with distinctly baculate, clavate, or gemmate supratectal elements. Four groups were recognized by these authors on the ba- sis of secondary variation in exine ornamentation. These groups showed no concordance with subge- neric or sectional classifications. Thus, G. pusillum and С. pyrenaicum were placed in the С. robertian- um group, and G. molle and G. divaricatum in the G. molle group. Blue pollen is the only feature known to be shared by sections Divaricata and Batrachioidea. The other sections of subgenus Robertium have yel- low (sects. Anemonifolia, Lucida, Polyantha, Rub- erta, and Unguiculata) or white pollen (sect. Tri- lopha) (Yeo, 1984: 13-17). All species of subgenus Erodioidea and most of subgenus Geranium have yellow pollen, though in the latter subgenus at least three species (G. dissectum L., G. pratense L., and G. richardsonii Fisch. & Trautv.) have blue pollen. Consequently, blue pollen is viewed most parsi- moniously as derived. Fruit. Geranium sects. Divaricata and Batra- chioidea belong to subgenus Robertium, which ex- hibits the “carpel-projection-type” of fruit dis- charge (Yeo 1984). Here, the whole mericarp is actively discharged by the explosive recoiling of the awn, which remains attached to the columella. Ac- cording to Yeo (1984), this discharge type is pre- sumably derived, the Erodium-type discharge being the primitive condition. In section Divaricata, de- crease in rostrum length reduces the effectiveness of the discharge mechanism. According to the out- group, this decrease is also viewed as derived. A rostrum that tapers gradually to the remains of the stigmas is the most frequent condition in those Ger- aniaceae exhibiting the Erodium-type discharge (see discussion following the generic description). The alternative condition, a ен rostrum abruptly narrowed at the apex, as in G. molle and С. mr is probably derived (Nieto Feliner & Aedo, 1995: 203). he mericarp surface is smooth in all species of section Batrachioidea except G. molle, which has transversely wrinkled mericarps, as do the two spe- 598 Annals of the Missouri Botanical Garden Figure 2. rse cuts of the meric ч in section Disaricata "атт (дей 3864 (MA)) е = exocarp, m = m 0 Е - , D, bar = 100 рт; E. bar = 10 jum; P penes = cies of section Divaricata. According to the out- group, transversely rugose mericarps are here in- terpreted as the derived condition. In both sections, the mericarps are usually covered by short hairs, but G. pyrenaicum exhibits two conditions: subsp. pyrenaicum (Fig. 2B) has hairy mericarps, while subsp. lusitanicum (Fig. 2A) has glabrous meri- carps. The mericarp surface is also virtually gla- brous in G. molle and G. aequale (Fig. 2C, 2D), with only a few cilia along the margins. According to the outgroup, glabrous mericarps are here inter- preted as the derived condition. arp, s = sclere SEM photomic rographs of Geranium mericarp. A-D. Surface ornamentation in sect. Batrachioidea. —A. : subsp. pyrenaicum (Granzow )). —D. 6. aequale (Airy Shaw € Nelmes s.n. leet): El A-7116 E. G. 2... (Bernouilli s.n. ( 69)). —F. 'hyma region, ie = endocarp. Scale А, B, ber 50 pm The mericarp wall in G. albanum is wider and has a more well-developed mesocarp than that of the other species studied here. Geranium divari- catum and species in section Batrachioidea have relatively thin mericarps due to no or limited de- velopment of the mesocarp, with at most a single layer of cells. The mericarps of all the studied spe- cies had well-developed sclerenchyma regions with crystals (Fig. 2E, 2F). The thicker mericarp of G. albanum could be related to a different germination strategy in this perennial species. Several authors have reported a relationship between dormancy and Volume 85, Number 4 1998 Aedo et al. 599 Geranium the permeability of seed testas and fruit walls (Al- dasoro et al., 1981; Rangaswamy & Nandakumar, 1985; Bewley & Black, 1994). Seeds. Seeds are more or less elliptical in out- line in section Batrachioidea, and obovate in sec- tion Divaricata. The seed-coat in both sections ap- pears smooth at a magnification of 30X, but SEM shows a reticulate surface due to the prominence of the outer and the middle layer of the outer in- tegument. The outer layer has cells with thickened walls and collapsed lumina, forming a polygonal structure. The seed-coat is usually brownish and bears scattered stomata. The cells of the inner part of the outer integument are strongly lignified and contain tannin and crystals. The next layer (the out- er layer of the inner integument) is also sclerified, but the cells are not so compacted, being prismatic with undulate anticlinal walls. In G. albanum, a species with thick mericarp, the testa is weaker because the cells of the outer layer of the inner integument are wider (са. 24 jum), almost cubic, and the lignified walls are more wide- ly separated (8-15 рт) than in the other species ere studied. Conversely, G. aequale, with seeds only partially covered by the mericarp, has the thickest testa (Fig. 3) in sections Batrachioidea and Divaricata (ca. 41 jum, vs. 25—30 jum in the other taxa). CHROMOSOME NUMBER The chromosome number of all species in Ge- ranium sect. Batrachioidea is 2n — 26 (see Ap- pendix 1). There has been some controversy in the case of G. pyrenaicum and G. pusillum, but Van Loon's (1984a, b) work has clarified the situation. All chromosome counts carried out to date for G. divaricatum are 2n — 28. However, for G. albanum, Warburg (1938: 145) reported п = 14 and Van a: 276) 2n — 20 (see Appendix 1). Ma- terial of both species was unvouchered and col- lected in botanical gardens. Thus, the chromosome number in section Divaricata is probably 2n — 28, but more counts should be done for G. albanum. According to Van Loon (1984b: 286), the basic chromosome number in Geranium is x — 14, as in most of the perennial species of the genus. The annual taxa, with various other base numbers, prob- ably evolved independently. In this context, the number 2n — 26 in section Batrachioidea could be seen as a derived character state. HYBRIDS Hybridization experiments in Geranium subg. Robertium have involved species of sections Ane- monifolia, Batrachioidea, Lucida, Ruberta, and Un- guiculata (Van Loon, 1984c; Widler-Kiefer & Yeo, 1987). No data are available for sections Divari- cata, Polyantha, and Trilop In section Batrac ma Abd hybrids have been described: G. Xoenense (said to be G. molle X G. pusillum); G. Xluganense (said to be G. molle X G. pyrenaicum); and G. Xhybridum (said to be G. pusillum X G. pyrenaicum). According to Van Loon (1984c), intraspecific crosses were usually highly successful in this section, but the only suc- cessful interspecific cross was that involving G. molle and G. brutium. Other crosses (G. pyrenaicum X G. brutium, G. pyrenaicum X G. molle) also pro- duced seeds, but the seedlings succumbed at an early stage. Thus, according to Van Loon's data, species of this section seem reproductively isolated. Having thoroughly studied original material and/ or original descriptions, we consider that G. luga- nense, G. oenense, and G. hybridum are probably not hybrids but synonyms of G. molle (the first two) or G. pusillum. Considering the difficulty in obtain- ing interspecific hybrids (Van Loon, 19842), the only successful cross, involving G. brutium and G. molle, supports our interpretation of G. brutium as a synonym of G. molle. DISTRIBUTION Geranium subg. Robertium is distributed widely in temperate regions from Macaronesia to Japan, and sections Trilopha and Ruberta reach tropical areas of central and east Africa. Section Divaricata comprises two species, with very different patterns of distribution. Geranium al- banum is endemic to the Caucasus and northern Iran, whereas G. divaricatum is distributed in a wide longitudinal range between Spain and the central Himalayas. At present, neither species has been reported as introduced in other areas of the world. The four species of Geranium sect. Batrachioidea are centered in Eurasia, between Macaronesia and the Himalayas, though all but G. aequale reach north Africa. They are spreading rapidly in tem- perate areas of North and South America, southern Africa, Australia, and Japan, where representatives of subgenus Geranium mainly grow. This process of colonization predominantly involves the three an- nual species, which occur in perturbed habitats, but also G. pyrenaicum. PHYLOGENETIC RELATIONSHIPS A cladistic analysis of Geranium sects. Divari- cata and Batrachioidea was carried out using a data 600 Annals of the Missouri Botanical Garden Figure 3. outer part of the inner inte gument more developed i wever, in: testa cells are wider in G. айанын (са. 24 о (Airy Shaw & Nelmes s.n. (MA-7123 ‚ bar = 50 pm; E, F, bar = set of 15 characters (Tables 1 and 2). The species used as outgroup was G. sylvaticum L. This decision is supported by the results of an rbcL-sequence data analysis, which places the clade composed of subgenus Erodioidea and subgenus Geranium as sister to subgenus Robertium (Price & Palmer, 199 e have selected G. sylvaticum, a member of subgenus Geranium, since in this species many characters of the in-group are applicable. Species in G. aequale . G. pusillum (Navarro et al. —D, E e albanum (Aedo 3864 (MA)). o1 = outer а nt; ii = inner integument. Scale ) um. Optic al (A-D) and SEM (E, F) photomicrographs of Geranium seed sections, showing the testa with the 1 рт) than the other species (25-30 шт). шт) than in pos ui T а (8-15 m). —A, Е 6. aequale ‚б. divaricatum (Sánchez of subgenus Erodioidea have quite different fruits, and many of the codified characters (characters 10, 1l, and 12) are inapplicable (Nieto Feliner & Aedo, 1995). One most-parsimonious cladogram was obtained, with length 20, consistency index (C.I.) 80, and retention index (R.I.) 81 (Fig. 4). Two well-supported clades were obtained in the cladogram, corresponding with sections Divaricata and Batrachioidea. In section Divaricata, mono- Volume 85, Number 4 1998 Aedo et al. Geranium 601 Table 1 ertium). Polymorphic, inapplicable, or missing data are coded as ‘?.’ Characters 1-15 are in Table 2. Data matrix used in the cladistic analysis of Geranium sects. Batrachioidea and Divaricata (subg. Rob- Acronym 1253 4250607.8 911 1111 012345 G. sylvaticum SYL 00000000?000000 G. aequale AEQ 2 0050 LUN TIC 0L LL 0 d 01 G. molle MOL 2 040, T XT реј 0 190 170 1 G. pusillum PUS 203 1840110011001 G. pyrenaicum subsp. pyrenaicum PYR l 0.1 уво 0.011.001 G. pyrenaicum subsp. lusitanicum LUS LO ТАО Т 0:10. LL 1-01 С. albanum ALB 0-11 "ОНО О Т1: 9? 9.2 01 ? G. divaricatum DIV 2-120 0070 7131/1 2 0 0 0 1 9 phyly is supported by three synapomorphies: (a) the incised margins of the cotyledons (character 2, see Table 2); (b) the inoperative fruit-discharge mech- anism (character 9); and (c) the obovate outline of the seeds (character 14). As previously described, this clade is also upheld by seedling structure. In section Batrachioidea, monophyly is also supported by three synapomorphies: (a) the obdeltate leaf seg- ments (character 4); (b) the presence of a petal claw (character 5); and (c) the chromosome number 2n = 26 (character 15). Geranium molle and б. aequale constitute а Table 2. Characters and character states used for cladistic analysis of Geranium sects. Batrachioidea and Divaricata (subg. Robertium). Characters Character states 1. Habit N . Cotyledon margins 3. Basal cauline leaves 4. Shape of leaf segments 5. Petal claw 6. Stamen pubescence on abaxial side --1 . Pollen color 8. Fruit discharge type 9. Fruit discharge mechanism . Fruit rostrum 11. Mericarp surface . Mericarp with longitudinal rib . Mericarp indument . Seed shape . Chromosome number perennial with well-developed horizontal rhizome perennial with poorly developed vertical rhizome — annual, without rhizome rii — alternate seed-ejection-type — carpel-protection-type — operative = inoperative = tapering gradually = narrowed abruptly moot = transversely wrinkled = Or OF OCF CF OF Cr Or OF CRM OF Or Or OF ONY со || 602 Annals of the Missouri Botanical Garden 11 -X— MOL 0 6 1013 | 111 = AEQ 95 1451115 o E си O 9 21111 1 = | Pus 9? 3 1112 1 lI 78 111 1 PYR 1 1 1 LUS 1 2914 Hoc 5 2 с | ||! 2 59 111 б Figu e 4 arra open bars are synapomorphies that rev elsew 3 -[—— ALB 1 The most-parsimonious cladogram of Miren sects. с ata and Batrachioidea. Solid bars are syn- ogram, double bars are parallelisms, and crosses are reversals (length 20, СЛ. = 80, КЛ. = 81). Se Table 1 for acronyms clade within section Batrachioidea supported by two synapomorphies (characters 6 and 10). The clade comprising G. pyrenaicum subsp. pyrenai- cum, G. pyrenaicum subsp. lusitanicum, and G. illum is supported by one synapomorphy (character 12). In this clade, G. pusillum appears as sister to the other two taxa. Monophyly of a group composed of sections Di- varicata and Batrachioidea is supported by only one character, pollen color. In the previously men- tioned tree, character 8 (fruit-discharge type) is ac- tually irrelevant, because it is shared by all sec- tions in the subgenus. However, in this study it was used to support the in-group. The emarginate petal apices also seem to support monophyly of this group, but we lack sufficient data to preclude their existence in the other sections. In contrast, blue pollen seems to be a derived feature useful at this level of analysis. No other evidence was obtained about the relationships between the two sections. A comprehensive study of subgenus Robertium could be expected to provide more information on this matter. TAXONOMIC TREATMENT Geranium L., Sp. Pl. 1: 676. 1753. TYPE: Gera- nium uate cum " A designated by Hanks & Small, 1907: 4). Annual, biennial, or perennial herbs, rarely shrubs. Leaves simple, usually palmately divided, sometimes entire or pinnately lobed, stipulate, pet- iolate; basal leaves usually forming a rosette; cau- line leaves opposite or alternate. Inflorescence ter- minal or axillary, cymose, bracteate. Flowers usually paired, actinomorphic, rarely somewhat zy- gomorphic. Sepals 5, imbricate, obtuse to caudate at apex. Petals 5, free, frequently emarginate at apex, sometimes clawed. Stamens 10, in two whorls, the outer one opposite to and the inner one alternating with the petals, all bearing anthers or, very rarely, 5 with staminodes; filaments broad, free Volume 85, Number 4 Aedo et al. 603 Geranium or united at the base. Nectaries 5, alternating with the petals. Ovary 5-locular, with 2 superposed ovules per locule, the style distinctly 5-fid. Fruit a schizocarp, long-beaked, splitting into five 1-seed- ed mericarps. Seed without or with very little en- dosperm; embryo with massive induplicate or con- voluted cotydelons. Geranium is divided into three subgenera, dis- tinguished by their fruit-discharge mechanisms (Yeo, 1984). The “seed-ejection-type,” which сћаг- acterizes subgenus Geranium, involves a single seed actively discharged by the explosive recoiling of the awn, which remains attached to the columella together with the mericarp. А second type of dis- charge, the “carpel-projection-type,” characterizes subgenus Robertium. Here, the explosive recurva- ture of the awn also acts as the propelling force, but in this case the whole mericarp, containing the seed, is dispersed, whereas the awn remains with the columella. Subgenus Erodioidea is identified by the *Erodium-type" of fruit discharge, in which the mericarp, including the coiled awn, is propelled over a short distance. Geranium subg. Robertium (Picard) Rouy, in Rouy & Foucaud., Fl. France 4: 94. 1897. Robertium Picard, Mém. Soc. Agric. Boulogne- sur-Mer 1: 134. 1837. TYPE: Geranium rob- ertianum (Greuter et al., 1994, Art. 22.5). Annual, biennial, or perennial herbs. Leaves pal- mately divided to the base or more shallowly divid- ed; cauline leaves opposite or alternate. Flowers usually actinomorphic, rarely somewhat zygomor- phic. Sepals erect or patent, sometimes longitudi- nally carinate. Petals rounded or emarginate at apex, = unguiculate; claw ecarinate or carinate. Stamens exserted or not; filaments glabrous to pi- lose; pollen yellow, blue, or white. Fruit discharge by carpel projection, each mericarp thrown off ex- plosively with the seed in it and the awn dropping away at the moment of explosion (discharge mech- anism sometimes inoperative); mericarps acute or obtuse, smooth, reticulate, ribbed or cristate. Cot- yledons entire or laterally incised. KEY TO THE SECTIONS OF GERANIUM subg. ROBERTIUM la. Leaves divided to the base. Glandular hairs of the inflorescence purple: more than half the length of the stamens ex- serted from throat of flower Geranium sect. Anemonifolia . Glandular hairs of the inflorescence wit colorless stalks and red heads; less than ioe the length of the stamens exserted fro throat of flower Geranium sect. Tubes ~ c lb. Leaves shallowly divided. 3a. Fruit discharge mechanism inoperativ Geranium sect. oe ae 3b. Fruit discharge mechanism operativ ue __ Geranium sect. Batrachioidea 4b. Pollen yellow, sometim ite. 5a. Calyx етти ара Geranium sect. Lucida 5b. Calyx not carinate. 6a. Mericarp apex obtuse; stamens e me T CNN Geranium sect. ides жа 6b. Mericarp apex acute; stamen not exserte 7a. Plants perennial ranium sect. Polyantha 7b. Plants ann ual Ge eranium sect. Batrachioidea W. D. J. Koch а Syn. Fl. Germ. Helv. Ed. 1 139. 1835. Geranium sect. Pyrenaica R. Knuth, in Engl., Pflanzenr. IV.129 (Heft 53): 46, 152. 1912, nom. illeg. TYPE: Geranium ин: Burm. f. (designated by Yeo, 1984: 15; see Aedo & Mufioz Garmendia, 1996: 104) Perennial or annual herbs; stems to 110 cm long, with simple or bifurcate monopodial branching, leafy, erect, decumbent or ascending, with patent eglandular and glandular hairs. Basal leaves in per- sistent rosettes; venation actinodromous, basal, per- fect, marginal; lamina orbicular or reniform in out- line, palmatifid, concolorous, hairy; segments 5—9, obdeltate, 3-12-lobed at apex; lower cauline leaves alternate or opposite; stipules lanceolate to ovate, papery, brown, pilose. Cymules solitary, arising from aerial stems; bracts lanceolate, sometimes lobed, papery, brown; peduncles present, with pat- ent eglandular and glandular hairs; bracteoles lin- ear to lanceolate, papery, brown; pedicels 2 per cy- mule, = ascending and often curved upward after anthesis, subequal, with patent glandular or eglan- dular hairs; peduncle and pedicel together very of- ten exceeding the subtending leaf. Sepals ovate, erect-patent at anthesis and erect in fruit, briefly mucronulate, marginally scarious; abaxial surface with eglandular or glandular hairs; adaxial surface glabrous, with a subapical tuft of hairs. Petals erect-patent, = obovate, emarginate, with a very short claw, without nectar passages, ciliate at base, with sessile glands on the adaxial surface, = pur- ple, without a dark basal spot. Stamens 10, both whorls bearing anthers or the inner one without an- thers; filaments lanceolate, expanded at base, per- sistent in fruit, with a conspicuous midvein, = cil- iate, usually pilose on abaxial surface, yellow with pink apex; pollen blue. Nectaries hemispherical, 604 Annals of the Missouri Botanical Garden glabrous. Stigmas purple. Fruit of the carpel-pro- jection-type, with discharge mechanism operative; mericarps smooth or transversely wrinkled, some- times with a longitudinal dorsal rib but never cris- tate, usually covering the seed completely, without a basal beak and without a callus, glabrous or hairy; rostrum not reduced, narrowed apically or not; stigmatic remains with 5 pilose lobes. Seeds ellipsoidal, smooth, brownish or reddish, the hilum YY, as long as the perimeter. Cotyledons entire. Chromosome number: п = 13, 26. Distribution. Africa and Macaronesia, Europe to central Asia and the Indian subcontinent, Aus- tralia, North America, southern South America, subantarctic and north-central Pacific Islands. e three most distinctive character states for Geranium sect. Batrachioidea are its obdeltate leaf segments, short petal claws, and chromosome num- ber of 2n — 26. The chromosome number is es- pecially relevant as a derived character, as it has not been found in any other section of subgenus Robertium. KEY TO THE SPECIES OF GERANIUM sect. BATRACHIOIDEA la. Stamens 10, the external whorl without anthers O E: pusillum lb. Stamens 10, both whorls bearing шн. ericarps rugose 2b. 2. smooth. . Plants annual; petals 3.5—4.5 RD 2. G. molle mm long aequale 3b. Plants atia p 7-11 mm long o 4. G. pyrenaicum 1. Geranium aequale (Bab.) Aedo, Anales Jard. Bot. Madrid 55: 466. 1997. Geranium molle var. aequale Bab., Man. Brit. Bot. Ed. 2: 65. 1847. TYPE: United Kingdom. England: near Leamington [52*15'N, 1°29'W], J. J. Murcott n. (lectotype, designated by Carolin (1965), E not seen). ее iens f. preusc = Abrom., Fl. Ost- & West- pre ett 1898. TYPE: Germany. 1. еп, Maz lebug, Plagne in Tannsee," [51°43' 10°43 El Abrom 1. (no authentic material loc cat ed; synonymy та to Yeo, 1984). Annual herb to 40 cm tall; stem erect or decum- bent, usually branched from the base, pilose, with ong eglandular hairs 1-1.2 mm long and short glandular and eglandular hairs < 0.5 mm long. Basal leaves in a persistent rosette; lamina 1.5-3 (-5) X 1.5-3.7(-5.8) em, divided for 0.6—0.75 of its length, pilose, with eglandular, appressed hairs; segments 7-9, 2-4 mm wide at the base, 3(—5)- lobed at apex; lower cauline leaves alternate; pet- ioles to 14 ст long, with patent, long eglandular hairs ca. 1 mm long and short glandular and eglan- dular hairs < 0.5 mm long; stipules 6-7 X 3-4 mm, ovate-lanceolate, sometimes lobed, pilose with eglandular hairs on abaxial surface, glabrous adax- ially. Bracts 2-4 X 1.5-2 mm, pilose with eglan- dular hairs on abaxial surface and on the margin, glabrous adaxially; peduncles 1—7 cm long, pilose, with eglandular patent hairs 1–1.7 mm long and short (< 0.5 mm) glandular and eglandular hairs; bracteoles 1.5-3 X 1-1.5 mm, lanceolate, some- times lobed, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adaxially; ped- icels 1-2.2 cm long, pilose, with eglandular, patent hairs 1–1.8 mm long and short (< mm) glan- dular and eglandular hairs. Sepals 3-5 X 1.5-2 mm, mucronulate (with mucro 0.1—0.2 mm long), with scarious margins 0.1 mm wide, with eglan- dular hairs 1-2 mm long and some shorter (< 0.5 mm) eglandular and glandular ones on the abaxial side, glabrous on the adaxial side. Petals 3.5—4.5 X 2-3 mm, emarginate (with notch 1 mm deep), with short claw, bright purple. Stamens 10, both whorls bearing anthers; filaments 3—4.5 mm long, lanceolate, glabrous except for a few ciliae on the proximal half; anthers 0.4—0.6 X 0.2-0.3 mm, pur- ple. Gynoecium ca. 5 mm long; stigma purple. Fruit 9-12 mm long; mericarps 1.4-1.5 X 1-1.1 mm, smooth, without longitudinal rib, not covering the seed completely, glabrous on most of the surface, densely ciliate at the base; rostrum 7-10.5 mm long, with a narrowed apex 1–1.5 mm, pilose (with erect-patent eglandular hairs 0.1-0.3 mm long); stigmatic remains ca. 1-2 mm long, with 5 hairy lobes. Seeds 1.6-1.7 X 0.9-1 mm, brownish, the hilum У;-У, as long as the perimeter. Chromosome number: 2n — 26. Figure 5. Distribution (Fig. 6). Northern, middle, and southwestern Europe; introduced in the northeast- ern United States and New Zealand (North I.); cul- tivated fields and dry places near villages, between 0 an Phenology. Flowering May-August. Representative specimens examined. NEW . Хо rth L, Colenso, 3924475, 17?4'E, 1821, Anonymous s.r K). BELG IUM. Liège, Rocherath, vallée du Төріне hs h, 560 m, 4 N, 6°18'Е, Fabri 857 (BR); Semois 9°53'N, 42457Е, Vinck 422 (BR); pr. Tintomoje, route 1, кы: Johomoje, 50°51'N, 5?28'E, Wilezek 1127 (K). DENMARK. Fn[?|termglatte, 1938, ми s.n. (C). GERMANY. SW of Saxony, 51%20'N, 12°25'Е, 1882, Anonymous s.n. (K); Same nsammlung des Ha Md gischen Staatsinstituts für angewandte pues 5333'N, 10%0'E, Bredemann & Nieser 50 (K). UNITED KING- DOM. England: Little Sark, Channel a 49°20'N, = Volume 85, Number 4 1998 Aedo et al. 605 Geranium ^W 2%; Кт, CASTILLO 97 Figure 5. Geranium aequale. —a. Hab EN ul it. —b. Leaf. —c. Peduncle. —d. Sepal. —e. Petal. —f. Stamen. —g. Fruit and sepals. —h. Mericarp. (Based on Airy [m & Nelmes s.n. (MA-71231).) 606 Annals of the Missouri Botanical Garden AC) Ps Аена m ін Figure 6. .S.A. and New Zealanc 2°22'W, Ballard & Gollon 228 (К); Andover, Hampshire, D 5'N. A orfolk, Buxton, 5224. (Ko: W Gloucester, 34, а А Coates, m. Cirenc diet: 100 m, Airy Shaw & Nelmes 45 (K, MA d ер ‘ester, 34, Avonmounth Docks, 51?29'N, 2?41' W, y 1933, Sandwith s.n. wi West Norfolk, Appleton, но N. 0*31'W, Hubbard 9243 (K). U.S. Wellesley, 27 May 1[?]48, . Massachusetts: Cummings s.n. (NY). ersey: Tom's River, 3 July m s.n. (NY); Morris Co., above Брата. M ackenz NY). New York: Long Island, Hewlett, Bickn ell phos (NY); Tompkins Co., lawn of East Roberts и Brock- port, Hammond 8256c (NY). P nia: Lancaster, July 1894, Bitner s.n. Oh ерее и Williamson s.n. (NY); Delaware River, N of Easton, Lancaster, 4 July 1890, Small s.n. (NY) Geranium aequale is close to С. molle, from which it is easily distinguished by its smooth, densely ciliate mericarps (those in G. molle are transversely wrinkled and sparsely ciliate at the ase). Moreover, the mericarps of G. aequale do not cover the seed completely, as in G. molle. The seeds of G. aequale have a thick testa, which may com- pensate for the slight protection provided by the mericarp. We were not able to find any intermedi- ates between G. aequale and G. molle. Geranium subg. Robertium exhibits several dif- ісагр orn rnamentation, which relevance. Consequently, we have decided to rec- ognize С. aequale as specifically distinct from С. molle. Geranium aequale cannot be considered a Natural distribution of Geranium aequale, based on herbarium records (also introduced in northeastern 1). variation included within the geographic range of G. molle, because it has a very different and smaller distribution area. 2. Geranium molle L., Sp. Pl. 682. 1753. TYPE: tab. 15 fig. 3-3a in Vaill, Bot. Paris. 1727 (lectotype, designated by Carolin, 1965: 332- 333). Geranium villosum Теп., Fl. o > is 1811-1815, om. illeg., non Mill. (1768 Шит pyrenaicum subsp. villosum (Ten.) Nyman, pen sp. Fl. Eur. 138. 8. Geranium molle "E villosum (Ten.) A. kid m molle v . Fl. bol Ed. 1: 371. 1913. TYPE: Italy. Pollino, "Tenore s.n. (lectotype, here des- ignated, МАР, the right-hand specimen; photocopy!). Geranium molle var. parvulum Теп., Syll. Pl. Fl. Neapol. 334. 1831. Geranium ied [c] parvulum (Теп.) Graebn., in Asch. & Graebn., Syn. Mitteleur. Fl. 7: 52. 1913. TYPE: Italy. “Calabria: Monteleone,” Te- nore s.n. (lectotype, here designated, NAP, the upper right specimen; photocopy!). Geranium villosum var. villosissimum Ten., Syll. Pl. Fl. apol. 334. 1831. Geranium molle var. arenarium A. Terrace. . Malpighia 4: 202. 1890, пот. illeg. TYPE: Italy. Monteleone, Tenore s.n, (lectotype, here designated, NAP, the middle specimen; photocopy!) Geranium molle var. album Picard, Mém. Soc. Agric. Bou- logne-sur-Mer 1: 129. 1837. Geranium molle [1] al- bum (Picard) 7. іп Asch. & Graebn., Syn. Mit- Muro Fl. 7: 52. 1913. TYPE: France. *Manchecourt, no authentic 5. located; synonymy ас- сы: to Каш 912). Geranium abortivum De Not. ex Ces., Bibliot. Ital. Giorn. 1: 838. in molle var. abortivum (De Not. ex Ces.) Nyman, Consp. Fl. Eur. 138. 1878 n — Volume 85, Number 4 Aedo et al. Geranium 607 TYPE: Italy. Prope Terranova in Sicilia, И, Bal- samo s.n. (lectotype, here designa ed, К. Maly, Verh. K.K. Zool.-Bot. Ges. 904. Geranium molle subsp. brutium (Gasp.) Graebn., in nee ch. & Graebn., Syn. Mi кісе Fl. 7: 53. 1913. ТҮРЕ: — Calabriae, Gasparrini s.n. (lectotype, here ISP М). Geranium DEM Ledeb., Fl. Ross. 1: 470. . Ge- ranium molle [b] oe eta poto 1 ch. & "Став :bn., Syn. Mitteleur. Fl. 7: 52. 1913. TYPE: реаты Lenkoran, Hansen s.n. Bou here designated, H!). Geranium stipulare Kunze, Flora 29: 698. 1846. Geranium i illk. & Lan . 1878. stipulare (Kunze) K. K. мај, Verh. К.К. Zoo Wien 54: 229. . Geranium molle [B] stipulare (Kunze) күге pig еч & Сга + . Mit leur. Fl. 7: 52. 19 13. Geranium molle subsp. stipu- lare (Көше Holmboe, ic Mus. ge 13: [6]. . TYPE: Spain. In are isthmi Gaditani co- piose, Kunze 2” "on Жы! K!; iso- lectotypes, ВМ!, W!). Geranium molle var. silio. рен Boiss., Fl. Orient. 1: anium pP TS dy (Boiss.) Posp., enl. 2: 30. 