The mammals of Paracou, French Guiana, a Neotropical lowland rainforest fauna. Part 1, Bats.

2 CONTENTS Abstract ....................................................................... 3 Re´sume´ ........................................................................ 4 Introduction .................................................................... 6 Acknowledgments ............................................................ 7 Site Description ................................................................ 8 Topography, Soils, and Vegetation .............................................. 8 Climate and Phenology ...................................................... 13 Bat Inventory Methods ........................................................ 14 Mistnetting ................................................................ 15 Searching for Roosts ........................................................ 17 Specimen Collection and Preservation ......................................... 19 Acoustic Survey ............................................................ 21 Scheduling and Personnel .................................................... 21 Systematic Accounts .......................................................... 22 Emballonuridae ............................................................. 23 Noctilionidae ............................................................... 42 Mormoopidae .............................................................. 45 Phyllostomidae ............................................................. 45 Desmodontinae ........................................................... 46 Glossophaginae .......................................................... 47 Phyllostominae ........................................................... 57 Carolliinae ............................................................... 94 Stenodermatinae .......................................................... 97 Furipteridae ............................................................... 128 Thyropteridae ............................................................. 129 Vespertilionidae ........................................................... 131 Molossidae ............................................................... 146 Analyses of Sampling ........................................................ 169 Sampling Results from Different Methods .................................... 169 Sampling Results in Different Habitats ....................................... 177 Sampling Results in Different Years ......................................... 181 Estimating Completeness ................................................... 182 Discussion .................................................................. 185 Taxonomic Composition and Biogeography ................................... 185 Species Richness .......................................................... 190 Trophic Guilds, Habitat Use, and Other Topics ................................ 192 Suggestions for Future Work .................................................. 198 Improving Inventory Efficiency ............................................. 199 Standards for Reporting Inventory Data ...................................... 200 Quantitative Methods for Diversity Comparisons .............................. 201 Prospects for Rapid Diversity Assessment .................................... 202 References .................................................................. 202 Appendix 1: Bats Previously Reported from French Guiana or Surinam, but not ..... Captured at Paracou ....................................................... 215 Appendix 2: Species Matrix for 14 Neotropical Rainforest Bat Inventories .......... 218 1998 3SIMMONS AND VOSS: PARACOU BATS ABSTRACT This report describes the results of bat inventory fieldwork at Paracou, a lowland rainforest locality in northern French Guiana. Working within a 3-km radius over the course of 168 sampling days from 1991 to 1994, we captured 3126 bats, of which about 78% were taken in ground-level mistnets, 10% in mistnets suspended above ground level, and 12% at roosts. We identified a total of 78 species, including 10 emballonurids, 2 noctilionids, 1 mormoopid, 49 phyllostomids, 1 furipterid, 1 thyropterid, 5 vespertilionids, and 9 molossids. Among our taxonomic results, we describe a new species of Micronycteris (sensu stricto) to honor Andre´ Brosset, pioneering monographer of rainforest bat faunas in India, Africa, and South America. In addition, we report the first records of eight other species from French Guiana: Centronycteris maximiliani,Peropteryx kappleri,Saccopteryx gymnura,Micronycteris homezi,Micronycteris schmidtorum,Molossops paranus,Molossus sinaloae, and Promops centralis. Most of these were previously known from Surinam, but the range extensions are significant for Saccopteryx gymnura (ca. 900 km), Micronycteris homezi (2200 km), and M.schmidtorum (1500 km). Altogether, the known bat fauna of French Guiana now consists of 102 species. The following significant taxonomic results are also reported herein. (1) Comparison of Paracou specimens referable to Peropteryx macrotis (Wagner) with the holotype of P.trinitatis Miller supports the conclusions of recent investigators that these taxa are separate species. (2) Morphological variation among specimens of small Choeroniscus from Paracou, together with examination of type material and a critical review of the literature, suggest that C.minor (Peters), C.intermedius (Allen and Chapman), and C.inca Thomas are conspecific; the oldest available name for the species is Peters’. (3) Glyphonycteris Thomas (including Barticonycteris Hill as a synonym), Micronycteris Gray (including Xenoctenes Miller as a synonym), and Trinycteris Sanborn are rediagnosed as distinct genera; Lampronycteris Sanborn and Neonycteris Sanborn, two other erstwhile subgenera of Micronycteris (sensu lato), should also be treated as full genera. (4) Micronycteris homezi Pirlot, based on a lost holotype and previously considered a nomen dubium, is redescribed and rediagnosed as a valid species. (5) Micronycteris megalotis (Gray) and M.microtis Miller are distinct species represented by sympatric collections from Paracou and other material similarly interpreted by recent investigators. (6) Mimon bennettii (Gray) and M.cozumelae Goldman are diagnosable as distinct species by consistent external and craniodental character differences. (7) Ectophylla H. Allen is rediagnosed to include Mesophylla Thomas in recognition of the sister-group relationship between E.alba H. Allen and E.macconnelli (Thomas). (8) The recent hypothesis that Sturnira lilium (E. Geoffroy) and S.luisi Davis are conspecific is rejected as implausible because of trenchant cranial character differences. (9) The Venezuelan and French Guianan specimens recently identified in the literature as Eptesicus andinus J. A. Allen are not conspecific with the holotype of that species; instead, examination of type specimens, other comparative material, and the primary literature suggests that this material is referable to E.chiriquinus Thomas. (10) All currently accepted synonymies for taxa included within Davis’s (1966) andinus group of Eptesicus are apparently incorrect; in our view, E.andinus is a senior synonym of E.montosus Thomas and E.chiralensis Anthony, whereas E.chiriquinus is a senior synonym of E.inca Thomas. (11) We review the contents of Cynomops Thomas, currently ranked as a subgenus of Molossops Peters, and tabulate diagnostic characters for the four species we regard as valid: M.abrasus (Temminck), M.greenhalli (Goodwin), M.paranus (Thomas), and M.planirostris (Peters). (12) Molossus barnesi Thomas is a valid species readily distinguishable from both M.molossus (Pallas) and M.coibensis J. A. Allen. Analyses of our sampling results indicate that (1) distinct sets of species are effectively sampled by different capture methods; (2) distinct sets of species inhabit different local habitats; and (3) increased sampling effort with any method generally results in more species, although the rate of accumulation declines with sample size (number of captures). Based on nonparametric statistical extrapolations, we estimate that the Paracou bat fauna probably consists of somewhere between 85 and 95 species; the more conservative richness estimator suggests that our inventory is perhaps about 90% complete. Judging from the known or inferred behaviors of the rare taxa (singletons and doubletons) in our data, most of the local species missing from this inventory are probably aerial insectivores, gleaning insectivores, or nectarivores. In terms of higher taxonomic composition, the bat fauna at Paracou is typical of those found throughout the humid Neotropical lowlands. A quantitative analysis of faunal similarity at the species level among 14 rainforest localities chosen as exemplars clusters the Paracou list with others previously reported from the Guiana sub- 4 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY region of Amazonia, next with lists from elsewhere in Amazonia, and lastly with Central American lists. Not surprisingly, pairwise similarity values show a positive correlation between faunal resemblance and geographic proximity within the Neotropical rainforest biome. Many (47%) of the bat species in the Paracou fauna are essentially pan-Neotropical in distribution and most of these are also known from habitats other than rainforest. The remaining species exhibit more restricted geographic distribution patterns, but true Amazonian endemics constitute only a minor fraction of the Paracou bat fauna. Species richness comparisons among inventory sites are complicated by problems of inconsistent methodology, habitat representation, and sampling effort. For example, the apparently exceptional diversity of emballonurids, phyllostomines, and molossids in the Paracou fauna is plausibly explained by our intensive use of elevated netting and roost surveys, and by prolonged effort, all of which factors act to reduce the well-known capture bias of ground-level mistnets (which consistently undersample these taxa in the short term). However, the low richness of carolliines and stenodermatines at Paracou by comparison with most other Amazonian (especially western Amazonian) localities is apparently real. The only approximately valid statistical comparison of species richness that we can make between sites based on published capture-frequency data suggests an increase of approximately 50% in understory bats from eastern Central America to Amazonia, but the real or artifactual nature of this estimated difference remains to be evaluated. A trophic classification of Paracou bats indicates that aerial insectivores are the most speciose feeding guild in the local fauna, followed by gleaning animalivores, frugivores, and nectarivores; ominivores, sanguivores, and piscivores are minor components. Patterns of differential habitat use among species within some feeding guilds can be inferred from our capture-frequency data, notably for aerial insectivores and frugivores. By contrast, gleaning animalivores appear to be largely restricted to primary forest, a puzzling phenomenon previously reported from other Neotropical rainforest localities. To facilitate future inventory fieldwork we provide (1) detailed descriptions of survey and capture methods, (2) illustrations of most local habitats recognized as distinct, (3) complete breakdowns of capture frequencies by method and habitat for each species, (4) photographs of numerous roosts at which bats were captured, and (5) descriptions and/or illustrations of useful characters for identifying species hitherto frequently confused in the field. Finally, we make recommendations for improving bat inventory efficiency, suggest minimal standards for reporting inventory data, urge the adoption of quantitative methods for intersite diversity comparisons, and comment on the prospects for rapid diversity assessment of rainforest bat faunas. RE ´SUME ´ Dans ce rapport nous de´crivons les re´sultats de l’inventaire du peuplement de chauves-souris effectue´ sur le terrain a` Paracou, un site de foreˆt hygrophile de basse altitude en Guyane franc¸aise. Dans un rayon de 3 km e´tudie´ pendant 168 jours d’e´chantillonnage de terrain de 1991 a` 1994, nous avons capture´ 3126 chauves-souris, soit 78% dans des filets pre`s du sol, 10% dans des filets e´leve´s suspendus au-dessus du sol, et 12% a` des dortoirs. Nous avons identifie´ 78 espe`ces au total, dont 10 emballonuride´s, 2 noctilionide´s, 1 mormoopide´, 49 phyllostomide´s, 1 furipteride´, 1 thyropteride´, 5 vespertilionide´s et 9 molosside´s. Parmi nos re´sultats taxonomiques nous de´crivons une espe`ce nouvelle de Micronycteris (sensu stricto) en l’honneur d’Andre´ Brosset, auteur de monographies innovatrices sur les faunes de chauves-souris des foreˆts hygrophiles de l’Inde, de l’Afrique et de l’Ame´rique du Sud. De plus, nous pre´sentons les premie`res donne´es concernant la pre´sence de huit autres espe`ces en Guyane franc¸aise: Centronycteris maximiliani,Peropteryx kappleri,Saccopteryx gymnura,Micronycteris homezi,Micronycteris schmidtorum,Molossops paranus,Molossus sinaloae et Promops centralis. La plupart de ces espe`ces e´taient de´ja` connues du Suriname, mais les extensions des aires de re´partition sont substantielles pour Saccopteryx gymnura (pre`s de 900 km), Micronycteris homezi (2200 km) et M.schmidtorum (1500 km). Dans ce travail, nous rapportons e´galement les importants re´sultats taxonomiques suivants. (1) La comparison d’exemplaires de Paracou se rapportant a`Peropteryx macrotis (Wagner) avec l’holotype de P.trinitatis Miller, confirme la conclusion d’auteurs re´cents, que ces taxons appartiennent a` des espe`ces diffe´rentes. (2) La variabilite´ morphologique de´celable parmi les exemplaires de petits Choeroniscus de Paracou, conjointement avec l’e´tude des types et avec une