1898. Geranium molle mt 5. in Hegi, Ill. Fl. . TYPE: Greece. Pro- e Mazeica, Arcadia, уч 3404 (lectotype, here spen d, G!). pu molle var. m Schur, Verh. Naturf. Vereins Brünn 15: 161. 1877. Geranium molle f. ann un (Sehun) 5 in E Ill. Fl. Mitt.-Eur. Ed. 1702. Geranium s [I] annuum EU Gra Bs in gd & Graebn., Syn. Mitteleur. Fl. 7: 52. 1913. TYPE: Austria. Auf Казепр леп i im Au- garten, Oktober, November 1872," Schur s.n. (no au- thentic ne located; synonymy according to Knuth, Geranium 1. var. г subperenne Schur, Verh. Naturf. Ver- eins Brünn 15: 161. 2 Geranium molle [II] sub- perenne (Schur) Grae. n Asch. € Graebn., Syn. 913. Geranium molle f sub- perenne (Schur) 2 жені in Hegi, Ill. Fl. Mitt.-Eur. Ed. 1, 4: )2. 1924. TYPE: Czech экеу ез “Bei Brünn die gewóhnliche Form, Mai-Jun o authentic material located; icem осон to Knuth, 1912). e Borbás ex Vau in W. D. , 1: 454. 1891. TYPE. Austria. Innsbruck, Hall, m Murr s.n. (lectotype, here designated, W!, the left-hand flowering speci- > Ф 3 T NE Ф о = © = < т Ф 2 Mt Geranium molle var. caespitosum N. Terracc., Nuov. Giorn. Bot 1907. Cadet molle [b] caespitosum (М. Terrace.) Graebn., in Asch. Graebn., Syn. Mitteleur. Fl. 7: 52. 1913. TYPE: It- q. Sin- jar Mt., Gulley of Dair Aasy, Al-Shehbaz, AL Maya h & Sharifi s.n. (holotype, BUH-30568 not seen). Annual herb to 45 cm tall; stem erect or decum- bent, usually branched from the base, pilose, with long eglandular hairs 1-1.7 mm long and short glandular and eglandular hairs < 0.5 mm long. Basal leaves іп a persistent rosette; lamina 0.9-4 x 0.9-5.2 cm, divided for 0.5—0.75 of its length, pilose, with eglandular appressed hairs; segments 7-9, 5 mm wide at the base, usually 3(—4)- lobed at apex; lower cauline leaves alternate; pet- ioles to 14 сш long, with patent, long eglandular hairs 1-1.5 mm long and short glandular and eglandular hairs « 0.5 mm long; stipules 6-9 x mm, ovate-lanceolate, sometimes lobed, pi- T with eglandular hairs on abaxial surface, gla- brous adaxially. Bracts 2-3 X 1.3-1.5 mm, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adaxially; peduncles 0.5-8 cm long, pilose, with eglandular patent hairs 1-1.8 mm long and short (< 0.5 mm) glandular and eglan- dular hairs; bracteoles 1-2 X 0.5-1.2 mm, lance- olate, sometimes lobed, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adaxially; pedicels 0.5-1.5 cm long, pilose, with eglandular patent hairs 1-1.8 mm long and short (< 0.5 mm) glandular and eglandular hairs. Sepals (1-)2.5-5.5(-6) X (0.9-)1.2-2.1(-2.5) mm, mu- cronulate (with mucro 0.1—0.2 mm long), with scar- ious margins 0.1—0.2 mm wide, with eglandular hairs 1-1.5 mm long and some shorter (< 0.5 mm) eglandular and glandular hairs on the abaxial side, glabrous on the adaxial side. Petals (3—)4.5—8.5 (-10.5) х (1.5-)2-5(-7) mm, emarginate (with notch 1-2.5 mm deep), with short claw, bright pur- ple. Stamens 10, both whorls bearing anthers; fil- aments 4—5 mm long, lanceolate, glabrous except for few cilia on the proximal half; anthers 0.7-1.5 X 0.3-0.5 mm, purple. Gynoecium 5-6 mm long; stigma purple. Fruit 8-14 mm long; mericarps 1.8— 2.] X 1.2-1.4 mm, transversely wrinkled, without longitudinal rib, covering the seed completely, gla- brous on the surface, with a few ciliae at the base; rostrum 6-11 mm long, with a narrowed apex 1–3 mm, pilose (with erect-patent eglandular hairs ca. 0.3 mm long); stigmatic remains ca. 1-2 mm long, with 5 pilose lobes. Seeds 1.4-1.8 X 1-1.2 mm, brownish, the hilum 1/6 as long as the perimeter. romosom n = 13; 2n = 26. Figure 7. 2. illustrations. Cavanilles (1787: tab. 83 g. 3); Reichenbach (1841—1842: tab. 191); Ross- ca (1952: pl. 34); Tokarski (1972: 66, pl. 22). e number Distribution (Fig. 8). Africa and Macaronesia, Australia, New Zealand, Europe; to western Asia and to the Indian subcontinent, North America, South America, subantarctic and north-central Pa- Annals o 608 f the Missouri Botanical Garden =, 4 ING ~ y ) TN i Mm q Ру LIS ж ALAS > Pa ma ы», текте x = y^ == vé ny > с ME cy 44. EA ye == -——— CASTILLO '97 Figure 7. Geranium molle. —a. Habit. —b. Leaf. —c. Peduncle. —d. Flower. —e. Sepal. —f. Petal. —g. Stamen. —h. Fruit and sepals. —i. Mericarp. (Based on Rigual s.n. (MA-371877).) Volume 85, Number 4 1998 Aedo et al. 609 Geranium Л A а Figure 8. Distribution of Geranium molle, based on herbarium records. cific Islands; also reported from Japan (Knuth, 1912: 58); cultivated and waste places, open hab- itats, dunes, dry grassland, or roadsides, between 0 and 1400 m. Additional maps. Meusel et al. (1978: 263); Hultén & Fries (1986: 635, map 1269). Phenology. Flowering February-August (Octo- ber-January in Southern Hemisphere). 2 specimens examined. ALGERIA. Ar- rew, Mun n. (К); Cherchell, 36°36'N, 2*12'E, 8 Маг. 1962, Charo s.n. (G). EGYPT. Quarry Bourg el Arab, Simpson 3277 (K); 21. East Sinai, Mar. 1950, Mei- nertzhagen s.n. (BM). LIBYA. Bughailan, Guichard 317 (BM). MOROCCO. 2 ua SW of El Jadida, roadside by coast, 10 m, Lambert 474 (BM). SOUTH AFRICA. Cape Peninsula, Salter 6855 (K). TUNISIA. Ain Sebaa to Jeb- bara beach, E of Tabarka, Davis 57759 (BM AUSTRALIA. Lord Howe .4 km N of Pine ~ 10 m, Johnson & Rodd 1210 (K). New South Wales: 6 mi. E of Scone towards Moonan Flat, 150 m, Coveny 3304 f F. W. Ф ш б 5 = Sy: ES un - а} alia: Р 5 (К). NEW ZEALAND. NEUE Mypres ban ec, m, 2. f. & Scheffer 624 (K). AUSTRIA. Tyrol sept., nnsbruck, Kerner s.n. (К). BELGIUM. Alleur, 50%41'N, 5°30’E, 20 June 1895, Polchet s.n. (BR). CYPRUS. Agios Philon, nr. Rizohanpzo, Davis 2209 (K). CZECH REPUB- LIC. Doksy, Bohemia bor., 270 m, 5 July 1980, Hadinec et al. s.n. (G). DENMARK. Brabrand W of Aarhus, 22 May 1968, Nielsen s.n. (MA-204309). FINLAND. Alandia, Lemland, in insula Slátholm, 9 June 1907, Florstróm s.n. (6. FRANCE. Corcega, Aitony valley, near Evisa, 21 Apr. 928, Edwards s.n. (BM); Andresselles, Pas de Calais, 1. 3816 (МА). GEORGIA. Abchasia, Suchumi, Mar- kovicz 2942 (С). GERMANY. Bad.-Wiirtt., Mindelsee, Un- terberger ат See, 6 May 1980, Anonymous s.n. (B) GREECE. Cholomondos mountains, Chalchidiki penin- 2. 5 May 1931, Chick 29a (K); Corfu, Feb. 1862, Mill n. (К). INDIA. Chamba, 32°N, 76°E, Clarke 23581 (К). IRA N. 15 km from Masiri to Basht, 700 m, Davis & ee hari 55856 (K). ISRAEL. Ghor, 4 km 5 Dair Alla, Masri Triangular, Al-Eisawi 1724 (BM). ITALY. oe bea R mous s.n. (K) coastal road from Alghero to Fertillia, Dunford Sicilia, Palermo, Todaro 1122 (K). A Post s.n. (К). NETHERLANDS. ажет nd Blea: m ger 1391 (L). NORWAY. Tisler, en af Slvolóe y 1865, Collet s.n. (K). PO D. . Fuckel s.n. (BM). PORTUGAL. Matosinhos, Pew osade, 13 Mar. 1977, Alexandre & Serra s.n. (MA-484473); Madeira. 5 S. Seb. up the valley, Lowe 139 (K); Azores, Fayal, 1 Godman s.n. (K). ROMANIA. Dobrogea, N von . ат Bahndamm, 30 m, 14 May 1976, Krendl s.n. ( 5 SIA. North Caucasus, Dagestan, Tarkitan и Machackaln, 42°53'N, 47°33'Е, Kuvaev 1-7 (LE); Russia Northwest: Lublinskoj gub., 60°19'N, 29%55'E, 18 Мау 1907, Gane- sin s.n. (W). SAUDI ARABIA. Jabal Aja nr. Hail off the Annals of the Missouri Botanical Garden Jaharah road, Collenette 8585 (K). SPAIN. Mon, San Mar- tín de Oscos, d m, 29TPH6993, Aedo et al. A226 (МА); Baleares, Alcu . 1899, Gandoger s.n. (W); Canary ~ 846, Воигреаи s.n. (K). SW DEN. Smolandia, 57%0'N, 15?0'E, М. J. Andersson sn. (MA-99920). SWITZERLAND. Lausanne, Aug. 1879, vrat & Barbey s.n. (K). TURKEY. A1 Edirne, 8 km W of Edirne, 100 m, Davis stage (K). UKRAINE. Krym, Gor- naja Grjada тежи m. Aija S. . Sevastopolja, 44°28'N, 34^8'E, Срејеу et p 366 (L E). UNITED KING- DOM. Scotland: Braemar, Mean 411 (K). YUGOSLA- VIA. Serbia: Jablanica, 2 . 1914, Maly s.n. (K). CANADA. British Col um ibia: Cadborough Bay, Van couver Island, Whiting & Stewart 431 (K). U.S.A. Arkan- sas: Marion C State Park, ca. 15 mi. SE Yell- ville, D'Arcy & Porter 4426 (MO). California: Mendocino Co., 2 mi. N Point Arena between Coast Highway and beac ~ 15 m, True 4222 (CAS). Georgia: Осопее Co., 3 mi. ESE of Farmington, Duncan 29037 (NY). Hawaii: Ha- кү PAR Paauhau, 1500 m, Hosaka 2203 (BISH). Idaho: Idaho i ringa and Or 4 по, са. 5 Henderson & Cholewa 6486 (NY). George Co., Beltsville, Hill 16730 (NY). Massachusetts: Nantucket Island, Siasconset, Mackeever 991 (BM). souri: McDonald Co., 1.5 mi. S of Goodam, T23N, R32W S 19, roadside park on W side of Hwy. 71, Summers 2944 (MO). New Jersey: Camden, T June 1876, Parker s.n. (NY). New York: Long Island, Hunters Point, 19 May 1880, Brown s.n. (NY). North Carolina: Avery Co., en- trance to Grandfather Mountain on US 221, 1400 m, Bouf- ford & Wood 23898 (MO). Ohio: на те 4 June 1886 er s.n. (NY). Oklahoma: 4.3 1 E of Eagletown, 2. 8336 (С). Oregon: Baker Co. pt campground along Pine Creek, below North Pine Creek, between Half- way ge Homestead, Cronquist 6543 (NY). Pennsylva- nia: Chester Co., Brookfield, July 1817, Canby s.n. : Blount Co., Mt. Nebo, Walland, Thomas 71177 (NY). Utah: Utah Co., Provo Bench near Pleasan View, Utah Ditchbank, 1500 m, Harrison 7543 (MO). Vir- ginia: Isle of Wight Co., Fort Boykin, 5 May 1991, Grimm s.n. (BM). Washington: King Co., 1 mi. N Snoqualmie Falls on the road between Falls City and 1. Cas- cade Mountains, 250 m, Anderson 2138 (MO). West Vir- ginia: Pendleton Co., Pike Gap rd., 0.5 mi. SE jet. of St. Rt. 28 at Circleville, Cusick 28120 (NY). Wisconsin: Olga, 10 June 1905, Engberg s.n. (NY). ARGENTINA. Entre Ríos: Gualeguay, Estancia San Ambrosio, 33?10'S, 59?14'W, Burkart 18088 (NY). CHILE. Concepción: región de Bíobio, La Posada, 17 km S Conc v са, 3625175, 73°3'W, и (W). FALKLAND IS. Byron Sound, West Falkland L, 7 Feb. 1912, q s.n. (K). URUGUAY. c. Cerro, 50 m, Herter 1312b (MO D 2 TE = = Ф Ж a e б Geranium molle is а very distinctive species, eas- ily identified by its transversely wrinkled mericarp, glabrous on the surface and sparsely ciliate at the base. It grows naturally almost throughout Europe, in the circum-Mediterranean area, Macaronesia, and central and western Asia. The eastern limit in Europe is not well known, because of the scarcity of herbarium material. In Asia, this species reaches the western Himalayas to 76°Е in India. It has been introduced in many temperate areas of North Amer- ica, South America, southern Africa, and Australia. Geranium molle shares some derived character states with G. aequale, such as glabrous stamen fil- aments, an abruptly tapered fruit rostrum, and gla- brous mericarp surface, supporting their close phy- logenetic relationship. А number of minor morphological variants of С. molle have been recognized in the literature, of which the most notable seems to be G. brutium. According to Webb and Ferguson (1968: 198), this is an eastern Mediterranean species similar to G. molle but frequently perennial (G. molle was con- sidered annual by these authors), with the lower- most inflorescence leaves shorter than the peduncle or slightly exceeding it (as opposed to considerably longer in G. molle) and with petals 6-11 mm long (3-7 mm long in С. molle). Davis (1967: 460) and Persson (1987: 547) considered G. brutium as a subspecies of G. molle, whereas Pignatti (1982: 10) preferred specific rank. All of these authors used the above-mentioned characters to recognize G. brutium. All the studied material identified (by several au- thors) as G. brutium is clearly annual, though var- iable in stature and robustness. This was already pointed out by Boissier (1867: 880, 882). Geranium molle subsp. sinjaricum, described by Al-Shehbaz et al. (1983: 353) from Iraq, was said to be peren- nial. Unfortunately, we were not able to examine ny original material on which this name was based, but all specimens studied from Iraq were annuals. The ratio between the length of the lowest inflorescence leaf and the peduncle varies consid- erably in G. molle, but independently of plant ro- bustness and petal length. This suggests that G. brutium has been distinguished from G. molle only because of its longer petals. However, some plants with long petals can be found throughout the geo- graphic range of G. molle, even in populations with mainly short petals. According to Yeo (pers. comm.) and our own observations, the earliest flowers usu- ally exhibit the longest petals, with petal length di- minishing as the season progresses. Moreover, the type specimen of G. brutium has petals 7.8 mm long, not far from the G. molle values. Consequent- ly, the forms with long petals are here not accorded taxonomic recognition. Sometimes it is possible to find depauperate plants of G. molle (up to 5 cm high), fertile but with the leaves not fully developed. Some such speci- mens [e.g., Florstróm s.n. (K) from Finland, or Davis & Bokhari 55856 (K) from Iran] have leaves with undivided lobes. However, no other character state is associated with this size reduction, which sug- Volume 85, Number 4 1998 Aedo et al. 611 Geranium gests that this form also does not deserve taxonomic recognition. Ger , Syst. Nat. Ed. 10: 1144. 1759 а Gaara parviflorum Curtis, Fl. Londin. 4(43): tab. 46. 1782, nom. illeg. Geranium parviflorum Chevall., Fl. Gén. Env. Paris Ed. 1, 2: 802. 1828, nom. illeg. TYPE: “Habitat in Anglia, Galia” [according to L., Sp. Pl. Ed. 2: 957. 1763] (lectotype, here designated, LINN-858.86; microfiche!). Geranium humile Cav., Diss. 4: 202, tab. 83 fig. 2. 1787. Steud., No- l. Geranium parviflorum var. humile (Cav) Gel Fl. Gén. Env. Paris Ed. 1, 2: TYPE: locality and collector unknown, ecimen annotated in Cavanilles's hand as “humile” (lectotype here designated, MA-475736!). Geranium dubium Chaix, Pl. Vap. 23. 1785. TYPE: “Circa pagos frequens” [in agro vapincense, aix s.n. (no 2. materia located; syn- on ae according to Knuth, 1 Geranium delicatulum uss., in Ten., Fl. Napol. 5: XII, 84. 1835-1836. 777 pusillum subsp. delicatulum (Ten. & Guss.) A. Terrace., Malphigia 4: 212. 1890. TYPE: Italy. Verne Tenore s.n. (lecto- type, о designated, NAP; photocopy!). Geranium pusillum var. elatum ran et les endroits her- beux," Picard s.n. (no authentic material located; synonymy 2. to Knuth, 19 Geranium circinatum Kanitz ["circinatnm" T Linnaea 32: no authentic material located; synonymy ac- cording B Knuth, 1912). Geranium pusillum ee кыана Schur, Enum. Transsilv. 137. 1866. Geranium pusillum f. ria Schur) E. in Hegi, Ill. Fl. Mitt.-Eur. Ed. 1, Geranium p | айап ea Cricket, in Asch. & Gra n. Mitteleu TYPE: e re Sándboden am Zibinflose bei Neppendorf. Jul.,” Schur s.n. (no AW Do located; synonymy according to Knuth, 1 Geranium peulopuilum Schur, Oesterr. Bot. Z. 18: 317. 186 anium pusillum [1 pseudopusillum (Schur) Graebn., Sn Asch. & Graebn., Syn. Mitteleur. Fl. 7: 42. 1913. TYPE: Austria. *Auf unbebauten steinig- sandigen Aeckern und Plätzen, unweit des Land- e Ж .n. (no e material located; a according to Knuth, 1912). Geranium pusillum var. iliorum Schur, Verh. Naturf. ereins Brünn 15: 163. 1876. Geranium pusillum [1] albiflorum (Schur) Graebn., in Asch. & Graebn., Syn. Mitteleur. Fl. 7: 41. 1913. TYPE: Austria. “In Obstgiirten bei Hermannstadt, eine Schattenform, Mai 1850,” Schur s.n. (no 21. те located; synonymy according to Knuth, 19 Geranium pusillum var. gracillimum ы. Verh. Naturf. Vereins Brünn 15: 162. 1876. Geranium pusillum f. жасалып (Schur) Gams, in Hegi, Ill. Fl. Mitt.-Eur. 704. 1924. к pusillum [1] gracil- limum (Schu r) Bp Asch. & Graebn., Syn. 24 Fl. 7: 42. 3. TYPE: Czech Republic. er змее b Brünn truppweise, Juni 1872, ” Schur no aut Кане Rem located; synonymy according to Knuth, 19 dg cu pusillum var. raja randum Schur, Verh. Naturf. s RR Brünn 15: 1876. Geranium pus- ae El maja randum (Schur) Graebn., in Mitte 4 a Knut Geranium по та var. rigidum Schur, Verh. Naturf. Ver- eins Brünn 15: 163. 18 124 material located; synonymy according to h, 1912). 4 4. Geranium pusillum |2] rigidum (Schur) Graebn., in Asch. raebn., Syn. Mitteleur. Fl. 7: 42. 1913. TYPE: Austria. “Auf steinig-san- digem Boden vor der Favoritenlinie in der Маће des Landgntes bei Wien, Mai 1867," Schur s.n. (no au- thentic losa located; synonymy according to Knuth, 1912). Geranium AnA Hausskn., tingen) Jena 3: 278 T Mitt. Geogr. Ges. (Thü- ). Geranium pusi condensatum Druce, Soc. Exch. Club Brit leles 5: 17. 1917. TYPE: United Kingdom. Englan Haven, Muddiford, Hants., 27 July 1916, ben s.n. (lectotype, here designated, ОХЕ!). Geranium nau var. tenuilobum Sennen, Pl. Espa 1927, no. 6058. 1928, in sched. TYPE: France. Cer- da us Мик и [42°29'N, 1°56’E], 7 July 1927, Sennen s.n. (lectotype, here designated, BC-825290 isolectotypes, BM!, MA-71059!, MA-71060!, D 708641, W!). Annual herb to 50 cm tall; stem erect or decum- bent, usually branched from the base, pilose, with short glandular and eglandular patent hairs (« 0.3 mm long). Basal leaves in a persistent rosette; lam- ina 1.5—3.8 X 1.5—4.8 cm, divided for 0.3—0.75 of its length, pilose, with eglandular, mem hairs; m wide at the base, 3—5-lobed at apex; lower cauline leaves opposite; petioles to 12 cm long, with short (< 0.3 mm) eglandular and glandular patent hairs; stipules 2—4 X 1–1.5 mm, lanceolate, sometimes lobed, pilose with eglandular hairs on abaxial surface, glabrous adaxially. Bracts 2—4 X 1-1.5 mm, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adax- ially; peduncles 0.5-3.2 cm long, pilose, with short < 0.3 mm) glandular and eglandular patent hairs; bracteoles 1.5-2 X 0.5 mm, linear-lanceolate; ped- icels 0.6-1.6 cm long, pilose, with short (< 0.3 mm) glandular and eglandular patent hairs. Sepals 34.5 2 mm, mucronulate (with mucro 0.1 segments 7, — mm long), with scarious margins ca. 0.1 mm wide, 612 Annals of the Missouri Botanical Garden with eglandular hairs ca. 1 mm long and some shorter (< 0.5 mm) eglandular and glandular hairs on the abaxial side, glabrous on the adaxial side. Petals 2-3 X 1-1.5 mm, emarginate (with notch 0.2-0.5 mm deep), with short claw, pale purple. Stamens 10, the inner whorl with filaments 1.2-1.5 mm long, lanceolate, pilose on the abaxial side, cil- iate on the proximal half; anthers 0.3 X 0.2 purple; external whorl with filaments 1 mm long, almost glabrous, lacking anthers. Gynoecium ca. 3 mm long; stigma 2” purple. Fruit 9-11 mm long; -1.9 -1.1 mm, smooth, with a mm, mericarps 1 longitudinal rib, covering the seed completely, pi- lose, with appressed-eglandular hairs to 0.2 mm long, with a few ciliae at the base; rostrum 7-9 mm long, obtuse eglandular and glandular hairs ca. 0.2 mm long); stigmatic remains 0.5-0.7 mm long, with 5 pilose lobes. Seeds 1.7-1.8 X 1-1.1 mm, reddish; hilum x as long as the perimeter. Chromosome number: — 26. Figure 9. Additional illustrations. Curtis (1782: tab. 46) [sub 6. parviflorum]; Reichenbach (1841-1842: tab. 190 fig. 4877); Gams (1924: 1703 fig. 1641); Ross-Craig (1952: pl. 35); Tokarski (1972: 68, pl. 28) at apex, pilose (with erect-patent, Distribution (Fig. 10). Europe to central Asia d the India frica, and 29) and Uruguay (Hetter, 1954); cultivated and waste places, open habitats, rocky slopes, and dry grassland, between 0 and 1900(-2500) m. Additional maps. Meusel et al. (1978: 263); Hultén & Fries (1986: 635, map 1270). Phenology. Flowering March-September (De- cember-January in Southern Hemisphere). MOROCCO. Re- 4?09' W, 1950 m, Representative specimens examined. fugio de Taffert, Atlas Medio, 33°38' М, Aedo 4 NEW ZEALAND. Wellington 174?47'E, 2 Apr. 1941, Healy s.n. (CH pital, Christchurch, 4323075, 172%42'E, (CHR). AFGHANISTAN. Магагі-І Sharif, N Afghanistan, fau- cibus fluvii Balkh supra Aq Kupruk, 700 m, 36%5'N, 66°52'Е, Rechinger 16298 (W). ARMENIA. Gdzi[?]rakaia, Such[?]ch, 9 July 1966, Anonymous s.n. (С). AUSTRIA. Austria к. өлік pr. Wildberg ad Lentiam urbem, 6 Sep. 884, 1. (BC- ege BELARUS. Prov. Minsk, 5 . 29°12’E, 1905, Bordzilowski s.n. M M. pnd 50°59'N, 4%50'E, 5 Sep 1940, Michelis s.n. (BR). BOSNIA AND HERZEGOVINA. a еј Gilliat-Smith 3321 (K). BULGARIA. Planities aciae, pr. Karlovo, 28 May 1975, Petrova s.n. (MA- 209947). CYPRUS. joper o Kotschy 704 (K). CZECH REPUBLIC. Moravia australis, Uhersky Ostroh, 6 June 1949, Podpera s.n. (K). DENMARK. Saltholm, engan 41°18'5, R); Burwood Hos- Healy 70.188 - 5538'N, 12?46'E, Hansen 46 (C). ESTONIA. Wormsoó, Hullo, 6 Aug. 1924, А (О. Е backecc, Nagu, 18 June y FRANCE. o uli да", 1889, Beaudouin s.n. (МА-71055). GERMANY. Bayern, Oberbayern, Griifelfing bei München, auf Brachückern, 28 June 1902, Dihm s.n. (MA-360452). GREECE. Pisoder- 1150 m, Ravi 23906 (K). ). IRELAND. Dublin, July 1903, Meade s.n. (K). ITALY. Aprutium, montis Magellae, valle Orfenta, 41738'N, 14°00’E, 200 m, July 1908, Guadagno s.n. (К). JAMMU-KASHMIR. Kashmir, East India Com- pany 323 Мы KAZAKSTAN. Almatinsk u., рг. Almati, 43°3'N, 76?56'E, Pavlova 32 (LE) NETHERLANDS. Ed d 52°54'N, 5?34'E, 4 July 1972, Slim s.n. (1). NORWAY. Holmsbó. 5954'N, 12?06'E, 29 June 1869, и (К). PAKISTAN. Chitral, Drosh, 35?52'N, 71%58'E, Toppin 9] (K). POLAND. Cracovia, pr. Zabier- zów, ad vicum Modlniczka Mala versus, 27 Apr. 1974, Sztyler & Tacik s.n. (MA-252530). PORTUGAL. Francoso, Sampaio 1909 (MA). ROMANIA. inis distr. Dolj, 120 m, 9 May 1931, D. Сти & M. Círtu s.n. (МА-252531). Brothers 216 (BM); Russia Central, Briansk, Pogar, in valle fluv. sudest prope Markovs k, 52°35'N, 33"15'E, 8 June 1980, Skvortsov s.n. (M); Russia 2 skovsk gub., Opochetskij, st. 27 5793, 28°35'E, Kuznetsova 461 (LE); Russia t, prope urb. Pskow, 5 diis 28°20'E, Andrejew Ino (G). SPAIN. Lérida, Alto Ará Bagergue, 1465 m, 31TCH2736, ч 2273 (MA). SW E. DEN. Gästrikland, Gävle, 60°40'N, 17°10'W, Nannfeldt 17173 (K). SWITZERLAND. Beat eee Kloster, 1160 m, 46743'N, 8°51'Е, 19 May 1920, Buh 12603). SYRIA. Kessab, 35°56'N, (G). TURKEY. Elmoli, 1200 m, Tenen 364 (K). TURK- MENISTAN. Ashabad, 37°58'N, 58?24'E, MT (G). UKRAINE. Kiew, jid Made ug Skvortsov s.n. (M); Krym, Severnoe С о ро ob- erezhe, 45734'N, 32%52'E, Корей 189 (L E). UNIT- ED KINGDOM. England: Aylexford, а Aug. 1902, Gregor s.n. (MA- 170873). UZBEKIS STA қалана. montes Keksuiski khrebet, іп vicinitati pagi Вгіс кие 1900 m, 41937 М, 70°5'E, 11 July 1973, Vasdk s.n. (С). CANADA. Alberta: Banff, 51? i0'N. 115° 1028 (NY). Macoun s.n. (NY). Ma ы 99*57' W, 2 788 (CAN). Ontario: Bruce е Lake, 4 152. Carroll Co., Elk Ranc (N | PUO Santa Cruz Co., Boulder Creek, 200 m Hesse 1118 (CAS). Colorado: St. Lupton, Johnston 555 34'W, Sanson Ida Illinois: Alton, along the track of the Illinois Terminal Railroad, E of Piasa Street, 38°53'N, 90711", 100 m, . 4336 (MO). Indiana: St. Joseph Co., E side Wolverton bog, along Road 23, 7 mi. SW of Sou Раст e den Rese , N Sho ore S , Bellevue Cemetery, N Danville, Cusick 30297 (NY). Maryland: Bladensburg, 38°56'N, 76°56' W, June 1879, Chickering s.n. (NY). Massachusetts: Mar- tha's Vineyard, Edgartown, MacKeever 445 (NY). Michi- Volume 85, Number 4 Aedo et al. 613 1998 Geranium (2, 4% pr Q 0 (У a — 118 we Me c Oy Y di тылы > O SS кү — Figure 9. Geranium pusillum. —a. Habit. —b. Leaf. —c. Peduncle. —d. Flower. —e. Sepal. —f. Petal. —g, h. Stamens. —i. Fruit and sepals. —j. Mericarp. (а-а, i, | based on Anttila s.n. (MA-180451); e-h based on Aedo 2371 (MA).) 614 Annals of the Missouri Botanical Garden ба t. а AS .. P Ы е г е dm | | | | — = q | ж | | | | | Fe NAO A EN "л лаш Ыш | > | N \ ques iere (/ | ` | © > On Y | " | 000 km P | E any ud | Dd | d Figure 10. Distribution of Geranium pusillum, based оп herbarium records. gan: Flint, Clarke s.n. "ab Missouri: Noel, Bush 5727 Y ead Mission, 47%55'N, 114^5'W, Sep. 1899, Blankinship s.n. (NY). Nebraska: Minden, 10 June 1931, Hapeman s.n. (MO). New Jersey: Camden, Parker s.n. (NY). New York: Albany, 16 June 1882, Dud- ley s.n. (CAS). Ohio: Euclid, 41°34'N, 81°33'W, Scair s.n. (NY). Oregon: Wallowa Co., Imnaha canyon, Peck 17528 (NY). "Meng been Allegheny Co., 1869, Porter s.n. (NY). South Dakota: pr. Brookings, 44^18'N, p Rs 26 July 1893, Willians s.n. (CAS). Tennessee: Dav Co., Belle Meade area, Kral 50321 (MO). Utah: Bidolph’s ashingt of Asotin, on bluffs along W side a Snake R., Hichco & Muhlick 21801 (NY). West Virginia: Smy th С ork Holston river, near Broad Ford, 750 m, 20 jue 1892, ЖИШШ si (МО). Wy- oming: Laramie Expt. Farm, 41%18'N, 105%35'W, Nelson 2038 (C S. ARGENTINA. Campo La Susana, 10 km de Peralta, 38°03'N, pres W, 320 m, Huidobro 1160 (NY). Geranium pusillum has only five anther-bearing stamens, which is the best character to identify this species. It has been frequently confused with G. pyrenaicum, a closely related perennial species with ten anther-bearing stamens. ЇЇ is also often confused with G. rotundifolium, another widespread annual species. However, G. rotundifolium, a mem- ber of subgenus Geranium, has typical seed-ejec- tion fruits, reticulate seeds, and ten anther-bearing stamens. Geranium pusillum is indigenous in the Eurasian portion of its range. The eastern limit of G. pusillum in Europe is not well known, because of the scar- city of herbarium material. In Asia, this species reaches the western Himalayas to 75%E in Jammu- Kashmir. Geranium pusillum has been introduced to many ji pie areas of North America, South America, and Australia. A search of 5 Linnaean Herbaria has yielded three sheets Е related to the protologue of G. pusillum: -858.85, LINN-858.86, and 4. is not a suitable choice, be- cause it represents Geranium molle. S-282.19 was annotated only by Linnaeus's son, hence is not rel- evant. Consequently, we prefer to select LINN- 858.86, annotated by Linnaeus himself, as lecto- type. The name С. pusillum has often been attributed to "Burm. f., Spec. Bot. Geran. 1759"; however, that work was published on on 17 a about two months after the Linnaean protologue. 4. Geranium pyrenaicum Burm. f., Spec. Bot. Geran. 27. 1759. TYPE: “Habitat in Pyren- aeis” (authentic specimens in G, lectotype not designated; see discussion below). Perennial herb with short vertical napiform rhi- zome; stem 15-70(-110) cm tall, erect, usually Volume 85, Number 4 1998 Aedo et al. 615 Geranium branched from the base, pilose, with long eglan- dular hairs 1-1.4 mm long and short glandular and eglandular hairs < 0.5 mm long. Basal leaves in a persistent rosette; lamina 2.8—6.2 X 2.5-7.5 ст, divided for 0.5—0.6 of its length, pilose, with eglan- dular, appressed hairs; segments 5-7, 4-8 mm wide at the base, (3-)7-10(-12)-lobed at apex; lower cauline leaves opposite; petioles to 25 cm long, with patent, long eglandular hairs 1-1.5 mm long and short glandular and eglandular hairs < 0.5 mm long; stipules 3-9 X 1–3.5 mm, lanceolate, some- times lobed, pilose with eglandular hairs on abaxial surface, glabrous adaxially. Bracts 4—7 X 1.5-2.5 mm, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adaxially; peduncles 1–3.8 cm long, pilose, with eglandular, patent hairs 1–1.3 mm long and short (« 0.5 mm) glandular and eglandular hairs; bracteoles 2.5-5 X 0.5-0.8 mm, lanceolate, sometimes lobed, pilose with eglandular hairs on abaxial surface and on the margin, gla- brous adaxially; pedicels 1—3 cm long, pilose, with short (< 0.5 mm) glandular and eglandular hairs and sometimes with eglandular patent hairs 0.7— 1.3 mm long. Sepals 3.5-5 X 1.6-2.5 mm, ovate, mucronulate (with mucro 0.2-0.3 mm long), with scarious margins ca. 0.1 mm wide, with short (< 0.5 mm) eglandular and glandular hairs on the ab- axial side, glabrous on the adaxial side. Petals 7— 11 . mm, emarginate (with notch 2-3 mm deep), with short claw, bright purple. Stamens 10, both whorls bearing anthers; filaments 4—5 mm long, lanceolate, pilose on the abaxial side, ciliate on the proximal half; anthers 1–1.2 X 0.5—0.7 mm, purple. Gynoecium 4—4.5 mm long; stigma purple. Fruit 18-20 mm long; mericarps 2.4-3.1 X 1.1- 1.4 mm, smooth, with a longitudinal rib, covering the seed completely, pilose, with appressed-eglan- dular hairs ca. 0.1 mm long or glabrous, not ciliate at the base; rostrum 10-15 mm long, obtuse at apex, pilose (with erect-patent, eglandular and glandular hairs ca. 0.1 mm long); stigmatic remains 1.5-1.8 mm long, with 5 pilose lobes. Seeds 2.2- 2.7 X 1.2-1.4 mm, brownish; hilum У—У, long. Geranium pyrenaicum is the only perennial spe- cies in section Batrachioidea. Geranium pusillum seems to be the closest relative of G. pyrenaicum. The mericarps of G. pyrenaicum are quite similar to those of G. pusillum: smooth, with appressed- eglandular hairs (except for subsp. lusitanicum), but slightly smaller. Moreover, G. pyrenaicum and G. pusillum share one derived character state: the presence of a dorsal, longitudinal rib on the meri- carp. Geranium pyrenaicum grows naturally in almost all of Europe, the Caucasus, Asia Minor, northern Iran, and northwest Africa. It has been introduced in some temperate areas of northeastern North America, and probably in northern Europe. Among the great number of morphological vari- ants formally described under Geranium pyrenai- cum, only that here segregated as subspecies lusi- tanicum seems of some importance. According to F. Jacquemoud (in litt.), two sheets of G. pyrenaicum are kept in the Burman herbarium at G. However, neither can be related unequivo- cally to the protologue because of the lack of dates, locality, and other relevant information. KEv TO THE SUBSPECIES OF GERANIUM PYRENAICUM la. ipe pilose; pedicels with hairs shorter КЕН И ЫК ЕН а. С. pyrenaicum subsp. ругепаїсит lb. 2. les pedicels usually with hairs —1.3 mm lon 4b С. pyrenaicum subsp. lusitanicum 4a. Geranium pyrenaicum subsp. pyrenaicum Geranium perenne Huds., Fl. Angl. Ed. 1: 265. 1762. Little-Chelsea,” collector unknown iw totype, here designated, LINN-858.80; microfic Geranium umbrosum Waldst. Kit., m Icon. Pl. Hung. 2: 131, tab. 124. 1803-1805. Geranium py- renaicum var. umbrosum (Waldst. & Kit.) DC., Prodr. 1: 643. 1824. Geranium pyrenaicum subvar. umbro- т (Waldst. & Kit.) N yman, Consp. Fl. Eur. 137. 1878. Geranium pyrenaicum [b] umbrosum (Waldst. & E ) Graebn., in Asch. & Graebn., Syn. Mitteleur. Fl. 7: 33. 1913. TYPE: Hungary. Kitaibel s.n. (lec- totype, here designated, B Geranium pyrenaicum var dca Picard, Mém. Soc. Agric. bna" а та 1: 131. 1837. ТҮРЕ: France. “а E e, sur iud versant М rempart, du cóté du champs de Foi t à la e St. Gilles, dans les ие и ;а Es sur Шы гетрагі, au- pres du Jardin des Plantes, et dunk le jardin méme, sur le petit rideau qui se trouve au-devant de la salle terial located; synonymy according to Knuth, 1912). Geranium minae шеш Pl. Rar. Sicil. 25. 1846. 514 (lectotype, here designated, PAL!, the left-hand specimen). Geranium pyrenaicum var. subvillosum Schur, Enum. Pl. Transsilv. 1 PE: Romania. "In den Weinbergen be е! Hamm И ғә =ч pa 3 according to Knuth, 1 Geranium pasé + al var. albiflorum Schur, Oesterr. Bot. . 18: 316. rum (Schur) Graebn., teleur. Fl. 7: 33. 1913. TYPE: Austria. * schattigen Rasenplützen іп Wäldern und Dici 616 Annals of the Missouri Botanical Garden den, häufig ist sie im Garten des К.К. Theresianums, wo ich nur diese beobachtet habe," Schur s.n. (no "am ME material located; synonymy according to Knuth, 1912) Geranium ate um var. [B] pilosum Rupr., Mém. Acad. at Pétersbourg, sér. 7, 15: 275. 1869. TYPE: 12. 15-22 бері... . in m. Bai Gora alt. 1140-900 hex.," Owerin s.n. (no authentic material ocated; synonymy according to Knuth, 1912). ahresber. Naturf dens 39: 56. 1885. TYPE: Svitsedand. 1 с (no authentic material cated; synonymy according to Knuth, 1912). Geranium pyrenaicum чинар, australe А. Terracc., Mal- Aider 4: 209. 1890. TYPE: Italy. Palermo alla Piz- Todaro 638 (lec OR hero designated, K!). Dioni pyrenaicum var. ens А. Terracc., Mal- pighia 4: 211. 1890. эана рутепаїсит [b] gra- cilescens (A. Terrace.) Grael jn. in Asch. € С Syn. Mitteleur. Fl. 1913. TYPE: ruzzo e · degli in monti romani, e qui e là sino ? Terracciano s.n. (no authentic material 'ated; senio according to Knuth, 1912). беш pyrenaicum var. patulivillosum Hausskn. & Bornm. ex Bornm., Mitth. Thüring. Bot. Vereins 20: 10. 1904—1905 5: Turkey. Pontus australis, Amasia, mte. Lokman, 9 May 1890, Bornmiiller 1974 (lec муре here designated, JE!). Geranium crinitum N. Terracc., Bull. „Orto Bot. Regia Napoli 3: 122. ©: Italy. Pisterola, ннан s.n. (lectotype, here desmaia NAP; ph otocopy! Geranium pyrenaic um [2] grandiflorum Se shur ex Graebn., & Graebn., Syn. Mitteleur. -E Geranium ipte ит [3] parviflorum Schur ex 17. in Asch. & Graebn., Syn. Mitteleur. Fl. 7: 33. 1913. T ie Romania: Schur s.n. (no suben 4 ted; synonymy according to Knuth, 1912). enm pyrenaicum var. malvaceum nen Bull. urith. Soc. Valais. Sci. Nat. 42: . TYPE: — or St-Pierre, Beauver 2. s.n. (lecto- ere designated, G!, the specimen on the lower pi Geranium pyrenaicum var. сое Sennen, РІ. 927, ү! XE: . (lectotype, Apa designated, in 2. 5!; iso- lec ciotypes, ВМ!, МА-71633!, оаа D Geranium dnte um f. Pull dum Gilm gp n, J. Bot. 70, Suppl.: 6. ер ТҮРЕ: uit: Hab. Mills aa ear Strangeways Research Hospital, Cam- bridge, "Gilmour & dnd s.n. (lectotype, here des- ignated, K!; isolectotype, W!). Geranium fiie um var. turolense Sennen, Diagn. о TYPE: Spain. Teruel, León s.n., РІ. aut no. “9773 (lectotype, here designated, 88765!). Geranium elbursense Gilli, Repert. Spec. Nov. Regni Veg. 46: 4 РЕ: Iran. Demawend, alm ober Rehne, 2640 m, 22 July 1936, Gilli s.n. (lectotype, here designated, W!). Stem (15-)25-50(-70) cm tall. Pedicels 1-3 cm long, pilose, with glandular and eglandular hairs shorter than 0.1 mm, usually without long eglandul B. hairs. Mericarps 2.4-3.1 mm long, pilose. Chromo- some number: n = 13; 2n = 26. Figure 11a-g. Ad- ditional illustrations. Cavanilles (1787: tab. 79 fig. 2); Reichenbach (1841-1842: tab. 192); Ross-Craig (1952: pl. 33); Tokarski (1972: 69, pl. 30). Distribution (Fig. 12). Europe, North Africa, the Caucasus, western Asia, and North America; also reported from Chile (Marticorena & Quezada, 1985: 47); waste places, n ди and forest margins, between О and 2 Additional maps. Meusel et al. (1978: s а & Fries (1986: 634, map 1268). Phenology. Flowering May-September. Representative specimens — MOROCCO. 71 m 5 Marrakech, m w Oukaimeden, 2520 m 31%12'N, 7°50'W, hich et al 9004 (BM). ALBANIA. M. Parmu, Alpes alb. sept., distr. Scutari, Baldaci 257 (BM). ARMENIA. Pr. oz. Sevan, 40%33'N, 44^57' E, Avetisian et al. s.n. (MA-252462). AUSTRIA. In ratis agri Vindobonensis, Kerner s.n. (K). BELGIUM. Caestert, 23 May 1950, Bakhuizen s.n. (К). BULGARIA. Rila pr. Samokov, 42°8' №, 23°33'Е, Petrova & Kozuharov 838 (MA). CZECH REPUBLIC. Moravia SW, Dacice, pago 10%13'E, Nielsen «с Pedersen 429 (MA). FINLAND. Aloa, 1878, Hollmén s.n. (MA-71628). FRANCE. Aveyron, St. Paul des-Fonts, cultivé de gr. orig. d'Igny (Seine-et-O.), May 1905, Coste s.n. (BC-825302). GERMANY. Bavaria, pr. Bamberg, 380 m, May 14, Harz s s.n. (BM). GREECE. О km NW Drama, Macedonia, 1000 m, Stainton 7704 (К). ш Erdely, Brafsó. 8 May 1898, Kuel s.n. (BC-12504). IRAN. Fao project Camp, 1520 m, 36°0'N, 53°0'E, Fishwick 8 (K). pag eae Dublin, Gam- ble 20055 (K). ITALY. Luca ania, M. Stace & Cotton s.n. (BM); M die Mountains, mt. Soro, 1710 m, 37%56'N, 14°41'Е, Ak- eroyd et al. 3744 ria Palermo, ca. 45 km S of Cefalù, SE of Rifugia Marini, 1600 m, Davis & Sutton 63866 (BM); in silvaticis о и Madonie, 1882, Citarda s.n. (K); m. Pizzuta, 1300 m, 7 June 1907, Lacaita s.n. (BM). NETHERLANDS. Culemborg, 51%57'N, 5?14'E, 5 June 1938, Van Soest s.n. (1). NORWAY. Akershus amt, Wer- enskiold 15298 (C). POLAND. Dittmannsdorf, pr. Walden- rg in Silesia, 50°46'N, 16?17' E, 400 m, 17 June 1884, Felsmann s.n. (BC-12492). ROMANIA. Transylvania, Schur 756 (K). RUSSIA. North Caucasus, Terek prov., 43°28'N, 44^11'E, 8 June 1911, Busch s.n. (W). SPAIN. Potes, bajada hacia Sotres, Castroviejo | ~ al. 4131EV (MA). SWEDEN. Uppsala, 5955'N, 17°38'Е, N. J. An- dersson s.n. (MA- ). ІТ Bern, Thring Н965.34 (К). SYRIA. Сїй el Hajar, June 1822, Ehrenberg s.n. (K). TURKEY. A4 Ilgaz Daglari, Karasu Valley, Gul- телег, S of Kastamonu, 1780 m, Cheese 1751 (К). UKRAINE. Krym, Bisetka Vetrov, 15 km NNE of Yalta, 1250 m, Chater 140 (BM). UNITED KINGDOM. Eng- land: Bakewell, NW Derbyshire, Bailey 277 (K). YU- GOSLAVIA. Montenegro: Crna Gora, Zablzak, lower Volume 85, Number 4 Aedo et al. 617 Geranium CASTILLO '97 сны A 1mm | «X «s A TN A ИЕКТЕН at Mr ty 1M УУ Ујед Figure 11. a-g. Geranium pyrenaicum subsp. pyrenaicum. —a. Нађи. —b. Pedicel. —c. Flower. —d. Sepal. —e. Stamen. —f. Fruit and sepals. —g. Mericarp. h, i. Geranium pyrenaicum subsp. lusitanicum. —h. Pedicel. —i. Meri- carp. (a, b, d-g based on Aedo 2084 (МА); c based on Aedo et al. CN340 (MA); h, i based on Losa s.n. (MA-71620).) 618 Annals of the Missouri Botanical Garden x ~ 1 ја \ C UY ) A T. М) ~ 4 - \ ^ SN o CN AAA CN INN PES < “21 ) aN = К] pw 4 J ) ON с, x 5 YN S 1 = pa nm V CONS ;| | ) | ~ 74 ~ lo 3 ajo Y | ‘ > < a. 3 ) 60° | ud N N I | =, e ЖЖ - (ен и $ ra Иоча Ci, е. @ 71.) AE ж” 30° ES \ \ | u > b а? y Y | » Uu o ) b A 0? $ VOU AY ме > N | Neg EN Y 4 ^] | t 4; 4 S Cx -- 2 Y Figure 12. Distribution of Geranium pyrenaicum subsp. ds. on herbarium records slopes of Savin Kuk, above Crno Jezero, 1750 m, Gardner 2441 (BM). CANADA. Ontario: Grey Со. W of Beaverdale, 44°24'N, 80238! W, Viem 1666 (CAN). Québec: Sil- 1 , 46'46'N, 71°15'W, Bernard ; w York: Bronx Co., New York Botanical Garden, Bronx Park. Gilly 426 (NY). 4b. Geranium pyrenaicum subsp. lusitanicum (Samp.) S. Ortiz, Anales Jard. Bot. Madrid 47: 244. 1990. Geranium pyrenaicum raga lusitan- . 1911. Gera- Samp., Fl. Port. Ed. 2: 331. 1947. TYPE: tugal. Castro Laboreiro, July 1903, Sampaio s.n. (lectotype, here designated, PO-4606!). Geranium pyrenaicum var. majus Pau ex Merino, Fl. Ga- licia 1: 283. 1905. TYPE: Spain. ad Rivas Pequeñas, Merino 305 (lectotype, here designated, LOU!). Stem (23-)40-75(-110) cm tall. Pedicels 1-2.5 mm long, pilose, with glandular and eglandular hairs shorter than 0.1 mm and usually also with pyrenaicum (dots) and subsp. lusitanicum (squares), based long eglandular hairs (0.5-)0.7-1.3 mm. Mericarps .6 mm long, glabrous. Chromosome number: 2n = 26. Figure 11h, i. Additional illustration. Or- tiz (1989: 242 fig. 1). Distribution (Fig. 12). Spain and Portugal; waste places, field margins, and forest margins be- tween 0 and 2000 m. Additional map. Ortiz (1989: 244 fig. 3). Phenology. Flowering June—July. они specimens examined. PORTUGAL. Вга- a, Paramio, Zeibe, 28 July 1971, Dias Pereira. s.n. (LISD Caldas do Gerez, June 1887, Murray s.n. (K); Cam- pea, 3 Aug. 1961, Rozeira et al. s.n. (PO-15435); Castelo de Vide, ИА 1908, Sampaio s.n. (PO-4609); Fafe, Armil, 8 May 1943, Barros s.n. (PO-30979); Macedo de Caval- 1943, Rozeira & Castro s.n. (MA-191371); 1906, Tavares s.n. (PO-4607); Meda, dentro do Castelo, 5 Oct. 1976, Costa s.n. (PO- 45102); Mogadouro a Azinhoso, Qta. da Nogueira, Barbosa & García 6754 (1151); Moimenta da Beira, Vila da Rua, beira da Ribeira, 26 June 1977, Costa s.n. (РО-28436); Póvoa de Lanhoso, Quintal de S. Geus, May 1908, Sam- paio s.n. (PO-4608); Sabugal, Quadrazais, junto ao rio Cóa na reserva, 2 June 1988, Ladero & Lousa s.n. (1151); Serra da Estrela, Fonte Paulo Martius, Rozeira s.n. (PO-15434); Volume 85, Number 4 1998 Aedo et al. 619 Geranium Tabuaço, Vale da Figueira, Barbosa «с García 8083 (LISI); Vinhais, Tuizelo, 28 May 1972, Marcos & Almeida s.n. (LISD; Castro Laboreiro, "ien acao, June 1903, Sampaio s.n. (PO-4606). SPAIN. Aule El Arenal, 1300 m, 12 Aug. 1986, Lucefio & Vargas s.n. (МА-407065); Vizcaya, Gor- bea, Guinea 4020 (MA); Burgos, La Revilla, Carazo, 1400 m, 30TVM6949, 6 July 1979, dis & Susana s.n. (M mayor, 17 May 1 944, 1972. Segura Zub (MA); Palen encia, De hesa de Montejo, valle de Tosande, О m, Monasterio 1341 (MA); Salamanca, Castañar de las Нола, Linares del Río Frío, 15 June 1974, Castro- viejo s.n. (МА-324277); Segovia, Altos del puerto de la Quesera, 1340 m, 24 June 1973, Gómez et al. s.n. (MA- 323972); Soria, Саћбп del río Lobos, 7 June 1980, Buades s.n. (MA-571625); Álava, Ichine, mt. Gorbea, Guinea 4019 (MA); Zamora, Rivadelago, barranco del Fornillo, June 1945, Losa s.n. (МА-71620). Geranium pyrenaicum subsp. lusitanicum com- prises plants from northwestern Spain and Portugal with glabrous mericarps and long, patent, eglan- dular hairs on the peduncles and pedicels. In this area, all specimens examined exhibit these fea- tures. However, eastward, in the eastern portion of the Cantabrian range and in the Iberian range, where the two subspecies occur sympatrically, the indumentum of the peduncles and pedicels is var- iable. This was pointed out by Ortiz (1989: 243), who proposed subspecific rank for these entities. Du Rietz (1930), for allopatric taxa merging mor- phologically where they come i (1986) described subspecies as “taxa believed to be allopatrically evolved from a common ancestor, not саи different to be recognized as spe- sulting from primary speciation at an " This could be the case in С. pyren- idering its geographical distribution and weak morphological divergence. We consider occasional specimens with glabrous mericarps from within the geographic range of C. pyrenaicum subsp. pyrenaicum to be discordant el- ements representing minor variation of no taxonom- ic relevance. However, if such individuals were confirmed as more common, the status of subsp. lusitanicum would have to be reconsidered. Geranium sect. Divaricata Rouy, in Rouy & Fou- caud, Fl. France 4: 88. 1897. TYPE: G. di- varicatum Ehrh. Perennial or annual herbs; stems to 60 cm long, with simple or bifurcate monopodial branching, leafy, erect, with patent eglandular and glandular hairs. Basal leaves in persistent or deciduous ro- settes; venation actinodromous, basal, perfect, mar- ginal; lamina pentagonal in outline, palmatifid, con- colorous, hairy; segments 5-7, rhombic, 7-15- lobed at apex. Cauline leaves opposite or alternate; stipules lanceolate, sometimes lobed, papery, brown, pilose. Cymules solitary, arising from aerial stems; bracts lanceolate, papery, brown; peduncles present, with patent glandular and eglandular hairs; bracteoles linear-lanceolate, sometimes lobed, pa- pery, brown; pedicels 2 per cymule, = ascending and often curved upward after anthesis, subequal, with patent glandular and eglandular hairs; pedun- cle and pedicel together very often exceeding the subtending leaf. Sepals ovate, erect-patent at an- thesis and erect in fruit, briefly mucronulate, mar- ginally scarious; abaxial surface with eglandular or glandular hairs; adaxial surface glabrous, with a subapical tuft of hairs. Petals erect-patent, * ob- ovate, emarginate, without claw, without nectar pas- sages, ciliate at base, with sessile glands on the adaxial surface, = purple, without a dark basal spot. Stamens 10, both whorls bearing anthers; fil- aments lanceolate, expanded at base, persistent in fruit, with a conspicuous midvein, sometimes cili- e, pale pink; pollen pink-purple. Fruit of the carpel-projection-type, with discharge mechanism inoperative; mericarps transversely wrinkled, sometimes cristate, covering the seed completely, without a basal beak and with- out a callus, hairy; rostrum reduced, obtuse at apex; stigmatic remains with 5 pilose lobes. Seeds ob- ovoid, smooth, brownish, the hilum 1/6 as long as the perimeter. Cotyledons laterally incised. Chro- mosome number: 2n — 20?, 28 Distribution. Southwestern Europe to central Asia and China. The most distinctive feature of Geranium sect. Divaricata is the inoperative fruit-discharge mech- anism. Other characters states, such as the incised margin of the cotyledons and the obovate seed out- line, also support this section as a natural entity. KEv TO THE SPECIES OF GERANIUM sect. DIVARICATA la. Plants perennial; mericarps with a longitudinal crest 5. G. albanu lb. Plants annual; mericarps without a longitudinal crest 6. G. divaricatum 5. Geranium albanum M. Bieb., Fl. Taur.-Cau- cas. 2: 137. 1808. TYPE: Georgia. “Ex Alban- ia Њепса, Wakiri" [Bakir district, pr. Signakh, 41°37'N, 45°54’E], Steven s.n. (lectotype, here designated, LE, photo!). Annals of the Missouri Botanical Garden Geranium cristatum Steven, Mém. Soc. Imp. Naturalistes Moscou 4: 50, tab. 5. 1813. TYPE: Georgia. Ju- charibasch үз na Stevin s.n. (lectotype, here designated, H Perennial herbs with rootstock ca. 6-8 mm diam., branched, with fusiform-swollen roots and remains of stipules and petioles at apex; stem 60 cm tall, erect, usually branched from the base, pilose, with long eglandular hairs 1-2 mm long, and short glandular and eglandular hairs < 0.5 mm long. Basal leaves in a deciduous rosette; lamina 2.3-5.2 X 3—4.5 cm, divided for 0.65—0.75 of its length, pilose, with eglandular, appressed hairs; segments 4—7 mm wide at the base, 7—9-lobed in distal half; petioles to 11 cm long, with patent long eglandular hairs 1-1.5 mm long and short glan- dular and eglandular hairs < 0.5 mm long; cauline leaves opposite; stipules 4—6 X 2—3 mm, pilose with eglandular hairs on abaxial surface, glabrous adaxially. Bracts 3—4 X 1.5-2 mm, lanceolate, sometimes lobed, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adax- ially; peduncles 1—4.8 cm long, pilose, with eglan- dular, patent hairs ca. 1 mm long and short (< 0.5 mm) glandular and eglandular hairs; bracteoles 3- 4 х l mm, pilose with eglandular hairs on abaxial surface and on the margin, glabrous adax- ially; pedicels 1—4 cm long, pilose, with eglandular patent hairs 1-1.5 mm long and short (< 0.5 mm) glandular and eglandular hairs. Sepals 6-9 x 2.5- 3 mm, mucronulate (with mucro 0.3-0.6 mm long), with scarious margins 0.1—0.2 mm wide, with eglandular hairs 1-1.3 mm long and some shorter (< 0.5 mm) eglandular and glandular hairs on the abaxial side, glabrous on the adaxial side. Petals 12-14 х 8-9 mm, emarginate (with notch ca. 1 mm deep), bright purple. Stamen filaments 5-6 mm long, with spreading glandular hairs and a tuft of eglandular hairs at the base on the abaxial side, without ciliae; anthers 1.5-1.6 X 0.8–0.9 mm, pur- ple. Gynoecium 6-7 mm long; stigma pink-purple. Fruit 15-17 mm long; mericarps 3-3.5 mm, with 3—4 transverse ribs and a longitudinal crest, pilose, with appressed-eglandular hairs up to 0.5 mm long, not ciliate at the base; rostrum 9-10 mm long, pilose (with erect-patent, eglandular and glan- dular hairs ca. 0.1 mm long); stigmatic remains 1— 2 mm long, with 5 pilose lobes. Seeds 2,8-3 X 1.8- 1.9 mm; hilum У, as long as the perimeter. Chro- mosome number: n — 14; — 20?. Figure 13. Additional illustrations. Karjagin (1955: 39 tab. 2); Grossheim (1962: 11 tab. 1 fig. 3); Tokarski (1972: 59, pl. 1). > Distribution (Fig. 14). Northern Iran and Cau- casus; meadows, stony places, and forest margins, between 100 and 2600 m. Additional maps. Grossheim (1962, map no. 4); Meusel et al. (1978: 263) Phenology. Flowering June-August. Representative specimens examined. AZERBAIJAN. Lankoran, 38°45'N, 48°50'E, 1836, Hohenacker s.n. (BM, M, W); Elisabethpol, Schuscha, pr. Chan-Kendy, 39°17'N, 46°23'E, June 1900, Fedossejew s.n. (LE). GEORGIA. Georg. Cauc., m. Wilnser, 1838, Kalm s.n. (G); in dumosis circa Telav, Kachetia, 41°55'N, 45?29'E, 20 June 1918, Pastuchov s.n. (K, W); Transcaucasia, Georgia Orient., Steppa Shiraki, m. Schavi-mta, 41°42'N, 46^15'E, 600 m, 8 May 1940, Sachokia s.n. (MA-575069); pr. Sihuscha Georg. Cauc., 1838, Hohenacker s.n. (K, M). IRAN. 59 km S of Shahi, 1500 m, 36°27'N, 52%51'E, Furse 2979 Қ, ХУ); Ardabil-Astara, 1200 m, 38?24'N, 48°52'Е, Bowles Scholarship Bot. Exped. 2312 (K); East Azerbaijan, W side of Hasi Amir Pass, on Russian border, 28 of Ardabil, 1600 m, Grant 16229 (W); Gilan, around the village Damash-E of Rudbar, 1700 m, 36°48'N, 49°23'Е, Ala 17143G (W); Gilan, in collibus 10-20 km W Astara, viam versus Heyran ducentem, 500 m, 38°22'N, 48°38'Е, Rechinger 39902 (В, С, W); Gorgan, 37%00'N, 54?30' E, Sharif 545 (W); Gozlu, 2. Koelz 16221 (ХУ); Guilan, 1. 741 (BM, Mazanderan, Haraz valley, Kare g. 100 m, 36? 11. 52°0'E, Wendelbo 584 (W); Ostan | a eae, 1070 m, 36?29'N, 51?9'E, E Schmid 6636 (G); Ostan 2, Dimelo, samt di versant caspien, peu en dessous de la crête, 2600 m, А Schmid 5986 (G, W); Ostan 2, entre Amol et Siavicheh, 1800 m, 36°28'N, 52°21'E, F Schmid 5853 (G, realis, in dumetis prope Rascht, : Bornmiiller & A. Bornmiiller 6507 (BM, С, К, W); Persia borealis, Elburz, Pole-Zan borealis, Ell Gauba 1453 (B); prov. Talysh et Korabach, m. Kohenaker, 38°29'N, yaks. 1838, Kalm s.n. (G). RUSSIA. North Caucasus, Dagestan, pr. urb. Derbent, 130 m, 42%3'N 48^17' E. Alexcenko 7437 (LE); distr. Kurinskij, pr. st. Div- iczi, 41°35'N, 47245"Е, Alexeenko 7428 (LE). —_~ ® Geranium albanum is a perennial species en- demic to northern Iran and the Caucasus. It has a singular mericarp, with a very thick wall and a well-developed dorsal crest. Features shared with G. divaricatum are the inoperative discharge mech- anism and the ribbed mericarp. The chromosome number in this species is not fully clarified. Warburg (1938: 145) and Van Loon (1984a: 276) have given different numbers, with only that of the first author concording with data for G. divaricatum. 6. ergo "ipsuin Ehrh., Beitr. Naturk. : 164. . Geranium winterlii Roth [“win- terli’ gb 2. Bot. , nom. illeg. TYPE: Hungary. Ehrhart Plant. Select. 69 (lectotype, here designated, M!). Geranium divaricatum var. ambiguum Rochel ex Schult., Oestr. Fl. Ed. 2, 2: 285. 1814. Geranium P e Volume 85, Number 4 Aedo et al. 1998 Geranium CASTILLO '97 НА s б \ \ | \ V | (i) NO NY Lo % \ AN (С | QS : Figure 13. Geranium albanum. —a. Habit. —b. Peduncle. —c. Sepal. —d. Petal. —e. Stamen. —f. Fruit and sepals. —g. Mericarp. (а-е based оп Kalm s.n. (С); f. р based on Schmid 6636 (G).) [B] ambiguum (Rochel ex Schult.) Graebn.. in Asch. Annual herb 20—50 ст tall; stem erect, usually & Graebn., Syn. Mitteleur. Fl. 7: 51. 1913. TYPE: Cultivated, Rochel 292 (lectotype, here designated, branched from the base, pilose, with long eglan- М!; isolectotype, W!). dular hairs 1-3 mm long and short glandular and 622 Annals of the Missouri Botanical Garden | к= C SP œ> Ne EN N | =. Em A Et PS . % ( EL. m N e AN ЈАТ ЧУ № 4 NC CMS vu N % 5 ) B > e = ^ L x / NM a • e " MES Wa / __ 40 p ae | ( PI ^“ s; А, А 4 e. | e | ж Ж МА “. | Б = ай pe e a de . m e \ 3 | e ы Figure 14. Distribution of Geranium albanum (longitude = 50°E; latitude = 40°N). Dots correspond to herbarium ) records, and squares indicate literature records from Grossheim (1962). eglandular hairs < 0.5 mm long. Basal leaves in a X persistent rosette; lamina 2.5-7 X 3.1–7.9 ст, divided for 0.65-0.85 of its length, pilose, with glandular and eglandular, appressed hairs; seg- ments 4-9 mm wide at the base, 7-12(-15)-lobed in distal half; petioles to 15 cm long, with patent long eglandular hairs 1-2.8 mm long and short glandular and eglandular hairs « 0.5 mm long; cauline leaves alternate; stipules 4—7 X 1-2 mm, pilose with glandular and eglandular hairs on ab- axial surface, glabrous adaxially. Bracts 4-5 X 1.5- 2 mm, linear-lanceolate, sometimes lobed, pilose with glandular and eglandular hairs on abaxial sur- face and on the margin, glabrous adaxially; pedun- cles 0.6—3.5 cm long, pilose, with eglandular patent hairs 1-3.5 mm long and short (< 0.5 mm) glan- dular and eglandular hairs; bracteoles 3-4 X 0. l mm, pilose with glandular and eglandular hairs on abaxial surface and on the margin, glabrous adaxially; pedicels 1–2.8 cm long, pilose, with eglandular patent hairs 1–2.5 mm long and short (< 0.5 mm) glandular and eglandular hairs. Sepals 4—4.5 X 1.8-2 mm, mucronulate (with mucro ca. 1 mm long), with scarious margins ca. 0.1 mm wide, with short (< 0.8 mm) eglandular and glandular hairs on the abaxial side, glabrous on the adaxial side. Petals 4.5-6.5 X 2.5—3 mm, emarginate (with notch ca. 1 mm deep), bright purple. Stamen fila- ments 1-2.5 mm long, pilose on the abaxial side, ciliate on the proximal half; anthers 0.4—0.6 X 0.3— 0.4 mm, purple. Gynoecium 3-3.5 mm long; stigma purple. Fruit 7-11 mm long; mericarps 2.8-3.5 X 2-1.8 mm, with 3—4 transverse ribs, without a lon- gitudinal rib or crest, pilose, with appressed-eglan- dular hairs up to 0.5 mm long, not ciliate at the base; long, pilose (with erect-patent, eglandular and glandular hairs ca. 0.1 mm long); stigmatic remains 0.5-1 mm long, with 5 pilose lobes. Seeds 2.4-2.6 X 1.3-1.4 mm; hilum 1/6 as long as the perimeter. Chromosome number: 2n — 28. Figure 15. Additional illustrations. Rei- chenbach (1841-1842: tab. 188 fig. 4873); Gams (1924: 1695, fig. 1637e—h); Tokarski (1972: 63, pl. 13). Distribution (Fig. 16). Europe to central Asia, China, and the Indian subcontinent; waste places, meadows, stony dry slopes, field margins, and shady wood borders, between 0 and 2100 m rostrum 5-8 Phenology. Flowering March-September. Representative specimens examined. | AFGHANISTAN. Farkhar-Tal, Takhar, 1250 m, 36?34'N, 69*51'E, Podlech 10484 (M). ARMENIA. Ararat, montes Gegamski khrebet, in vicinitate ruinarum pagi Akhkeng, 2100 m, 39?47'N, Volume 85, Number 4 Aedo et al. 623 1998 Geranium 22-2. rele uu AL 7 t CLAN PA al OS 7-2-- т. з. T^ < I P Lal. ed (212, T Pr WA JE CASTILLO '97 ыш 15. Geranium divaricatum. --а. Habit. —b. Leaf. —c. Stipule. —d. Peduncle. —e. Sep g. Stamen. —h. Fruit and sepals. —i. Mericarp. —j, К. Seeds. (a, c-g based on Popov € мине n. E "mo b E on Fritzsche s.n. (BC-825281); h-k based rs Koch 46/348 (MA).) Annals of the Missouri Botanical Garden Figure 16. 44^46'E, 10 July 1975, Vasák s.n. (B). AUSTRIA. Lienz, Tirolis, 46°15'N, 12750'Е, 19 Мау 1869, Gander s.n. (К). BELARUS. Minsk, ki ni, Mozyrskago. 53%54'N, 27°34'E, 1902, Bordzilowski s.n. (LE). BULGARIA. In graminosis ad Sofia, June 1924, Stefanoff s. n. (BM). CHI- NA. у X ; Tian-schan, Ak- -tau, 43% O'N, 84%0'E, 15 May Vve 2. y s.n. (МА-71170). FRANCE. Cerdagne, vallée de Carol à Porta, 1500 m, 13 Aug. 1916, Sennen s.n. (MA- | 2221 GERMANY. Pr. Francfort-sur- l'Oder, 52?20'N, 2'E, 27 June 1847, Buek s.n. (K). “ЕСЕ. Mac та 2. Sandwith 970 (K). HUNG . Pes Cspel, 19 June 1900, Degen s. n. (6). INDIA. Сал, W Ha: 2000 m, 32°N, 76 Кап 2067 > dbi NW Himalaya, A 882 (K). Мама, nr. Karend, W of Kermanshah, 2000 m, 31°28’ 54°5 54 E. Fur rse 1892 (K). IRAQ. Jebel Sinjar, Mosul liwa, 1000 m, Gillet 11134 d. ITALY. Oulx, ad saepes secun- Ferrari & Vallino s.n. (K). JAMMU- KASHMIR. Preslang near Pahlgam, Kashmir, 4371" N, 75°25'Е, Stewart 21676 ia, Wernoje, 43°17'N, caba s.n. (K). KYR- Fergana, distr. Osch. pr. Gule za, 40°21'N, 73°26'Е, 30 May quos Tranzschel s.n. (LE). POLAND. Breslau, 51%5'N, 17?0'E, June 1860, Uechtriz s.n. (W). ROMANIA. 15. Dolj, pr. Timbu- Distribution of Geranium divaricatum, based on herbarium records. resti, 70 m, 9 May 1971, D. Сти & M. Сти s.n. (MA- 252485). RUSSIA. North Caucasus, Dschmagat-Tal nor- dóstlich Teberda, 43°28'N, 41%46'E, Stohr 9 (В); Russia Central, Kursk, Bielgorod, 50%38'N, 36°36'Е, 5 June 1900, Sukaczew s.n. (LE); Russia East, Ве idm. omous SSR, 2” rajon, Userganskaya, 54^N, 56°Е, 7 241 (LE) Russia South, Saratov, Sarepta, 8°31'N, 44°29'Ẹ, 1 June 1894, Becker s.n. (M). SLO- VARIA. In mte. Zobor, Nitriam, 482217, 18°7'Е, 1836, Láng s.n. (K). SPAIN. Granada, Sierra Nevada, loc. du- metes inter Cerro Trevenque et Aquilones de Dilar, 1800 m, Porta & Rigo 104 (K). SWEDEN. Sodermanland, Nacka, 4 Sep. 1915, Vestergren s.n. (W). SWITZERLAND. Cantou des Grisons, Engadine, ruine Steinsberg Ardez, 1490 m, 22 July 1933, Huber s.n. (BC-79174). TADZHIK- : 67°37'Е, 23 May . (LE). TURKEY. Babadagh Dobrud- scha, bei UA. 36°32'N, 29°10'Е, Sintenis 524 (K). TURKMENISTAN. Ashabad, in angustiis Karanky, 37°58'N, 58?24' E, Litwinow 1129 (G з). UKRAINE. Char- Коу pr. Walki, 5020 36°15'E, Lindeman s.n. (W). UZ- BEKISTAN. Pr. Tashkent, 41° 16'N, 69°13'E, Kuschake- witz s.n. (K). The most distinctive feature of this annual spe- cies is the transversely ribbed mericarp, which has thin walls and no longitudinal crest. The eastern limit of Geranium divaricatum in Volume 85, Number 4 1998 Aedo et al. 625 Geranium Europe is not well established, because it is quite difficult to obtain material from Russian herbaria. In Asia, this species reaches the Chinese Xinjian (84^E) and the western Himalayas to 76°E. It is known from Sweden by only one collection, which could be an introduction; the nearest locality is in Germany, almost 1000 km southward. DuBioUs NAMES yo brutium [b] micranthum N. Terracc., Bull. O . Regia Univ. Napoli 3: 122. 1913. TYPE: Italy «Pistercla a Signa sopra Vallesega,” N. > s.n. (no authentic material located). = С. m T ne. Chenevard, Bull. Herb. hs 3: 427. 1903. TYPE: Switzerland. *Crocefisso, Mt n. Giorgia" Chenevard s.n. (no authentic material located). = G. molle? Geranium molle f. candidum Beck, Fl. Nieder-Osterreich. 892. TYPE: Austria. *Auf bebauten und wüs- ten Stellen, unter Buschwerk hie und da um Wien und bis gegen Baden, bei Laxenburg, Hainburg, Melk, Schenkenbrunn, Retz. V-IX,” Beck s.n. (по authentic material located). = G. molle? Geranium Mp var. pM Viv. [* Lager ol Fl. Liby Spec. 39. . TYPE: Libya. *H. in totá Cy renaicá, ud s.n. qu authentic material locat- ed). = G. molle? Gerantum n var. grandiflorum Vis., E Dalmat. 3: 212. 