re´vision critique de la litte´rature, sugge`rent que C. minor (Peters), C.intermedius (Allen et Chapman) et C.inca Thomas sont conspe´cifiques; le plus ancien nom disponible pour l’espe`ce est celui 1998 5SIMMONS AND VOSS: PARACOU BATS de Peters. (3) De nouvelles diagnoses sont donne´es pour Glyphonycteris Thomas (qui inclut Barticonycteris Hill comme synonyme), et pour Trinycteris Sanborn; Lampronycteris Sanborn et Neonycteris Sanborn, deux anciens sous-genres de Micronycteris (sensu lato) devraient e´galement eˆtre conside´re´s comme des genres. (4) Micronycteris homezi Pirlot, fonde´ sur un holotype perdu et conside´re´ pre´ce´demment comme un nomen nudum, est rede´crit et diagnostique´a` nouveau comme une espe`ce valable. (5) Micronycteris megalotis (Gray) et M.microtis Miller sont des espe`ces distinctes repre´sente´es par des exemplaires re´colte´s en sympatrie a` Paracou et par d’autre mate´riel, ce qui est interpre´te´ de fac¸on similaire par des chercheurs re´cents. (6) Mimon bennettii (Gray) et M.cozumelae Goldman peuvent eˆtre diagnostique´es comme deux espe`ces distinctes graˆce a` des diffe´rences de caracte`res externes et craniodentaires. (7) Le genre Ectophylla H. Allen est diagnostique´a` nouveau pour y inclure Mesophylla Thomas, ce qui reconnait ainsi les rapports d’espe`ces-soeurs entre E.alba H. Allen et E.macconnelli (Thomas), cette dernie`re appele´e auparavant Mesophylla. (8) L’hypothe`se, re´cemment e´mise, que Sturnira lilium (E. Geoffroy) et S.luisi Davis sont conspe´cifiques est rejete´e comme e´tant peu plausible a` cause de diffe´rences marque´es des caracte`res craniaux. (9) Les exemplaires du Ve´- ne´zue´la et de Guyane franc¸aise re´cemment de´termine´s dans la litte´rature comme appartenant a` Eptesicus andinus J. A. Allen ne sont pas conspe´- cifiques avec l’holotype de cette espe`ce; en revanche, l’e´tude des exemplaires types, de mate´riel comparatif additionnel, et de la litte´rature originale, sugge`rent que ce mate´riel peut eˆtre rapporte´ a` E.chiriquinus Thomas. (10) Toutes les synonymies accepte´es a` l’heure actuelle pour les taxons inclus au sein du groupe d’Eptesicus appele´ groupe andinus par Davis (1966) semblent eˆtre incorrectes; d’apre`s nous, E.andinus est un synonyme plus ancien d’E.montosus Thomas et de E.chiralensis Anthony, tandis que E.chiriquinus est un synonyme plus ancien d’E.inca Thomas. (11) Nous re´visons la teneur de Cynomops Thomas, actuellement classe´ comme un sous-genre de Molossops Peters, et pre´sentons sous forme tabulaire les caracte`res diagnostiques des quatre espe`ces que nous conside´rons comme valables: M. abrasus (Temminck), M.greenhalli (Goodwin), M.paranus (Thomas), et M.planirostris (Peters). (12) Molossus barnesi (Thomas) est une espe`ce valable, aise´ment diffe´renciable a` la fois de M. molossus (Pallas) et de M.coibensis J. A. Allen. L’analyse de notre e´chantillonnage indique que: (1) des assemblages d’espe`ces distincts sont effectivement e´chantillonne´s par diffe´rentes me´thodes de capture; (2) des assemblages d’espe`ces distinctes occupent des biotopes diffe´rents au sein de notre zone d’e´tude; et (3) un effort accru d’e´chantillonnage, quelle que soit la me´thode, produit en general davantage d’espe`ces, bien que le taux d’accumulation baisse avec une augmentation de la taille de l’e´chantillon (nombre de captures). Sur la base d’extrapolations statistiques non-parame´triques, nous estimons que le peuplement de chauves-souris de Paracou contient entre 85 et 95 espe`ces; l’estimation conservatrice de cette richesse sugge`re que notre inventaire est peut-eˆtre a` 90% complet. Si nous nous basons sur les comportements, connus ou infe´re´s, des taxons rares (repre´sente´s dans notre e´chantillonnage par des exemplaires uniques ou des doublets), la plupart des espe`ces qui manquent encore a` cet inventaire sont probablement des insectivores ae´riens, des insectivores glaneurs ou des nectarivores. En ce qui concerne sa composition taxonomique suprage´ne´rique, la faune de chauves-souris de Paracou est typique de celles que l’on trouve dans l’ensemble des re´gions humides ne´otropicales de basse altitude. Une analyse quantitative de la ressemblance faunistique au niveau spe´cifique parmi 14 sites choisis comme exemples de foreˆt hygrophile, place la liste de Paracou, d’abord parmi d’autres mentionne´es auparavant de la subre´gion guyanaise de l’Amazonie, ensuite parmi des listes provenant d’autres re´gions de l’Amazonie, et enfin parmi des sites d’Ame´rique centrale. Les valeurs des indices de similarite´, compare´es deux a` deux, montrent une corre´lation positive entre la ressemblance faunistique et la proximite´ge´ographique des sites au sein du biome foreˆt hygrophile ne´otropicale. De nombreuses espe`ces de chauves-souris du peuplement de Paracou (47% du total) ont une re´partition essentiellement pan-ne´otropicale, et la plupart d’entre elles sont connues d’habitats autres que la foreˆt hygrophile. Les autres espe`ces montrent des sche´mas de distribution ge´ographiquement plus restreints, mais les vrais ende´miques amazoniens ne constituent qu’une fraction mineure du peuplement de chauves-souris de Paracou. Les comparaisons de richesse spe´cifique parmi les sites inventorie´s pre´ce´demment sont rendues complique´es par des proble`mes de me´thodologie ine´gale, de re´presentation d’habitats et d’effort d’e´chantillonnage. Par exemple, l’explication de la diversite´, a` premie`re vue exceptionnelle, des emballonuride´s, phyllostomine´s et molosside´s dans la faune de Paracou, est rendue plausible par notre emploi intensif de filets e´leve´s et par notre e´tude de dortoirs, ainsi que par un effort prolonge´, me´thodes d’e´chantillonnage qui toutes re´duisent la tendance bien connue de surcapturer lorsque les filets sont pose´s pre`s du sol, ce qui donne, dans 6 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY le court terme, un sous-e´chantillonnage de ces taxons. Toutefois, la faible richesse des carolliine´s et des sternodermatine´s a` Paracou, compare´e a` celle de la plupart des autres localite´s amazoniennes (spe´cialement d’Amazonie occidentale), ne paraıˆt pas artefactuelle. La seule comparaison statistique a` peu pre`s valable que nous pouvons faire de richesse spe´cifique entre sites, base´e sur des donne´es de fre´quences de captures dans la litte´- rature, sugge`re que la proportion de chauves-souris du sous-bois augmente d’environ 50% entre l’Ame´rique centrale orientale et l’Amazonie, mais la nature re´elle ou artifactuelle de cette diffe´rence estime´e reste a`eˆtre e´value´e. Une classification trophique des chauves-souris de Paracou indique que les insectivores ae´riens sont la guilde du peuplement local contenant le plus d’espe`ces, suivie par les animalivores glaneurs, les frugivores et les nectarivores; les omnivores, sanguivores et piscivores ne repre´sentent que des fractions mineures. Les patrons d’utilisation diffe´rentielle des habitats au sein de certaines guildes alimentaires peuvent eˆtre de´duits a` partir de nos donne´es de fre´quences de captures, notamment pour les insectivores ae´riens et les frugivores. Par contre, les animalivores glaneurs semblent eˆtre largement limite´s a` la foreˆt primaire, un phenome`ne de´concertant, de´ja` mentionne´ auparavant pour d’autres localite´s de la foreˆt hygrophile ne´otropicale. Dans le but de faciliter de futurs inventaires de terrain: (1) nous donnons des descriptions de´taille´es de nos me´thodes d’e´tude et de capture, (2) nous illustrons la plupart des habitats locaux reconnus comme distincts, (3) nous indiquons, pour chaque espe`ce, la re´partition comple`te des fre´- quences de capture par me´thode et par habitat, (4) nous presentons des photographies des nombreux dortoirs ou` nous avons capture´ des chauves-souris, et (5) nous de´crivons et/ou illustrons des caracte`res utiles pour la de´termination d’espe`ces qui ont e´te´ fre´quemment confondues jusqu’a` pre´sent sur le terrain. Finalement, nous faisons des recommandations visant a` ame´liorer l’efficacite´ des inventaires de chauves-souris, nous sugge´rons des niveaux minimaux pour les rapports de donne´es d’inventaires, nous conseillons vivement l’adoption de me´thodes quantitatives pour des comparaisons de diversite´ inter-sites, et nous faisons des remarques au sujet de l’e´valuation rapide de la diversite´ des faunes de chauves-souris en foreˆt hygrophile. INTRODUCTION The most species-rich mammalian faunas in the world are probably those that inhabit the lowland rainforests of certain parts of Central and South America, but compelling evidence to support this conjecture is sparse. Much of what we know about Neotropical rainforest mammal diversity comes from just a handful of sites where large species lists were built up over many years as byproducts of field research on behavior, community ecology, demography, and other natural history topics (e.g., Anthony, 1921; Enders, 1935; Glanz, 1982; Janson and Emmons, 1990; Handley et al., 1991; Timm, 1994; Hutterer et al., 1995). Such haphazardly collected data are difficult to analyze for completeness, and estimating species richness at most sites is therefore a matter of more-orless educated guesswork (Voss and Emmons, 1996). For both research and conservation purposes it would be useful to know how much time and effort and what combination of inventory methods are necessary to obtain reasonably complete species lists of mammals at rainforested Neotropical localities. Unfortunately, the current literature does not provide an adequate basis for making such estimates. Among other problems, few inventory reports have included sufficient methodological information to know exactly what was done, the local habitats actually sampled for mammals are seldom identified explicitly, relevant measures of sampling effort are rarely provided, and the frequency data necessary to judge inventory completeness are almost never published. Species identifications are another major impediment to interpreting inventory results. Most genera of Neotropical rainforest mammals have never been revised, or were revised long ago from sparse and geographically scattered material. Much of the secondary literature commonly consulted for specimen identification (e.g., keys, checklists, and field guides) therefore reflects taxonomic conventions that have not been tested by recent analyses of character data. For many genera, reliable identifications can only be obtained by consulting the primary literature 1998 7SIMMONS AND VOSS: PARACOU BATS and by directly comparing voucher specimens with types. Unfortunately, published species lists from rainforest mammal inventory projects are seldom accompanied by sufficient documentation to judge the reliability of the identifications they contain. To provide a baseline for more effective research on Neotropical rainforest mammal diversity, we looked for a study site where we could return year after year to obtain results from sustained sampling effort using the full range of methods required for taxonomically comprehensive surveys (Voss and Emmons, 1996). We were attracted to French Guiana because of its unique combination of extensive primary rainforest in close proximity to modern transportation and communication networks. At Paracou—a research area with primitive but adequate living quarters in the midst of a large tract of forest only a few kilometers by excellent roads from post offices, telephones, fax machines, markets, and hospitals—we found an ideal location for our project. This is the first of two monographs based on our fieldwork at Paracou, which began in 1991 and continued to 1994. Herein we describe the research area, explain the methods used to sample the bat fauna, and report our bat-sampling results. We document all taxonomic identifications by reviewing the relevant literature, by providing tables of external and craniodental measurements of voucher specimens, and by detailed comparisons with types and other relevant material in cases where we encountered significant problems. We summarize field observations for each species by capture method and habitat, and we remark noteworthy ecological differences between congeners or other closely related taxa. We analyze our sampling results to assess methodological bias, habitat differences, temporal variation, and completeness. We select 13 other Neotropical rainforest bat inventories as comparative exemplars, and we discuss these together with the Paracou fauna in terms of taxonomic composition, biogeography, species richness, and trophic structure. Finally, we offer suggestions for future work based on the outcome of our field and museum research with Paracou bats. ACKNOWLEDGMENTS We thank the administrative staff of CIRAD in French Guiana, especially V. Favrichon, O. Hamel, J.-F. Julia, and O. Laroussinie, for their permission to work at Paracou, for space to store our field equipment in Kourou, and for numerous courtesies that made our fieldwork more productive and enjoyable. We are also grateful to G. Dubost for his initial invitation to work at Paracou, and for his advice and encouragement throughout the course of our fieldwork. P.