1851, . illeg., non Viv. (1824). TYPE: Italy. *Hab in peut sus et ruderatis circa Zara, Se ben- ico, Traù, Spalato, R hentic material located). = G. mo Geranium moll var. grandiflorum Lojac., Malpighia 20: 194 m. illeg., non Viv. (1824). TYPE: Italy. “In herb idis Pa өн, Herb. Pan.!,” Lojacono s. n. (no authentic material located). — Geranium п. Env. Paris Ed. 1, 2: 802. 1828. TYPE: France. "Commun dans les bois et в décombres, " Cheval- r s.n. (no authentic аа locate molle? Geranium molle subsp. pollinense A. Tenens Malpighia : 198. 1890. Geranium pollinense N. Terracc. ex alle Neviere ed all'A racciano s.n. (no authentic material located). = б. molle? Geranium molle var. o Boenn. ex Rchb., Fl. Germ. Excurs. 778. 1832. agas pusillum f. suaveolens Boca, ex Rchb.) Gams, in Hegi, Ill. Fl. Mitt.-Eur. Ed. 1, 4: 1702. 1924. Caium molle |1] suaveolens n ex Rchb.) Graebn., in Asch. & Gr аен , Syn. Mitteleur. Fl. 7: 52. 1913. TYPE: Ger- man "Auf ebautem Boden, Schutt, an Mauern, Planken" (no authentic material located). = С. mol- le? Geranium molle f. pinguis К. Мају, Verh. К.К. Zool.-Bot. Ges. Wien 54: 229. 1% guis (K. Mal$) Graebn., : Mitteleur. Fl. 7: 52. 1913. e PE: Yugoslavia. “No- vibazar: kljeznicatal zwischen Prijepolje und Jabuka (Weisbach)," Weisbach s.n. (no authentic material lo- cated). = С. molle? Geranium molle var. montanum A. Terrac. ex N. Terracc., Bull. Orto Bot. Regia Univ. Napoli 3: 123. 1913. TYPE: Italy. “Acquanera,” N. 2. s.n. (по authentic material located). = 6. т Lum molle [II] tenuisecta A. Terrace. ех P cond Asch. & Graebn., Syn. Mitteleur. Fl. 7: 52. 1913, TYPE: Italy. "Nei colli sino a che sui monti. . .," Terracciano s.n. (no authentic material 2. . тойе? Geranium molle [I] triviale А. Terracc. ех Graebn., in sc Graebn., Syn. Mitteleur. Fl. 7: 52. 1913. TYPE: Italy. *Porto d'Anzio," A. Terracciano s.n. (no authentic material located). — Geranium punctatum Kanitz, Linnaea 32: 569. 1863, nom. Шег., non Andrews (1799). TYPE: men “Hab- pro G. um- ‚=, = cated). = 6. molle? ан pusillum var. exsertum Peterm., Fl. Lips. Ex- rs. 512. 1838. TYPE: Germany. “... ad praedium Pfaffendod. ad pagos Leutzsch, pons Reudnitz, ad oppidum Delitzsch etc.," Petermann s.n. (no authen- tic material located). = G. pusillum? Geranium pusillum var. luxurians А. Terracc., Malphigia 4: . 1890. TYPE: Italy. “... che ho qui di Carls- baad e dell’ Alpe Gebbo (1165 m) in val Cairasca ss . Terracciano s.n. (no authentic material located). = G. pusillum? бағатын pusillum f. subcalvum Casp. ex Abrom., Fl. Ost- lestpreussen 154. 1898. TYPE: Germany. “West- gea Fl. zw. Kamin u. Obkaser Mühle, Abhang m Mochelsee (Km. 70),” p s.n. (no authentic о located). = G. p Geranium pusillum var. FOU E a Oesterr. Bot. Z Saut.) Graebn., i Fl. 7: 42, 1913. TYPE: Italy. “Bozen: an Weinbergs- mauern in St. Johann, Guntschna, St dalena, bei Caslar,” Sauter s.n. (no authentic material located). = G. pusillum? Geranium неку var. depilatum Sommier & Levier, p. S.-Peterburgsk. Bot. Sada 16: 102. I сей depilatum (Sommier & Levier) Grossh., Grossh. & Schischk., Sched. Herb. Pl. Or. Esse - . 1928. TYPE: Georgia. “Rekom, (Lojka)." Lojka s.n. (no authentic material located) = б. pyrenaicum subsp. pyrenaicum’ Geranium pyrenaicum var. ex Malpighia E Ка 1906. ТҮРЕ: ltaly. * exsicc. pl. Piz . Herb. Pan.!,” Lojacono s.n. (no authentic сени: bosoi. = = С, pyrenaicum subsp. pyrenaicum? Geranium pyrenaicum var. montanum А. Terracc., Mal- pighia 4: 208. 1890. TYPE: Italy. “Monte Pollino all'Afforcata N. Terracciano!, sommità del Coccorello in Abruzzo Cherici! Velino Sanguinetti!,” N. Terrac- ciano s.n. (no authentic material located). — G. pyr- enaicum subsp. p pyrenaicum Geranium pyrenaicum var. mutilum Beck, Fl. Nieder-Os- terreich. 563. 1892. Geranium poe [4] ти- tilum (Beck) Graebn., in Asch. & Graebn., Syn. Mit- teleur. Fl. 7: 33. 1913. TYPE: pase “Bei Hütteldorf, Laxemburg,” Beck s.n. (no authentic ma- terial located). = ] Geranium pyrenaicum nse Bot. Regia Univ. Napoli 3: 122. 1913. TYPE: ‘Telly. “5. Vito urine e nei pressi del Piscone di Pister- 626 Annals of the Missouri Botanical Garden ola," N. Terracciano s.n. (no authentic ешн lo- cated). = С. pyrenaicum subsp. pyrenaicum ExcLUDED NAMES — ber Ehrenb. ex R. Knuth, in Engl., Pflan- IV.129 (Heft 53): 152. 1912, nom. nud., pro зни Geranium calabrum Теп., пот. nud., in sched. (NAP: pho- loco Geranium p €— var. tenuisectum Sennen, Pl. Es- agne E no. 2575 (1916), nom. inge in sched. (BC-825282). Geranium molle subf. abortiva (De Not. ex Ces.) A. Ter- 2. 1890, nom. inval. (see Greu- et al., 1994, Art. 33.5). Geranium me f. albiflorum R. Uechtr., С Schles. Ges. Vaterl. Cult. "60 ): 254. 1883, п пи 4. ub var. caucasicum Regel ex Woronow, in | Busch & Fomin, FI. Qum "Crit. 7(3): 66 1909, nom. nud., pro Geranium т. уаг. diffusum. Ten. ex A. Terracc., Mal- pighia 4: 202. 1890, nom. Geranium molle var. е Ten. e ex КА. Terracc., Malpighia 4: 202. 1890, nom. Geranium molle f. 1. А. Terracc., Malpighia 4: pel 1890, nom. inval. (see Greuter et al., 1994, 33.5) piii molle var. graecum А. Terracc., Malpighia 4: . 1890, nom. nud. Geranium molle var. ia Gasp. ex UNS Consp. Fl. 38. 1878, nom. nud., pro бл. molle var. ove Borbás, Oesterr. Bot. Z. 40: 382. 1890, nom. nud. Geranium molle var. maritimum Lojac., Malpighia 20: 194. 1906, nom. nud., pro syn. Geranium molle var. montanum A. Terracc., Malpighia 4: 2. 1890, nom. Geranium molle be. normale А. Terracc., ра )2. 1890; nom. inval. (see Greuter et a 1994, Art 24.3). Geranium molle f. pygmaea A. Terracc., Malpighia 4: 202. , nom. inval. (see Crue et al., 1994, Art 33.5). Geranium molle f. sepincola A. Terracc., Malpighia 4: a 202. 1890, nom. inval. (see Greuter et al., 1994, Geranium malls f. tenuisecta A. Terracc., Malpighia 4: 99—200, E 1890, nom. inval. TR Greuter et al., 1 Art Geranium solle Г un A. Terracc., 00, 202. Hits nom. inval. ( 1994, Art Geranium molle = pee A. алд ех Gortani & М. Gortani, Fl. Friulana 2: 300. 1906, nom Geranium molle var. typicum Pu Fl. ка Küstenl. 31. 1898, nom. inval. (see Greuter et al., 1994, Malpighia 4: 199— see Greuter et al., . 24.3). Geranium molle f. villosissima A. Terracc., Malpighia 4: )2. 1890, nom. inval. (see Greuter et al., 1994, Art 5). ко molle var. vulcanicum А. Terracc., Malpighia 4: . 1890, nom. nud. ш. multiflorum Lang ex bs ш Oesterr. Bot. Z. 18: 317. 1868, nom. nud., Geranium novum Winterl, denen p Bot. Univ. Hung., fig. 2. 1788, nom. inval. (see Greuter et al., 1994, .6). omuium pseudovillosum Schur E ае зан Enum. Pl. Transsilv. 921. 1866 ro syn. Geranium pusillum var. albiflorum Opiz. m 47. 1852, nom. nu Geranium pusillum var. album Lindm., Bot. Soc. Exch. Club Brit. Isles 7: 766. 1925, nom. nud. Geranium pusillum f. major A. Terracc., Malpighia 4: 212. 1890, nom. inval. (see Greuter et al., 1994, A 33.5). Geranium pusillum f. minor А. Terracc., dee eo 212. 1890, nom. inval. (see Greuter et al., ‚ Art in Авс Geranium pusillum [a] genuinum 1 Gra . 1913, nom. inval. , Syn. Mitteleur. Fl. 7: 4 Gan. E et al., 1994, Art Geranium pusillum subf. humifusa A. Тепас cc., Malpighia 4: 212. 1890, nom. inval. (see Greuter et al., 1994, Geranium pusillum f. humile Bueck ex Prahl, Krit. Fl. Schlesw.-Holst. Ed. 1 2: 37. 1889, nom. nud., pro syn. Geranium pusillum subsp. normale A. Terracc., Malpighia 4: 212. 1890, nom. inval. (see Greuter et al., 1994, ‚2 nim animad f. algeriensis A. Terracc., Malpighia 4: 211. 1890, nom. inval. (see Greuter et al., 1994, Art. 33.5). Geranium pyrenaicum var. diffusum Ten. ex 2 a “4 Malpighia 4: 211. 1890, nom. nud., Geranium m var. grandiflorum Schur, Verh. Na- turf. Vereins Briinn 15: 161. 1876, nom. nud. Geranium pyrenaicum var. heterotrichum а. пот. nud., in sched. (PH!) nn pyrenaicum f. ma A. Terracc., Malpighia 4: 211. 1890, nom. inval. (өсе Greuter et al., 1994, Art. 33.5). Geranium pyrenaicum f. minor А. Terracc., ds 208. 1890, nom. inval. ee Greuter et al., 1994, Art 33.5). Geranium pyrenaicum subsp. normale A. Terracc., Mal- pighia 4: 211. 1890, nom. о aes Coun et al., 1994, A З). Geranium pea var. parviflorum je Verh. Na- urf. Vereins Brünn 15: 161. 1876, n nud. Geranium pyrenaicum f. sicula A. Terrac ~ Malpighia 4: 4: 211. 1890, nom. inval. (see Greuter 4 al, 1994, Art Geranium VA cii var. typicum Woronow, in Kusn., Bus . n, Fl. Cauc. Crit. 3(7): 56. m nom. inval. (see Greuter et al., 1994, Art. 24. Geranium Кем [a] typicum Woronow e ndo in Asch. & Gra ‚ Syn. Mitteleur. Fl. 7: 33. 1913, nom. inval. | и et al., 1994, Art. 24.3). Geranium pyrenaicum var. velutinum Buhse, Aufzühl. Transkauk. 48. 1860, nom. nu Geranium subdivaricatum Schur, Verh. Naturf. Vereins rünn 15: 160. 1877, nom. inval. (see Greuter et al., 1994, Art. 34. Geranium sillosum е albiflorum a РІ. orn no. 2994. 1917, nom. nud., in sched. (BM!, W!). Geranium villosum var. gracile Sonnen. РІ. b ek. no. 2994. 1917, nom. nud., in sched. (BM!, W!). Literature od Aedo, C. 1996. Revision of Geranium subgenus Erodioi- dea Teil Syst. Bot. Monogr. 49: 1-104 Volume 85, Number 4 1998 Aedo et al. Geranium — & F. Muñoz Garmendia. 1996. Some notes on the sectional nomenclature of Geranium (Geraniaceae). Tax- on 45: 104-106. Aldasoro, J. 1. А. Matilla & С. Nicolás. 1981. Effect a . (Copenhagen) 53: 139-145. . Al-Khakani & A.-R. A. Al-May- . New or noteworthy taxa for the flora of Iraq. Candollea 38: 349-358 Alves, M. C. € M. T. Leitáo. 1976. Contribugao para o conhecimento citotaxonomico des s de Rd XIII. Bol. Soc. Brot. ser. 2, 50: 2 Arohonka, T. 1982. Chromosome counts of vasc "m plants of ба island ve in Nauvo, SW Finland. Ann. Univ. O., -12. Baltisberger. M. js Cytological investigations of some Greek plants. Fl. Medit. 1: 157-173. : Miam J. D. & M. Black. 1994. Seeds. Physiology of Development and Germination, ed. 2. Plenum Press, ew pu . Bócher & K. Larsen. 1958. Experimental and c stc 5. on plant species. IV. jv ыы in short-lived herbs. Biol. Skr. 10(2): 1 Boissier, É. 1867. Flora orientalis, Vol. 1. H ak Basel, Genève and Lyon. Britton, N. L. & H. A. Brown. 1913. An Illustrated нав of the Northern шы States and Canada, ed. 2. D ublications, Хау T K. P. 198 9. Chromosomenzahilen von Tor rod zen aus Hessen, 4. Hess. Florist. Briefe 38: 11-14. Carlquist, S. € D. R. Bissing. 1976. Leaf 54 оҒ На- walian geraniums | relation to ecology and taxonomy. Biotropica 8: 248-259. Carolin, R. C. 1965. The genus Geranium L. in the south western Pacific area. Proc. Linn. Soc. New South Wales 87. Monadelphiae classis disserta- tiones decem. Quarta dissertatio botanica. F. A. Didot, Paris. Chatterjee, A. & A. K. Sharma. 1970. ж ege d study in Geraniales. Мис leus (Calcutta) 13: 179-20( Curtis, W. 1782. Flora Londinensis, Vol. 6. W. ms Lon- 1 Чоп. Dahlgren, К. V. 1943. Svedjenüvan (Geranium bohemi- cum) och brandnüvan oo lanuginosum). Svensk Bot. Tidskr. di 127-16 Davis, P. H. 1 . Geranium L. In: P. H. Davis, J. Cullen ¿ J. E. Coode, Flora of Turkey 2: au Univ. Press, E Mri 970. Geranium sect. Tuberosa, revision and evo- Mus. кл Israel J. Bot. 19: 91—113. Dersch, G. 4. Uber einige Chromosomenzühlungen an 2. Blütenpflanzen. Philippia 2: 75—82. Du Rietz, G. E. 1930. The fundamental units of biological taxonomy. Svensk Bot. Tidskr. 24: TE Felsenstein, J. 1985. Confidence limits on phylogenies: n approach using the bootstrap. о 39: 783- 791. Fernández Casas, Ј., S. Pajarón & М. L. Rodríguez Pas- cual. 1978. Números cromosómicos para la flora españ- ola. 60—65. Lagascalia 8: 109-112. Fisnzén, R. & L.-À. Gustavsson. 1983. сосе num- ers in flowering n from the high mountains of Ster- ea Ellas, Greece. Willdenowia 13: 10 1-106. Fritsch, R. M. 1973. IOPB chromosome number reports XLI. Taxon 22: 460—461 Gadella, Th. W. J. & E. Kliphuis. 1966. Chromosome m of flowering plants in the 2” II. Proc. . Ned. Akad. Wetensch. С 69: Gallen N. 1988. Recherche sur ho de la flore or- ophile du Maroc: Étude caryologique et a pode A Inst. Sci. Univ. Mohammed У, Sér 35: Gams, B. iod Geranium L. In: G. Hegi. Illustrierte Flora von RT -Europa 4(3): 1668-1716. J. F. Lehmann unic Gauger, W. 1937. Ergebnisse einer zytologischen Mi эя suchung der Familie der Сегатасеае. I. Planta 26: 529-531. Greuter, W., F. R. Barrie, H. M. Burdet, W. G. Chaloner. — Қ 3 E ‘aoe O: таш = ы z >” n = c r ! r 44) = ВЕР ©. ~ os c 3 т Ф E International Code of erant Nomenclature (Tokyo ode). Regnum Veg. 131: 1-389. Grossheim, А. А. 1962. Flora m ed. 2, Vol. 6. Acad. Sci. к Moscow-Leningr. & J. K. Small. 1907. башына: In: N. L. Britton Н L. M. Underwood, North American Flora, Vol. 25: 3-24. New York Botanical Garden, New York. Heitz, E. 1926. Der Nachweis der Chromosomen. Ver- gleichende nar - ші ihre Se a und Form im Pflanzenreich I G. 1954. iod eum del QR G. Hetter, Hickey, 1 à 973. оа of the architecture of ІТ leaves r. J. Bot. 60: 17-33. Hill, L. M. 1989. IOPB n ee data 1. Int. Organ. Pl. Biosyst. beer 13: 17-19 Hultén, E. & M. Fries. 1986. At los of North European pem Plants, Vol. 2. Koeltz Scientific Books, Кбп- . Malecka, R. Izmaitwow, Н. Wcislo, R. Cza- pik & K. Musial (Editors). 1996. Further studies : У зар пе numbers of diee angiosperms Part XX Acta Biol. Cracov., Ser. 38: 9-27. Karjagin, I. I. 1955. Flora К dta. Vol. 6. AN Azer- АЕТ. SSR Press, Baku. Knuth, R. 1912. Geraniaceae. r, Das Plan- 2. г 129 (Ней 53): 1-631. n ү мека a Ге! Lidén. M. 1986. Synopsis of Fumarioideae (Papaveraceae) with a monograph of the tribe Fumarieae. Opera Bot. 33. -- 4 88: | уе, А. & E. Kjellqvist. 1974. Cytotaxonomy of Spanish plants. IV. ан Caesalpinaceae—Asteraceae. Lagascalia 4: 153-2 & D. Love. ia ion taxonomic ical studies on bo- real plants. II. Some new chromosome numbers of Scandinavian P Ark. Bot. BIA: 1 -22. « 956. Cytotaxonomical conspectus of the Icelandic боса, Acta Horti Gothob. 20: 65-291 . Part 4. Act Nin M. dn өзге 1985. F de la flora vascular de Chile. Gayana, Bot. 42: 1- Meusel, H., E. Jäger, S. Rauschert & E. и 1978. 628 Annals of the Missouri Botanical Garden Vergleichende Chorologie der PEE UE Flora -Karten-, Vol. " >. Fischer, Je Mizianty, M., Z. Mirek & L. m 1983. Chromosome numbers of б, 2, plants (Part 4). Acta Soc. Bot. Poloniae 52: 205-214. Mulligan, G. A. 1959. Chromosome numbers of Canadian weeds. II. Canad. J. Bot. 37: 81—92. Natarajan, G. 1978. IOPB а number reports LXII. Taxon 27: 526— Nieto Feliner, G. & C. re do. 1995. A cladistic analysis of Geranium subgen. Erodioidea (Picard) Yeo (Gerani- vd Bot. J. Linn. Soc. 119: 195-212. iz, S. 1989. Caracterización taxonómica de las pobla- ibérico-occidentales de Geranium pyrenaicum . (Geraniaceae). Anales Jard. Bot. Madrid 47: 42—244. Persson, J. 1987. Geranium L. In: А. Strid, Mountain Flo- ra of Greece 1: 538—549. Cambridge Univ. Press, Cam- bridge, S ndon, New York, New Rochelle, Melbourne Sy Pignatti, S , 1982. Flora d'Italia. Edagricole, Bologna Pólya, L. 1950. Chromosome numbers = ee plants II. Ann. Biol. Univ. Debrecen. Price, К A. & J. D. Palmer. 1993. 2. relation- of the Geraniaceae and Geraniales from rbcL se- Nia comparisons. Ann. Missouri Bot. Gard. 80: –671 Rangaswamy, N. S. & L. Nandakumar. 1985. Correlative studies on seed coat structure, chemical composition, and impermeability in the legume Rhynchosia minima. Bot. Gaz. 146: 50 2 Reichenbach, H. G. L. manicae et - ae, Vol. : йел УСУ S. 1952. Drawings F British Plants, G. Bell & А | Sanderson, M. Ј. 1. Phylogenetic relationships ir North Americ cds L. (Fabaceae). Syst. Bot 4—430. 1841- чем: Icones Florae ай Hofmeister, Leipzi Part VL Semerenko, L. V. 1985. Chromosome numbers in some species of flowering plants of Byelorussian flora. Bot. е (Moscow & Leningrad) 70: 992—994. [In Rus- Shaw, R. J. 1952. A cytotaxonomic study of the genus Geranium in the Wasatch region of Mis and Utah. Madrofio 11: 297—304. Skaliriska, M., J. Jankum € Н. Weisto (Editors). 1976. Further studies in chromosome numbers of Polish an- giosperms. Eleventh contribution. Acta Biol. Cracov., Ser. Bot. 19: ad , E. Pogan & R apik (Editors). 1978. Further studies in chromosome peint of Polish angiosperms. 1 contribution. Acta Biol. Cracov., Ser. Bot. 21: Sato PJ. « S. Blackmore. 1991. Geraniaceae Jn: . Blackmore, The Northwest European s Flora 6: vs ls Elsevier, Amsterdam, London, New ork, and Toky« Strid. A. 1980. Chromosome number reports LXIX. Taxon 29: 709-711. R. Franzén. 1981. IOPB chromosome number reports LXXIII. Taxon 30: 829—842. Swofford, D. L. 1993. PAUP: aid gai Analysis Using Parsimony, version 3.1.1. Computer program distributed d ` Ilinois Natural Voie: Survey, Champaign, Il- dx e G. 1934. Die Bedeutung der Polyploidie für die Verbreitung der Angiospermen, erläutert an den Arten Schleswig-Holsteins, A Ausblicken auf andere Floren- gebiete. Bot. Jahr st. 67: 1–36. Tokarski, M. 1972. 1... and taxonomical anal- ysis of the fruits and seeds of the European and Cau- casian species of the genus Geranium L. Monogr. Bot. 36 Tolivia, D. & J. Tolivia. 1987. Fasga: А new polychromatic method for шша and D^ rential staining of plant tissues. 1 uj й. Van Loon, Ј. С. from Europe, ТЕ The e pe de species. Proc. Kon. Ned. Akad. Wetensch. С 87: 2 7. а Вене rs in Geranium from Euro I. The annual species. Proc. Kon. Ned. Akad. . С 87: 279-296. . 1984c. Hybridization experiments in Geranium. Genetica 65: 167- & A. K. Van Setten. 1982. e oor — Eiir LXXVI. Taxon 31: ———, T. V. J. Gadella & E. e n. е к іп some flowering fou from southern France. Acta Bot. Neerl. 20: 157-166 Warburg, E. F. 1938. Taxonomy and relationship in the Geraniales in the light of iheir cytology. New Phytol. 37: 130- Watrous, L. E. & Q. L. Wheeler. 1981. The outgroup com- parison method of character analysis. Syst. Zool. 30: 1— Webb, p. A. & J. K. Ferguson. 1968. Geranium L. In: 1 Heywood, N. A. Burges, D. M. Moo ш S. M. Walters & D A. Webb, Flora —199. Cambridge Univ. Press, Cam- Widler Kiefer H. & P. F. Yeo. "i Fertility relationships of Geranium Vari tied aes s. Ruberta, Апето des lia, Lu- и and l id üculata. Pl. гей Evol. 155: 283— o, P. F. 1973. The biology and systematics of um sections Anemonifolia Knuth and Ruberta Dum. Bot. J. Linn. Soc. he 285—346. ————. 1984. Fruit-discharge-type in Geranium (Geran- iac ind i use in 1. ation and Ив evolutionary im- plic ations. a J. Linn. Soc. 89: 1-36. —. 1990. Тһе ste ation of канк то ia In: Р. Vorster, Eo 'eedings of the International Geraniaceae js чода Univ. Stellenbosch, Stellenbosch, South Afric in south-west A revision of Geranium L. China. UT. Bot. 49: 123-211 APPENDIX 1 Chromosome numbers of Geranium sects. Batrachioidea and Divaricata Geranium sect. Batrachioidea G. aequale, ЕГ = 26 (Gauger, 1937: 529). у. molle, n = 13 (Warburg, 1938: 142); 2n = 26 ge 1937: 529; Ta 1938: 142; Live € Live, 1956: Alves € Leitáo, 1976: 233; Skaliriska et al., Natarajan, 1978: 529; Májovsky, 1978: 25; "ifs & — 1983: 104; Van Loon, 1984b: 295; Hill, 198 G. pisillan 2n = 26 (G sauger, жш 529; Lóve & Lóve, 1945: ^ólya, 1950: 51; Shaw, 1952: 299; Live & Lóve, pos 209; Gadella & Pe 1973: 460; Májovsky, 1974: 10: Alves & Leitão, 1976: Volume 85, Number 4 Aedo et al. 629 98 Geranium 233; Skaliriska et al., 1976: он Fernández Casas et al, dubium Chaix ........................ 611 1978: 109; Arohonka, 1982: 5; Van Loon, 1984b: 295; elbursense Gilli ........................ 616 Buttler, 1989: 13); 2n — hod (Warburg, 1938: 142). humile Cat- eis is ы МА c 611 G. pyrenaicum subsp naicum, п = 11-12 (Heitz, hybridum Havsekn. ИИ 599, 611 1926: 642, 678; Tischler, 1934: 10); n = 13 (Galland, (еіосашіоп Ledeb. ...................... 607 1988: 144); 2n — 20 (Chatterjee & Plut. 1970: 183); lu, e СһепеуагЧд................. 625 2n = 26 (Gauger, 1937: 530; Májovsky, 1974: 11; Skal- ifiska et al., 1978: 42; Strid, 1980: 710; Strid & Franzén, 1981: p Van Loon & Van Setten, 1982: 591; Van Loon, 1984a: 277; Semerenko, 1985: 993; Galland, 1988: 144; T oe dr 2. 2n - 28? (Warburg, 1938: 151; al., : 159; ташу et а 1983: 208). С. 5... ла и 5. = 26 (Alves & Lei- tão, 1976: 232; Van Loon, 1984a: Geranium sect. Divaricata G. albanum, n — 14 (Warburg, 1938: 145); 2n — 20 (Van а: 276). G. divaricatum, 2n — 28 (Májovsky, 1974: 10; Dersch, 1974: 77; Van Loon, 1984b: 294; Jankun et al., 1996: INDEX TO SCIENTIFIC NAMES ccepted names are in roman type; 2 main entry for dani is in boldface. Synonyms are in itali Ger 4, 595, 596, .. 599, 602, 603 ain р 1. Yeo 94, 596, 597, 600, 603 subg. Geranium .. 594, 595, 596, 597, 599, 600, 614 subg. Robertium (Picard) Rouy ... 594, 595, 596, 597, 599, 600 , 603, sect. Anemonifolia R. Knuth ...... . 597, 599 sect. Batrachioidea W. D. J. Koch 594, 595, 596, 598, 599, , 601, 602, 615 3, 604, sect. Divaricata Rouy ... 594, 595, e. e 598, 599, 600 619 sect. Polyantha Reiche .......... 97, ' 599 sect. Pyrenaica R. Knuth ............... 603 sect. Ruberta Dumort. .......... 594, 597. 599 sect. Trilepha Yeo .......... 594, 595, 597, 599 ect. Шош (Boiss.) Reiche ...... 594, 595 597, 599 abortivum De Not. ex Сез. ................ 606 aculeolatum Oliv. ...................... 596 597, 598, 599, 601, 610 banum M. Bieb. ... 595, 597, 598, 599, 619. 620 bifidum E ex R. Knuth i bohemicum brutium Gasp; [b] dao ҖЫ №. Terrae; с b ика 25 calabrum Ten. ........................ 626 circinatum Kanitz ...................... 611 crinit М. DOG. «cuv CERES dS 616 cristatum eo ae oe ee Ы 620 delicatulum Ten. & Guss. ................. 611 ps (Sommier & Less) Grossh. ........ 625 dissectum L. ......................... 597 597, 598, 599, 619, 620, 624 var. ambiguum Rachel ex Schult. 620 var. tenuisectum Sen [B] ambiguum (Rochel е ex Schult.) Graebn. 626 620—621 599, lusitanicum (Samp.) Samp. ex J. M. M. Lopes ... 618 60 macropetalum (Boiss. Posp. ............... 7 ] A касни IIT 615 Еа КГ" 594, 595, 597, 598, 599, 601, 606, 610, 614 subsp. brutium (Gasp.) Graebn. ........... 607 subsp. normale A. Terrace. .............. 626 subsp polline e А. Terrace. A 625 шет sinjaricum Al-Shehbaz € аай oo. 22252 Bee os 607, 610 4. stipulare (Kunze) Holmboe .......... 607 subsp. ient (Ten) A. Тетасс........... 606 var. abortivum (De Not. ex Ces.) Nyman ..... 606 var. pin Ба о ена а 604 var. album A tee as 64 Kee a ees 606 var, annuum Del a 607 var. arenarium TACO: vico ra 606 var. brutium (Gasp.) K. Maly ............. 607 var. caespitosum М. Terrace. ............. 607 var. caucasicum Regel ex Woronow ......... 626 var. diffusum Ten. ex A. Terrace. .......... 626 var. elatum Ten. ex A. Terrace. ........... 626 var. graecum А. Теттасс................. 626 var. grandiflorum Lange ................ 607 var. grandiflorum Lili ................ 625 var. grandiflorum Vis. ................. 625 var. grandiflorum Му................... 625 var. lucanum Gasp. ex Nyman ............ 626 var. macropetalum Boiss. ............... 607 var 12. ПИКИ o oh bey eee eds 626 var. maritimum Lojac. ................. 626 var. minus Chevall. ................... 625 var. montanum A, Terrac. ех N. Terrace. ..... 625 var. montanum is Terrace. .............. 626 var. parvulum. Tem, cuerda rr as eas 606 var. stipulare (Ku nze) ie AE 607 var. suaveolens Boenn. ex Rchb. ........... 625 var. subperenne 5сһиг.................. 607 var. typicum Posp. .................... 626 ar. villosum (Теп.) Cout. ............... 606 var. vulcanicum A. Terrace. .............. 626 subvar. macropetalum (Boiss. )Gams ........ 607 f albiflorum R. Uechtr. ................ 626 f. annuum (Schur) батз................ 607 candidum Beck .................... 625 f. wlabrate A. Termace.. usus ee куьз 626 f. pinguis К. Malý .................... 625 f. preuschoffü Abrom. .................. 604 f. pygmaea A. Теттасе. ................. 626 f. sepincola A. Тетасс.................. 626 f. stipulare (Kunze) К. Maly ............. 607 f. subperenne (Schur) Gams .............. 607 f. tenuisecta А. Terrace. ................ 626 f. trivialis A. Terrace. .................. 626 villosissima А. Terrace. ............... 626 subf. abortiva De Not. ex Ces.) A. Теггасс. ... 626 [a] triviale A. Terrace. ex Gortani & M. Gortani .. 626 [b] caespitosum (М. Terracc.) Graebn. ....... 607 [b] leiocaulon (Ledeb.) Graebn. ........... 607 uis (K. Мају) Graebn. ............ 625 [B] stipulare (Kunze) Стаеђп.............. 607 630 Annals of the Missouri Botanical Garden [c] parvulum (Ten.) Сгаебп............... 606 subsp. lusitanicum (Samp.) S. Ortiz .... 597, 598, [I] annuum (Schur) Graebn. .............. 607 602, 615, 618, 619 [I] triviale A Horis ex Gortani & subsp. normale А. Terrace. .............. 626 M. Gortani) Graebn. ................. 625 subsp. pyrenaicum Burm. f. ........ 597, 598, ІШІ subpereme (Schur) Graebn. ........... 607 602, 615, 619 [II] tenuisecta A. Terrace. ex Graebn. ....... 625 subsp. villosum Mri Nyman ............ 606 [1] album (Picard) Graebn. .............. 606 var. albiflorum Schur .................. 615 [1] suaveolens (Boenn. ex Rchb.) Graebn. ..... 625 var. depilatum х & Levier .......... 625 ЕР ч ех Schur’ сељанка e ко 626 var. diffusum Ten. ex A. Terrace. .......... 626 novum Winterl ........................ 626 var. gracilescens A. Terrace. ............. 616 оепе s Hol ex Hallier: 6.44545 ore ae 599, 607 var. grandiflorum Schur ................ 626 iflarum Curtis ................... 611, | var. heterotrichum Sennen ............... 626 parviflorum Chevall. .................... var. leiocarpum Guss. ex Lojac. ........... 625 var. humile (Cav.) Chevall. .............. E var. longepedicellatum Sennen ............ 616 perenne Huds. 6x56 59% 5%%55%5:%%%%Ұ 5а 615 var. lusitanicum (Samp.) Samp. ........... 618 pollinense №. Terrace. ех A. Terrace. ......... 625 var. majus Pau ex Merino ............... 618 pratense Le s eues а пик те ee ee 597 var. malvaceum Beauverd ............... 616 pseudopusillum Schur ................... 611 var. minae (Tineo) Nyman ............... 615 pseudovillosum Schur .................... 626 var. montanum А. Terrace. .............. 625 punctatum Kanit$ ...................... 625 var. mutilum Beck .................... 625 pusillum L......... 597, 599, 602, iden 614, 615 var. parviflorum Schur ................. 626 subsp. delicatulum ie bi Guss.) A. Terrace. . Du var. patulivillosum Hausskn. & Bornm. i normale A. Terrace, .............. 626 ex Bomm. «ac Rcg ЕЕ балық Vr CR d 616 var. albiflorum Opiz UN XA ERE E E E 626 var. [B] pilosum Кирг. ................. 616 var. albilorum Schur .................. 611 var. pumilum Рїсагд................... 615 var. album Lindm. .................... 626 var. subvillosum Schur ................. 615 var. axilliflorum Schur ................. 611 var. turolense cal нама ER 616 var. condensatum Druce ................ 611 var. typicum Woronow .................. 626 var. elatum Picard .................... 611 umbrosum (Waldst. & Kit) DC. ........ 615 var. exsertum Peterm. .................. 625 var. Lie Buhse 424.24. енен 626 var. gracillimum ©сһиг................. 611 subvar. umbrosum (Waldst. & Ки.) Nyman .... 615 var. humile (Cav.) Steud. ................ 611 f. algeriensis A. Terrace. ................ 626 var. luxurians А. Terrace. ............... 625 f. maior А. Terrace. ................... 626 var. majus-grandifolium SG ИШЕ ese e ек кж жуз 611 f. minor A. Terracc. ................... 626 var. rigidum ©$сһиг.................... 611 f. pallidum Gilmour & Stearn ............ 616 var. tenuilobum Sennen ................. 611 f. sicula A. Terrace. ................... 626 var. villosum F. Saut. .................. 625 гаса lusitanicum бапр.................. 618 f. axilliflorum (Schur) Gams ............. 611 1] albiflorum (Schur) Graebn. ............ 615 f. gracillimum (Schur) Gams ............. 611 2] grandiflorum Schur ex Graebn. ......... 616 f. humile Bueck ex Prahl ............... 626 3] parviflorum Schur ex Graebn. .......... 616 f. major A. Terrace. ................... 626 4] mutilum (Beck) Graebn. .............. 625 f. minor A. Terrace. ................... 626 a] typicum Graebn. ................... 626 f. rigidum (Schur) Gams ................ 611 b] gracilescens (А ign ) Graebn. ....... 616 f. suaveolens (Boenn. ex Rchb.) Gams ....... 625 b] murense N. Теггасс.................. 625 f. beal Casp. ex s PE 625 b] umbrosum (Waldst. & Kit.) Graebn. ...... 615 subf. humifusa А. Тетасс. .............. 626 rhaeticum Вгйррег...................... 616 «| genuinum Graebn. ................. 626 ric 2 Fisch. & Trautv. .............. 597 ВІ axilliflorum (Schur) Graebn. ........... 611 robertianum [......................... 603 ?] villosum (F. Saut.) Graebn. ............ 626 rotundifolium los: uou SOGAR CR ARRIUS CR 614 1] albiflorum (Schur) Graebn. ............ 6ll stipulare Kunze... 607 1] gracillimum (Schur) Graebn. ........... 611 subdivaricatum Schur ek ye eee 626 2 sciri коега (Sc bun Graebn. ...... б11__зу/хайсит1........................ 600, 602 2] rigidum (Schur) Graebn. ............. 611 sia to Waldst. & Kit. ................. 615 b] ax jn rum (Schur) A deux ак villosum Ten. ......................... 606 B] c (Kanitz) Graebn. ........... 611 f. albiflorum Sennen ................... 626 II] pseudopusillum уе 'hur) uo Ке” 611 var gracile Sennen ................... 626 pyrenaicum Bur , 595, 597, 598, 599, var. villosissimum Ten. ................. 606 us 614, 615,619 шии Roth ......................... 620 subsp. australe A. Terrace. .............. 616 Robertium Picard ...................... 603 REVISIÓN Y ANÁLISIS Fernando О. Zuloaga?, Osvaldo CLADÍSTICO DE Morrone?, Andrea S. Vega? y Liliana M. Giussani? STEINCHISMA (POACEAE: PANICOIDEAE: PANICEAE)! RESUMEN En el presente tratamiento se realiza una revisión de las especies del género Steinchisma Raf. (Poaceae), pertene- ciente a la tribu Paniceae de la subfamilia Panicoideae. Se estudiaron seis especies de este género, incluyendo el análisis exomorfológico, anatómico y cladístico de las mismas. El género Steinchisma se caracteriza por incluir cii perennes, que habitan en lugares abiertos en campos o bordes de ríos y arroyos; las plantas poseen lígulas membraná- anatomía intermedia entre las especies Kranz y non-Kranz. Se discuten, por medio de un análisis cladístico, la monofilia del género y su relación con subgéneros de Panicum, como así también con géneros afines de la tribu Paniceae. Se incluye una clave de las especies analizadas, una descripción anatómica del género, fotografías de antecio superior de dos eapecies, descripciones morfológicas, ilustraciones y mapas de distribución de los diferentes taxones. Se realiza una nueva combinación, S. stenophylla (Hack.) Zuloaga & Morrone ABSTRACT e genus Steinchisma Raf. (Poaceae: Panicoideae: Paniceae) is revised. Six species are treated in this work, in which exomorphological and anatomical characters are analyzed cladistically. Steinchisma c omprises perennial species that grow as or near borders of rivers and streams. The plants possess шы ligules, leaves lanceolate lower palea as long as the upper anthecium, expanded at maturity, lower flower present, with three stamens, or absent: indurate upper с and а basic chromosome number of x = 10. А similar anatomical pattern, intermediate between Kranz and non-Kranz species, was found in all species. A cladistic analysis of Steinchisma was conducted in order to test its monophyly, and жн ст with the subgenera of Panicum. A key to the precios is ee as well as an anatomical description of the genus an photomicrographs of two species; illustrations, and distribution maps are Ее. м all six species of Steinchisma. А new combination is pn Steinchisma stenophylla (Hack.) гри & Morro Steinchisma fue establecido como género por Ra- — antecio superior verrugoso en toda su superficie, finesque en el año 1830, sobre la base de Panicum siendo, de acuerdo a este autor, la mayoría de sus hians Elliott. Nash (1903) distinguió a Steinchisma ^ especies intermedias entre las Kranz y las non- de Panicum por poseer pálea inferior expandida a Kranz, y mencionó que $. cuprea (Hitchc. & Chase) la madure a espiguilla. Hitchcock y Chase W. V. Br. (= Panicum cupreum Hitchc. & Chase) (1910, 1915), Pilger (1931, 1940) y Hsu (1965) pudiera ser una especie C,. Brown transfirió P. de- consideraron especies de Steinchisma dentro dela сїрїепз Nees ex Trin, P. cupreum y P. exiguiflorum sección Laxa (Hitchc. & Chase) Pilg. de Panicum. Griseb. al género Steinchisma. Bouton et al. (1981) Brown (1977) consideró a Steinchisma a nivel ge- estudiaron la vía fotosintética de varias especies de nérico, distinguiéndolo de Panicum por incluir es- Steinchisma (bajo Panicum), y señalaron la afinidad pecies con inflorescencias laxas a contraidas, sin de las mismas con especies de la sección Laxa de espiguillas dispuestas en racimos unilaterales, con Panicum, incluyendo caracteres exomorfológicos y ' Los autores desean expresar su agradecimiento a Vladimiro Dudás por la realización de las ilustraciones y el rmado de los mapas. El trabajo de campo fue posible gracias a subsidios de la National Geographic Society, números 5334-94, 5765-96 y 6042-97. * Instituto de Botánica Darwinion, Labardén 200, Casilla de Correo 22, San Isidro 1642, Argentina. ANN. MISSOURI Bor. GARD. 85: 631-656. 1998. 632 Annals of the Missouri Botanical Garden el námero básico de cromosomas. Renvoize (1982) estableció el género Plagiantha, al cual distinguió de Panicum por poseer espiguillas dispuestas oblí- cuamente en los pedicelos y lemma inferior mem- branácea; no relacionó este género con Steinchis- ma. Gould y Shaw (1983) trataron a Steinchisma a nivel genérico, mencionando como caracteres di- ferenciales la pálea inferior epe y el antecio superior con papilas verrugosas en toda su super- ficie. Brown et al. (1985) produjeron híbridos arti- entre especies de isma, com aa (Döll) Renvoize uo Р. lla Doll) y S. hians (Elliott) Nash (bajo P. hians), con especies de la sección Laxa. Clayton y Renvoize (1986) revalidaron Steinchisma, separándolo de Panicum por la pálea expandida a la madurez de las espiguillas. Renvoize (1987) transfirió Panicum spathellosum a Steinchisma. Zuloaga (1987a) ubicó a Steinchisma como un subgénero de Panicum, caracterizándolo por incluir especies con inflorescencias laxas, espiguillas gla- bras, con gluma inferior 3-nervia, % а Y del largo de la espiguilla, gluma superior y lemma inferior 5-nervia, pálea inferior expandida a la madurez y antecio superior con papilas verrugosas en toda su superficie. Un criterio similar adoptó Webster (1988), quien mencionó que no existen suficientes diferencias para distinguir a Steinchisma como un género independiente, aunque indicó que el aná- lisis de especies sudamericanas podría justificar esta separación. Posteriormente, Zuloaga et al. (1993), al realizar un estudio fenético de las es- pecies americanas de Panicum, trataron a Stein- chisma como subgénero, estableciendo relaciones entre éste y el subgénero tipo. Estos autores in- dicaron que Steinchisma comparte ciertos caracte- res con el subgénero tipo, como mesofilo compacto, células del mesofilo radiadas y células tipo “em- palizada" ausentes en el mesofilo. A la vez, Zuloaga et al. ) también relacionaron a Steinchisma con secciones del subgénero non-Kranz Phanopy- rum (Raf.) Pilg., en particular con aquellas que in- cluyen especies con пйшего básico de cromosomas = 10, una similar nerviación de las brácteas de la espiguilla y rango de variación de isótopos de carbono. Watson y Dallwitz (1992) trataron a Steinchisma con cuatro especies, a nivel de género, señalando no obstante que su posición taxonómica es arbitra- ria, pudiendo incluirse también como un subgénero de Panicum. Caracterizaron al mismo por tener in- florescencias laxas, con pálea inferior endurecida y expandida a la madurez y antecio superior endu- Plagiantha fue separado рог recido. es- tos autores por comprender una especie anual, non- Kranz, con espiguillas comprimidas dorsiventral- mente, lemma inferior biaquillada, 4-nervia y pálea inferior endurecida y expandida a la madurez. En la presente contribución se ha efectuado un análisis exomorfológico e histofoliar de las especies de Steinchisma. Con estos datos, sumados a la in- formación citológica, se llevó a cabo un análisis cladístico, a fin de poner a prueba la monofilia de este taxón, y discutir su ubicación dentro de la tri- bu Paniceae en relación con los subgéneros de Panicum y otros géneros de la tribu. A partir de este análisis cladístico se establecen las relaciones filogenéticas de las especies del género. MATERIALES Y MÉTODOS ESTUDIO HISTOFOLIAR Se obtuvieron cortes transversales y epidermis a la altura del tercio medio de la penúltima lámina de la innovación fértil. Se utilizó material prove- niente de ejemplares de herbario, previamente tra- tado con Contrad 70 (Schmid & Turner, 1977) du- rante 24 a 48 hs a 20°С, o material fresco fijado en FAA. Los cortes transversales se hicieron a mano alzada, previo tratamiento con HF al 5% du- rante 24 hs. Para la obtención de las epidermis, se siguió el método de Metcalfe (1960). Los cortes fue- ron coloreados con azul de metileno al 1% y con safranina al 1% en alcohol 80° o con safranina- alcian blue y montados en gelatina-glicerina. Para la observación de las células clorenquimáticas, se realizaron macerados siguiendo el método de Jef- frey (Sass, 1940). Para la identificación de los cuerpos de sílice y células suberosas, se utilizó respectivamente fenol líquido (Metcalfe, 1960) y Sudán III (Sass, 1940). La determinación de los plástidos de almidón, y su distribución, se realizó mediante unas gotas de so- lución iodo-iodurada (Sass, 1940). Para las des- cripciones histofoliares, se adoptó la terminología propuesta por Ellis (1976, 1979). Las observacio- nes anatómicas fueron hechas con un microscopio fotónico Wild M20 con cámara de dibujo. Las di- secciones fueron estudiadas con un microscopio es- tereoscópico Wild M5 con cámara de dibujo. Las fotomicrografías fueron tomadas con un equipo au- tomático Nikon FXA, con cámara fotográfica DX- DB2 35 mm, y la película utilizada Kodak T-MAX de 100 ASA. Para la obtención de fotomicrografías de epidermis abaxiales de la lemma y pálea del antecio superior, se empleó un microscopio electró- nico de barrido Jeol JSM-25 SII, perteneciente Volume 85, Number 4 1998 Zuloaga et al. g 633 Steinchisma (Poaceae) Tabla 1. Lista de los caracteres y estados de los caracteres usados en el análisis cladístico. 1. Duración: О = anual; 1 = perenne 2. Rizomas largamente јој id os: О = ausentes; 1 = presentes 3. Láminas: О = planas; 1 = filiform 4. Inflorescencias: О = laxas; 1 = contraidas 5. Ramificaciones: О = ы: 1 = no unilaterales 6 distal de las ra 7. Espiguillas те 0 = oblicuas; 1 = no оМіспа 8. Gluma superior: 0 = %—М del largo de la ici 1 9 11. Nerviación de la lemma inferior: O = 12. Flor inferior: O = neutra; 1 = estaminada . Ejes de las ramificaciones: О = espiguillas no agrupadas sobre las ramas; 1 = espiguillas agrupadas en la porción mas = %-И del largo de la espiguilla . Desarrollo de dh pálea inferior a la madurez: 0 = nulo; 1 = expandida apicalmente; 2 = expandida lateralmente 10. Textura de la lemma inferior: О = papirácea; 1 = herbácea 3-nervia; ] — 2-4- inde 2 = 5-11-nervia 13. Textura del antecio superior: О = sin papilas simples; 1 = con papilas simples en el apice; 2 — con papilas simples en toda la superficie 14. Textura del antecio superior: О = sin papilas 2 | — con papilas verrugosas en toda la superficie 15. Color del antecio superior: 0 = castaño; 1 = 16. Androceo de la flor 22 ge de A баған funcionales: 0 = 3; 1 = 2; 2 = 0 17. Células fusoides: О = presentes; 1 = ntes 18. Nümero de células = бота entre oie О = 2-3; 1 = 5-7; 2 = más de 7 19. Mesofilo: 0 = lax = compacto 20. Vaina 2. 0 — sin cloroplastos; 1 — con cloroplastos по especializados; 2 — con cloroplastos espe- cializados 21. Valor promedio °C: 0 = —9 a —12%е; 1 = 22. Митего cromosómico básico: 0 = x = 10: 1 = —20 a —30%o a la Facultad de Odontología (Universidad Nacional de Buenos Aires, Argentina). ESTUDIO EXOMORFOLÓGICO El mismo fue realizado sobre la base de mate- riales pertenecientes a los siguientes herbarios: B, ? Я CEN, СЕРЕС, COL, CORD, CTES, Е, С, СН, IAN, IBGE, К, LIL, LP, LPB, M, MEXU, МО, MY, NY, P, PORT, В, RB, 5, SI, SP, UB, USM, UTMC, US, VEN, W. Dentro del material examinado de cada especie, sólo se citan ejempla- res representativos de cada país; una lista completa de los especímenes numerados estudiados, se en- cuentra ordenada alfabéticamente por coleccionista al final del texto. Con un asterisco (*) se sefialan los ejemplares empleados en el estudio histofoliar y con dos asteriscos (**) aquellos especímenes uti- lizados en el análisis de la lemma y pálea superior: Chase 10847**, Davidse 31539*, Ekman 6064*, 7400*, Harley et al. 19384**, Hitchcock 23356*,**. Joergensen 2439*, León 18579**, Lies- ner & Holst 21288*, Pringle 3449**, Reeder & Reeder 4466**, Steinbach 2654*, Zuloaga et al. 2330*, 3108*, **, 3161*, 3214*, 3244*, 3305*, **, Zuloaga & Deginani 462*, Zuloaga & Morrone 3012*, 3074*, 4660*, 4706* ANÁLISIS CLADÍSTICO Para el análisis filogenético se emplearon 22 ca- racteres exomorfológicos (vegetativos y reproducti- vos), histofoliares y citológicos, los que fueron ana- lizados sobre la base de material de herbario (ver Apéndice 1) y completados con datos bibliográfi- cos. Seis de éstos (9, 11, 13, 16, 18, 20) son ca- racteres multiestados y fueron tratados como no aditivos (no ordenados). La lista de los caracteres y la codificación de sus estados se muestran en la Tabla 1. La Tabla 2 contiene la matriz de datos (especies por caracteres). El análisis cladístico fue llevado a cabo utilizan- do la opción de enumeración implícita (ie*) del programa Hennig86 versión 1.5 (Farris, 1988) y MSWAP+ del programa NONA versión 1.6 (Golo- boff, 1993). Para el análisis de la distribución de los caracteres en los cladogramas, obtenidos a par- tir de las rutinas del Hennig86, y la generación de los mismos, se empleó el programa CLADOS (Міх- on, 1993). NONA fue usado también para calcular el “Bremer support” (Bremer, 1988, 1994). Fue uti- lizado para determinar en los árboles subóptimos cuántos pasos adicionales eran necesarios para co- lapsar un nodo. Taxones del grupo interno. El género Panicum es el centro de un complejo de taxones dentro de Annals of the Missouri Botanical Garden Tabla 2. mórficos, con ? los estados de caracteres no codificables. Matriz de datos utilizada para el análisis de Steinchisma; entre corchetes “[]” se indican binomios poli- Otachyrium a Panicum subg. Panicum nicum secc. Laxa Steinchisma decipiens Steinchisma exiguiflora Steinchisma hians Steinchisma spathellosa Steinchisma stenophylla e 11 11111111222 1 2 3 4 5678901 23456789012 1 0 [01][01]1100200 11000120011 1) [01] o 0 1111012[01]1010101201 1 0 0 0 0111010[01]2010020010 0 0 0 0 1101211 10110120010 1 0 0 1 1111210 00111111110 1 1 0 1 1111110 00111111110 1 0 1 0 1011210 0011111111? 1 0 0 0 1011210 00111111110 1 1 0 0 1111110 10112111110 1 1 1 0 1111110 10110111110 la tribu Paniceae (Hsu, 1965; Clayton & Renvoize, 1986). Panicum posee una amplia distribución con cerca de 500 especies que habitan en regiones tem- pladas a tropicales en ambos hemisferios (Clayton & Renvoize, 1986). Hasta el presente no se han llevado a cabo estudios filogenéticos de Panicum, taxón con presencia de un notable polimorfismo, estando este género dividido en varios subgéneros, o secciones, los que podrían ser considerados gé- neros diferentes (Crims, 1991). Este marcado po- limorfismo podría indicar que Panicum no es mo- nofilético, y posiblemente no podrían ser usados caracteres para definir el género. En el presente análisis se incluyeron, para poner a prueba la mo- nofilia de Steinchisma, subunidades monofiléticas del género Panicum como taxones terminales si- guiendo el criterio de Nixon y Davis (1991). Dichas subunidades corresponden al subgénero Panicum y la sección Laxa del subgénero Phanopyrum, de Panicum, y al género monotípico Plagiantha. Las especies de Steinchisma fueron incluidas como ta- xones terminales a fin de reconocer el patrón filo- genético del géne Zuloaga et al. (1993) realizaron un análisis fe- nético estableciendo las relaciones entre los sub- géneros de Panicum. Estos autores sefialaron que Panicum subgénero Panicum es un grupo homo- géneo claramente separado del resto de los sub- géneros. Este subgénero se asemeja a Steinchisma por caracteres anatómicos del mesofilo y la vaina externa de los haces vasculares. El subgenéro Pan- icum incluye aproximadamente 100 especies, ha- 4. analizado, en el presente estudio, 26 es- Este subgénero posee cinco secciones (Zuloaga 1987a), pero la clasificación infragenéri- no ha sido examinada en este trabajo. ү кы et al. (1985), Zuloaga (1987a), Webster (1988) y Zuloaga et al. (1993) relacionaron asimis- mo a Steinchisma con Panicum sección Laxa por caracteres de la espiguilla, como el largo de la glu- ma inferior, la nerviación de las brácteas y la pre- sencia ocasional en P. laxum Sw. de una pálea in- ferior expandida y papilas verrugosas en la superficie del antecio superior. Zuloaga et al. (1992) revisaron la sección Laxa e incluyeron en la misma 13 especies, que comparten los siguientes caracteres: рша membranácea, espiguillas dis- puestas unilateralmente, con la gluma inferior 1— 3-nervia, gluma superior y lemma inferior 5(-7)- nervias, antecio superior con papilas simples re- gularmente distribuidas en toda su superficie, un número básico de cromosomas x = 10, y por incluir especies non-Kranz, con células fusoides en el me- sofilo. De esta sección se revisaron cinco especies. Plagiantha es exomorfológicamente similar a Steinchisma; ambos géneros poseen la pálea infe- rior expandida a la madurez, papilas verrugosas so- bre la superficie del antecio superior y un número 10. Plagiantha di- fiere por incluir plantas anuales, con espiguillas oblicuas sobre los pedicelos, con la lemma inferior 2-4-петуіа, deprimida en su porción central; es además un género non-Kranz. La única especie de Plagiantha fue examinada en este trabajo. básico de cromosomas de x Taxón del grupo externo. Clayton y Renvoize (1986), en su esquema de relaciones dentro de la tribu Paniceae, presentaron a Otachyrium Nees como el grupo hermano de Steinchisma y Plagiant- ha, por la presencia de la pálea inferior expandida. Consecuentemente, este género es utilizado como grupo externo en el presente análisis, siguiendo el criterio de Nixon y Carpenter (1993). Otachyrium es un grupo monofilético, definido por poseer la gluma inferior y superior subiguales, cortas, Y5—4 del largo de la espiguilla, dejando al descubierto el dorso del antecio superior, siendo este castaño Volume 85, Number 4 1998 Zuloa aga 635 ЛЕ (UE ИЯ oscuro a la madurez (Sendulsky & Soderstrom, RESULTADOS MORFOLOGÍA Y CARACTERES TAXONÓMICOS Hábito. cespitosas a largamente rizomatosas (en el último caso en $. decipiens, S. spathellosa y S. stenophylla), con cañas decumbentes hacia la base a erectas. Los entrenudos son cilíndricos, huecos y glabros, mien- tras que los nudos son comprimidos y glabros. Las lígulas son membranáceas en la base y cortamente ciliadas en el ápice. Las láminas van desde planas a filiformes, en este último caso en S. exiguiflora y S. stenophylla. El género incluye especies perennes, Inflorescencias. Las inflorescencias son laxas a contraidas. Inflorescencias laxas se presentan en S. exiguiflora, S. hians, S. spathellosa y S. stenophylla; las ramificaciones son divergentes del raquis, ha- llándose, en S. exiguiflora y S. hians, los ejes des- nudos en la base, con las espiguillas aproximadas hacia la porción superior. Steinchisma cuprea y S. decipiens poseen inflorescencias contraidas, con ejes cortos aproximados al eje principal. Las espiguillas se encuentran, en todas las es- pecies, dispuestas en pares en ramificaciones no unilaterales. Espiguillas. Las espiguillas son bifloras, dor- siventralmente comprimidas, glabras, con una flor inferior estaminada a neutra y una flor superior per- fecta. La gluma inferior es típicamente 3-nervia, variando en longitud desde % a % del largo de la espiguilla. La gluma superior es 3—5-nervia, obtusa pálea inferior endurecida y expandida a la madu- rez, con las alas escabrosas a cortamente pilosas; sobresale a lo ancho de la espiguilla en S. hians, S. exiguiflora y S. cuprea, mientras que sobresale a lo largo en S. decipiens, S. spathellosa y S. stenophy- lla; la flor inferior puede ser neutra a estaminada. El antecio superior es ovoide, cartilaginoso, pajizo, glabro y cubierto de papilas verrugosas en toda su superficie. Flor. Es importante resaltar la variación pre- sente en el androceo en las diferentes especies de Steinchisma. E] flósculo inferior es neutro en S. cu- prea, S. decipiens, S. exiguiflora y S. hians, y es- taminado en S. spathellosa y S. stenophylla. Por otra parte, el flósculo superior posee androceo con dos estambres en las primeras cuatro especies antes ci- tadas, tres estaminodios en S. spathellosa y tres es- tambres desarrollados en S. stenophylla. En todos los casos se observan dos estilos y dos estigmas plumosos, y dos lodículas truncadas y conduplica- as. Vega (1996) citó la presencia, como carácter úni- co, en S. spathellosa, de flor inferior estaminada y flor superior pistilada, describiendo a la especie como la única diclino-monoica dentro de Panicum. En el resto de las especies de Steinchisma se halla andromonoecia en 8. stenophylla, con flor inferior estaminada y flor superior perfecta, mientras que en los cuatro taxones restantes son monoclinos, con la flor inferior neutra y la superior perfecta, siendo este caso similar al que ocurre en Panicum elep- hantipes Nees ex Trin. (Urbani, 1990). Textura y Ornamentación del Antecio Superior (Fig. 1). hallan papilas verrugosas, dispuestas regularmente, sobre la superficie de la lemma y pálea superior; un carácter similar se presenta en Plagiantha te- nella Renvoize. Papilas similares se han hallado hacia el ápice de la pálea superior en especies del subgénero Panicum (Zuloaga, 1987a, b; Zuloaga & Morrone, 1996) y en ejemplares aislados de Pani- cum laxum, perteneciente a la sección Laxa (Zu- loaga et al., 1992) En todas las especies del subgénero se ANATOMÍA FOLIAR Caracteres histofoliares en corte transversal (Fig. . Transcorte: lámina en transcorte en forma de “V” abierta, con un ángulo menor o mayor de 90° entre ambas láminas; zonas costales e intercostales adaxiales y abaxiales manifiestas, con aspecto mo- niliforme; zonas costales opuestas a los haces vas- culares de primer y segundo orden, de 114-156 um de espesor; zonas intercostales no asociadas a haces vasculares, de 52-109 рт de espesor. Haz vascular medio: formado por un haz vascular de primer orden solitario (en S. decipiens, S. hians, S. exiguiflora y S. stenophylla), asociado a dos ha- ces vasculares de segundo orden, en S. cuprea, o bien asociado, en 5. spathellosa, a dos haces vas- culares de primer y segundo orden con una pro- yección abaxial y adaxial formada por extensiones de células parenquimáticas de mayor diámetro que el de las células de la vaina parenquimática. Distribución de los haces vasculares: haces vas- culares de primer y segundo orden equidistantes de ambas epidermis o levemente desplazados hacia la epidermis abaxial; haces vasculares de primer orden contiguos separados por (1-)3-5 haces vas- culares de segundo orden. Estructura de los haces vasculares: haces vascu- 636 Annals of the Missouri Botanical Garden igura 1. 2 MEB de antecios superiores de especies E d hisma. US). —A. talle Bote 3449, e de la Webs con D. Steinchisma hians (C je 10847, US). —( —D. Detalle de la lemma, con papilas verrugosas. lares de primer orden trabados, de contorno elíp- tico, metaxilema formado por dos vasos de contorno poligonal a circular y diámetro mayor que el de las células de la vaina parenquimática con las que se encuentra en contacto; laguna protoxilemática pre- sente; vaina mestomática contínua; vaina paren- quimática discontínua, interrumpida por el escle- rénquima hacia la cara abaxial; haces vasculares de segundo orden trabados, de contorno elíptico o circular, con floema y xilema distinguibles. Vainas de los haces vasculares: haces vasculares de primer y segundo orden rodeados por la vaina mestomática y la vaina parenquimática; vaina mes- tomática contínua, células de paredes engrosadas y lumen pequeño, las paredes radiales derechas y las tangenciales arqueadas; vaina mestomática de los p yapilas verrugosas. — A, B. Steinchisma cuprea e A, con papilas verrugosas. C . Porción media de la pálea y d de la lemma, con papilas verrugosas. haces vasculares de primer orden formada por 18— 26 células, 10—13 parenquimática de los haces vasculares de primer orden formada por 9-11 células, 6–7 en los de se- en los de segundo orden; vaina gundo orden, con cloroplastos de posición centrí- peta Esclerénquima: pobremente desarrollado, de po- sición subepidérmica, discontínuo, formando gru- pos densos de fibras asociados a los haces vascu- lares y al margen de la lámina. Mesofilo: células clorenquimáticas nodulares, de disposición compacta y radiada alrededor de los haces vasculares; 5-7 células clorenquimáticas (52-156 jm) entre la ‘es. vasculares contiguos; “arm cells,” células fusoides y células distintivas Kranz ausentes. Volume 85, Number 4 1998 Zuloaga et al g Х 637 Steinchisma (Poaceae) 50 um Anatomía foliar en corte transversal y epidermis de especies de Steinchisma. A, B. Steinchisma hians Figura 2. (Zuloaga 3108, SI) segundo orden al. E, F. Steinchisma con haces vasculares de primer y segundo orden (Zuloaga Células epidérmicas adaxiales: células bulifor- mes presentes en las zonas intercostales, en forma de abanico, en grupos de 4—6 células, siendo la célula central de mayor tamaño que las laterales, no asociadas a parénquima incoloro; epidermis for- mada por células papilosas, con aguijones, ganchos y micropelos bicelulares; macropelos presentes, con un cojín basal de células epidérmicas sobree- . —E. Transcorte de una 3161, SI). —F. Epidermis abaxial (Zuloaga 3074, SI). levadas (en S. cuprea y S. exiguiflora), o ausentes en las restantes especies. Células epidérmicas abaxiales: epidermis abaxial formada por células de paredes arqueadas, no pa- pilosas, con ganchos, aguijones y micropelos bice- lulares; macropelos y células buliformes ausentes. Epidermis abaxial en vista paradermal (Fig. 638 Annals of the Missouri Botanical Garden Tabla 3. Números cromosómicos en especies del género Steinchisma. úmero cromosómico Referencias bibliográficas Bouton et al., P ohl € Davidse, 1971 en etiqueta del ejemplar Reeder & Reeder 4466 Dubcovsky & Zuloaga, 1991 1981; Dubcovsky & Zuloaga, 1991 Brown, 1948 Parodi, 1946; Brown, 1951; Núñez, 1952; Gould, 1968; Dav- idse & Pohl, 1972; Bouton et al., 1981; Dubcovsky & Zu- loaga, 1991 Especie Steinchisma cuprea 2n = 20 Steinchisma decipiens n = 10 2п = 20 Steinchisma hians n = 10 2п = 18 2п = 20 Steinchisma spathellosa 2n = Steinchisma stenophylla Bouton et al., Morrone et al., 1981; Dubcovsky & Zuloaga, 1991 2). Zonación: zonas costales e intercostales dis- tinguibles. Células largas intercostales: rectangula- res, más de tres veces más largas que anchas, de paredes anticlinales longitudinales moderada a pro- fundamente onduladas, las periclinales derechas a oblicuas, no papilosas. Células cortas intercostales: solitarias, entre células largas. Aparatos estomáti- cos: tipo “panicoide,” de 31-35 X 24.7-26 jum, dispuestos en 1—4 hileras longitudinales en las zo- nas intercostales y separados entre si por 1(–3) cé- lulas largas interestomáticas; células suberosas o pares sílico-suberosos aislados; células subsidiarias triangulares o en forma de domo. Аршуопез у Gan- chos ausentes o presentes en las zonas costales y/o intercostales. Micropelos: bicelulares, de 65-88 шт de largo, presentes en las zonas intercostales entre células largas; célula distal generalmente caediza, de paredes delgadas y ápice agudo; célula basal persistente, de paredes levemente engrosa cropelos: ausentes. Cuerpos silíceos costales: hallé riformes, longitudinalmente alargados, asociados a células suberosas formando hileras longitudinales contínuas. Cuerpos silíceos intercostales: ausentes. pidermis adaxial en vista paradermal: presenta células largas con papilas anchas, una por célula, de posición distal y sin paredes engrosadas. Apa- ratos estomáticos dispuestos en 2-3 hileras longi- tudinales en las zonas intercostales. Aguijones, gan- chos y micropelos bicelulares frecuentes, similares a los presentes en la epidermis abaxial. Macropelos unicelulares con cojín basal de células epidérmicas sobreelevadas. DISCUSIÓN El género Steinchisma se caracteriza por incluir especies intermedias entre las especies Kranz y non-Kranz. Anatómicamente, las especies son Kranz pero tienen un número menor de organelas en la vaina parenquimática externa (Brown et al., 1985); las organelas son de posición centrípeta en la vaina. Además, las vainas vasculares se encuen- tran a una distancia mayor que la presente en las especies Kranz: el número de células del mesofilo varía entre 5 y 7, siendo de 2 a 4 en los taxones Kranz, y usualmente de más de 7 en las especies non-Kranz (Ellis, 1988). El patrón api es in- termedio entre las plantas Kranz y non-Kranz Brown & Brown, 1975; Kanai & bu 1975; Ku & Edwards, 1978; Ku & Kanai, 1976; Morgan & Brown, 1979; Brown et al., 1985; Oguro et al., 1985); no obstante, todas las especies poseen un valor de isótopos de carbono ('*C) correspondiente al rango de los taxones С,, el que varía entre —22 y —38%o; el valor varía entre —9 y – 18%о en los taxones C, (Brown, 1977). Esta variación se pre- senta pues la materia orgánica de las especies C, es mucho más baja en contenido de "C; esto se debe a la acción no discriminatoria de la enzima ribulosa difosfato carboxilasa en las especies C}. Brown (1977) publicó los siguientes recuentos de isótopos de carbono para especies de Steinchisma: S. cuprea: "С —26.9%е; S. decipiens: С —26.7%о; S. exiguiflora: °C -28.1%о, S. hians "С —26%o. Para S. spathellosa se registró un valor de "C —27.1%е (В. Н. Brown, com. pers.), siendo el mis- mo de "С —27.5%е en S. stenophylla (R. H. Brown, com. pers.). Para Plagiantha tenella existe un re- cuento de РУС de —29.09%е (К. H. Brown, com. pers.). ~ NÜMEROS CROMOSÓMICOS Steinchisma posee un nümero cromosómico bá- sico de x — 10, de acuerdo a los recuentos reali- zados hasta el momento (Tabla 3). El género se caracteriza por incluir especies di- ploides, con excepción de S. spathellosa, ánica es- pecie hexaploide de Steinchisma. Volume 85, Number 4 Zuloaga 639 ga eta Steinchisma Poaceae) —— Otachyrium Subgén. Panicum Sect. Laxa NM ————— Plagiantha S. spathellosa | | | | S. stenophylla 2 — | S. exiguiflora | S. hians S. cuprea | | S. decipiens Figura 3. Cladograma de consenso estricto obtenido a partir de 2 cladogramas igualmente cortos. Los números indicados sobre las ramas corresponden a los valores del “Bremer support.’ ANÁLISIS CLADÍSTICO Los patrones filogenéticos obtenidos, a partir de los datos de la matriz (Tabla Se obtuvieron dos árboles igualmente parsimonio- sos con una longitud de 34 pasos, un índice de consistencia de 0.82 y un índice de retención de l excluir las autapomorffas de los taxones terminales (caracteres 5, 8, 10, 15, 17, 21), el ín- dice de consistencia es de 0.78. El árbol de con- senso estricto de los cladogramas (Fig. 3) define a Steinchisma como un grupo monofilético y a Pla- giantha como su grupo hermano. Los dos cladogra- mas originales difieren en la posición de Panicum subgénero Panicum y de la sección Laxa. En uno de ellos (Fig. 4A), la sección Laxa es el grupo her- mano del clado que une Steinchisma con Plagiant- ha, y Panicum es el grupo hermano del resto. En el otro cladograma (Fig. 4B), ambos taxones, sec- ciones Laxa y Panicum, forman un clado monofi- lético, sustentado por poseer pálea inferior hialina o ausente (9). Al calcular el “Bremer support” para los clado- gramas (Bremer, 1988, 1994), y luego de varias ho- ras de cómputos, se obtuvieron 12,941 árboles, para 15 pasos adicionales de largo. Los resultados se muestran en el árbol de consenso estricto, donde el clado Steinchisma necesita dos pasos adicionales para colapsar el nodo. Las restantes ramas del árbol de consenso tienen un soporte igual a uno. El bajo ), fueron similares. valor soporte de las ramas es típico para los análisis basados en un conjunto de datos morfológicos (Ka- ris, 1 El С" análisis soporta la hipótesis que con- sidera a Steinchisma como un grupo monofilético sustentado por dos sinapomorfías: el патего de cé- lulas clorenquimáticas entre haces vasculares (18) el cladograma de la Figura 4A, el clado formado por S. spathellosa y S. stenophylla es basal dentro del género, y está sustentado por la presencia de argos rizomas hojosos y cundidores (2). Estas es- pecies crecen preferentemente en cursos de aguas turbulentas, en orillas de arroyos y ríos, entre las piedras. Steinchisma spathellosa presenta la nove- dad de poseer estaminodios en la flor superior (16). Un segundo clado está constituido por S. exiguiflo- ra-S. decipiens, sustentado por dos sinapomorfías: flor inferior neutra (12) y flor superior con dos es- tambres (16). Steinchisma exiguiflora y S. hians for- man, asimismo, un clado por tener espiguillas dis- tribuidas en la porción superior de las ramas (6). El clado S. cuprea y S. decipiens posee inflorescen- cias contraidas (4). En el segundo cladograma (Fig. 4B), Steinchisma presenta la misma topología; estos cladogramas difieren en la evolución del carácter 9. En ambos cladogramas se observa que Plagiant- ha se comporta como el grupo hermano de Stein- chisma, en una posición inclusiva entre este último taxón y Panicum más la sección Laxa, sustentando este hecho la hipótesis que excluye a Steinchisma de Panicum. Si bien el resultado de este análisis también podría llevar a la inclusión de Plagiantha y Steinchisma dentro de Panicum, se considera que este último género es polifilético y la inclusión de Plagiantha y Steinchisma harían al mismo aún más heterogéneo La evolución de la flor en Poaceae ha sido sujeta a numerosas investigaciones (Arber, 1934; Clifford, 1961; Anton & Connor, 1995). En la tribu Paniceae las espiguillas son bifloras, con una flor superior perfecta, con un pistilo y tres estambres, y una flor inferior estaminada, con tres estambres (Clayton & Renvoize, 1986). En el presente análisis se observa el siguiente patrón en el sistema floral de Stein- chisma: la condición andromonoica es primitiva, y a partir de la misma deriva la condición diclino- monoica, presente en 5. spathellosa; en esta espe- cie, por reducción de los estambres de la flor su- perior, se halla una flor superior pistilada y una flor inferior estaminada. A la vez, en el clado S. exi- guiflorum-S. decipiens se presenta la condición mo- noica, la que sería derivada por reducción de la Annals of the 640 Missouri Botanical Garden “80% LL а ѕиәгагэәр ` Idi2ep `$ 4-11 9 eeJdno 'S T Hi 912, suely '5 71; L | LLL inBixo * жаа | eol б S € 026181 | ey¡Aydoua]s ‘$ Ha кй 2162 7161 esojjayjeds ‘$ * too LLLI eujueiDe[d AHHH HHH шие! Id LLZ L S,0L,8 2 020 exe] 298 13 5 21515 шпоиға 'иобапо 10210: 2212026181 LL L шпилузејо p BIDUI]SISUOS әр IMPUI) JI :97 () (погоиздал әр 2orpur) YI “pE (рпизиој) q `ѕегѕе[дошоц seroe4 seireq ѕеү *serpourodeurs ивјеџов se1dau SELVA uoo «орвоірш залојовлео SO] ‘(Z e[qe]) sa121»eeo әр zew рү әр лига e sopruajqo sosoruouirsaed ojuoui[enzr сешелдореј "я “үр enr LL V sueidi2ep '5 easdna ‘Ss T 0-0 1— 9121 6 suely ‘5 4, | |9 LLL елојипб!хо ‘$ AHH © 026181 | 2 Ацаоиој5 ‘$ H, LET ё |ә 7,616 esojjoujeds ‘< 9| 0 200 sm eujueiDe[d AHHH 22 6241 beet exe] '1205 diHHI- ©1018 2 415, 5 02102 windiueg ‘uabqns -HHHHeE јава 90975 120261 81 LL 2 шпилузејо x 6 Volume 85, Number 4 Zuloaga et al. 641 Steinchisma (Poaceae) flor estaminada inferior, que es neutra, permane- ciendo la flor superior perfecta. Cabe destacar que esta flor superior, en el clado antes mencionado, posee dos estambres de posición lateral, siendo el frontal ausente. Este carácter ha sido citado como excepcional dentro de la tribu Paniceae (Simon & Weiller, 1995) DISTRIBUCIÓN Y ECOLOGÍA Las especies de Steinchisma se distribuyen des- de Estados Unidos de América hasta la Argentina. Steinchisma hians es la especie que posee una mayor distribución, hallándose en los Estados Uni- dos de América, México, Guatemala, Honduras, Nicaragua y América del Sur, desde Colombia has- ta Argentina. Steinchisma cuprea es una especie endémica de México, donde crece en campos entre los 2100 y 2600 m s.m. Por otra parte, S. exigui- flora se halla ánicamente en las islas del Caribe, desde las Bahamas hasta Jamaica y Haití. Stein- chisma decipiens, S. stenophylla y S. spathellosa son exclusivas de América del Sur; S. spathellosa tiene una distribución restringida, encontrándos esde Santa Catarina, en ose a 10 Argentina. Steinchisma stenophylla es una especie que crece en bordes de arroyos y ríos de la Guyana Venezolana, y diversas áreas del cerrado en Brasil, encontrándose en Bahia, Minas Gerais y Pará. Fi- nalmente, S. decipiens también crece en bordes de ríos y arroyos, desde el nordeste de Brasil hasta Bolivia, Paraguay, Uruguay y Argentina; se ha ha- llado ocasionalmente en Colombia y Venezuela. Es interesante destacar que, de acuerdo al re- CLAVE PARA DIFERENCIAR LAs EsPECIES DE STEINCHISMA sultado del análisis cladístico, los clados resultan- tes dentro del género, formados por S. spathellosa + S. stenophylla, S. decipiens + S. cuprea, y S. hians + S. exiguiflora, poseen una distribución alo- pátrica, de acuerdo a lo antes expuesto y lo que se observa en las Figuras 5 y 6. TRATAMIENTO TAXONÓMICO Steinchisma Raf., Bull. Bot., Geneva 1: 220. 1830. Panicum subg. Steinchisma (Raf.) Zu- loaga, en Soderstrom et al., Grass Syst. Evol. : Steinchisma hians (Elliott) ћ [= Рем ин hians Elliott Inflorescencias laxas a contraidas, con espigui- llas congestas en las ramificaciones. Espiguillas glabras, con gluma inferior 3-nervia, Y a % del largo de la espiguilla. Gluma superior y lemma in- ferior 3-5(-7)-nervias. Pálea inferior conspicua, ex- pandida a la madurez a lo ancho y/o a lo largo de la espiguilla; flor inferior estaminada o neutra. An- tecio superior ovoide a elipsoide, con papilas ve- rrugosas dispuestas regularmente en hileras longi- tudinales, con aguijones hacia el ápice de la lemma o sin los mismos. Plantas cespitosas, corta a lar- gamente rizomatosas, a largamente rizomatosas, con lígulas membranáceas, cortamente ciliadas a laci- niadas, láminas lanceoladas a filiformes, planas. Género con 6 especies americanas, distribuidas desde los Estados Unidos de América hasta la Ar- gentina. Habitan comúnmente en lugares húmedos y abiertos, en bordes de arroyos, pantanos o zonas inundables, desde el nivel del mar hasta los 2600 m de elevación. la. Inflorescencia contraida, espiciforme, con ramificaciones de аан orden adpresas al eje principal y espi- guillas cubriendo desde 2a. Esp An n cafias erectas; Méxic base los ejes de las ramificacio as de 2.6–3.2 mm de ке con pálea (е dunt: a lo ancho a la madurez; plantas . cuprea 2b. Espiguilla: de 1.8-2.4(-2.8) mm de je con pálea inferior no sobresaliendo a lo ancho a la madurez: antas largamente rizomatosas, con сай lb. Inflorescencia laxa, con las [^ie 'aciones ml hacia la porción basal. 3a. Espiguillas de 1 1.2- 1.4 mm de largo; islas del ( 3b. Espiga de (1. ntas con cañas rígidas, ma ñas decumbentes; Colombia a Argen Ma. atras M S. aribe 7-)2.4-3.6 mm de largo; Unidos de América a la Argentina. estamente ramificadas en los nudos basales y medios; o presentes; láminas filiformes, de 0.2-0.5(-2) mm de ancho; flor superior con tres estambre S. 5 5. decipiens amificaciones de segundo orden distantes entre si y divergentes del raquis: eje de 5. exiguiflora inflorescencias 4b. Pana con cañas herbáceas, no rígidas, no ramificadas en los nudos basales y medios; inflorescencias елері ylla axilares ausentes; láminas lanceoladas, de 2-10 mm de ancho; flor superior con dos estambres o tres estamin РА odios. за. Plantas de 15-60 cm de alto, cortamente rizomatosas; espiguillas de (1.7-)2.2-2.6 mm de largo. con pálea inferior sobresaliendo a lo ancho de la espiguilla; flor inferior neutra; flor superior c ‘on dos be S. hians 5b. Plantas de 60-120 cm de alto, largamente rizomatosas; espiguilla de 3-3.6 á mm de bas con lea inferior no sobresaliendo a lo ancho de la espiguilla; flor inferior estaminada; flor super con tres estaminodios gae m Annals of the 642 Missouri Botanical Garden “pyAydouays "$ А оѕорәцтраѕ “с *naojyimzixo “с *suaidioop `$ *no4dno пшвтузитоје әр ugionquisi( 00: "d | о | ийм ымыы ыйыы эша = wx000! 008 009 OOF 002 0 6 ensiy Volume 85, Number 4 1998 Zuloaga et al. Steinchisma (Poaceae) | ж Steinchisma hians | \ d \ Figura 6. Distribución de Steinchisma hians. 1. Steinchisma cuprea (Hitchc. & Chase) W. V. Br., Mem. Torrey Bot. Club 23: 20. 1977. Pa- nicum cupreum Hitchc. & Chase, Contr. U.S. Natl. Herb. 15: 120, fig. 113. 1910. Panicum hians var. purpurascens Scribn., Proc. Acad. Nat. Sci. Philadelphia 1891: 296. 1891. TIPO: México. México: Flor de María, wet hollows of plains, Pringle 3449 (holótipo, US-80756!). Plantas perennes, cespitosas, cortamente rizo- matosas, con cañas de alto, erectas, simples, paucinodes; entrenudos cilíndricos, hue- cos, glabros, pajizos a violáceos; nudos comprimi- dos, castafios, glabros. Vainas estriadas, aquilladas, las basales más largas que los entrenudos, las su- periores más cortas, glabras, los bordes membra- náceos. Lígulas de ca. 0.6 mm de largo, membra- náceas, cortamente ciliadas en la porción superior; cuello glabro. Láminas de 8-25 X 0.2-0.4 cm, li- near-lanceoladas, con los bordes involutos, de base redondeada y ápice subulado, pilosa en la cara ada- xial, con pelos largos, delgados, cara abaxial glabra о esparcidamente pilosa hacia la base. Inflorescen- cias terminales largamente exertas; pedánculo de 17-20 cm de largo; cilíndrico, glabro, panojas de 4—13 X 1-2 cm, contraidas; eje principal anguloso, escabriásculo; pulvínulos glabros; ramificaciones de primer orden alternas, las inferiores de 2-7 cm de largo, contraidas a ocasionalmente algo diver- gentes, el resto ascendentes, distanciadas entre si, adpresas al eje principal; ejes de las ramificaciones y pedicelos triquetros, escabriásculos; espiguillas sobre ejes de tercer orden hacia la porción superior 644 Annals of the Missouri Botanical Garden de las ramas. Espiguillas de 2.6-3.2 X 0.8-1 mm, largamente elipsoides, glabras, pajizas a violáceas a la madurez; gluma superior y lemma inferior su- biguales o la gluma superior ligeramente menor y no cubriendo el ápice del antecio superior. Gluma inferior de 1-1.4 mm de largo, 8 a menos de % del largo de la espiguilla, ovada, aguda, 3-nervia. Glu- ma superior de 2.5—3.1 mm de largo, aguda, 5-ner- via. Lemma inferior de 2.5—3.1 mm de largo, glu- miforme, 5-nervia. Pálea inferior de 2.5-3 X 1 mm, elíptica, expandida y ensanchada a la madurez, rí- gida, pajiza o con tintes violáceos, glabra, los bor- des escabriásculos; flor inferior neutra. Antecio su- perior de 2.4 X violáceos, plano-convexo, glabro, con papilas ve- rrugosas distribuidas regularmente en toda su su- perficie; lemma apiculada, hialina en el ápice; lo- dículas 2, de 0.2 mm de largo, conduplicadas, truncadas; estambres 2, anteras de О. largo. Cariopsis de 1.5 X 0.6 mm, largamente elip- soide, castaño, hilo oblongo; embrión menos de Y del largo de la cariopsis. mm, ovoide, pajizo o con tintes Distribución y ecología. México, en los estados de Durango, Jalisco, México, Puebla, Querétaro, Veracruz y Zacatecas, sobre suelos hümedos; co- mún a orillas de lagunas у ríos. Se encuentra entre los 2100 y 2600 m s.m. Material representativo citado. MÉXICO. Durango: near El Salto Los Angeles, small stream road between Durango and Mazatlán, Beetle M-7708 (MO); 3 mi. E of El Salto, 8400 ft., Reeder & Reeder 4466 (US). Jalisco: 25 mi. S of Guadalajara, Reeder 2330 (US). México: To- luca, 9000 ft., p aedi & Leavenworth 1925 (US). Puebla: Laguna San Baltasar, 2140 m, Nicolás 217, 5879 (P, US), s.n. (MEXU); Paso de Arc eod Oliva 58 (US); San Fe ands Orte de Puebla, 2300 m, Boege 2543 (MEXU). Querétaro: cerca de San E 8 km al E de Amealco, sobre la carretera a Aculco, 2600 m, Rzedowski 48632 (MEXU). cruz: near El Puerto, above Acul- tzingo, 7650 ft., Sharp 44753 (MEXU). Zacatecas: 38 km al W de Jalpa, sobre la carretera a Tlaltenango, 30 km del entronque con la carretera Jalpa-Juchipila, 2550 m, Rzedowski & McVaugh 1024 (US). ~ Steinchisma cuprea, especie endémica de Méxi- co, comparte con S. decipiens, especie de América del Sur, inflorescencias contraidas, con las ramas adpresas al eje principal; se diferencia de esta es- pecie por incluir plantas cespitosas, con espiguillas de mayor tamaño, con la pálea inferior expandida a lo ancho a la madurez. A su vez, se distingue de S. hians, especie de amplia distribución, por incluir esta última plantas con inflorescencias laxas, abier- tas, y espiguillas menores, hasta de 2.6 mm de lar- go. 2. Steinchisma decipiens (Nees ex Trin.) W. V. Br., Mem. Torrey Bot. Club 23: 20. 1977. Pa- nicum decipiens Nees ex Trin., Gram. Panic. 227. 1826. TIPO: Brasil. Minas Gerais: in hu- midis arenosis pr. Agua Quenti, же s.n. (holótipo, LE!; e B!, US-97 , US- 2903516!; foto del holótipo, K!). . 7. Panicum decipiens Nees en Mart., Fl. Bras. Enum. Pl. 2(1) 193. 1829, nom. illeg, non Nees i 2-2 TIPO: Brasil. juni Gerais: “ o Minan rum versus Paranan, ubi ad mozam itur," Martius s.n. «ет ва М!; isótipo, US- 2903515). Planta perenne, largamente rizomatosa, con ri- zomas hojosos; cañas de 15—70 cm de alto, erectas, geniculadas y arraigadas o no en los nudos inferio- res, paucinodes, simples; entrenudos cilindricos a comprimidos, huecos, glabros; nudos glabros, blan- quecinos a violáceos. Vainas de 2-8 cm de largo, comprimidas, esparcidamente pilosas a glabras, pa- piráceas, verdosas a violáceas, de bordes hialinos, los superiores cortamente ciliados. Lígulas de 0.2— 0.4 mm de largo, membranáceas, cortamente cilia- das en la porción superior; cuello blanquecino, gla- bro. Láminas de 5-25 X 0 7 cm, linear-lan- ceoladas, con los bordes involutos, de base angosta y ápice acuminado, verde oscuras; nervio medio conspicuo a inconspicuo; cara adaxial con pelos tuberculados caducos; cara abaxial pilosa a glabra. Inflorescencias corta a largamente pedunculadas; pedúnculo hasta de 40 cm de largo; panojas, de 6— 13 X 0.5-2 cm, contraidas, espiciformes, con ra- mificaciones de segundo orden adpresas al eje prin- cipal, la inferior algo distanciada y divergente; eje principal anguloso, liso, glabro; ramificaciones al- ternas; pulvínulos glabros; espiguillas en pares o solitarias sobre ramificaciones cortas de tercer or- den; pedicelos claviformes, escabriásculos. Espi- guillas de 1.8-2.4(-2.8) X 0.6 mm, largamente elipsoides, glabras, verdosas a violáceas. Gluma in- ferior de 1.2-2 mm de largo, Y a % del largo de la espiguilla, de ápice obtuso a acuminado, 3-nervia, os nervios anastomosados hacia el ápice. Gluma superior de 1.5-2(-2.3) mm de largo, de ápice ob- tuso a agudo, dejando al descubierto la porción su- perior del antecio, 5-nervia. Lemma inferior de 1.8— 2.3(-3) mm de largo, lanceolada, 3-5-пегуіа, glumiforme. Pálea inferior de 1.8-2.4(-2.8) X 0.6 mm, lanceolada; quillas escabriúsculas; flor inferior neutra. Antecio superior de 1.6-2.2(-2.6) X 0.5-0.6 mm, largamente ovoide, glabro, blanquecino o con tintes violáceos hacia el ápice, acuminado; lodí- culas 2, de 0.3 mm de largo, conduplicadas, trun- cadas; estambres 2; anteras de 0.2—0.4 mm de lar- go. Cariopsis largamente obovoide, castafio, de 1.3 Volume 85, Number 4 Zuloaga et al. 645 1998 Steinchisma (Poaceae) шо | Figu Steinchisma d Aid 1805, SP). —A. Hábito. —B. Esp БР vista P —C. Espiguilla vista del ED. ы la gluma infer inferior, vista ventral de la pálea inferio rior con alas. —E. Antecio superio visto del lado de la pálea. pin 5. ровон con dos lodículas, dos bs y x MD. 646 Annals of the Missouri Botanical Garden X 0.6 mm; hilo oblongo; embrión % del largo de la cariopsis. Distribución y ecología. Especie sudamericana, hallándose desde Colombia y el norte del Brasil hasta Paraguay, Uruguay, Bolivia y la Argentina. Habita en suelos hámedos en bordes de pantanos o lagunas o a lo largo de cursos de agua sobre suelos arenosos. Se encuentra desde el nivel del mar hasta aproximadamente los 1600 m s.m. Material de lin aio citado. ARGENTINA. Corrientes: de Ituzaingó a Villa Olivari, Zuloaga et al. 3305 (SI). Misiones: a de Apóstoles a Azara, Zu- а « 1. 544 (SI, US). Salta: Alemanía, Venturi ), US). Tucumán: Los Chamicos, Venturi 2785 . BOL IVIA. Santa Cruz: Terebinto, Steinbach a Bahia: 19.5 km SE of town of Morro A 052 road to Mundo Novo by the Rio Ferro Doido, isi et al. 19384 (СЕРЕС, K, MO, P, e Vila = Rio de Contas, middle co a Almas, Harley et al. 19647 (CEPEC, K, MO, P, US). Distrito. Feder : Reserva Eco- logica do IBGE, Corrego Ronc ape 2. 436 (51). Goiás: m S of Cristalina, ele et al. 9 К, МО, NY, SP, UB, US). Mato Gross Sul: dd d Grande and үүнү 400 m, Chase 10925 (RB, US). Minas Ge- rais: 45 km SE of Belo Horizonte, Serra do bii Mise et al. 19571 (F, NY, UB, US); 35 km S of Gouvea, km 243 on MG 259, Anderson et al. 35147 (F, MO, NY SP, UB); Serra do Cipó, 1220 m, i et al. 36395 (MO, NY, UB, US). Parana: 10 km Palmas, Smith et al. 15626 (NY, P, RB, SI, US). Rio Leopoldo, Dutra 422 (SI); Toms, Mens et al. 2326 (CEN). Santa Catarina: Mun. ‚ 10 km S of Horizonte, Smith & Klein 15584 (К, MO, P. RB, US). Sao Paulo: Fazenda Campininha, just N of rio Moji-Guacu, ca. pe km NNE of Padua Sales, Ейеп «с Ейеп 2667 (MO, S UB, US). COLOMBIA. Meta: 20 km SE of Villavicencio, 480 m, Alston 7578 (COL, US). Norte de Santander Welten García & ae 2 (COL). PARAGUAY. mbay: Parque Nacional Cerro Corá, Hahn 2533 (MO): ); Sierra de Шш Hassler 101 10 (G, K, LIL, NY, P, US). ig be in ee си us Ipacaray, Hassler 1. (С, US). Conc n: zwischen Río Apa und Río - daban, Fiebrig 1507 (С, К, US). Cordillera: Coil de Altos, Hassler 3658 (G ). Guairá: Villa Rica, Balansa 60 (G, K, P). Paragu dini et al. 9038 (SI). : epos 25 km N of Castillos, Bartlett 21348 (SI). Ta- mbó: camino a Rivera, 32 km de Tacuarembó, Ca- бы & Zuloaga 32427 (SI). VENEZUELA. Bolívar: Hato Divina Pastora, Gran Sabana, Tamayo 2901 (US, VEN). Exomorfológicamente, esta especie es afín a S. spathellosa, con la que comparte la presencia de largos rizomas en la base. Ambos taxones son fre- cuentes en bordes de cursos de agua. Steinchisma spathellosa se distingue por incluir plantas de ma- yor tamaño, con inflorescencias laxas con espigui- llas de 3-3.6 mm de largo; las espiguillas llevan en la flor superior tres estaminodios (en lugar de los dos estambres desarrollados que están presentes en S. decipiens). Steinchisma decipiens es comün en cerrados del Brasil, donde se extiende desde Bahia hasta Rio Grande do Sul, llegando hasta Paraguay y el nor- deste de Argentina. Por el contrario, se ha hallado esporádicamente en Venezuela, Colombia y el no- roeste de la Argentina. 3. Steinchisma exiguiflora (Griseb.) W. V. Br., Mem. Torrey Bot. Club 23: 20. 1977. Panicum exiguiflorum Griseb., Cat. Pl. Cuba 234. 1866. А. Rich., en Sagra, . 11: 305. 1850, nom. illeg., non (P. Beauv.) Raspail (1825). TIPO: Cuba: sin localidad, de la Sagra s.n. (holótipo, P!) Panicum pres var. variegatum Griseb., Cat. Pl. Cuba 233. . TIPO: Cuba: sin loc liad, шы 3450 lio. GOET? no visto; isótipos, G!, 5470!, NY!, Р!, US-80733!, 4211. Nd леш Hack., Oesterr. Bot. Z. 51: 370. 1901. : Bahamas. Fortune Island, inter frutices, Eg- gers 3978 (holótipo, W!). Planta perenne, cespitosa, cortamente rizomato- ва; cañas de (7-)40-60 cm de alto, erectas, ocasio- nalmente decumbentes, simples o ramificadas; en- trenudos de 1-3(-8) ст de largo, cilíndricos, glabros; nudos pajizos, comprimidos, glabros. Vainas de 2-4 cm de largo, estriadas, ciliadas hacia los bordes superiores, glabras en el resto de la su- perficie. Lígulas de 0.3 mm de largo, membraná- ceo-ciliadas, hialinas, cuello pajizo, glabro. Lámi- nas de 3-4.5(-10) х 0 9 cm, lineares a filiformes, involutas, pilosas en la cara adaxial, gla- bras en la cara abaxial, de ápice subulado, los bor- des lisos. /nflorescencia exerta; pedúnculo de (2-) 10 cm de largo, cilíndrico, glabro; panoja de (2-) 5-8 X (1-)2-7 ст, laxa, ocasionalmente contraida; eje principal anguloso, glabro; ramificaciones pri- marias alternas, distantes y divergentes del eje principal; pulvínulos glabros; raquis de las ramificaciones de 0.5-3.5 cm de largo, glabras, li- sas; pedicelos de 0.1-0.4 cm de largo, solitarios, claviformes, glabros. Espiguillas de 1.2-1.4 X mm, elipsoides, glabras, pajizas a verdosas o con tintes violáceos. Gluma inferior de 0.5—0.6 mm de largo, % del largo de la espiguilla, 3-nervia, agu- da, glabra. Gluma superior de 0.8-1 mm de largo, más corta que la lemma inferior y dejando al des- cubierto el ápice del antecio superior, 3-nervia, ob- tusa. Lemma inferior de 1-1.2 mm de largo, 3-ner- via, aguda. Pálea inferior de 1.2 X 0.6 elíptica, ensanchada a la madurez, llegando hasta mm, 1.4 mm de ancho, papirácea, glabra, los bordes es- Volume 85, Number 4 1998 Zuloaga 647 ga eta Steinchisma "iu oe cabrosos; flor inferior neutra. Antecio superior de 1.2 X 0.5 mm, ovoide, papiloso, glabro; flor supe- rior perfecta; lodículas 2, truncadas, conduplica- das; estambres 2; anteras de 0.4 mm de largo; es- tigmas plumosos. Cariopsis de 0.8 X 0.4 mm, largamente elipsoide, castafia; hilo oblongo; em- brión menos de % del largo de la cariopsis. Distribución y ecología. Especie exclusiva de las islas del Caribe, hallándose desde las Bahamas hasta Haití; se encuentra en sabanas hümedas, lle- gando desde el nivel del mar hasta 400 m de ele- vación. Material representativo citado. BAHAMA ISLANDS. Acklin Island, near Snug Corner, Correll & Proctor 48888 (MO, NY, US); Great Inagua, about 1 mi. E of Matthew Town, Correll 41685 (NY). CUBA. 12 km E of Baragua, Hitchcock 23356 (US). Habana: Guanabacoa, Hitchcock 23241 (US); foot of Jiquima hill, Madruga, León 14690 (US). Isla de la Juventud: Siguaena, Britton et al. 15381 NY, P, US); San Pedro and lind. Britton & Wilson riente: Boe. banks of Río Canto, т 2443 (US); Holguín, foot of Cerro de Fraile, p 0 (NY, US); Bayate, Sabana sg л ir an US E rd del Río: savannas ne a de E Viejo, between Guane and Remates, . Killip. 32314 (05); 13 km S of Pinar del Rio Ns 23278 (US); Sabana de San Julián, S . León & Roca 700 (US); Laguna de Piedras, ES of. joe т León 18579 (US); Santa Cruz de los Pinos, Finca Mamey, Ekman s.n., Amer. Gr. Hb. 702 (G, NY, US, W). Villa Clara: 3 km E of Santa Clara, Howard 4343 (NY, US); Sabana de Ma- nacas, León 9277 (NY, US); sin localidad, Wright 756 (С, MO), 3450 (W). REPÜBLICA DOMINICANA. Santo Do- mingo: vicinity of Ciudad Trujillo, Allard 14276 (US). HAITÍ. Massif du Nord, Gros-Morne, M. Bellance, Ekman 4916 (US). JAMAICA. Clarendon: 0.8 mi. by road E of Toll Gate, Proctor 37804 (MO Davidse (1994) sinonimizó S. exiguiflora (Pani- cum exiguiflorum) con S. hians (P. hians), sin es- tablecer cual fue el criterio que llevó a esta deter- minación. Sin embargo, la especie se distingue de S. hians por tener hojas lineares a filiformes y es- piguillas de 1.2-1.4 mm de largo. Se separa de 8. stenophylla, taxón suramericano que también posee láminas filiformes, por incluir esta áltima especie plantas de mayor porte, de 30-120 cm de alto, con láminas hasta de 30 cm de largo y espiguillas de 2.4—3.2 mm de largo, con la flor inferior estaminada y superior perfecta con tres estambres. 4. Steinchisma hians (Elliott) Nash, en Small, Fl. S.E. U.S. 105. 1903. Panicum hians Elliott, Sketch Bot. S. Carolina 1: 118. 1816. TIPO: Estados Unidos de América. Virginia: "in pi- netis humidis" (holótipo, CHARL no visto; isó- tipo y foto del holótipo, US-80696!). Figura 8. Panicum milioides Nees ex Trin., Gram. Panic. 225. 1826. Panicum milioides Nees en Mart . Bras. Enum. T 2(1): 175. 1829. TIPO: Brasil. foie et Oei- s, Bahia and Piauhy,” Martius s.n. (lectótipo, aquí Сет, LE-TRIN!; isolectótipos, М!, US). dad, зе n. (holótipo, B!; isótipo, US-2830915!). Panicum oblongiflorum Desv., “Habitat in Carolina," sin colector, s.n. (holótipo, Р!). Рапісит jejunum Trin., Мет. Acad. Imp. Sci. Saint-Pé- tersbourg, Sér. 6, Sci. Math., Seconde Pt. Sci. Nat. 2: 103. 1836. TIPO: Estados Unidos de América. “Louisiana,” sin colector, s.n. (holótipo, LE-TRIN!). Panicum hians var. pallescens Dóll, en Mart., Fl. Bras. 2(2): 240. 1877. TIPO: Brasil. Minas Gerais: Lagoa nta, Warming s.n. (? no visto). Panicum milioides var. filifolium R. A. Palacios, en Bur- > pto. La Paz, ruta 126, km 53, Burkart 21073 (holótipo, SI!). Planta perenne, cespitosa, cortamente rizomato- sa, con cafías cm de alto geniculadas y arraigadas hacia la base, luego erguidas, simples a ramificadas en los nudos inferiores y ramificadas o no en los nudos superiores; entrenudos de 2-10 cm de largo, comprimidos, huecos, glabros; nudos gla- bros, oscuros, comprimidos. Vainas de 1.6–8 cm de largo, estriadas, esparcidamente híspidas a glabras, un borde pestañoso, el restante membranáceo. Lí- gulas de 0.4—0.8 mm de largo, membranáceas, cor- tamente ciliadas en la porción superior; cuello gla- bro, de color castaño claro. Láminas de 4.5-2 .2-0.5 cm, linear-lanceoladas, planas o con los bordes involutos, de base angostada y ápice larga- mente atenuado, pilosa hacia la base en la cara adaxial y los bordes inferiores, el resto de la su- perficie glabra. Inflorescencias terminales exertas; pedúnculo de 4—15 cm de largo, cilíndrico, glabro; panojas de 5-20(-25) X 4-10 cm, laxas, difusas, con ramificaciones de segundo orden alternas, di- vergentes y distanciadas entre si; eje de las rami- ficaciones desnudos en la porción basal; ramifica- iones de tercer orden cortas, con espiguillas apareadas o solitarias sobre pedicelos cortos; eje principal anguloso, escabriúsculo; pulvínulos cas- taño-claros, esparcidamente pilosos a glabros; eje de las ramificaciones escabroso; pedicelos de 0.7-2 mm de largo, claviformes, escabriúsculos. Espigui- llas de (1.7-)2.2-2.6 X 0.6-0.9 mm, largamente elipsoides, biconvexas, glabras, verdosas o con tin- tes violáceos. Gluma inferior de 0.8-1.2 mm de lar- go, ovada, Y del largo de la espiguilla, 3(-5)-ner- via; nervio medio escabriúsculo, el ápice agudo. luma superior de 1.6-1.8 mm de largo, general- mente más corta que el antecio superior, 5(—7)-ner- via, el nervio medio escabriúsculo. Lemma inferior de 1.8-2.5 mm de largo, glumiforme, 5(-7)-nervia. 