-M. Forget kindly helped us get started in 1991. Special thanks are due to P. Petronelli, a gifted amateur naturalist and long-time resident at Paracou, who generously shared his special knowledge of the forest besides repairing almost everything that broke or malfunctioned around camp. Without his help it would have taken us much longer to learn the trail system, identify plants, find bat roosts, and generally use our time efficiently in the field. Whatever measure of success we achieved at Paracou is in large part owed to him. For their assistance with bat inventory fieldwork we thank Darrin Lunde, who was always there to help from 1991 to 1993; to Suzanne Smith, who competently bore the brunt of the mistnetting effort in 1993; to Roland Kays, who climbed trees and assisted in many other ways in 1993; to Andrea Peffley, who did almost everything almost every day in 1994; and to Nancy Voss, whose stalwart assistance helped make our last field season the most productive ever. Amy Berkov and Louise Emmons also helped us net bats in 1993 and 1994, respectively. Back at the museum we were grateful for the assistance of Eleanor Sterling who first set up the Paracou database, to Darrin Lunde and Burton Barnett who completed and proofed it, and to Tenley Conway who patiently made numerous revisions and printouts as our species identifications were refined. Darrin Lunde, Eleanor Sterling, and Andrea Peffley cataloged specimens, pulled skulls, and curated the collections. Pat Brunnauer resourcefully tracked down dozens of references in the AMNH library, and Don Clyde always found the rest somewhere else. Peter Goldberg skillfully print- 8 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY ed all of the photographs we took in the field, and Pat Wynne did all of the splendid artwork. For the loan of specimens or for their hospitality while we visited their respective institutions (abbreviated as below), we thank P. D. Jenkins (BMNH); D. A. Schlitter and S. B. McLaren (CM); L. R. Heaney and B. D. Patterson (FMNH); L. Granjon and M. Tranier (MNHN); V. Pacheco (MUSM); M. D. Engstrom (ROM); and M. D. Carleton, L. K. Gordon, and C. O. Handley, Jr. (USNM). Critical feedback on the first draft of this report was provided by L. H. Emmons, M. D. Engstrom, E. K. V. Kalko, and B. D. Patterson, all of whom took time from their busy research schedules to help improve our text. Franc¸ois Vuilleumier very kindly translated our abstract. Our field research was initiated with funds from the Department of Mammalogy and the research administration of the American Museum of Natural History. Funding for the last two field seasons was provided by grant number 4991–93 from the National Geographic Society. We thank both organizations for their generous support. We dedicate this report to our late friend and colleague, Karl F. Koopman, whose patience, help, and encouragement were essential at many points in our museum research. The systematic accounts would not have been the same without him, and we are saddened that he did not live to see them completed. SITE DESCRIPTION TOPOGRAPHY, SOILS, AND VEGETATION The collections and observations reported below were obtained within the Domaine Experimental Paracou (hereafter, Paracou), a large tract of land administered as a research concession by the Centre de Coope´ration Internationale en Recherche Agronomique pour le De´veloppement (CIRAD). From a field camp at 5 8 16 9 31 0 N, 52 8 55 9 25 0 W (approximately 12 km SSE of Sinnamary and 33 km WNW of Kourou; figs. 1, 2), we worked in a roughly circular area with a radius of about 3 km. 1 The local terrain consists of low, densely forested hills drained by the headwaters of four streams: Crique Mogestern and Crique Verlot (both tributaries of the lower Sinnamary River), and Crique Parakou and Crique Malmanoury (which flow separately into the Atlantic Ocean). The minimum elevation above sea level within our study area is about 4 m, the maximum about 45 m (IGN, 1991). 1 We estimated the coordinates of our camp as the mean of four readings obtained in 1994 with a Panasonic KX-G5500 GPS receiver using the WGS84 (standard) map datum. A few bats collected under old highway bridges crossing Crique Renner (ca. 7.2 km to the NE) and Crique Paracou (ca. 7.1 km N) are reported among our voucher material, but all of these belong to species found within our 3-km sampling radius. Local soils are the result of in situ weathering from quartzites and schists of the Bonidoro series (Bariteau, 1993; Durrieu de Madron, 1993). Over most of our study area, the substrate is an acidic brownish or reddish sandy clay (mixed with ironstone concretions on some ridgetops and hillsides), but a few patches of almost pure white sand also occur (Barthes, 1991). Although heavily weathered quartzite boulders are occasionally exposed in streams, there are no large rocky outcrops or caves in the region. Local watercourses are all small (mostly 5 m or less in width), shallow (usually 1 m or less in depth), and heavily shaded, with transparent, tea-colored (‘‘black’’) water flowing slowly over predominantly sandy beds. Most of this landscape is covered with primary rainforest 2 (fig. 3), but the new asphalt highway from Kourou to Sinnamary cuts 2 No macroscopic soil charcoal was observed when D. P. Lunde and R. W. Kays dug 55 pits, each about 40 cm deep and 30 cm in diameter, to install pitfall traps in 1993 (Voss et al., in prep.). We likewise saw no charcoal in the spoil from numerous soil sample cores made in a 25-ha plot (Parcel 16) by a forestry research class in 1994. Soil charcoal has been interpreted as evidence of prehistoric agriculture at other Neotropical rainforest localities (e.g., by Lovejoy and Bierregaard, 1990; McDade and Hartshorn, 1994), and its absence suggests that the mature forest at Paracou has not been cleared for centuries. 1998 15SIMMONS AND VOSS: PARACOU BATS Therefore, we freely changed our equipment and modified our procedures from season to season (sometimes from day to day) as suggested by our own accumulating results, prompted by recommendations from our colleagues, or demanded by changing climatic and phenological circumstances in the field. MISTNETTING At the begining of our survey (in 1991) we used standard-weight 70-denier (70 d) mist nets to catch bats. We later (1992) switched to light-weight 30and 50-denier (30 d and 50 d) nets and used them almost exclusively thereafter. The nets we used at Paracou were 2.6 m high and came in 6-, 13-, and 18-m lengths; the size of the nylon mesh was 36 mm. G ROUND -L EVEL N ETTING : To catch bats at ground level, 0–3 m by our convention, we mounted nets on poles (usually made from saplings cut and trimmed in situ) thrust into the soil and braced by guys of braided nylon cord (fig. 7). To minimize disturbance of the understory vegetation that might alarm bats patrolling familiar territory, we usually tried to set nets along preexisting trails. Even so, some clearing was usually necessary to avoid entangling nets in trailside vegetation, and to leave enough space for people to work on each side. Lightweight (30 d) nets are so fragile and hard to clean of entangled litter that it is generally also necessary to sweep the ground clean of leaves and twigs under each net. As a general rule, nets must be moved every night because bats quickly learn to avoid them. Netting effectively on a nightly schedule therefore required at least two persons: one to stay in camp and process the last night’s catch, and the other to survey the forest for new net locations, clear new net lanes, and set up for the next night’s work. We chose net locations to sample as many forest microhabitats as possible: well-drained sites, palm swamps, streambeds, unusual plant associations, etc. We often opened nets shortly before dusk (when it was still light enough to read easily) because some bats emerge from their roosts in the late afternoon. Once the nets were opened we stayed with them constantly because large bats chew holes in nets unless they are promptly removed, and because small bats often escape from nets unless they are immediately seized. We regularly kept nets open from dusk to dawn in 1991 and 1992, but netting after 22:00 hours was seldom productive, so we usually closed our nets before midnight in subsequent years. In addition to passively waiting for bats to fly into our nets, we used the Audubon Bird Call (fig. 8) to attract them. This simple device produces high-pitched (but not ultrasonic) squeaks resembling (among other things) the distress calls of stenodermatines, which sometimes flock to the sound in large numbers (Handley et al., 1991). Of greater interest for inventory work, some phyllostomines that are otherwise hard to capture (e.g., Chrotopterus auritus,Glyphonycteris daviesi,Tonatia schulzi,Vampyrum spectrum) are likewise attracted, perhaps because they are interested in distressed bats, singing orthopterans (Tuttle et al., 1985), or other small animals with high-pitched vocalizations as prey. We first used Audubon Bird Calls sparingly in 1992, but in later years we frequently used them when bored by the absence of bat activity around our nets. Bats were removed from nets as soon as possible after capture and placed in cloth bags closed by drawstrings. When too many bats were captured to place each in its own bag, we often put two or more conspecifics together, but we never knowingly mixed species in bags. In general, we bagged all captured bats whether or not they were required as specimens in order to prevent recaptures on the same night; after the nets were closed, unwanted bats were checked for correct identification and sex, and then released. Bags containing captured bats were always suspended from cords tied horizontally between nearby trees to keep them away from prowling opossums. E LEVATED N ETTING : To catch bats in the subcanopy (fig. 9) and in the canopy itself (fig. 10), we mounted our nets on slender poles, the sawn-off top and bottom ends of which were wrapped with duct tape (to prevent splitting) and fitted with small eyebolts. Swivel-snaps were used to attach the poles to a length of 1 ⁄ 8 -inch (ca. 3 mm) braided nylon line that formed a continuous running 16 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 7. Ground-level mistnets (furled in this daytime photograph) in swampy primary forest; portions of three nets (arrows) supported by guyed poles are shown. We caught 30 species of bats in 11 nights of ground-level netting at this site from 1991 to 1994, including our only examples of Glyphonycteris daviesi. The most commonly captured species here were Carollia perspicillata,Artibeus jamaicensis,A.obscurus,Phyllostomus elongatus,Rhinophylla pumilio,Tonatia saurophila, and Trachops cirrhosus. 1998 17SIMMONS AND VOSS: PARACOU BATS Fig. 8. The Audubon Bird Call (available from R. W. Eddy, Box 0172, Newington CT 06131) consists of a pewter key in a hollow wooden cylinder; twisting the key produces highpitched squeaks that can be used to attract bats to nets. Handley et al. (1991: fig. 12-1) provided a spectrographic analysis of the sounds produced by this useful device, all of which are in the audible range. loop by means of which the net could be raised or lowered (flagpole fashion) at each end. The running loops were threaded through overhead supports (eyebolts or brass rings), the positions of which determined the height to which nets could be raised. When trees on either side of a suitable canopy gap or flyway were close together (without enough room between their crowns to suspend a net) and not too large ( # 35 cm diameter), we climbed them using French tree-climbing spikes (Model 502, available from ETS Lacoste, 24160 Excideuil, France; see Mori, 1987: fig. I-3A) and screwed large eyebolts directly into the trunks. To fix supports across larger gaps, we used the giant slingshot designed by Munn (1991) to shoot lead fishing weights tied to heavy monofilament line (on a spin-casting reel) into or over the crowns of trees on either side. 4 Brass rings (through which the running loops were first threaded) were then drawn up on braided line; our simplest net rigs used a single topline to hold the rings in place. In addition to having suitable overhead supports on each side of a canopy gap, elevated net sites must be clear enough at 4 D. P. Lunde and R. W. Kays constructed and tested our line-shooting equipment and helped install the first canopy nets at Paracou. ground level that nets can be raised and lowered without obstruction. Unfortunately, most natural gaps in the canopy are treefalls, which are always littered with prostrate trunks and limbs. Roads and clearings provide canopy gaps that are generally free of such obstacles (fig. 11), but manmade openings in the forest are often bordered by young trees that are not tall enough to support high nets. These constraints, together with the labor required to rig elevated nets, precluded frequent site changes in sampling the canopy bat fauna at Paracou. We measured the approximate height of elevated nets by attaching a tape measure to the bottom of one of the net poles; capture height was then recorded as the 3-m interval from the bottom to the top of the net (i.e., 10–13 m for bats caught in a net with lowermost pole ends 10 m above the ground). However, we found it essential to keep a sharp eye on elevated nets as they were lowered to remove entangled bats because lowflying species were often captured in the process; capture height data from carelessly monitored nets could therefore overestimate the vertical activity range of understory species. SEARCHING FOR ROOSTS Many rainforest bats that are hard to catch in mistnets can easily be collected or observed at their diurnal roosts. We found some roosts by chance, when bats flew up suddenly as we passed an unsuspected retreat in the forest understory. In such cases, we stepped a few meters away and stood quietly until the animals returned (usually in just a few minutes). Most roosts, however, were discovered by deliberate searching along trails. Among other potential roost sites in primary and secondary forest, we searched beneath undercut streambanks, under the buttresses of fallen trees, inside hollow logs, inside hollow standing trees, under loose bark on standing trees, and in foliage (especially among the leaves of large monocots such as palms, Heliconia spp., Phenakospermum guyannense, Philodendron spp.). Although we sometimes made special efforts to find particular roost types (e.g., leaf tents or hollow trees), we never ignored other kinds of roosts encoun- 18 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 9. Subcanopy of well-drained primary forest at Paracou, here viewed from about 20 m above the ground (photograph by R. W. Kays). The subcanopy is much less cluttered by leafy vegetation than the undergrowth, which may explain the higher capture rates of aerial insectivores like Cormura brevirostris,Peropteryx kappleri,Saccopteryx bilineata, and S.leptura at this level. At least 16 species of phyllostomids are probably also active in this habitat, although most of our elevated net captures of them were over roads, not inside the forest. tered as we searched. In addition to looking for natural roosts, we investigated reports of bats roosting in houses and storage sheds, and we searched every culvert under roads within our sampling radius. Most bats were located visually in their diurnal retreats, and some with distinctive posture, coloration, or other external characters could be reliably identified without capture. Likely patches of subcanopy foliage were scanned with binoculars, which were also useful for examining well-lit roosts in the forest understory that could not be closely approached without frightening the inhabitants. Roosts under fallen trees, inside hollow logs, and other dark refugia were examined with a headlight (which leaves the hands free to push aside obstructing vegetation, wield a net, aim a gun, or take notes). Short-handled butterfly nets were sometimes useful for capturing roosting bats, but in many situations shooting (with number 12 shot in .22 caliber or .410 loads) was the only option. Occasionally, roosting bats could be heard fluttering inside tree holes, hollow logs, or other places where they could not be seen or reached. In such situations, we plugged all but one entrance, around which we built a mistnet enclosure to catch the animals as they emerged at dusk. We used short (6 m) 70-d mistnets wrapped around upright poles to enclose the roost opening on all sides, and then thatched the enclosure with palm fronds to prevent emerging bats from escaping upward. Such enclosures usually captured every emerging bat. To obtain data on roost occupancy in successive years, we assigned every roost a unique number; durable roosts (those that could be expected to last for more than a few months, e.g., hollow logs, tree cavities) were identified by nailed aluminum tags. We tried 1998 19SIMMONS AND VOSS: PARACOU BATS Fig. 10. View over the forest canopy from the crown of an emergent tree, about 45 m above the ground (photograph by R. W. Kays). The unobstructed airspace above the trees is probably used mostly by species of Diclidurus,Eptesicus,Lasiurus, and molossids, although most of our captures of these taxa were made closer to the ground in nets suspended across a road through the forest or over roadside puddles. Many phyllostomids are probably active in the canopy itself, although we captured only Phyllostomus hastatus,Ametrida centurio, and Chiroderma villosum in our highest nets. to revist each durable roost every subsequent field season and record the species and numbers of bats in residence, but some roosts were destroyed when new land was cleared for agricultural experiments and others could not be found again. Some foliage roosts were revisited several times per field season following removal of the original occupants to determine whether leaf-tents used by one species are ever subsequently colonized by another. To explore the effects of observer bias in searching for roosts, and to obtain data on roost density, we systematically surveyed a 25-ha plot of primary forest (Parcel 16 of CIRAD’s forestry research program), which was conveniently divided by a square grid of surveyed trails into 1-ha quadrats (fig. 12). Four searchers with widely varying prior experience were assigned quadrats by lottery in 1994. Each 1-ha quadrat was then surveyed for roosts by one searcher who first walked the perimeter trails (totaling 400 m) and then the trailless interior; the interior was searched in four parallel transects spaced 20 m apart. Thus, no point within any quadrat was more than 10 m from a searched path or transect. Searchers were instructed to investigate all likely roosting sites and to flag all discovered roosts so that they could be subsequently revisited and confirmed by the most experienced searcher. Because Parcel 16 was set aside for nondestructive research, it was not possible to determine the identification of all roosting bats found in this survey. SPECIMEN COLLECTION AND PRESERVATION We preserved voucher specimens using standard procedures described and illustrated by Handley (1988). In collecting specimens, our objective was to preserve enough examples of each species to assess the taxonomic status of the Paracou population with 20 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 11. A narrow dirt road across which we raised nets to catch high-flying bats. The road is bordered by young secondary growth and selectively logged forest, but primary forest occurs within about 50–100 m on either side. 1998 21SIMMONS AND VOSS: PARACOU BATS Fig. 12. Map of a 25-ha plot of primary forest (Parcel 16) divided by a grid of surveyed trails into 1-ha quadrats (left) and the plan used to search each 1-ha quadrat for bat roosts (right). Two trails on the east side provided access from a dirt road ca. 0.5 km away. Most of the surveyed habitat was welldrained forest, but quadrats 5, 11, 12, 16, 17, 21, 22, and 25 also contained swampy forest. The standard search procedure for each quadrat is explained in the text. respect to populations sampled at other localities. Usually, we limited our removal to about 20 individuals (10 males and 10 females) per species, enough to obtain meaningful estimates of the range of variation and central tendency in each sex for characters of taxonomic interest. For open populations of most rainforest bats, destructive sampling on this scale is probably trivial by comparison with natural demographic processes. 5 5 Leigh and Handley (1991) estimated a population density of about 200 Artibeus jamaicensis per square kilometer of lowland forest in central Panama. Because their study site was rich in fig trees, the primary food resource of A. jamaicensis, this estimate is perhaps on the high side for the species. However, A.jamaicensis is larger (45 g) than most Neotropical rainforest bats, whose caloric requirements can probably be obtained in smaller home ranges. On the assumption that 100 individuals/km 2 is a conservative average value for many stenodermatines, we could expect about 2800 individuals of each species within our study area (ca. 28 km 2 ). An annual survivorship of 64% (the highest figure suggested by Leigh and Handley for A. jamaicensis) would then imply about 1000 natural deaths per species per year at Paracou. Presumably, such mortality is annually compensated by recruitment in all species that maintain stable local populations. ACOUSTIC SURVEY In the last year of our work at Paracou (1994), we used a custom-made electronic bat detector borrowed from Elisabeth K. V. Kalko (University of Tuebingen, Germany, and Smithsonian Tropical Research Insitute, Panama) to record the echolocation calls of bats foraging along roads and in manmade and natural clearings. The signals were amplified, fed into a transient recorder, and read out at 1 ⁄ 15 reduced speed on a Sony Walkman Professional cassette tape recorder. We made 198 recordings over the course of 10 nights between 9 October and 30 October 1994. Analyses of these recordings are currently in progress, and the results will be presented elsewhere. SCHEDULING AND PERSONNEL We worked at Paracou from 1991 to 1994, with the dates, personnel, and methods listed below. Although all personnel helped to some extent with the bat inventory, some were primarily engaged in other activities; only those chiefly responsible for the listed 22 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY methods are identified parenthetically. Italics indicate the authors of fieldnotes (deposited in the AMNH Department of Mammalogy archives) that should be consulted for detailed information about relevant methods. 1991: 26 June–27 August; Darrin P. Lunde, Nancy B. Simmons, and Robert S. Voss; ground-level netting (Simmons) and searching for roosts (Voss). 1992: 20 October–18 November; Darrin P. Lunde, Nancy B. Simmons, and Robert S. Voss; ground-level netting (Simmons, Voss) and searching for roosts (Voss). 1993: 20 July–15 September; Roland W. Kays, Darrin P. Lunde, Suzanne E. Smith, and Robert S. Voss; ground-level and elevated netting (Smith, Voss) and searching for roosts (Voss). 1994: 19 September–13 November; Louise H. Emmons, Andrea L. Peffley, Nancy B. Simmons, Nancy A. Voss, and Robert S. Voss; ground-level and elevated netting (Peffley, Simmons, N. A. Voss, R. S. Voss), searching for roosts (Peffley, Simmons, N. A. Voss, R.S.Voss), and ultrasonic census (Peffley, Simmons). SYSTEMATIC ACCOUNTS The following accounts are arranged systematically by family (and by subfamily within Phyllostomidae), and then alphabetically by genus and species. Unless noted otherwise, the nomenclature and taxonomic sequence used herein follows Koopman (1993). The organization of most accounts employs several standard subheadings. V OUCHER M ATERIAL : This section summarizes information about specimens collected, which are listed by sex and museum catalog number. Individuals measured for tables are indicated with asterisks. I DENTIFICATION : In the field, we made extensive use of Emmons’ (1990) generally excellent descriptions of external characters to make preliminary identifications. Although subsequent museum study of voucher specimens confirmed most of our field determinations, we also encountered numerous taxonomic problems that had to be resolved to justify the nomenclature adopted below. In cases where identification was relatively straightforward, we simply list the technical references we consulted at the beginning of each account and remark any noteworthy points in which our material differs from published descriptions. Species identifications involving significant taxonomic problems are treated at length, with reviews of the relevant literature and analyses of character variation in the specimens examined. We also discuss currently recognized subspecies in this section, although we do not use trinomial headings below. Measurements of voucher specimens and comparative material are provided in tables accompanying each species account. These morphometric summaries include only measurements of adults with closed epiphyses. Linear measurements of external and craniodental dimensions are reported in millimeters (mm); weights are reported in grams (g). The following measurements were taken for all or some of the species treated below. Total length: Distance from the tip of the snout to the tip of the last caudal vertebra. Tail length: Measured from the point of dorsal flexure of the tail with the sacrum to the tip of the last caudal vertebra. Hindfoot length: From the anterior edge of the base of the calcar to the tip of the claw of the longest toe. Ear length: From the notch to the fleshy tip of the pinna. Forearm length: From the elbow (tip of the olecranon process) to the wrist (including the carpals). This measurement is made with the wing at least partially folded. Tibia length: From the proximal end of the tibia to the posterior base of the calcar. Thumb length: From the metacarpal-phalangeal joint to the tip of the claw. Greatest length of skull: From the posteriormost point on the occiput to the anteriormost point on the premaxillae (excluding the incisors). Condyloincisive length: From the posteriormost point on the occipital condyles to the anteriormost point on the upper incisors. Condylocanine length: From the posteriormost point on the occipital condyles to the anteriormost point on the upper canines. Lacrimal breadth: Greatest breadth across the lac- 1998 23SIMMONS AND VOSS: PARACOU BATS rimal ( 5 anteorbital) ridges, when present and well defined. Postorbital breadth: Least breadth across the frontals posterior to the postorbital processes or bulges. Zygomatic breadth: Greatest breadth across the zygomatic arches. Braincase breadth: Greatest breadth of the globular part of the braincase. Mastoid breadth: Greatest cranial breadth across the mastoid region. Maxillary toothrow length: From the anteriormost edge of the canine crown to the posteriormost edge of the crown of M3. Breadth across molars: Greatest breadth across the outer edges of the crowns of the upper molars. Breadth across canines: Greatest breadth across the outer edges of the crowns of the upper canines. To confirm identifications of problematic species, we compared our voucher material to museum specimens from other localities, and we often consulted literature descriptions of types that we did not personally examine. We identify the museums in which vouchers and other specimens are preserved by the following abbreviations: AMNH, American Museum of Natural History (New York); BMNH, British Museum of Natural History (London); CM, Carnegie Museum of Natural History (Pittsburgh); FMNH, Field Museum of Natural History (Chicago); MNHN, Muse´um National d’Histoire Naturelle (Paris); MUSM, Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (Lima); RMNH, Rijksmuseum van Natuurlijke Historie (Leiden); SMN, Staatliches Museum fu¨r Naturkunde (Stuttgart); USNM, National Museum of Natural History (Washington, D.C.); ZMH, Zoologisches Museum Hamburg (Hamburg). F IELD O BSERVATIONS : This section summarizes our capture records and roost observations, and includes discussion of habitats, behavior, and other natural history topics. In reporting our observations for each species, we first list capture frequencies by inventory method, and then provide a breakdown of capture frequencies by habitat for each method. If every recorded capture is known to represent a unique individual, we state the number of individuals captured. Although we did not mark released bats, the number of individuals known to have been captured is occasionally larger than the number of vouchers we preserved because only one individual of each sex was released for some species. If more than one individual of either sex were released, we state the number of captures recorded and note that these ‘‘possibly’’ or ‘‘probably’’ (a subjective judgment) include some recaptures. For each species caught in elevated nets, we summarize capture height records as the interval from the bottom of the lowest elevated net in which the species was taken to the top of the highest net (e.g., 10–21 m for a species taken once in a net suspended 10–13 m above the ground and again in a net 18–21 m above the ground). We also note any conspicuous differences in mistnet capture-habitat or capture-height frequencies between congeners or other closely related species. We next provide information about roosting habits, especially roost type (hollow log, tree cavity, leaf-tent, etc.), roosting habitat, numbers of individuals in roosting groups observed, and the age and sex of captured roosting-group members. In reporting habitat data for each species, we use the following categories: (1) welldrained primary forest, (2) swampy primary forest, (3) creekside primary forest, (4) treefall openings in primary forest, (5) manmade clearings (roadways, gardens, plantations, etc.), (6) closed-canopy secondary growth (including selectively logged forest), and (7) roadside puddles. The term ‘‘primary forest’’ (without modifiers) includes categories 1–4, whereas categories 5–7 are often referred to collectively as ‘‘modified habitats.’’ EMBALLONURIDAE We captured 10 emballonurid species at Paracou, three of which are here reported for the first time from French Guiana. The species we identified from morphological voucher material represent the genera Centronycteris,Cormura,Diclidurus,Peropteryx (including Peronymus), Rhynchonycteris, and Saccopteryx. Published records from elsewhere in French Guiana and Surinam include another four emballonurid species that might also occur in our study area (appendix 1). 24 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Centronycteris maximiliani (Fischer) Figure 13 V OUCHER M ATERIAL : 1 male (AMNH 267397); see table 1 for measurements. I DENTIFICATION : Previously regarded as monotypic (e.g., by Koopman, 1993), the genus Centronycteris was recently shown to contain at least two species by Simmons and Handley (1998). According to these authors, the known range of Centronycteris maximiliani includes southern Venezuela, the Guianas, and northern, eastern, and southeastern Brazil, whereas C.centralis Thomas (1912b) occurs in Central America, Colombia, Peru, and on both sides of the Andes in Ecuador. Both species are known from only a few specimens, however, and their true geographic ranges may be considerably larger. In fact, because no known biogeographic barriers separate the ranges of C.maximiliani and C. centralis, these bats might occur sympatrically almost anywhere in Amazonia. Our Paracou voucher is the first known specimen of C.maximiliani from French Guiana. Measurements and descriptions of Centronycteris maximiliani from the Guianas and elsewhere were provided by Thomas (1913), Husson (1962, 1978), Williams et al. (1983), and Simmons and Handley (1998). The latter authors summarized craniodental characters and measurements that distinguish C.maximiliani from C.centralis. Our Paracou voucher formed much of the basis of their description of C.maximiliani. F IELD O BSERVATIONS : Our single individual of this species was shot in the late afternoon (about 16:00 hours) as it perched about 3 m above the ground on the underside of a large (ca. 20 3 30 cm) unmodified leaf of a small melastomataceous tree beside a trail in well-drained primary forest. Cormura brevirostris (Wagner) V OUCHER M ATERIAL : 17 females (AMNH *266009, *267070, 267076, *267078, *267389, *267391, *267823, *267824, *267825, *267826, *267828; MNHN *1995.823, *1995.824, *1995.825, *1995.826, *1995.833, *1995.834) and 18 males (AMNH *265994, *266002, *266003, *267069, *267071, *267074, *267075, 267079, *267390, *267394, *267822, *267827; MNHN *1995.827, *1995.828, 1995.829, *1995.830, *1995.831, *1995.832); see table 1 for measurements. I DENTIFICATION :Cormura brevirostris can be distinguished easily from other emballonurids by the unique location of the antebrachial wing sac (Sanborn, 1937: fig. 37; Jones and Hood, 1993: fig. 1). Craniodental characters and measurements of C.brevirostris from the Guianas and other parts of South America were discussed by Sanborn (1937), Husson (1962, 1978), Brosset and CharlesDominique (1990), and Jones and Hood (1993). Husson (1962) provided a revised description of Cormura and a detailed discussion of Myopteryx pullus, which he and all subsequent authors have considered to be a junior synonym of C.brevirostris. No subspecies are currently recognized (Jones and Hood, 1993; Koopman, 1994). Our specimens from Paracou conform to previous descriptions of Cormura brevirostris and fall within the range of size variation previously documented in Surinam and French Guiana. F IELD O BSERVATIONS : We recorded 42 captures (possibly including some recaptures) of Cormura brevirostris at Paracou: 7 in ground-level mistnets, 12 in elevated mistnets, and 23 at roosts. Of the seven groundlevel mistnet captures, one was in welldrained primary forest, three were in creekside primary forest, two were in a treefall opening in creekside primary forest, and one was in a manmade clearing. Eight elevatednet captures were between 10 and 21 m above a narrow dirt road, and four were made between 10 and 38 m over a treefall gap in well-drained primary forest. We encountered 10 different roosting groups of Cormura brevirostris at eight unique roost sites (one site was revisited twice). Five roost sites were on the dark undersides of fallen trees (e.g., fig. 14), but one roosting group was found clinging to an unmodified leaf of Phenakospermum guyannensis (fig. 15), another occupied a shallow cavity in the base of a living tree suspended by its roots over a stream, and another was found under a concrete bridge. In addition to the bridge roost, two natural roosts (both in primary forest) were over small streams, but 1998 31SIMMONS AND VOSS: PARACOU BATS Fig. 16. Roost successively occupied by three species of emballonurids in well-drained primary forest at Paracou. The dark concave underside of this broken trunk (arrow) sheltered roosting groups of three Peropteryx macrotis in 1991, four P.kappleri in 1992, and three Cormura brevirostris in 1994; the roost was found vacant in 1993. 32 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 17. Roost of Peropteryx leucoptera and Furipterus horrens in well-drained primary forest at Paracou (some understory vegetation was selectively removed for photography). Two individuals of P. leucoptera were found roosting in a horizontal chamber (arrow) between two buttresses of this fallen tree in 1992, and one F.horrens occupied a similar chamber on the other side at the same time. This roost was vacant in 1993 and was destroyed by a treefall in 1994. Deep recesses between the buttresses of fallen trees like this provide roosts for several bat species besides the two found here, including Peropteryx kappleri,Choeroniscus minor, and Lonchophylla thomasi. However, whereas emballonurids usually occupy dimly lighted sites between lateral buttresses, glossophagine roosts are always on the darker underside (where they can be seen only by crawling underneath such trees with a headlight). 1998 33SIMMONS AND VOSS: PARACOU BATS Fig. 18. Roost of Peropteryx leucoptera at the margin of an experimental plot in well-drained forest at Paracou. Felled by cuts through the buttresses about 1 m above the ground, this tree was the only roost of P.leucoptera that we found unobstructed by dense understory vegetation. Seven or eight bats were observed roosting in the cave-like recess (arrow) formed by the standing base and the fallen trunk. mura brevirostris in 1992, and we once found Furipterus horrens simultaneously roosting beneath the same fallen tree as P. leucoptera, but in a different chamber separated by large buttresses. Otherwise, this species was not observed to share roost sites with other bats. On several occasions we saw bats with white wings, almost certainly Peropteryx leucoptera, flying regular beats 1–2 m above the ground at dusk in our camp clearing. Peropteryx macrotis (Wagner) Figures 19, 20 V OUCHER M ATERIAL : 2 females (AMNH *266006, *267396) and 3 males (AMNH *266005, *266007; MNHN *1995.843); see table 4 for measurements. I DENTIFICATION : Brosset and Charles-Dominique (1990) recognized two small Peropteryx species in French Guiana that they identified as P.macrotis (originally described by Wagner [1843] based on a Brazilian specimen from Mato Grosso) and P. trinitatis (described by Miller [1899] based on four specimens from Trinidad). Previously, Sanborn (1937) and Goodwin and Greenhall (1961) had concluded that trinitatis was no more than subspecifically distinct from macrotis because specimens referable to these taxa are similar and were not then known from sympatry. Although Handley (1976) subsequently reported that P.macrotis and P.trinitatis occur sympatrically in Venezuela, he provided no discussion of diagnostic characters. In consequence, most recent authors (e.g., Jones and Hood, 1993; Koopman, 1993, 1994) have continued to recognize trinitatis as a subspecies of P.macrotis. According to Brosset and Charles-Dominique (1990), Peropteryx macrotis and P. trinitatis can be distinguished in French Guiana based on size (e.g., forearm length 43.0– 34 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 19. Dorsal (A), ventral (B), and lateral (C) views of the skull of Peropteryx macrotis (AMNH 266005; male) from Paracou. Scale bar 5 5 mm. 1998 35SIMMONS AND VOSS: PARACOU BATS Fig. 20. Close-up lateral views of the upper right dentition of (A)Peropteryx macrotis (AMNH 267396; female) from Paracou and (B) the holotype of P.trinitatis (AMNH 7496; female) from Trinidad. Occlusal views of the upper right dentition (C)P.macrotis (AMNH 267396; female) and (D) the holotype of P.trinitatis (AMNH 7496; female) are shown below. Note the species difference in morphology of the anterior upper premolar. Scale bars 5 5 mm. 36 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY 47.0 mm, maxillary toothrow length 5.3–6.0 mm in P.macrotis, forearm 39.5–40.4 mm, maxillary toothrow 4.7–4.8 in P.trinitatis), skull shape (shorter, more rounded in P.trinitatis), and ear and tragus shape (both narrower in P.trinitatis). They also observed differences in sexual dimorphism (females larger than males in P.macrotis but not in P.trinitatis) and social behavior (social groups of 1–4 individuals in P.macrotis, much larger [ . 100] in P.trinitatis). Unfortunately, all of their observations concerning P.trinitatis were based on a single colony from which only two specimens (one male and one female) were collected (Brosset and Charles-Dominique, 1990). Our voucher material from Paracou appears to represent the same species that Brosset and Charles-Dominique identified as Peropteryx macrotis despite some discrepancies in measurements. Whereas all measurements of the two females in our sample (table 4) fall within the range of variation they reported for female P.macrotis (op. cit.), the forearm measurements of our three males (39.7, 40.1, and 42.2 mm) correspond more closely to values they reported for P.trinitatis. Examination of skull shape and ear morphology, however, suggest that all of our specimens represent P.macrotis sensu Brosset and Charles-Dominique. Therefore, if there are two diagnosable taxa of small Peropteryx in French Guiana, it appears that males cannot be distinguished on the basis of forearm measurements alone. To test our identification of the Paracou specimens, we compared our series with the adult female holotype (AMNH 7496) of Peropteryx trinitatis. The forearm of the holotype measures 41.9 mm, much smaller than those of either of our Paracou females. Although the skull of the holotype is badly damaged, the rostrum and toothrows remain largely intact (fig. 20). The maxillary toothrow length of AMNH 7496 is 5.21 mm, and the breadth across the upper molars is 5.74 mm, values that likewise fall well below those of our female Paracou vouchers (table 4) and of the female specimen identified as P.macrotis by Brosset and Charles-Dominique (1990). We also noted one qualitative difference between the holotype of trinitatis and our voucher material: the anterior upper premolar is peglike and lacks well-defined anterior and posterior cusps in trinitatis, but this tooth is larger and has a distinct posterior cusp in the Paracou specimens. Paratypes (AMNH 7493, 7494, 7495) and topotypes (AMNH 175556, 175558, 175559) of trinitatis resemble the holotype in size and also have tiny, peglike anterior upper premolars. According to Carter and Dolan (1978), the holotype of Peropteryx macrotis is an uncatalogued adult female preserved in alcohol at the Naturhistorisches Museum Wien. Unfortunately, the skull is missing and presumed lost. However, the forearm measurement of the holotype reported by Carter and Dolan, 45.8 mm, corresponds closely with that of our female specimen from Paracou (45.0 mm) and with those identified as P.macrotis by Brosset and Charles Dominique (43.9– 47.0 mm). These measurements are larger than, and do not overlap with, measurements previously reported for females of P.trinitatis (e.g., by Goodwin and Greenhall, 1961). We therefore follow Brosset and CharlesDominique in referring the larger of the two small Peropteryx from French Guiana to P. macrotis. F IELD O BSERVATIONS : We captured only five individuals of Peropteryx macrotis, all of them at roosts. One roosting group of three bats (two adult males and one adult female) was found beneath the broken trunk of a fallen tree cantilevered about 1.5 m above the ground in well-drained primary forest (fig. 16). Another group of three individuals (of which only one adult male and one subadult female were captured) was found in a large hollow log (ca. 60 cm in inside diameter), also in well-drained primary forest. Rhynchonycteris naso (Wied) V OUCHER M ATERIAL : 4 females (AMNH *265985, *265988, *267373; MNHN *1995.844) and 6 males (AMNH *265981, *265986, *265987, *267372; MNHN *1995.845, *1995.846); see table 7 for measurements. I DENTIFICATION : Sanborn (1937) and Jones and Hood (1993) discussed cranial and dental characters of Rhychonycteris naso, and Husson (1962, 1978) provided a detailed description and comparative measurements. 