648 Annals of the Missouri Botanical Garden Figura 8. Steinchisma hians E Hass ler 123, del lado de la gluma d june Pálea inferior n. pálea. —H. Flor vista hilar. Pálea inferior de 2-2.5 X 0.7-1.2 mm, oblonga, papirácea, sobresaliendo a lo largo y ancho de la espiguilla a la madurez; alas manifiestas, cortamen- te escabriásculas en los márgenes; flor inferior neu- tra. Antecio superior de 1.7-2.1 X .6 mm, lar- gamente ovoide, pajizo, papiloso, con aguijones en el ápice de la lemma, el resto de la superficie gla- bra, de ápice agudo a acuminado; lodículas 2, ca. 0.2 mm de largo, conduplicadas; estambres 2; an- SI; D-I, Burkart 17513, SI). —A. Hábito iguilla vista del lado de la gluma superior. —D. Esp F. em visto del lado de la lemma. —G. pier con is ea dos estambres y dos estigmas. —I. Cariopsis, vista escutelar. —J. Cariopsis, . —B. Espiguilla vista iguilla, vista lateral. — ntecio superior visto del lado de la teras de 0.4—0.6 mm de largo. Cariopsis de 1.1–1.2 x 0 m, largamente ovoide, castaña, hilo oblongo; embrión poco menor de la mitad del largo de la cariopsi а. y ecología. Se halla en Estados Unidos de América, México, Centroamérica, Co- lombia, Brasil, Bolivia, Paraguay, Uruguay y la Ar- gentina. Crece en lugares abiertos y húmedos desde el nivel del mar hasta los 2600 m s.m. Volume 85, Number 4 1998 Zuloaga et al. Steinchisma (Poaceae) 649 Material representativo citado. ESTADOS UNIDOS DE AMÉRICA. Florida: gia: along Ogeechee Texas: Dallas, Reverchon 1680 (P). GUATEMALA. Hue- h of Huehuetenango, 1800 m by 12017 ү: MO). ! CO. Chiapas: 9 $ 1 Mexican Highway 1 зке m, Breedlove & Davidse 54947 (MO). Gueiajuáto: Sant e Juventino Ro- a Guanajuato, oak forest, Beetle M-7323 (MO). km adelante del crucero de Arandas, carretera po 1780 m, Guzmán et al. 956 (MEXU). inity of Morelia, N of Loma del Zapote, 1950 1 m, pem 5657 (MEXU, MO, US). NICARAGUA. Estelí: Mesas Moropotente, 13?14'N, 86^16'W, 1100- 1300 m, Davidse et al. 30617 (MO). ARGENTINA. Buenos Aires: Pdo. La Plata, Elizalde, Cabrera 7417 (MO, SI, US). Chaco: Dpto. Bermejo, Las Palmas, Joergensen 2439 (MO, SI). Córdoba: San Teo- doro, Stuckert 21459 (G), 21494 (G). Corrientes: Dpto. Berón de Astrada, 46 km W de Itá Ibaté, Valencia, Ahu- тада 417 (CTES, F, МО, 51); Ruta Nac. 40, 6 km S entrada a Garruchos, Zuloaga et al. 3108 (SI). Distrito Federal: Barrancas al sur, Venturi 160 (G). Entre Ríos: 11 vieja, al. 261 (SI, US). Jujuy: Quebrada de Tiraxi, camino a Tiraxi, Zuloaga & Morrone 3012 (MO, SI). Misiones: Dpto. 21. San José, Escuela Agrotécnica Pascual Gentilini, adde et и 28516 (SI), 29066 (51. Salta: Dpto. Anta, La La | 1297 (MO). Santiago del Estero: Dpto. Belgrano, 10 km de Bandera a Pinto, Cristóbal 47 (CTES). Tucumán: Dpto. Leales, Сћаћаг Pozo, Venturi 480 (BAA, LIL, US), 1635 (BAA, LIL, US). BOLIVIA. Beni: Espíritu, Beck 5314 (LPB, 51, US), 5017 (LPB). Tarija: Guerra Huaico, 16 km SW de bns Coro 1373 (LPB). BRASIL. Bahia: Alagoinhas, Chase 8118 (MO, NY, US). Goiás: entre Bra- silia y Niquélandia, Pires et al. 9676 (UB). Mato Grosso do Sul: Campo Grande, Chase 10847 (IAN, RB, US). Minas Gerais: Serra do Curral, SE of Belo Horizonte, 1000 m, Chase 8969 (US). Paraná: 8 km NE of the Pa- raná-Santa Catarina жыра at the Rio Negro, 820 m, Da- p et al. 11035 (MO, NY). Piauí: Mun. Urucuf, ca. 11 m SW of Urucuí, in of Rio Urucuí-Preto, Ейеп & Eiten 4522 (NY, US). Rio Grande do Sul: 60 km W of Passo Fundo along Hwy. BR-285 to Vacaria, at intersec- tion of highway with the Rio Ligeiro, 800 m, Davidse et al. 11157A (MO). Santa Catarina: 19 km al S de Abe- lardo Luz, Smith & Klein 11518 (NY, US). COLOMBIA. undinamarca: Finca San Rafael, 2600 m, García Ba- rriga 10773 (US). Valle: Cartago, Santa Ana de los Ca- balleros, 950 m, Cuatrecasas 23035 (F, US). PARAGUAY. Alto Paraguay: Puerto Casado, Hartley 118 (US). Alto Paraná: Irala, Montes 11054 (US). Amambay: Sierra de Amambay, Hassler 10783 (G, K, NY, P, US). Boquerón: Puerto Casado, Hartley 118 (US); Ruta Trans Chaco, 8 km SE de Mariscal Estigarribia, Schinini & Palacios 25780 (MO). Central: Asunción, Balansa 59 (С, K, P, US); lago Ipacaray, Hassler 12433 (С, NY); Itá sies 1. del Río Paraguay, Schinini 6294 (NY). Conce tre el Río Apa y el Río Aquidabán, Fiebrig 4776 (P. "5093 (F, G). Cordillera: San Bernardino, Hassler 123 (G, SI). Guairá: Villa Rica, Joergensen 3543 (BAF, F, MO, NY SI, US). Paraguarí: Paraguarí, Lindman А1887 (P). Pre- sidente Hayes: С 25°12'S, 57738'W, Schinini 26706 (MO), 26764 (MO). San Pedro: Alto Paraguay. Colonia Primavera, Woolston G.66 (NY, SD. G.49 (NY), G.80 (NY), G.105 (NY). URUGUAY. Artigas: Bella Unión, Herter s.n. (US); vicinity of Artigas, Beetle & Ro- sengurtt 1029 (MO). Cerro Largo: Río Negro y Arroyo Palleros, cerca de Melo, Rosengurtt 263 (US). Durazno: Río Yí, Herter 548 (MO, US). Florida: costa del Río Santa Lucía, Lombardo 3050 (US). San José: monte de Santa Lucía, 1-1926, Lombardo s.n. (US). Soriano: Juan Jack- son, Gallinal et al. B-244 (US). Tacuarembó: Pozo, Hon- do, 6 km de Tambores, Cabrera & Zuloaga 32356 (SD. Steinchisma hians es la especie con más amplia distribución y variabilidad dentro del género; es un elemento frecuente en campos húmedos desde Es- tados Unidos de América hasta la Argentina, ca- racterizándose, junto con S. cuprea y S. exiguiflora, por tener la pálea inferior conspicuamente expan- dida a lo ancho a la madurez de la espiguilla. Se distingue de S. cuprea por sus panojas laxas, di- fusas, y de S. exiguiflora por el mayor tamafio de plantas, inflorescencias y espiguillas. 5. Steinchisma кон (Doll) Renvoize, Kew Bull. 42: 921. 7. Panicum үн sum Dóll, еп Mart., ый Bras. 2(2): 241. 1877. TIPO: Brasil: sin localidad, Sellow s.n. bale tipo, B!; isótipos, BAA!, K!, US-81127!). Fi- gura 9 Panicum schenckii Hack., Osterr. Bot. Z. 51: 426. 1901. TIPO: Brasil. Santa Catarina: ltajaí, prope Blume- nau, Schenck 579 (holótipo, W!; isótipos, BAA!; US!). Panicum turfosum Mez, Bot. Jahrb. Syst. 56, Beibl. 125: 1921.TIPO: Paraguay. Alto Paraná: Alto Paraná, 1909/1910, Fiebrig 6471 (holótipo, М!; isótipos, BAA!, G!, K!, Ш, SI!, US-81159!). Planta perenne, herbácea, largamente rizomato- sa, con rizomas hojosos; cañas de 60-120 cm de alto radicantes y ramificadas en los nudos inferio- res, luego erguidas; entrenudos de 3.5-20 cm de largo, glabros, comprimidos, huecos; nudos glabros, comprimidos, violáceos. Vainas de 5-14 cm de lar- go, mayores o menores que los entrenudos, glabras, verdosas a violáceas, con los bordes membraná- ceos, pestañosos en su parte superior. Lígulas de mm de largo, membranáceas, cortamente laciniadas o ciliadas en la parte superior, con pelos por detrás en la base de la lámina; cuello glabro, pajizo. Láminas de 13-35 X 0.3-1 cm, linear-lan- ceoladas, planas o con los bordes involutos, de base angostada, el ápice largamente atenuado, pilosas hacia la base de la cara adaxial, glabras en el resto e la superficie, los bordes lisos. Inflorescencias ter- minales exertas, pedúnculos de 10-30 cm de largo, 650 Annals of the Missouri Botanical Garden A Figura 9. Espiguilla vista del lado de la gluma inferior. — vista dorsal, encerrando 3 estambres. —F. Pálea qud vista ventral, encerrando 3 estambres. D Pálea inferior madura. —H. Antecio superior visto del lado de la lemma. —1. Antecio superior visto del lado de Іа pálea. —J. Flor superior con dos lodículas, tres estaminodios y dos E álea 2. Volume 85, Number 4 1998 Zuloaga 651 С Ағ 84 cilíndricos, glabros; panojas de 10-28 X 2-15 cm, axas, difusas a contraidas; ramificaciones de se- gundo orden divergentes del raquis, alternas u sub- opuestas, lisas, onduladas, las de tercer orden cor- tas, con espiguillas apretadas sobre los ejes; eje principal escabroso, anguloso, pulvínulos glabros; pedicelos de 0.4—2 mm de largo escabriásculos a glabros, claviformes. Espiguillas de 3-3.6 X 0.7— 0.9 mm, largamente elipsoides, glabras, verdosas o con tintes violáceos. Gluma inferior de 1.2-1.6 mm de largo, ovada, % del largo de la espiguilla, de ápice obtuso, 1-3-nervia, el nervio medio escabro- so hacia la porción superior. Gluma superior de 2.2-2.8 mm de largo, llegando o no a cubrir al antecio superior, 3—5-nervia; nervio medio escabro- so. Lemma inferior de 2.8—3.3 mm de largo, 3-5- nervia, de ápice agudo. Pálea inferior de 2.5-3.3 X 0.6-0.8 mm, papirácea, con alas manifiestas, no sobresaliendo a lo ancho a la madurez, los bordes cortamente ciliados a glabros, el ápice obtuso; flor inferior estaminada; anteras 3, de 1.2-2 mm de lar- go, anaranjadas; lodículas 2. Antecio superior de x mm, largamente ovoide, glabro, membranáceo, blanquecino, papiloso, de apice agudo; lemma 3-nervia; lodículas 2, нбай: estaminodios 3, de 0.4—0.6 mm de largo. Cariopsis de 1.6-1.8 X 0.7 mm, elipsoide, castaña; hilo oblongo; embrión menos de la mitad del largo de la cariopsis. Distribución y ecología. Brasil, Paraguay y la Argentina, llegando hasta la provincia de Buenos Aires. Habita en forma de densas matas en bordes de arroyos y ríos, en suelos rocosos, desde el nivel del mar hasta aproximadamente 1000 m s.m Material representativo citado. ARGENTINA. Bue- nos Aires: Pdo. Berisso, Isla Santiago, Cabrera 3368 (C, SI, SP). Distrito Federal: Palermo, Hicken s.n. (SI- 13493). Entre Ríos: Salto Grande, Renvoize et al. 2965 (MO, SI); Dpto. Uruguay, ruta 14, N de Concepción del Uruguay, Troncoso et al. 2396 (SI). Misiones: De Após- toles a Concepción de la Sierra, Arroyo Las Tunas, Zu- loaga et al. 3244* (MO, 51); Jardín América, Zuloaga & Deginani 462 (LP, MO, SI); Dpto. Montecarlo, Puerto Pi- ray, Renvoize et al. 3191 (K, MO, NY, SI). BRASIL. Pa- гапа: Mun. Sáo Jorge do Oeste, Rio Iguacú, Salto Osorio, Hatschbach 20555 (K). Rio Grande do Sul: Pelotas, Cos- ta Sacco 363 (US). Santa Catarina: above Ibirama, Smith et al. 7600 (K, NY, RB); Fazenda Campo Sáo Vicente, 24 km W of Campo Eré, Smith & Klein 13841 (NY, R, SI). PARAGUAY. Alto Paraná: Arala, Montes 9896 (LIL). p Piribebuy, Salto Piraretá, Degen 1369 (MO). í: Salto Piraretá, del & Vervoorst 508 (LIL); eris Cristal, Hahn 2611 (PY, En esta especie se observa una reducción en el tamaño de las plantas en su límite austral de dis- tribución. Es así que los ejemplares de Entre Ríos y Buenos Aires son más pequeños, con inflorescen- cias algo más contraídas. Vega (1996) indicó que existe una particular reducción de las estructuras florales de S. spathellosa; esta autora describió a la especie como diclino monoica, señalando que exis- te una reducción de la fertilidad masculina en las flores superiores de cada espiguilla, conservándose tres estaminodios en forma vestigial. 6. Steinchisma stenophylla (Hack.) Zuloaga & 901. TIPO: Brasil. Goiás: Paranana, 28 Mayo 1895, Glaziou 22534 (holótipo, W!; isótipos, G!, K!, P!). Figura 10. Panicum goyazense Mez, Bot. Jahrb. Syst. 56, Beibl. 125: 4. 1921. TIPO: Brasil. Goiás: sin localidad, Gardner 4067 (holótipo, B!; isótipos, BAA!, K!, US-80683!, foto del holótipo, US-80683!). Plantas perennes, cortamente rizomatosas, con cañas de 30-120 cm de alto, erectas, decumbentes a geniculadas y marcadamente ramificadas en los nudos medios y basales, multinodes, entrenudos hasta de 15 cm de largo, cilíndricos, rígidos, gla- bros, huecos; nudos glabros. Vainas hasta de 11 cm de largo, estriadas, más cortas que los entrenudos, pajizas, persistentes sobre los entrenudos basales, glabras, un borde membranáceo, el restante corta- mente pestañoso a glabro. Lígulas de 0.4—0.7 mm de largo, membranáceas, cortamente laciniadas en la porción superior; cuello glabro. Láminas de 9— 16(-30) cm X 0.2-0.5 mm (con los bordes invo- lutos), hasta 2 mm de ancho cuando abierta, an- gostada hacia la base, el ápice largamente subulado, aquilladas; cara adaxial densa a espar- cidamente pilosa, con pelos largos blanquecinos, más densos hacia la región basal, a glabra; cara abaxial glabra. Inflorescencias largamente exertas; pedánculo de 10—23 cm de largo, cilíndrico, gla- bro; panojas de 2-15 X 2-10 cm; eje principal an- guloso, due iens pulvínulos glabros; ramifica- ciones de primer orden divergentes, ocasionalmente adpresas, sna hacia la base, alternas y dis- tanciadas entre si; ejes de las ramificaciones y pe- dicelos triquetros, escabriásculos; espiguillas sobre ejes de segundo orden hacia la porción superior de las ramas; panojas axilares presentes, similares a la panoja terminal. Espiguillas de 2.4—3.2 X 0.5- 0.8 mm, largamente elipsoides, biconvexas, gla- bras, verdosas o con tintes violáceos; gluma supe- rior y lemma inferior subiguales, o la gluma supe- rior menor que la lemma inferior, dejando al descubierto el ápice del antecio superior. Gluma inferior de 1-1.2 mm de largo, % o poco más del largo de la espiguilla, ovada, aguda a obtusa, 3- 652 Annals of the Missouri Botanical Garden Figura 10. Steinchisma stenophylla (Zuloaga & Morrone 4660, SI). —A. Hábito. —B. Detalle de la región ligular. —C. Espiga vista del lado de la gluma inferior. —D. Espiguilla vista del lado de la gluma superior. —E. Espiguilla, vista latera ` Pálea inferior madura, vista dorsal. —G. Pálea inferior madura, vista ventral. —H. Antecio superior visto del dad de la lemma. —I. Antecio superior visto e lado de la pálea. —J. Pálea superior con tres estambres. --К. Cariopsis, vista escutelar. —L. Cariopsis, vista Volume 85, Number 4 Zuloaga 653 ga eta Steinchisma ЖЕ RUN nervia. Gluma superior de 2.2-3 mm de largo, 7, del largo de la lemma inferior, 5-nervia, obtusa. Lemma inferior de 2.2-3 mm de largo, 3—5-nervia. Pálea inferior de 2.2-2.8 X 0.6-0.8 mm, expan- dida a la madurez y sobresaliendo a lo largo de la espiguilla, papirácea y con márgenes escabrosos; or inferior estaminada; estambres 3; anteras de 1- 1.6 mm de largo, lodículas 2, conduplicadas. An- tecio superior de 2.2-2.8 X ovoide, cartilaginoso, pajizo, verrucoso, finamente escabroso en el ápice; lodículas 2, ca. 0.5 mm de largo, conduplicadas; estambres 3; anteras de 1.2— 1.3 х 0.5 mm, de contorno oblonga; hilo oblongo; embrión menos de 0.6 mm, largamente 1.4 mm de largo. Cariopsis de % del largo de la cariopsis. Distribución y ecología. Se halla en Venezuela, en el departamento de Amazonas, y en Brasil, en cerrados en los estados de Bahia, Minas Gerais y Pará. Crece en márgenes arenosos y rocosos de ori- llas de ríos y arroyos; llega desde 200 m hasta los 1250 m s.m Material d citado. BRASIL. Bahia: Serra do Rio de Conta N of the town of Rio d Contas in flood x of the Rio Brumado, 980 m, Harley et al. 15498 (P); Serra da Agua de Rega, Rio Riachao, ca. 27 km N of Seabra, road to Agua de Rega, 1000 m, Irwin et al. 31028 (MO). Minas Gerais: Rio das oie Glaziou 20110 (P); ca. 18 km W of Сгао Mogol, 950 Irwin et al. 23589 (MO, US); Cardeal Mota, Fazenda Mon. jolos, 19°15'N, 43*40'W, Zuloaga & Morrone 4660 (МО, SI); rodovia де Cardeal Mota a Conceigao do Mato Dentro, BR-010, Serra do Cipó, km 117, Córrego Vitalino, 19°20'N, 43730", 1320 m, Zuloaga & Morrone 4706 (MO, SI). Pará: Caripi, Spruce 63 (M. W), 76 (P). VE- NEZUELA. Amazonas: vicinity of Rio Coro-Coro, near Airport of Yutaje, 250 m, 5°35'N, 66^10'W, Liesner et al. 10956 (MO, NY, VEN), sand island in Cafio Asisa, Cowan E Wurdack 31556 (NY, US); Caño Yutaje, Serrania Yutaje, 250 m, Maguire 32509 (NY, US); Río Orinoco, Río Cu- % mucunuma, Culebra rapids, d et al. 30350 (NY, US, VEN); л Atures, Río Coro-Coro, W of Serranía Yutaje, 6 km N of settlement of 2. M 66°07'W, 320 m, tiie & Holst 21288 (MO, SI); Dpto. Atabapo, Cafio Negro, rfo arriba desde la confluencia con el Rfo Cunucunuma, 200 m, Steyermark et al. 126266 (MO, NY, VEN); Dpto. Atures, Rfo Coro-Coro, Ы of Serranía de Yu- taje, З km N of settlement of Ушаје, 200 m, 5°38’N, 66?07' W, Holst & Liesner 3094 (MO, sl. VEN). Steinchisma stenophylla es una especie afín a S. spathellosa, de la cual se distingue por sus láminas filiformes, espiguillas menores y por tener flor su- perior perfecta, con tres estambres desarrollados. Esta especie posee una distribución disyunta, ha- llándose en cerrados de Brasil y en la Guayana venezolana. Literatura Citada Anton, А. M. & Н. E. Connor. 1995. Floral biology and reproduction in Poa (Poeae: Gramineae). Austral. J. Bot. 43: 577—599. Arber, уд 1934. Тһе Gramineae. Cambridge Univ. Press, Cambri pr Bouton, J. H rown, J. K. Bolton & R. Campagnoli. 1981. doe. + of grass species differing in car- bon Ta fixation pathways. Pl. Physiol. (Lancaster) 67: 433—4 Bremer, K. + The limits of amino acid sequence data in DT phylogenetic reconstruction. Evolution 03. 42: i: 994. ME support and tree stability. Cladis- tics 10: 295- Brown, R. H. "e WM Brown. 1975. Photosynthetic char- 5. of Panicim milioides, a 2. with reduced photorespiration. Crop Sci. (Madison) 1 . Bouton, P. T. Evans, H. E. 7. & L L. Rigsby 1985. Photosynthesis, morphology, leaf anato- my, and cytogenetics of s asma, с, апа С/С, iol. (Lancaster) 77: 653—658. Brown, W. V. 1948. A Де урна, study in the Gramineae. Amer. E Bot. 35: 382-395. . Chromosome numbers pe some Texas grass- es. Ball, Ж чеч Bot. Club 78: 292— he Kranz syndrome d its 22. іп grass udi dud Mem. Torrey Bot. Club 23: 1-97. Clayton, W. . A. Renvoize. pam . graminum. ew Bull., Additional Ser. 13: Clifford, H. T. 1961. Floral E uin in Фе family Gra- mineae. Evolution 15: 455-460. Davidse, G. 1994. Panicum. Pp. 302-318 en G. Davidse, M. Sousa S. & A. O. Chater (editores), Flora Mesoa- mericana, Vol. 6, Alismataceae a Cyperaceae. Univer- sidad Nacional уинн de México, México, D.F; Missouri Botanical algae St. Louis; The Natural His- tory Museum, London. R. W. Pohl. 1972. Chromosome numbers and notes on some Central American grasses. Canad. J. Bot. 50: 1441-1452. rich ap & F. O. Zuloaga. 1991. Nümeros cromosó- micos de especies nopee ricanas de psc. (Poa- eae: {кын Bol. Soc. Argent. Bot. 2 Ellis. R. P. 1976. A procedure = andan compar- ative leaf anatomy in the P e. I. The leaf bl: ^d as viewed in Epia section. : Bothalia 12: 65- 19 е for standardizing compa ә leaf Do pe Poaceae. IL e EDEN as seen in surface is Bothalia 12: 64 Leaf anatomy and 22... | js ath (Palicaae: Panisoddsas fric — lo. = 86. Reference, version 1.5. Ј. 1 ‚ Port Jefferson Station, New Goloboff. P 1993. NONA vesión 1.6, Pro grama y docu- mentación е por el autor. PU. Argentina. Gould, F. W. 1968. Chromosome numbers of Texas grass- es. Canad. d Bot. 46: 1315-13 R. B. Shaw. 1983. Grass Systematics, 2nd. ed. exas A & M Univ. Press, College Sta Hitchcock. A. S. & A. Chase. 1910. The North American species E Pao Contr. U.S. Natl. Herb. 15: 1-396. 5. Tropical North American species of Pa- nicum. verd U.S. Natl. Herb. 17: 459-539. Hsu, C. C. 1965. The classification of Panicum (Grami- Annals of the Missouri Botanical Garden neae) and its allies, with special reference to the char- acters of lodicule, style-base and lemma. J. Fac. Sci. о = a с == © Ф . Kashiwagi. 1975. Panicum milioides, a gramineae plant having Kranz leaf Э он Се photosynthesis. Pl. Cell Physiol. 16: 669—6 Karis, P. O. 1995. Cladistics of the subtribe Nee EON 72 Si Bot. 20: 40-54. . S. B. € G. E. Edwards. 1978. Photosynthetic ef- гор of Panicum hians and Panicum milioides in m to C, and C, plants. РІ. Cell Physiol. 19: 665- Eum R. Ka anai. 1976. Distribution of enzymes relat- ed to C, and C, гаг of photosynthesis between mesophyll and b an in pend ides vocate € 50. Anatomy of Monoc vois еони. I. Gra- ae. NT Univ Press, Oxford. Morgan. J. A R. H. Brown. 1979. 1. іп grass species you in carbon dioxide fixation path- ways. $^ кые 2. 257-262. Мо Bs „ЈН. ziker, F. О. — & A. Escobar. К Мени а Os en Pan eae sudameri- anas 5 (Parca ee Darwinian 33: 53-60. Nash, G. eae. Pp. 49- en J. K. Small аш A of De i ма vehe States. New Nixon, к. С. 1993. Clados, versión 1.38 [32 bit] beta ver- sion. Programa de с ЕМ ión distribuido por el autor. & J. 1. Davis. 1991. Polymorphic taxa, missing values and cladistic aaa Cladistics 7: 233-241. & J. M. Carpenter. 1993. On outgroups. Cladistics 9: 413-426. Núñez, O. 1952. Investigaciones cariosistemáticas en las gramíneas argentinas de la Uds Paniceae. Revista Fac. Agron. Univ. Nac ne eR -255. Ні a 1985. Sompatie С. C s in 1. Ann. B Parodi. L .R 946. Gramíneas bonariensis. Clave para la 2 de los géneros y enumerac ión de las es- 08 па ten Gattungen. Notizbl. Вог. Сап. Berlin- Dahlem 11: 240—245. 940. hw did III: Unterfamilie Panicoideae. .E . Ргап per Die natürlichen Planzenfamilien Ed. 2, 14e: 1-208. Engelmann, Leip- Pohl. E a & G. Davidse. 1971. Chromosome numbers о 221 => Brittonia 23: pou 324. к S. A. 1982. ew genus and several new spe- = of grasses from Da (Brazil). Kew Bull. 37: 323- —— New grasses from Paraná, Brazil. Kew Bull. ts #0 –925. Sass, J. E. 1940. Elements of Botanical Microtechnique. McGraw Hill, pu e and London 1 р 1977. Contrad 70, an effective softener of herbarium material for anatomical study. Taxon 26: 551-556. Sendulsky, T. & T. R. Soderstrom. 1984. Revision of the South American genus Otachyrium (Poaceae: Panicoi- deae). Smithsonian Contr. Bot. 57: Simon, B. K. & C. M. lined 1995. (НЕ а new grass genus (Po oaceae: Paniceae) dn south-eastern Queensland. 224 369-3 Urbani, M. 1990. Citología y método de reproducción de Panicum elephantipes (Gramineae). Bol. Soc. Argent. Bot. 26: 205-208 Vega, A. S. 1996. Biología reproductiva de Panicum spat- hellosum (Poaceae: Panicoideae: Paniceae). Darwiniana : 199-211. Watson, L. . 1992. The Grass Genera of the World. CAB та ЙЫ Press, Cambridge. Webster, R. D. 1988. Genera of the North American Pa- eae (Poaceae: Panicoideae). Syst. Bot. 13: 57 Zuloaga, F. O. 1987a. Systematics of the New World spe- cies of Panicum (Poaceae: Paniceae). Pp. 287-306 en . Soderstrom, K. W. Hilu, C. S. Campbell & M. E. Barkwarth (editores), Grass . and Evolution. Smithsonian Institution Press, Washington, 87b. A revision of Panicum subgenus Panicum section 2. (Poaceae: Paniceae). Ann. Missouri Bot. Ga 5 74: 463-478. . Morrone. 1996. Revisión de las especies 5. de Рапісит subgénero Panicum sección Panicum (Poaceae: Panicoideae: Paniceae). Ann. Mis- souri Bot. Gard. 83: 200—280. . Ellis & O. Morrone. 1992. A revision of Bano um “subgenus Phanopyrum section Laxa (Poaceae: Panicoideae: Paniceae). Ann. Missouri Bot. Gard. 79: 770-818. — ———, J. Dubcovsky & O. Morrone. 1993. Infrageneric phenetio sen in genus Panicum (Poaceae: Pani coideae: Paniceae): A numerical analysis. Canad. J. Bot. 71: 1312-1327. Indice de colecciones numeradas Cada especimen es citado por el a pelli do del s primer кА ие еп el sm Steinchisma hi 5. Steinchisma spathellosa (Dall) ia oize 6. Ve pe Ар Minis ) Md & Morrone Abrell 52 (5); Aguilar 746 (4), 999 (4); Ahumada, O. 17 (4), 785 (4), 1151 7 ke (4), 1493 (4), 2023 (4) 2274 (4), 2752 (4), 3201 (4); Allard, S. қ; 14276 (3); Allem, А. 1306 (4), dd ), 1861 (4), 1914 (4); Alonso, J. 37 (4); Alston, | | С. 7578 (2); үң А. 1086 (6); Anderson, W. R. о hg d (2), 3639 ^ As Araujo 359 (4); Arbo, M. и гзепе, Bro. С. 3221 (4), 5314 (4), 5017 (4); Posila; A. 1029 (4). 1439 (4), 1959 (4), 7323 (4), 7708 (1); Bertoni, M. 2290 (5), 4629 (5); Black, С. A. 2152b (2), 51-10985 (2), 51- 11069 (2); Boege 2543 (1); Boelcke, O. 1483 (4); Brace, L. J. K. 4164 (3), 4380 (3); Breedlove, D. E. 54947 (4); Britton, N. L. beue (3), 14699 (3), 15381 (3); Brown, R. ; . А. E. 26 (5), 271 (5), 285 (5), 1721 (5), 3349 (4), "3024 (5), 5016 (4), 12760 (4), 12953 (5), 14063 (4), 15040 (5), 17513 (4), 18584 (4), 20173 (4), 20296 (4), 21597 (4), 22877 (4), 24090b (4); Burman, A. G. 87 (2), 645 гађтега, A. L. 3368 (5), 5392 (4), 7417 (4), 26569 (4), 27321 (4), 28161 (4), 28368 (4), 28516 (4), 28629 (4), 29066 (4), 32356 (4), 28631 (5), 28634 (2), 28666 (5), 28934 (5), 29472 (5), 32427 (2); Сапо 1846 (4); Сћаве, А. 8068 (4), 8073 (4), 8118 (4), 8591 (2), 8634 (2), 8727 (2), 8969 (4), 9105 (6), 9417 (2), 10279 1^ (2), 10283 (2), Volume 85, Number 4 Zuloaga et al. Steinchisma (Poaceae) 655 10286 (2), 10332 (2), 10501 (2), 10847 (4), 10854 (2), 10925 (2), 10941 (4), 10990 (4), 11242 (2), 11759 (2), 11815 (6); Clayton, D. W. 4284 (2), 4473 (4); Coro 1373 8888 (3), 41685 (3); Corte 1026 (4); Cowan, R. S. 31556 (6); Cristó- bal, C. 47 (4); Cuatrecasas, J. 23035 (4); Curtiss, A. H. 5534, 6836 (4); Cutler, D. 2005 (2). Davidse, G. 10597 (2), 10613 (2), 11035 (4), 11157A (4), 30617 (4), 31539 (4); Degen, R. 1369 (5); Dombrow- ski, L. T. D. 443 (2); Dusén, P. 279a (2), 3658 (2), 13230 (2), 17571 (2); Dutra, J. 422 (2). Eiten, G. 1973 (2), 2667 o 4522 (4), 4773 (4), 4776 (4), 5093 (4); Ekman, E. L. 572 (3), 573 (3), 639 (5), 2443 (3), 4916 (3), a (3), 7400 (3), 7580 (3), 12259 (3), Amer. Gr. Hb. 702 (3). Fiebrig, K. 7 (2), 5403 (5); Filgueiras, Т. S. 2078 (2), 2169 (2), 3014 (2 Galli, I. 242 (4); Gallinal B-244 (4); García Barriga, H. 10773 (4), 15552 (4); García 2 (2); Gautier 4 (4); Glaziou, A. F. 15620 (2), 16619 (2), 17639 (2), 20110 (6), 22535 (2); Guaglianone, E. R. 261 (4), 1170 (5); Guzmán 956 (4). , W. J. 2533 (2), 2611 (5); iie R. 15498 (6), bos 9), 19384 (2), 19647 (2); Hartley 118 (4); Hassler, 123 (4), 2694 (4), 3658 (2), 10110 y 10783 (4), 11643 (4), 11907 (2), 12347 (2), 12433 (4); Hatschbach, G. 2875 (2), 4321 (2), 12888 (2), 15169 (5), 15238 (2), 20555 (5), 22820 (2), 32995 (2), 33573 (2); Herter, d 548 (4); Hicken, C. M. 5450 (5); Hitchcock, A. S. (3), 23241 (3), 23278 (3), 23356 (3); Holst, B. ee ls Howard 4343 (3); Huber, O. 4322 (6). Ibarrola, A. 361 (4); Idinael 597 (4); Imaguire 1207 (2); Irwin, Н. S. 9751 (2), 12591 (2), 19571 (2), 23589 (6), 25628 (2). E (6). 7 (4); Joergensen, P. 2439 (4), 3543 (4); Joly, A. B. a Killip, E. P. 32314 (3 ), 43980 (3); Klein, R. M. 3593 (2). 5900 (2). 2. (2), 11875 (5); КопіпсК, 4е 121 (4); 652 (4), 24553 (4), 24666 (2), 25716 Y . C. 1925 1); León, Bro. 911 (3), 5608 (3), 6377 (3), 6654 (3), 7008 (3), 9277 (3), 14690 (3), 18579 (3), 20249 (3); Liesner, R. 10956 (6), 17513 (6), 21288 (6); Lillo, M. 3921 (4); Lindman, C. A. M. A1887 (4); Lombardo, A. 3050 (4); Lorentz, P. G. 615 (4); Lossen, W. 298 (4); Luna 883 - 3 (2); Maguire, В. 30350 (6), 32509 (6); joue Ce R. 4351 (4), 4835 (4); Mat- - E 128 ; Meyer, T. 364 (4), 5439 (2); Molina 7 (4); Mad 1745 (4); Montes, J. E. 612 (4), 3305 ip БА» (4), 9896 (5), 11054 (4), 15268 Кү? 15325 (4); Morel, I. 707 (4), 2128 (4); Morello, J. 375 dam O. 444 (2); Morton 2996 (3); Mülgura, M. E. p 3 Nash, С. V. 213 (4), 1331 S, 1450 (3); Node Bro. 217 (1), 5879 (1); Nicora, E. 604 (4), 2670 (4), 3043 (4), 4659 (4), 17694 (4); . С. 2069 (2). T Parodi, L. R. 3943 (4), 5648 (2); Pedersen, T. M. 1297 (4), 1374 (4), 7092 (2), 10240 (4); Pereira 6730 (4); Pflanz, C. 4016 (4); Pickel, D. B. 5198 (2); Pinto, P. 279 (4); Pires, J. M. 9676 (4); Pohl, R. W. 12017 (4), 12162 65 (4); Praderi 536 (4); Prance, G. T. P24905 (6); Puiggari, J. I. 3245 (2) — = R. 6157 (2), 6959 у (5), 7888 (2). 11955 (4 ); Бейне, s. A. 2965 (5), 2967 СА aor (5), 3151 (2), 3191 (5), 3629 (4); Reverchon, Ј. ríguez 74 (4), 378 (5); Rojas, T. 113 (4), a 48 T Hoana 204 (2); Pass Ре B. B-263 (4), B-5379 (4), B-5505 (4); Rúgolo, Z. E. (2); Rzedowski, J. 1024 (1), 48632 (1). Schinini, A. 6294 (4), 7434 (2), 8323 (2), 12796 (4), 12818 (4), 13177 (2), 25780 (4), 26706 (4), 26764 (4); Schulz, A. G. 3845, 5779, 10310 (4), 14847 (4), 15768 (4), 17462 (4); Schwacke, C. А. W. 13790 (2); Schwarz 5116 (4), 6751 (4), 7675 (4), 10726 (5); Schwindt 914 (4); Sendulsky, T. 146 (2), 418b (2), 1883 (2); Sergues 2566 (3); Scipione 264 (4); Shafer 10742 (3); Sharp 44753 (1); 2” va, Т. 263 (6), 436 (2); Skvortzov, B. 64 (2); Smith, L. 7600 (5), 8294 (2), 4), 9947 (4), 10928 (4), 1 (4), 11934 (4), 12594 (5), 13064 (4), 13841 (5), 13985 (2), 15489 (2), 15626 (2), 15584 (2), 16048 (2), 2 (4); Sparre, B. 508 (5), А 6 (6); St. Hilaire, А. 416 (2); s (4). T": би 24089 (4), 26636a (4), ; Steinbach, J. 2 2); Steyermark, J. 32077 (4), 48148 (4). 