1998 37SIMMONS AND VOSS: PARACOU BATS Measurements of specimens previously collected in French Guiana were reported by Brosset and Charles-Dominique (1990). No subspecies are currently recognized (Koopman, 1994). Our voucher specimens conform to previous qualitative descriptions of Rhynchonycteris naso and measurements fall within the range of size variation previously reported for the species. However, our morphometric data (table 7) somewhat exceed the range of size variation known from the Guianas, representing specimens both slightly smaller and slightly larger than those reported by Husson (1962, 1978) and Brosset and Charles-Dominique (1990). F IELD O BSERVATIONS : We captured 10 individuals of Rhynchonycteris naso, all of which were found roosting over water in large (3–4 m in diameter) metal culverts or under concrete bridges. The five roosting groups we observed ranged in size from three to about nine individuals. One group of three bats consisted of two adult males and one adult female, and one group of about nine contained at least two adult males, but our information about the age and sex composition of the other groups is too incomplete to be informative. Saccopteryx References useful for identifying species of Saccopteryx include Sanborn (1937) and Jones and Hood (1993); both papers provide keys based on external and craniodental morphology. Although three size-graded species (from large to small, S.bilineata,S.leptura, and S.canescens) are commonly collected sympatrically in Amazonia, the small species at Paracou is S.gymnura, previously unknown from the Guianas. Saccopteryx bilineata (Temminck) V OUCHER M ATERIAL : 11 females (AMNH *265963, *265965, *267058, *267060, *267377, *267379, *267842; MNHN *1995.847, *1995.848, *1995.849, *1995.850) and 11 males (AMNH *265962, 265964, *265966, *267057, *267064, *267374, *267378, *267840; MNHN *1995.851, *1995.852, *1995.853); see table 8 for measurements. One individual of unknown sex (AMNH 266977) was recovered from the crop of a bat falcon. I DENTIFICATION : Descriptions and measurements of Saccopteryx bilineata from the Guianas and elsewhere were provided by Thomas (1904), Sanborn (1937), Goodwin and Greenhall (1961), Husson (1962, 1978), and Brosset and Charles-Dominique (1990). There is considerable disagreement concerning trinomial nomenclature. Thomas (1904) explicitly recognized only two subspecies, S. b.bilineata in northern South America (including Trinidad), and S.b.centralis in Central America. Subsequently, Sanborn (1937) argued that size variation in this species was continuously distributed and precluded delin- 38 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY eation of subspecies in the absence of other significant characters; in his revision, the names insignis,perspicillifer, and centralis were therefore regarded as strict synonyms of S.bilineata. Although many authors have followed Sanborn’s (1937) opinion (e.g., Husson, 1962, 1978), others have not. Cabrera (1958), for example, recognized S.perspicillifer (from Trinidad) as a distinct species, and Goodwin and Greenhall (1961) later treated perspicillifer as a valid subspecies of S.bilineata. Most recently, Koopman (1994) recognized S.b.bilineata as ranging from tropical Mexico throughout Central America and tropical South America except northern Venezuela, Trinidad, and Tobago, and S.b.perspicillifer as occuring in northern Venezuela, Trinidad, and Tobago). Recognition of perspicillifer as a taxon distinct from bilineata is apparently based entirely on size, with specimens from northern Venezuela, Trinidad (the type locality), and Tobago supposedly being larger and more robust than specimens from elsewhere in the Neotropics (Miller, 1899; Goodwin and Greenhall, 1961). However, this putative geographic pattern is not supported by available data. Instead, our personal observations and data reported by Thomas (1904), Sanborn (1937), Husson (1962, 1978), and Brosset and Charles-Dominique (1990) indicate that specimens from eastern Ecuador, Guyana, Surinam, French Guiana, Brazil, and Bolivia exhibit similar forearm, skull, and dental measurements as specimens from northern Venezuela, Trinidad, and Tobago. Thus, although significant geographic variation may exist among some populations of Saccopteryx bilineata, none of the subspecies 1998 39SIMMONS AND VOSS: PARACOU BATS Fig. 21. Roost of Saccopteryx bilineata and Micronycteris hirsuta in creekside primary forest at Paracou. Small roosting groups of S.bilineata occupied the half-open basal part of the central cavity (large arrow) where they were just visible without artificial illumination in 1992 and 1994. A roosting group of M.hirsuta occupied the completely enclosed dark interior (small arrow) several meters above the colony of S.bilineata in 1992. In 1993 this roost was occupied only by Carollia perspicillata. 40 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY traditionally recognized by authors appear to be justified by the evidence at hand. Our voucher material from Paracou conforms with previous descriptions of the species, and measurements fall within the range of size variation previously documented from the Guianas. F IELD O BSERVATIONS : We recorded 34 captures (possibly including some recaptures) of Saccopteryx bilineata: 14 in ground-level mistnets, 11 in elevated mistnets, and 9 at roosts; additionally, 1 specimen was recovered from the crop of a road-killed Falco rufigularis (which had also eaten a specimen of S.leptura). Of the 14 ground-level mistnet captures, 4 were made in well-drained primary forest, 2 in swampy primary forest, 3 in creekside primary forest, and 5 in manmade clearings. Six S.bilineata were taken in nets suspended 10–21 m over a narrow dirt road, two were netted 10–13 m over a treefall gap in creekside primary forest, and three were netted 7–10 m above the ground in the subcanopy of swampy primary forest. Six of our mistnet captures were made before dark, between 18:00 and 18:35 hours. We found five roosting groups of Saccopteryx bilineata at Paracou. All of these occupied more-or-less vertical cavities (by contrast with the essentially horizontal chambers typically used by Cormura brevirostris and Peropteryx spp.) with not-quite-completely dark interiors (the roosting bats were usually just visible without illumination), but other roost characteristics differed. One roost was in a 2-m-high chimney, open above and below, formed by the fused buttresses of a large tree (fig. 3); another was in the semicylindrical basal opening of a central cavity that extended far up into the trunk of a much smaller tree (fig. 21); two were in the open bases of hollow trees big enough for a grown man to stand inside; and the fifth was an open cleft, 12 m above the ground at its lowest point, in the trunk of another big tree (accessed by climbing a rope). Roosting groups that we were able to count varied from one to five individuals. Two individuals that we found roosting alone were both adult males. One group of five that we captured in its entirety consisted of two adult males, two adult females, and one juvenile. Saccopteryx gymnura Thomas Figure 22 V OUCHER M ATERIAL : 2 males (AMNH *267843; MNHN *1995.862); see table 8 for measurements. I DENTIFICATION : Characters and measurements of Saccopteryx gymnura were discussed by Thomas (1901b) and Sanborn (1937). No subspecies are current recognized (Koopman, 1994). Our material of Saccopteryx gymnura from Paracou represents a range extension of approximately 900 km for this species, which was previously known only from a few localities along the lower Amazon in Brazil (Koopman, 1994). Despite this considerable geographic hiatus, the Paracou specimens conform closely with published descriptions of the holotype (Thomas, 1901b; Sanborn, 1937) and are indistinguishable from other Brazilian material referable to S.gymnura (AMNH 93519, 93520; USNM 392995, 460080). The only species that might be confused with Saccopteryx gymnura is S.canescens. Although similar in size, these tiny bats can be distinguished unambiguously by (1) dorsal fur color (dark brown in gymnura, brown with strong grayish or yellowish frosting in canescens), (2) dorsal fur markings (stripes absent or very faint in gymnura, a pair of white stripes always visible and often bright in canescens), (3) ventral fur banding (unicolored dark brown in gymnura, bicolored black or dark brown with tan tips in canescens), (4) origin of the posterior part of the wing membrane (from the metatarsals in gymnura, from the tibia in canescens), and (5) length of the forearm (33.5–35.3 mm in gymnura, 35.8–40.8 mm in canescens). Although there is some species overlap in length of the maxillary toothrow, our specimens of gymnura have shorter toothrows (4.6 mm) than those previously reported for specimens of canescens (4.9–5.0 mm) from French Guiana (Brosset and Charles-Dominique, 1990). F IELD O BSERVATIONS : Both of our specimens of Saccopteryx gymnura were captured in ground-level mistnets, one in a small clearing bordered by well-drained primary forest (fig. 2), the other over a narrow dirt 1998 47SIMMONS AND VOSS: PARACOU BATS French Guiana by Brosset and Charles-Dominique (1990). F IELD O BSERVATIONS : All of the five Desmodus rotundus we captured at Paracou were taken in ground-level mistnets. Two individuals were taken in well-drained primary forest, two in swampy primary forest, and one in a manmade clearing. Diaemus youngi (Jentink) V OUCHER M ATERIAL : 1 female (AMNH *268571) and 2 males (AMNH *266347; MNHN *1995.984); see table 11 for measurements. I DENTIFICATION : Descriptions and measurements of Diaemus youngi appear in Husson (1962, 1978), Swanepoel and Genoways (1979), Hall (1981), Koopman (1988), and Greenhall and Schutt (1996). Some authors place D.youngi in the genus Desmodus (e.g., Handley, 1976; Emmons, 1990), but we retain Diaemus as a distinct genus following Cabrera (1958), Husson (1962, 1978), Jones and Carter (1976), Hall (1981), Koopman (1993, 1994), and Greenhall and Schutt (1996). Although Thomas (1928c) named D. y.cypselinus based on material from Peru, Husson (1962: 198) noted that cypselinus was probably ‘‘not subspecifically different from the typical form.’’ Subspecies of Diaemus youngi have not been recognized by most authors (e.g., Cabrera, 1958; Koopman, 1993, 1994; Greenhall and Schutt, 1996). Our material of Diaemus youngi from Paracou conforms closely with previous qualitative descriptions of the species. Although forearm and cranial dimensions of our specimens are considerably smaller than those reported for specimens from Mexico, Costa Rica, Peru, Venezuela, and Trinidad, they are only slightly smaller than those previously reported from the Guianas (Thomas, 1928b; Goodwin and Greenhall, 1961; Husson, 1962; Hall, 1981; Brosset and Charles-Dominique, 1990; Greenhall and Schutt, 1996). The significance of this variation is not clear, but we note that the Guyanese holotype of D.youngi (RNH 12088) is one of the smallest specimens hitherto reported in the literature (Husson, 1962; Carter and Dolan, 1978). Thus, if subspecies are recognized in some future revisionary study, our material would presumably be referred to the nominate form. F IELD O BSERVATIONS : Two of the three Diaemus youngi we captured at Paracou were taken in ground-level mistnets, one in well-drained primary forest, and the other in a manmade clearing. The third specimen was captured in a mistnet suspended 17–20 m above a narrow dirt road. GLOSSOPHAGINAE We captured five glossophagine species at Paracou, but one or two additional species might also be expected in our study area (appendix 1). Great care must be exercised in identifying glossophagines in the field, as even some genera can be hard to distinguish externally until familiarity is gained with the nuances of character expression in this group. Anoura caudifera (E. Geoffroy) V OUCHER M ATERIAL : 1 male (AMNH *267290); see table 12 for measurements. I DENTIFICATION : Keys to the species of Anoura were provided by Tamsitt and Nagorsen (1982) and Handley (1984), but new species have been described subsequently (Molinari, 1994) and additional taxa still await description (Emmons, 1997). Handley (1984) and 48 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Molinari (1994) are especially useful references because they tabulated measurement data and provided detailed morphological comparisons among species that could easily be confused. Other useful references containing descriptions and measurements of A.caudifera 6 are Husson (1962, 1978), Tamsitt and Valdivieso (1966), Taddei (1975b), Swanepoel and Genoways (1979), Williams and Genoways (1980a), Webster and Jones (1984), and Brosset and Charles-Dominique (1990). No subspecies of A.caudifera are currently recognized (Koopman, 1994; Molinari, 1994). Although our single specimen agrees in most respects with previous descriptions of Anoura caudifera, several of its measurements fall in the zone of morphometric overlap between A.caudifera and a smaller species recently described by Molinari (1994), A.luismanueli. Like two individuals of caudifera reported from Surinam by Williams and Genoways (1980a), the Paracou specimen lacks a distinct tail enclosed within the uropatagium, and thus total body length is reduced. Forearm length of the Paracou specimen likewise falls at the bottom end of the range of variation previously reported for A. caudifera (e.g., by Tamsitt and Valdivieso, 1966). Molinari (1994) suggested that Tamsitt and Valdivieso may have included specimens of luismanueli or another small undescribed Anoura in their samples, but Molinari’s estimate of the range of forearm length in caudifera differs little from those of previous authors. Our specimen has a forearm slightly smaller than Molinari reported for true caudifera (33.6 mm as compared with 34.6), a difference that we do not regard as significant given the range of variation commonly seen within other phyllostomid species. The Paracou specimen also has a shorter maxillary toothrow than do specimens of caudifera previously reported from French Guiana and Venezuela, but in this respect it resembles Surinamese material measured by Husson (1962, 1978), Williams and Genoways (1980a), and Molinari (1994). Comparisons of our voucher to Molinari’s 6 We follow Handley’s (1984) spelling of the specific epithet, which Koopman (1993) and most other authors have incorrectly spelled ‘‘caudifer.’’ (1994) descriptions of Anoura luismanueli and A.caudifera are otherwise consistent with its identification as a small example of caudifera. The Paracou specimen has a longer calcar and broader uropatagium than seen in luismanueli, and it lacks a continuous, dense fringe of hairs along the uropatagial edge. Instead, the free margin of the uropatagium is lightly furred near the midline but is naked laterally, a trait diagnostic of caudifera (Molinari, 1994). Other aspects of the pelage of the Paracou specimen agree with Molinari’s description of caudifera with the exception of the color of the bases of the hairs in our specimen, which are grayishwhite rather than yellowishor creamy-white as he described. Given the range of variation in basal fur color that we have observed in other species, we interpret this as within-species variation. F IELD O BSERVATIONS : Our single example of Anoura caudifera was taken in a groundlevel mistnet in creekside primary forest. Choeroniscus minor (Peters) Figure 24 V OUCHER M ATERIAL : 8 females (AMNH *266120, *266121, *266123, *266377, *267153; MNHN *1998.668, 1998.669, *1998.670) and 5 males (AMNH *266124, *267946, *267947, *267948; MNHN 1998.671); see tables 13 and 14 for measurements. I DENTIFICATION : The complex taxonomic history of Choeroniscus, particularly of the smaller species originally placed in the genus Choeronycteris, is unfortunately relevant for understanding the correct identification of our Paracou vouchers. Choeronycteris minor was first described by Peters (1868) based on a specimen from Surinam. Although Goodwin and Greenhall (1961) stated that the type was destroyed in World War II, Husson (1962) reported that Peter’s original specimen was, in fact, still intact at the Stuttgart museum. Husson (1962) published measurements and a photograph of the holotype (SMN 441), and also provided a detailed description. A second small species, Choeronycteris intermedia, was described by Allen and Chapman (1893) based on specimens from Trinidad, and Tho- 1998 49SIMMONS AND VOSS: PARACOU BATS mas (1912a) named a third, Choeronycteris inca, based on a specimen from Peru. Thomas (1928a) subsequently subdivided Choeronycteris, placing minor,intermedia, and inca in his new genus Choeroniscus. Considerable confusion has surrounded the use of these three specific epithets (Thomas, 1912a; Sanborn, 1954; Goodwin and Greenhall, 1961; Husson, 1962; Koopman, 1978; and Williams and Genoways, 1980a), much of it the result of small available sample sizes and marked sexual dimorphism (Genoways et al., 1973; Koopman, 1978). The latter factor is crucial because the holotypes of minor and inca are males, while that of intermedius is a female. Conflicting reports concerning the morphology and sex of the supposedly lost holotype of minor only served to compound the real biological problem, whether these names really represent three or fewer species. There is general agreement that Allen and Chapman (1893) named their Tinidadian material as a new species in part because several measurements of minor were incorrectly reported by Peters (1868). Thomas (1912a: 404) observed that The Trinidad bat described by Allen and Chapman as Ch.intermedia is very probably the same as Ch.minor, as those authors were deceived by Peter’s impossible measurement of 11 mm. for the calcar, an organ drawn as about 5 mm. long in the more recently published plate of the latter species . . . Thomas (1912a) then described inca based on dental proportions (length and width of the premolars and molars) in which it supposedly differed from minor. He also noted that the skull of inca was larger and broader than that of minor, but this comparison was apparently based on a specimen of minor other than the holotype, as measurements of the latter do not confirm his observation (table 13). Most recent authors have followed Koopman (1978) in regarding inca as a junior synonym of minor, and intermedius as a potentially distinct species. Koopman’s (1978: 8) conclusion that two species could be recognized was based on observations concerning dental variation and length of the rostrum: Study of a number of skulls of Choeroniscus from Trinidad, British Guiana, Brazil, Peru, and Ecuador [has] convinced me that two species can be distinguished . . . by rostral length, which is reflected in maxillary tooth row length. Thomas (1912[a]) distinguished inca from minor solely by the relative sizes of the premolars and molars. On the basis of these skulls, I can see considerable variation in this character, but no clearcut distinction between two forms. The species with the longer rostrum would then be called C.minor (inca a synonym) and the species with the shorter rostrum would stand as C.intermedia [sic]. Koopman (1994) gave the range of minor as extending from coastal Ecuador to the mouth of the Amazon, northward to eastern Venezuela, and southward to northwestern Bolivia. Although most specimens of small Choeroniscus recently collected in the Guianas have been identified as intermedius (e.g., by Brosset and Charles-Dominique, 1990), a long-muzzled female from Guyana (AMNH 140471) was referred to minor by Williams and Genoways (1980a). In attempting to identify our material from Paracou, we examined the holotype of intermedius; Husson’s (1962) measurements, photographs, and description of the holotype of minor; Thomas’s (1912a) description of the holotype of inca; specimens referred to intermedius and minor in the AMNH; and measurements of additional specimens referred to intermedius and minor by Genoways et al. (1973), Genoways and Williams (1980a), Ochoa and Ferna´ndez (1982), Brosset and Charles-Dominique (1990), and Anderson (1997). Although we kept males and females separate, and investigated variation in size, skull shape, and dental morphology, we found no consistent differences among named forms of small Choeroniscus. For example, dental morphology in our series of 13 specimens from Paracou is highly variable, with some individuals having much shorter or longer teeth than others, others having narrower or broader teeth, etc. These differences form no consistent pattern and do not appear to be correlated with body size or with rostral proportions. Because our sample includes individuals with both ‘‘minor-type’’ and ‘‘inca-type’’ teeth (sensu Thomas, 1912a), as well as others with intermediate conditions, we follow Koopman (1978, 1993, 1994) in concluding that minor and inca are conspecific. Dental morphology has never been used to separate minor (or inca) from intermedius. 50 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 24. Ventral views of the skull of six individuals of Choeroniscus minor from Paracou showing intraspecific variation in craniodental morphology. (A) AMNH 266377 (female); (B) MNHN 1995.1271 (female); (C) AMNH 266123 (female); (D) AMNH 267948 (male); (E) AMNH 267946 (male); (F) AMNH 267947 (male). Note variation in relative size of the premolars and molars, skull length and breadth, rostral proportions, and form of the palate. This series includes individuals with ‘‘minor-type’’ 1998 51SIMMONS AND VOSS: PARACOU BATS ← dentitions (short premolars and long molars; e.g., E), others with ‘‘inca-type’’ teeth (premolars and molars subequal in length; e.g., D), and dental intermediates (e.g., A). It also includes large individuals with a relatively long (‘‘minor-type’’) rostrum (e.g., C), smaller specimens with a shorter (‘‘intermediustype’’) rostrum (e.g., A), and intermediates (e.g., B). Scale bar 5 10 mm. Instead, measurements of the skull, rostrum, and maxillary toothrow appear to be the only characters so employed in recent years (Koopman, 1978; Williams and Genoways, 1980a). Rather than confirming that minor and intermedius are distinct species, our comparisons of measurement data show the opposite. Whereas measurements of our female specimens from Paracou broadly overlap those of the female holotype of intermedius, measurements of our males overlap those of the male holotypes of minor and inca (tables 13, 14). For both sexes, measurements of our sample correspond closely to those reported for a large series (10 males and 26 females) of topotypical intermedius from Trinidad (Genoways et al., 1973). Measurements of other specimens from Venezuela and the Guianas (reported by Genoways and Williams [1980a], Ochoa and Ferna´ndez [1982], and Brosset and Charles-Dominique [1990]) and Bolivia (Anderson, 1997) further fill the morphometric gap that Koopman (1978) observed between his small series of minor and intermedius. Considering the possiblity that our largest voucher specimens 52 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY might represent minor and the smallest intermedius, we carefully examined variation in external and craniodental morphology within our series, but ultimately found no evidence that the Paracou material represents more than one species. In our view, specimens identified as intermedius in the literature simply represent the lower end of a normal range of within-species size variation, while those identified as minor and inca represent the upper end of this range. We therefore conclude that these epithets represent a single species for which the oldest available name is Choeroniscus minor (Peters, 1868). Recognition of subspecies appears to be unjustified based on the data currently at hand. F IELD O BSERVATIONS : We captured 13 individuals of Choeroniscus minor at Paracou, of which 7 were taken in ground-level mistnets and 6 at roosts. Three individuals were netted in swampy primary forest, three in creekside primary forest, and one in a manmade clearing. One roosting group of six bats was found on the dark underside of a fallen tree with large buttresses (similar to that shown in fig. 17) in well-drained primary forest; of this group, one adult male and four adult females were captured, and one bat escaped. Another roosting bat, a solitary adult male, was found hanging beneath the undercut bank of a dry streambed in well-drained primary forest. Glossophaga soricina (Pallas) V OUCHER M ATERIAL : 10 females (AMNH *267134, *267137, *267138, *267448, *267449, *267953, *267956, 267958; MNHN 1998 53SIMMONS AND VOSS: PARACOU BATS *1995.1040, *1995.1041) and 25 males (AMNH *266090, *266091, *266092, *266093, *266094, *266095, 266099, *267139, *267140, *267949, *267950, *267951, 267954, *267955, 267957, *267959; MNHN *1995.1042, *1995.1043, *1995.1044, *1995.1045, *1995.1046, *1995.1047, 1995.1048, *1995.1049, *1995.1050); see table 15 for measurements. I DENTIFICATION : The most useful reference for identifying species of Glossophaga is Webster’s (1993) revision, which includes a key as well as detailed descriptions and comparative measurements. Alvarez et al. (1991) provided a good description of G.soricina, but the principal reference for the species is also Webster’s revision. Five subspecies of G.soricina are currently recognized, of which three occur in South America: G.s. handleyi (Mexico throughout Central America to northern and western Colombia), G.s. soricina (South America east of the Andes from Colombia and Venezuela in the north to Paraguay and northern Argentina), and G. s.valens (drier areas of western Ecuador and Peru) (Alvarez et al., 1991; Webster, 1993; Koopman, 1994). Measurements of specimens taken in the Guianas were reported by Husson (1962, 1978), Brosset and CharlesDominique (1990), and Webster (1993). Our Paracou material conforms in all respects with Webster’s (1993) description of Glossophaga soricina soricina, and their measurements (table 15) fall within the range of variation previously reported for G.soricina in Brazil and the Guianas (Husson, 1962, 1978; Taddei, 1975b; Brosset and Charles-Dominique, 1990; Webster, 1993). F IELD O BSERVATIONS : We recorded 56 captures (possibly including some recaptures) of Glossophaga soricina at Paracou, of which 45 were in ground-level mistnets and 11 were at roosts. Seven of the mistnet captures were in well-drained primary forest, 11 were in swampy primary forest, 3 were in creekside primary forest, and 24 were in manmade clearings. The proportional difference in capture-habitat frequencies between this species and Lonchophylla thomasi, the only other common glossophagine at Paracou (see below), is noteworthy: G.soricina was more commonly netted in modified habitats and L. thomasi in primary habitats than would be expected if these sympatric nectarivores did not differ in habitat use (table 16). Although this result is consistent with Brosset and 54 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Charles-Dominique’s (1990) characterization of G.soricina as a species of modified biotopes, we note that specimens are in fact known from ecologically pristine localities in French Guiana (e.g., USNM 