126198 (6), uA s ); Stuckert, T. 16812 (4), 18736 (4), 21459 (4), 21494 (4); Swallen, J. R. 3849 (5), 7098 (4), 7164 (2), 7613 (4), Ne (2), Tamayo, F. La (2), 3209 Y (2); p 4889 (4); Tra- , 5 (3); | a S. G. 939 (2), 2848 (4); Haus: I J. AS ); Troncoso, N. 5 2728 (4), 2755 (4), 3412 (4). (Ле, Е. 1961 (2). Valla, J. Т 63 (4); Valls, J. F. 1721 (4), 2326 (2), 4761 (2), 4769 (2); Venturi, S. 160 (4), 480 6), 1635 (4), 2629 (2), 2727 (2), 2785 a ee (2), 9932 Williams, T. A. 22091 (4); Woolston, E C.66 (4 ), G.49 (4), G.80 (4), G. Te y? Wright, C. 756 (3), 3450 (3). Zardini, E. M. 7443 (5), 9038 (2); Zuloaga, F. O. 13 (4), 425 (2), 462 (5), 544. (2), 1985 (4), 2330 (4), As (4 ), 3074 (5), 3094 (2), 3095 (2), 3108 (4), 3161 (5), 3214 (2), 3244 (5). 3305 (2), 3314 (4), 3617 (4), 4179 (4), 4660 (6), 4706 (6). APÉNDICE I. LISTA DE MATERIAL ANALIZADO PARA EL Es- TUDIO CLADÍSTICO. Panicum subgénero Panicum sección Panicum P. bergii. ARGENTINA. Buenos Aires: Krapovickas 2906 (MO, SI). Chaco: Renvoize et al. 3583 (MO, SI). BOLIVIA. La Paz: Beck & Haase 9921 (K). P. capillare. CANADA. Nova Scotia: Fernald & Long 19754 (US). ESTADOS UNIDOS DE AMÉRICA. Arizona: Kearney Peebles 12868 (US). P. capillarioides. UNIDOS DE AMÉRICA. Texas: Tharp 43066 (MO). MÉXICO. o León: Hitchcock 5547 (US). San Luis Potosí: Rzdewosky 4618 (US). P. hallii. ESTADOS UNIDOS DE AMÉRICA. Arizona: Hitchcock 3706 (US); Griffiths & Thornber 238 (US). MÉXICO. Agu i (MO). CUBA. Cienfuegos: Jack NY, na: León 14181 (MO, US). Р асып COLOMBIA. Magdalena: Smith 2152 (F, С, СН, К, MO, NY, Р, US). COSTA RICA. Guanacaste: E & Davidse 11291 (F, MEXU). HONDURAS. Cortés Nelson et al. 5621 (MO). P. lepidulum. GUATEMA XIC ientes: Reeder ee E, > CESA. Feuillet 4442 (US). HONDURAS. Morazán: Ко- dr(guez 3512 (US). 656 Annals of the Missouri Botanical Garden Panicum subgénero Panicum sección Dichotomiflora P aquaticum. BRASIL. Davidse et al. 12034 (MO). MÉ- CO. Quintana Roó: Sousa 11207 (MEXU). P. barto- wense. BAHAMAS. Grand Bahama: Correll & Popenoe 46671 (MO). P. dichotomiflorum. VENEZUELA. Anzoá- tegui: Burkart 17289 (SI). — ag od 66 (US). P. elephantipes. ARGENTINA. Bue Aires: Zuloaga 3084 (SD. BRASIL. Mato Grosso: “Rondon s.n. (SD. P. sublaeve. EL SALVADOR. La Libertad: Pohl 12774 (MO). MÉXICO. Chiapas: 7. & Davidse 54548 . P. vaseyanum. MÉXICO. Aguas Calientes: Hitch- cock 7491 (US), 7485 (US. Panicum subgénero Panicum sección Rudgeana P. campestre. BRASIL. Mato Grosso: Chase 10790 (RB). Sáo Paulo: Lofgren 212 (RB). P. cayennense. COLOM- BIA. Cauca: Lehmann 5268 (US), 5269 (US). MÉXICO. Campeche: Reeder & Reeder 6111 (MEXU). GUATE- MALA. Izabal: Le Doux et al. 106 (NY). P. cervicatum. VENEZUELA. Amazonas: Huber 862 (MO), 1351 (MO). P. ligulare. BRASIL. Distrito Federal: Filgueiras 1261 (SI). P. rudgei. COSTA RICA. Puntarenas: Zamora et al. 1184 (MO). VENEZUELA. Amazonas: Maguire 29424 (NY), Cowan 31486 (US) Panicum subgénero Panicum sección Urvilleana P. chloroleucum. ARGENTINA. Catamarca: Cabrera et E = 3 т е Cabrera et d 31731 (SI). P. race- . ARGENTINA. Buenos Aires: Cabrera 5535 (51); Zuloaga res (SI). P a ARGENTINA. s Aires: Cabrera 6620 (SI). Mendoza: Boelcke et al. 15797 (5 I). Panicum subgénero Panicum sección Virgata — — Ф P. amarum. ESTADOS UNIDOS DE AMÉRICA. Ala- bama: Deramus 726 (MO). MÉXICO. Campeche: Saver et al. 3350 (US). P. glabripes. ARGENTINA. Corrientes: Tressens et al. 414 (SI). BRASIL. Paraná: Jonsson 1124a SI). P. tricholaenoides. ARGENTINA. Chaco: Joergensen 2430 (MO). Corrientes: Schinini 16095 (SI). P. virgatum. MÉXICO. Chiapas: Шарық 9199 (MO). Quintana Roó: Durán & Егреје! 551 (M ~ Panicum subgénero Phanopyrum sección Laxa P. hylaeicum. GUATEMALA. Alta Verapaz: von Tuerck- heim 1254 (US). HONDURAS. Comayagua: Williams & Williams 18435 (US). MÉXICO. Chiapas: Breedlove & Davidse 54040 (MO, US). P. laxum. ARGENTINA. Bue- nos Aires: Parodi 4662b (BAA). BOLIVIA. Beni: Beck 3228 (SI). BRASIL. Acre: Croat & Rosas 62653 (SI). P. pilosum. ARGENTINA. Corrientes: Krapovickas et al. 24307 (CTES, SI). BRASIL. Amapá: Rabelo et al. 3319 (MO); Curucá Ore Body, Cowan 38177 (NY). COSTA RICA. Alajuela: Pohl & Davidse 11254 (US). P. polygo- natum. BOLIVIA. Cochabamba: Steinbach 484 (MO, NY), Cárdenas 700 (US). 7703 (M, MO, P). COLOMBIA. Amazonas: Black & Schultes 46—122 (US). 21. Gutiérrez & Barkley 17C172 (LIL, SI, US). P. stevensianum. BRASIL. Per- nambuco: Chase 7717 (US). COLOMBIA. Casanare: Blydenstein s.n. (51, US). CUBA. Habana: Ekman 11516, 13093 (US) Plagiantha tenella. BRASIL. Bahia: Harley et al. 16769 (K), Pinto 127/80 (US); Zuloaga et al. 4773 (IBGE, MO, SD, Zuloaga et al. 4775 (SI), Zuloaga et al. 4808 (51), 4813 (SI). ANNALS OF THE MISSOURI BOTANICAL GARDEN: CHECKLIST FOR AUTHORS 1. General Instructions O Text is in English or Spanish on numbered pages. O Manuscript is typed on one side of nonglossy 8% x 11 in. paper. O At least 1 in. is left as margin all around, except on the first page, which has 3 in. left blank at the top. Three copies of double- or triple-spaced printed manu- script, including abstract, legends, tables, specimen lists, Literature Cited, and footnotes, are enclosed. Manuscript is also submitted on MS-DOS/WINDOWS 3% in. diskette, as WordPerfect? [preferable] file. Right-hand margin is not justified, and words are not broken there. Special typefaces are not used. Common Latin words or phrases are not italicized (e.g., et al., i.e., etc.). Only names at the rank of genus and below are itali- cized. Correct accents, umlauts, and other diacritical marks have been included. All figures and tables are cited in the text and are numbered in the order in which they are to appear. Photocopies of the figures are included with each copy f the manuscript. "E dec ts sensu, Cd hf 2. Style Recent issue of the Annals is used as a model. Vien Manual of Style, latest edition, is used as a reference. [343 3. First Page O Footnotes are typed as double-spaced paragraphs on the first page. The first footnote contains acknowledg- ments, including information on granti herbaria that loaned specimens, and t ng agencies, e name of the otes. No other footnotes occur, except in tables, where unavoidable. 4. Abstract (ЈА опе- paragraph abstract precedes text. Papers in n English abstract in addition to a (1 para agraph) and includes brief statements about ihe papers intent, materials and methods, results, and significance of findings. (= Ег Ф ® E. Л E w б > т © © 5 EN т о 5. Tables O Tables are neat, double-spaced, and easily understood rather than long and com [0 Tables do not contain ve or horizontal lines. [Ed- itor and/or printer will add them as needed. Captions are typed double-spaced as paragraphs at the tops of the tables. O Each table starts on a separate sheet and is double- spaced. 6. Abbreviations L] Periods are used after all abbreviations (which are сне хсері metric measures, compass direc- and herbarium designations. tes are given as part of collection information, three-letter month abbreviations are used, except for months with four letters, which are spelled out in full. О States are not abbreviated, and cities are spelled out. [St., n St. Louis, is acceptable O Periodicals are abbreviated am 776 to B-P-H (Bo- tanico-Periodicum-Huntianum) and to B-P-H/S (Bo- tanico-Periodicum-Huntianum/Supplementum ). [] Authors’ names are abbreviated according to Brummit & Powell's Authors of Plant Names [0 Book titles are abbreviated according to Taxonomic Lit- erature, edition 2, but with initial letters capitalized. Book titles are spelled out in the Literature Cited O Herbaria are abbreviated according to Index Herba- riorum, edition eight (8) 7. Taxonomic Treatment L] Species entries are organized as follows: Heading, ver- nacular name(s), Latin diagnosis (if necessary), de- sc ription, distribution, summary, 5 'imens examined, discussion. The discussions are жылк and follow the same order, e.g., diagnostic characteristics, ipe li from similar species, variation, distributi ogy, nomenclature and typification, uses. [0 One paragraph per basionym is pe as follows: Taxon author, literature citation, type citation, e.g., Pleuroth- yrium amplifolium (Mez) Rohwer, Mitt. Inst. Allg. Bot. Hamburg 20: 43. 1986. Nectandra amplifolia Mez, Ar- beiten Kónigl. Bot. Gart. Breslau 1: 131. 1892. TYPE: Brazil. Rio de Janeiro: Alto =. sahé, Glaziou 17731 (holotype, B; isotypes, B, G, K, NY, P). [0 Lectotype designations are ты together with an ion and e mitted is making the lectotypification, the бане" "ћете designated" is use O Exhaust points are used for specimens аш and ше not seen are indicated as such (e.g US not se O A brief ed diagnosis for each new taxon is provided rather than a complete Latin descriptio For species with infraspecific taxa: To and discussion are composite (incorporating all infraspe- cific taxa) and parallel with other species descriptions. Descriptions of infraspecific taxa are parallel with one another (in the same species). All synonyms are listed under the appropriate infraspecific taxon O Descriptions: Descriptions are parallel, within a given rank. All measurements are metric. Hyphens are used for parenthetical extremes: “peduncle (8.2-)14.3- 658 Annals of the Missouri Botanical Garden 28.0(-31.9) cm long," unless intermediate values are not expected: ovary with (2)4(6) locules. Lengt width are е in the following manner: lamina 36.4— 82.8 X .8 ст. When 5. nomina nuda, misapplied names апа excluded names are included in the discussion follow- ing the description, or at the end of the paper, but are not part of the formal synonymy. O 8. Specimens Examined O If many ко, were examined, those cited in the text are limited to ca. 1% manuscript pages. An index to specimens examined is placed at the end of the paper, following the Literature Cited. It is ar- ranged alphabetically by collector, followed by collec- tion number, followed by the number of the taxon in the text. Names (including initial(s) of first а: second collector are provided, “et al.” if three or mo Specimens are cited in the text as follows: Additional specimens examined (or Selected specimens examined) MEXICO. Oaxaca: Sierra San Pedro Nolesco, Talea, 12?37"N, 85°14" W, 950-1100 m, З Feb. 1987 (fl), Jer- gensen 865 (BM, G, K, US). [Dates and reproductive status are optional but are omitted from longer lists.] Countries are run together in the same paragraph, e.g., COUNTRY А. Major political MON . . COUN- TRY B. Major political division . Separate par- · used for major nta regions within 21 ES hs are major political divisions. 9. Specimen Vouchers and Genetic Sequences O If paper presents original data, associated herbarium dent on the paper, reference to the original wild source may be required.] Vouchers are also cited from com- mon names and uses taken from specimen labels. Herbarium vouchers state the collector and number, herbarium in which the voucher is located, and a clear annotation that the material represents the voucher for the study in question. Nucleic acid or protein sequences corresponding to equal or greater than 50 nucleotides are entered into an appropriate cais bank accession numb [Long sequences еш 0 two pages) will not be rou- tinely publishe Author accepts АВТ ТЕ for establishing the ac- curacy of information provided. О О О 10. Keys Keys are clear and have been checked carefully for consistency with the descriptions. Leads of each cou- plet are parallel. O Dichotomous keys are indented. O Infraspecific taxa are keyed separately, not in species keys. 11. Literature Cited O The Literature Cited contains all references cited in the text O АП entries in the Literature Cited are cited in the text. O Spelling of author(s) name(s) and years of publication have been double-checked. All entries have been verified against original sources, especially journal titles, accents, diacritical marks, and spelling in languages other than English O Periodicals are listed as follows: author's last name, initial(s). Year. Full title of article. Journal abbreviated as in B-P-H/S. Volume: pages. No parenthetical part numbers after volume numbers are given unless each part is paginated separately. O For more than one author, this style is followed: au- thors last name, initial(s), second иеде initial(s), last name & third author's initial(s), last name. L] Books appear as follows: author's A. name, initial(s). Year. Full Unabbreviated Title (edited by Editor), 3rd ed., Vol. 2. 2 City of Publication O Citations of work “in prep.,” unpublis hed theses and dissertations, and EE references to inaccessible sources have been eliminated or kept to a minimum. They are not included in the Literature Cited. O 12. Illustrations O Author(s) name(s) and figure number(s) are written in pencil on the back of each figure or plate, on both originals and review copies O Scale bars appear on illustrations, photographs, and m O Magnifications/reductions are not indicated in cap- tions All laive materials are mounted on stiff card- oard no thicker than !4 in. О ration are presized to fit either column width (2% in. or ca. 68 mm) or full page width (5% in. or ca. 140 mm), or eer no larger than 11 X 16 in. (= 29 X 40 cm). [Maximum size for printed illustrations is 5% x 8% in.] =н or delicate figures are submit- ted as photost O Figures are ху Е in Arabic numerals іп the order of their citation in the text. Parts of figures are labeled with capital letters. [.] Photographs are crisp black-and-white prints. L] Figures are grouped into composite plates when pos- sible; edges of photographs are abutted. L] No stripping is inserted between plate or figure seg- | O O O ents. [Printer will insert stripping Edges of figures are squa Maps include reference to s adde and longitude and are bounded by a fine border. [0 Scanning electron micrographs are free of conspicuous charging. O Axes on graphs are all labeled O Captions provide all explanatory text. No text appears on the figures. Captions are separate from other text, one paragraph for each group of figures, and following the style in current issues of the Annals О Symbols on maps are legible and reduction has been taken into consideration. Electronic Artwork—Printer's Specifications 1. We accept line art, halftones and color figures on disk or via electronic delivery. Please include a printout or email of the file directory that includes the file name(s), size and kind of file . You should discuss figure sizing specifications with the journal prior to disk submission, and the figures on disk should be the final size desired. . Graphic files should be submitted in TIFF or EPS ipe: format. Other applications that are acceptable ar follows: Adobe Photoshop; Corel Draw. Adobe usa: tor; Adobe Freehand. . All line art should be scanned at 1200 d.p.i. All grays- cale and color images should be scanned at 350 d.p.i. Color images should be submitted in CMYK mode, not RGB or Indexed Color mode . Include the screen and printer fonts used within the graphic file. Use only Adobe Type 1 postscript fonts. True Type fonts and System “bitmap” fonts should not be used. . If the files have been compressed, please indicate what compression program has been used. . The disk submission should be accompanied by a hard copy of the figure that exactly matches what is on the disk in the event that the disk is unuseable. ~ w ~ сл с - Acceptable Electronic Artwork Media 3%" floppy disks; Syquest cartridges; EZ Flyer; ZIP disks; JAZ disks; Magneto Optical disks; CD ROM; Artwork may also be submitted via email attachment or КТР. Please contact Tara Schuley, e-mail address: tschuleyO allenpress.com for more information. 13. What to Submit O Full mailing and email addresses, telephone and fax numbers are submitted in cover letter. O Three copies of manuscript are bad О Copies/photocopies of original figures are enclosed with first submission. Copies of half-tone figures (pho- tographs) and plates must be of sufficiently good qual- ity to show all the detail of the originals. Digital or dye-sublimation proofs of electronic graphic files are submitted. [Send originals or photostats of originals when requested—usually when manuscript is accept- [1 An MS-DOS/WINDOWS 3% in. diskette, as Word- Perfect? ш file, is submitted in addition to the printed documen SCOPE OF THE ANNALS The Annals publishes original articles in systematic bot- any and related fields. Papers whose purpose is the estab- lishment of new nomenclature entities in vascular plants and bryophytes are not accepted. Rather, they should be submitted to Novon for consideration. (Novon instructions are available from the managing editor. CONDITIONS FOR PUBLICATION Authors are urged to contact the managing editor prior to submitting manuscripts. Manuscripts sent already re- viewed and revised are generally subject to the full normal review process initiated her e. Manuscripts not properly having access to crucial inne ideale Communi- cation with the editor at all points of the publication pro- cess is encouraged. PAGE CHARGES Page costs are $80 per page, although charges may be reduced under certain circumstances. Corrections in proof will be billed nonnegotiably to the author at $3 per line changed. CORRESPONDENCE my McPherson, Tel: bey 577-5112 e anaging Editor Missouri Botanical Garden Press Department 1 amcpherson@ P.O. Box 299 rschctr.mobot.org St. Louis, MO 63166-0299, Updated 10/98 U.S.A Volume 85, Number 4, PP. 531-659 of the ANNALS OF dg bor ud JRI BOTANICAL GARDEN as published on December 31, ANNALS OF THE MISSOURI BOTANICAL GARDEN VOLUME 85 1998 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. This volume has been printed on 704 Vintage Gloss. This is an acid-free paper designed to have a shelf-life of over 100 years. Vintage Gloss is manufactured by the Potlatch Paper Company. Photographs used in the ANNALS are reproduced using 300 line screen halftones. The binding used in the production of the ANNALS is a proprietary method known as Permanent Binding. The ANNALS is printed and distributed by Allen Press, Inc. of Lawrence, Kansas 66044, U.S.A. € Missouri Botanical Garden 1998 ISSN 0026-6493 VOLUME 85 AEDO, CARLOS, JUAN JOSÉ ALDASORO & CARMEN NAVARRO. Taxo- nomic Revision of Geranium Sections Batrachioidea and Divaricata (Ger- aniaceae) AEDO, CARLOS. (See Juan José Aldasoro, Carlos Aedo & Carmen Navarro) ALDASORO, JUAN JOSÉ, CARLOS AEDO & CARMEN NAVARRO. Pome Anatomy of Rosaceae Subfam. Maloideae, with Special Reference to Ру- rus ALDASORO, JUAN JOSÉ. (See Carlos Aedo, Juan José Aldasoro & Carmen avarro ALISCIONI, SANDRA S. (See Osvaldo Morrone, Fernando O. Zuloaga, Mirta O. Arriaga, Raúl Pozner & Sandra S. Aliscioni) ANGIOSPERM PHYLOGENY GROUP, THE. [Káre Bremer, Mark W. Chase, eter F. Stevens et al.] An Ordinal Classification for the Families of Flowering Plants ARRIAGA, MIRTA. (See Osvaldo Morrone, Fernando O. Zuloaga, Mirta O. Arriaga, Raúl Pozner & Sandra S. Aliscioni) AUSTIN, MIKE P. An Ecological Perspective on Biodiversity Investigations: Examples from Australian Eucalypt Forests BASKAUF, CAROL J. & SHARON SNAPP. Population Genetics of the Cedar- glade Endemic Astragalus bibullatus (Fabaceae) Using Isozymes ______ BAYER, RANDALL J. € JULIAN R. STARR. Tribal Phylogeny of the Aster- aceae Based on Two Non-coding Chloroplast Sequences, the trnL Intron and trnL/trnF Intergenic Spacer BENÍTEZ DE ROJAS, CARMEN & WILLIAM G. D'ARCY. The Genera Ces- trum and Sessea (Solanaceae: Cestreae) in Venezuela BERNHARDT, PETER. (See Richard R. Clinebell II & Peter Bernhardt) ___ BERNHARDT, PETER. (See Peter Goldblatt, Peter Bernhardt & John C. Man- ning) BERNHARDT, PETER. (See Peter Goldblatt, John C. Manning & Peter Bern- BETZ, R. F. (See M. L. Bowles, J. L. McBride & R. F. Betz) BITTNER, R. TODD & DAVID J. GIBSON. Microhabitat Relations of the Rare Reed Bent Grass, Calamagrostis porteri subsp. insperata (Poaceae), with Implications for Its Conservation BOWLES, M. L., J. L. MCBRIDE & R. F. BETZ. Management and Restoration Ecology of the Federal Threatened Mead's Milkweed, Asclepias meadii (Asclepiadaceae) BOWLES, MARLIN L. (See Diane L. Tecic, Jenny L. McBride, Marlin L. Bowles & Daniel L. Nickrent) BRADFORD, JASON C. A Cladistic Analysis of Species Groups in Weinmannia (Cunoniaceae) Based on Morphology and Inflorescence Architecture ___ 1998 594. 518 69 BREMER, KARE. (See The Angiosperm Phylogeny Group) CHASE, MARK W. (See The Angiosperm Phylogeny Group) CHASE, MARK W. (See Owi I. Nandi, Mark W. Chase & Peter K. Endress) CLARK, DEBORAH А. Deciphering Landscape Mosaics of Neotropical Trees: GIS and Systematic Sampling Provide New Views of Tropical Rainforest Diversity CLINEBELL II, RICHARD R. & PETER BERNHARDT. The Pollination Ecol- ogy of Five Species of Penstemon (Scrophulariaceae) in the Tallgrass Prairie CRAWFORD, DANIEL J., ELIZABETH J. ESSELMAN, JENNIFER L. WIN- DUS € CAROL 5. PABIN. Genetic Variation in Running Buffalo Clover (Trifolium stoloniferum: Fabaceae) Using Random Amplified Polymorphic DNA Markers (RAPDs) D'ARCY, WILLIAM G. (See Carmen Benítez de Rojas & William G. D'Arcy) ENDRESS, PETER K. (See Owi I. Nandi, Mark W. Chase & Peter K. Endress) ESSELMAN, ELIZABETH J. (See Daniel J. Crawford, Elizabeth J. Esselman, Jennifer L. Windus & Carol S. Pabin) GIBSON, DAVID J. (See Todd R. Bittner & David J. Gibson) GOLDBLATT, PETER, JOHN C. MANNING & PETER BERNHARDT. Adap- tive Radiation of Bee-pollinated Gladiolus Species (Iridaceae) in South- ern Africa GOLDBLATT, PETER, PETER BERNHARDT & JOHN C. MANNING. Pol- lination of Petaloid Geophytes by Monkey Beetles (Scarabaeidae: Rute- linae: Hopliini) in Southern Africa GIUSSANI, LILIANA M. (See Fernando O. Zuloaga, Osvaldo Morrone, Andrea S. Vega & Liliana M. Giussani) HAUK, WARREN D. A Review of the Genus Paragonia (Bignoniaceae) ___. HAVENS, KAYRI. Selected Proceedings from the 1997 Midwestern Rare Plant Conference. Introduction HAVENS, KAYRI & DOUGLAS L. HOLLAND. Factors Affecting Reproduc- tive Success in a Rare Grass, Calamagrostis porteri subsp. insperata ___ HOLLAND, DOUGLAS L. (See Kayri Havens & Douglas L. Holland) _____ HONG DE-YUAN, PAN KAI-YU & YU HONG. Taxonomy of the Paeonia delavayi Complex (Paeoniaceae) |... 449594092 HONG, YU. (See Hong De-yuan, Pan Kai-yu & Yu Hong) JENNINGS, MICHAEL D. (See J. Michael Scott & Michael D. Jennings) ___ JOHNSON, 5. D., К. E. STEINER, V. B. УНТЕНЕАР & 1. VOGELPOEL. Pollination Ecology and Maintenance of Species Integrity in Co-occurring Disa racemosa L.f. and Disa venosa Sw. (Orchidaceae) in South Africa _ MANNING, JOHN C. (See Peter Goldblatt, John C. Manning & Peter Bern- hardt) 126 81 273 137 81 69 492 215 492 MANNING, JOHN C. (See Peter Goldblatt, Peter Bernhardt & John C. Man- ning) MCBRIDE, J. L. (See M. L. Bowles, J. L. McBride & R. F. Ве) _______ MCBRIDE, JENNY L. (See Diane L. Tecic, Jenny L. McBride, Marlin L. Bowles & Daniel L. Nickrent) MONRO, ALEX К. € PETER J. STAFFORD. A Synopsis of the Genus Echi- nopepon (Cucurbitaceae: Sicyeae), Including Three New Таха _______ MORRONE, OSVALDO, FERNANDO O. ZULOAGA, MIRTA O. ARRIAGA, RAÜL POZNER & SANDRA 5. ALISCIONI. Revisión Sistemática y Análisis Cladístico del Género Chaetium (Poaceae: Panicoideae: Pani- ceae) MORRONE, OSVALDO. (See Fernando O. Zuloaga, Osvaldo Morrone, Andrea S. Vega & Liliana M. Giussani) MUÑOZ, JÉSUS. Materials Toward a Revision of Grimmia (Musci: Grimmi- aceae): Nomenclature and Taxonomy of Grimmia longirostris _________ MUÑOZ, JÉSUS. A Taxonomic Revision of Grimmia subgenus Orthogrimmia (Musci, Grimmiaceae) NANDI, OWI L, MARK W. CHASE € PETER К. ENDRESS. A Combined Cladistic Analysis of Angiosperms Using rbcL and Non-Molecular Data Sets NAVARRO, CARMEN. (See Juan José Aldasoro, Carlos Aedo & Carmen Na- varro NAVARRO, CARMEN. (See Carlos Aedo, Juan José Aldasoro & Carmen Na- varro) NICKRENT, DANIEL L. (See Diane L. Tecic, Jenny L. McBride, Marlin L. Bowles & Daniel L. Nickrent) ORTIZ, SANTIAGO, JUAN RODRÍGUEZ-OUBINA & MESFIN TADESSE. A axonomic Revision of Dicoma (Asteraceae: Cichorioideae: Mutisieae) for the Horn of Africa PABIN, CAROL S. (Daniel J. Crawford, Elizabeth J. Esselman, Jennifer L. Windus & Carol S. Pabin) PAN KAI-YU. (See Hong De-yuan, Pan Kai-yu & Yu Hong) PHIPPS, J. B. Synopsis of Crataegus Series Apiifoliae, Cordatae, Microcarpae, and Brevispinae (Rosaceae subfam. Maloideae) POZNER, RAÚL. Revisión del Género Cucurbitella (Cucurbitaceae) _______ POZNER, RAÚL. (See Osvaldo Morrone, Fernando O. Zuloaga, Mirta O. Arri- aga, Raál Pozner & Sandra S. Aliscioni) RICHARDSON, P. MICK. New Tools for Investigating Biodiversity, the 43rd Annual Systematics Symposium of the Missouri Botanical Garden. Intro- duction RODRÍGUEZ-OUBINA, JUAN. (See Santiago Ortiz, Juan Rodríguez-Oubiña & Mesfin Tadesse) SCOTT, J. MICHAEL & MICHAEL D. JENNINGS. Large-area Mapping of Biodiversity 215 110 977 251 404 631 352 367 137 440 SNAPP, SHARON. (See Carol J. Baskauf & Sharon Snapp) STAFFORD, PETER J. (See Alex K. Monro 4 Peter J. Stafford) |... STARR, JULIAN R. (See Randall J. Bayer & Julian R. Starr) STEINER, K. E. (See S. D. Johnson, K. E. Steiner, V. B. Whitehead & L. Vogelpoel) STEVENS, PETER F. (See The Angiosperm Phylogeny Group) TADESSE, MESFIN. (See Santiago Ortiz, Juan Rodríguez-Oubiña & Mesfin Tadesse) TECIC, DIANE L., JENNY L. MCBRIDE, MARLIN L. BOWLES & DANIEL L. NICKRENT. Genetic Variability in the Federal Threatened Mead's Milkweed, Asclepias meadii (Asclepiadaceae) as Determined by Allozyme Electrophoresis TUOMISTO, HANNA. What Satellite Imagery and Large-scale Field Studies an Tell about Biodiversity Patterns in Amazonian Forests |... VEGA, ANDREA S. (See Fernando O. Zuloaga, Osvaldo Morrone, Andrea S. Vega & Liliana M. Giussani) VOGELPOEL, L. (See S. D. Johnson, K. E. Steiner, V. B. Whitehead & L. Vogelpoel) WHITEHEAD, V. B. (See S. D. Johnson, К. E. Steiner, V. B. Whitehead & L. Vogelpoel) WINDUS, JENNIFER L. (See Daniel J. Crawford, Elizabeth J. Esselman, Jen- nifer L. Windus & Carol S. Pabin) ZULOAGA, FERNANDO O., OSVALDO MORRONE, ANDREA S. VEGA & LILIANA M. GIUSSANI. Revisión y Análisis Cladístico de Steinchisma (Poaceae: Panicoideae: Paniceae) ZULOAGA, FERNANDO O. (See Osvaldo Morrone, Fernando O. Zuloaga, Mir- ta O. Arriaga, Raál Pozner & Sandra S. Aliscioni) 404 Flora of the Venezuelan Guayana Located in the southeastern half of Venezuela, the Venezuelan Guayana is the core area of what has been called *The Lost World." The area is dominated by massive table mountains known as tepuis and includes many endemic species and genera, with much of the area still in pristine condition. There are nearly 10,000 species in the flora area, and over half will be illustrated by line drawings. Volumes 3 and 4 of the Flora of the Venezuelan Guayana are now available from Missouri Botanical Garden Press: Berry, P. E., B. K. Holst, and K. Yatskievych, editors. Flora of the Venezuelan Guayana. Volume 3, Araliaceae-Cactaceae. 1997. ISBN 0-915279-46-0. 774 pp. 1113 species treated. 628 line drawings. $67.95. Volume 4, Caesalpiniaceae-Ericaceae. 1998. ISBN 0-915279-52-5. 799 pp. 1329 species treat- ed. 621 line drawings. $67.95 Also still available: | Volume 1, Introduction (includes Vegetation Map and Topographical Map). 1995. ISBN 0- 88192-313-3. 320 pp. of text, plus 44 pp. of color plates, 10 b/w photos, 51 line drawings. $52.95. Volume 2, Pteridophytes, Spermatophytes (Acanthaceae-Araceae). 1995. ISBN 0-88192-326- 5. 706 pp. 1285 species treated. 618 line drawings. $67.95. Vegetation Map and Topographical Map, 2-map set: rolled and shipped in tube $17.00; or professionally folded $15.00. | Price No. Total Volume 1 (& 2-map set) 932% с Make checks payable to: Volume 195 m Missouri Botanical Garden Press Volume 3 TE чажын 4344 Shaw Blvd. Volume 4 46195 —— St. Louis, MO 63110-2991 2-map set, rolled ; uU. Ga 2-тар set, folded $15.00 Add for shipping and handling: Check one: Visa ——— MasterCard Within the U.S. “Саға No. — — —— Exp. date for the first book об LLL... тате Or order using Visa or MasterCard: for мек жене book $1.00 қ Жа 314-577-9534 Canada and Mexico fax. 314-577-9591 fot the first book > MORES A - e-mail, mbgpress@mobot.org for each additional book ¡ $1.50 DAPR тала ыссы Ship to: for each additional book X00. .— Air rates available on request Total enclosed i $. yi Comes ee салаа , Pan Kai-yu & Y