548471, 548472, collected by L. H. Emmons at Saut Parare´ in 1984). Of the three roosts of Glossophaga soricina we found at Paracou, two were in tree cavities (figs. 25, 26) and one was under under a concrete bridge. The bridge roost contained many G.soricina, perhaps hundreds (of which seven males and one female were collected as vouchers), as well as roosting groups of Rhynchonycteris naso and Carollia perspicillata. One tree cavity roost occupied by an adult male-female pair of G.soricina (fig. 26) was also shared with C.perspicillata; the other tree cavity roost was occupied a solitary adult male. Lichonycteris obscura Thomas V OUCHER M ATERIAL : 1 female (AMNH *267960); see table 15 for measurements. I DENTIFICATION : Two color morphs occur among specimens of Lichonycteris, a dark brown form that agrees with Thomas’ (1895) original description of L.obscura, and a pale brown form that is often identified as L.degener Miller (1931). However, the taxonomic status of the latter is problematic due to confusion regarding patterns of variation in both craniodental and pelage characters (Gardner, 1976; Hill, 1985; Ochoa et al., 1993). Pending a revision of the genus, we follow Gardner (1976), Hill (1985), Ochoa et al. (1993), and Koopman (1993, 1994) in regarding degener as a junior synonym of obscura. Descriptions and comparative measurements of L.obscura so defined can be found in Husson (1962, 1978), Gardner (1976), Swanepoel and Genoways (1979), Hill (1985), Brosset and Charles-Dominique (1990), and Ochoa et al. (1993). Our chocolate-brown specimen agrees closely with previous descriptions of Lichonycteris obscura, and falls within the range of variation in measurements previously reported in the literature. F IELD O BSERVATIONS : Our specimen was captured in a ground-level mistnet in welldrained primary forest. Lonchophylla thomasi J. A. Allen V OUCHER M ATERIAL : 15 females (AMNH *266100, *266107, *266108, *266109, *267147, *267148, *267149, *267451, *267452, *267939; MNHN *1995.1096, *1995.1097, *1995.1098, *1995.1099, *1995.1100) and 23 males (AMNH *266101, *266102, *266103, *266104, *266105, *266106, *266110, *266113, *266114, *266116, *266117, 267150, *267940, *267941, *267942, *267943; MNHN *1995.1101, *1995.1102, *1995.1103, *1995.1104, *1995.1105, 1995.1106, *1995.1107); see table 17 for measurements. I DENTIFICATION : The most useful reference for identifying species of Lonchophylla is Taddei et al. (1983), who provided a key and discussed variation. We consulted descriptions and comparative measurements of Lonchophylla thomasi in Husson (1962, 1978), Hill (1964, 1980), Gardner (1976), Taddei et al. (1978, 1983), Swanepoel and Genoways (1979), and Brosset and Charles-Dominique (1990). No subspecies of L.thomasi are currently recognized (Taddei et al., 1983; Koopman, 1994). Our voucher material, one of the largest series available from a single locality, conforms in all respects to previous descriptions of Lonchophylla thomasi. F IELD O BSERVATIONS : We recorded 55 captures (possibly including some recaptures) of 1998 55SIMMONS AND VOSS: PARACOU BATS Fig. 25. A large kimboto, Pradosia cochlearia (Sapotaceae), a common canopy emergent in welldrained primary forest at Paracou that often has rotted cavities sheltering bats and other mammals. In 1991 the cavernous lower chamber of this tree (arrow) contained a solitary Glossophaga soricina, but in 1992 we found a roosting group of four Micronycteris megalotis and another consisting of one Carollia perspicillata in the same space; in 1993 and 1994 this roost was occupied by small groups of C.perspicillata only. 56 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 26. Roost of Glossophaga soricina,Phyllostomus elongatus, and Carollia perspicillata in welldrained primary forest at Paracou. This tree cavity, about 60–70 cm in diameter and several meters high, contained about 20 C.perspicillata (two of which, both adult females, were collected as vouchers) and one adult male P.elongatus in 1992. In 1993 the same space contained about 10 C.perspicillata and 2 G.soricina (an adult male-female pair). It is possible that this roost was used by Vampyrum spectrum in 1991, when RSV observed two huge bats vocalizing loudly and flying around the opening at night; however, the roost was vacant when we climbed the tree to investigate two days later. 1998 63SIMMONS AND VOSS: PARACOU BATS Fig. 28. Dorsal (A), ventral (B), and lateral (C) views of the skull of the holotype of Micronycteris brosseti (MNHN 1995.1030; male). Scale bar 5 10 mm. 64 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY *266034, *266035, *267419), all of which were likewise collected at Paracou. See tables 22 and 23 for measurements. R EFERRED M ATERIAL : In addition to the type series from Paracou, we also refer nine specimens to Micronycteris brosseti from four other South American localities as follows: PERU, Loreto, Puerto Indiana, Rı´o 1998 65SIMMONS AND VOSS: PARACOU BATS Amazonas near mouth of Rı´o Napo (AMNH 73495, 73496, 73498, 73499); Peru, Loreto, Jenaro Herrera (MUSM 5528); Peru, Loreto, Rı´o Yavarı´ Mirı´m, Quebrada Esperanza (FMNH 89100, 89101, 89102); BRAZIL, Sa˜o Paulo, Rio Juquia´, Barra (FMNH 92997). One of these specimens (MUSM 5528) was previously identified as M. schmidtorum by Ascorra et al. (1991a). D ISTRIBUTION : Although specimens of Micronycteris brosseti are currently known only from eastern Peru, French Guiana, and southeastern Brazil, the species could reasonably be expected to occur throughout Amazonia and the Brazilian Atlantic forest. E TYMOLOGY : We name this species to honor Andre´ Brosset in recognition of his many important contributions to knowledge of tropical bat faunas in India, Africa, and South America. In particular, Brosset and Charles-Dominique (1990) provided an indispensable guide to the bat fauna of French Guiana, as the tattered covers, creased pages, and annotated margins of our well-thumbed field copy eloquently attest. D IAGNOSIS : A small Micronycteris with dark brown dorsal fur and pale (gray or buff) ventral fur; dorsal hairs bicolored, 7–8 mm long in shoulder region, with white base comprising ⅓ – ½ of each hair; fur on outside of medial third of pinna short ( # 3 mm) and dense; pinnae large with rounded tips, connected across crown of head by low interauricular band with shallow midline notch; thumb small; second phalanx of wing digit IV shorter than first; calcar longer than foot; mastoid breadth less than zygomatic breadth; crown height of upper incisors not reduced; no gap present between outer upper incisor and canine; P3 and P4 subequal in anteroposterior length, height of P3 slighly less than that of P4; P4 with small lingual heel and poorly developed lingual cusp; M1 narrower than M2; no noticeable gap between posterior edge of cingulum of M2 and anterolingual edge of M3 when toothrow seen in occlusal view; lower incisors small and bilobed; p2 and p4 subequal in anteroposterior length and height, p3 smaller in both dimensions; coronoid process high, upper margin of ascending process with steep slope (25–30 8 ). Of the characters listed above, four are particularly useful for field identification of M.brosseti: color of the ventral fur (pale gray or buff), length of the fur on the leading edge of the pinna (short, # 3 mm), relative length of the first and second phalanges of wing digit IV (second shorter than first), and the calcar (longer than foot). Within the genus Micronycteris, this combination of traits is unique to M.brosseti. M EASUREMENTS : A summary of measurements of Micronycteris brosseti, including those of the type series and of all other referred specimens, is provided in table 23. D ESCRIPTION AND C OMPARISONS :Micronycteris brosseti requires close comparison with six other like-size congeners with which it might be confused, namely M.homezi,M. megalotis,M.microtis,M.minuta,M.sanborni, and M.schmidtorum. By contrast, M. hirsuta stands out from all other members of the genus by its larger size and unique lower incisor morphology (see table 24 and species account for hirsuta below). The following paragraphs therefore omit hirsuta from further consideration and focus only on the remaining seven species. Micronycteris brosseti is one of the smallest members of the genus, with mean values for most of its measurements falling well below those of other species found at Paracou (tables 23–25). Despite some morphometric overlap between larger individuals of brosseti and a few smaller specimens in the congeneric series measured by Simmons (1996b: table 2), measurements are clearly useful for distinguishing brosseti from most other Micronycteris species where they occur sympatrically. At Paracou, the only species from which brosseti is indistinguishable in univariate morphometric comparisons (tables 23–25) is microtis. At Paracou, brosseti is consistently smaller than megalotis in body weight (females only), ear length, thumb length, greatest length of skull (males only), braincase breadth (females only), mastoid breadth, zygomatic breadth (females only), maxillary toothrow length, and breadth across molars (males only). Similarly, brosseti is smaller than homezi (known only from one male; see species account below) in body weight, ear length, forearm length, tibia length, and in 66 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY all craniodental measurements. In comparison with minuta,brosseti is consistently smaller in body weight, ear length, greatest length of skull (females only), condyloincisive length (females only), postorbital breadth, braincase breadth (females only), mastoid breadth, maxillary toothrow length (females only), and breadth across molars (females only). Likewise, brosseti is smaller than schmidtorum in many measurements, including body weight, ear length, tibia length, thumb length, and all craniodental measurements. The entire dorsal pelage of brosseti is composed of bicolored hairs with white bases and brown tips. In the upper back region, the white bases comprises approximately one-third to one-half of the length of each 1998 67SIMMONS AND VOSS: PARACOU BATS hair. This is comparable to the pattern seen in microtis,megalotis,schmidtorum, and homezi, in which the white bases comprise one-fourth to one-half of each hair on the upper back. In contrast, the white bases are much more extensive in minuta (one-half to two-thirds of each hair) and sanborni (twothirds to three-fourths of each hair). The length of the dorsal fur over the upper back in brosseti is approximately 7–8 mm, comparable to that of microtis and homezi. In contrast, minuta and sanborni have shorter fur over the shoulders (5–7 mm), while megalotis and schmidtorum have longer fur (8–11 mm). The ventral fur of brosseti (including that covering the neck and throat) is either pale gray or pale buff, whereas that of megalotis 68 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY and microtis is brown (approximately the same shade as the dorsal fur). The underparts of schmidtorum and minuta are also pale (gray or buff), like brosseti, but sanborni has bright white ventral fur. All of the Paracou specimens of brosseti have a pale gray venter, as do some specimens from Peru (e.g., MUSM 5528). However, some older and somewhat faded Peruvian specimens of brosseti have pale buff venters (e.g., AMNH 73495–73499, collected in 1926). It is not clear if the buff color is the result of postmortem alteration or within-species variation. Both ventral fur-color variants can be observed among museum series of other species (e.g., minuta and schmidtorum), so this variation (whatever its cause) is evidently not peculiar to brosseti. The fur on the outside of the medial third of the ear pinna is short ( # 3 mm) and dense in brosseti, similar to the condition seen in microtis,minuta, and sanborni. By contrast, the auricular fur of megalotis,schmidtorum, and homezi is much longer (5–8 mm). As in all other species of Micronycteris, the ears of brosseti are connected across the crown of the head by an interauricular band of skin. In brosseti, this band is low and nearly horizontal in profile, and is marked at the midline by a shallow notch; a similar condition is seen in megalotis and microtis. This morphology contrasts sharply with the interauricular band in minuta,sanborni, and homezi, which is much higher and divided by a deep median notch into two roughly triangular flaps (see species account for homezi below). The interauricular band of schimidtorum is of intermediate height with a moderately deep median notch. The second phalanx of wing digit IV is shorter than the first in brosseti, as it is also in megalotis,microtis,schmidtorum, and homezi. By contrast, the first and second phalanges of digit IV are subequal in minuta and sanborni. The calcar is distinctly longer than the hindfoot in brosseti, resembling the condition seen in megalotis,microtis, and schmidtorum. The calcar is approximately equal in length to the hindfoot in sanborni, however, and the calcar is shorter than the foot in minuta and homezi. As noted by Simmons (1996b), cranial morphology is remarkably similar in all species of Micronycteris. Like other members of the genus, brosseti has a relatively long rostrum, no anteorbital inflation, a large braincase, and shallow basisphenoid pits. Mastoid breadth is less than zygomatic breadth in brosseti, as it is likewise in megalotis,microtis, and schmidtorum; mastoid breadth is greater than zygomatic breadth in minuta, sanborni, and homezi. Like all other congeneric species, brosseti has a dental formula of I 2/2, C 1/1, P 2/3, M 3/3 3 2 5 34. The crown height of the upper incisors is not reduced, and no gap is present between the outer upper incisor and the canine. In this respect, brosseti resembles most species of Micronycteris; the contrasting morphology is seen only in sanborni. Although differences in the upper premolar dentition are subtle, variation in this region provides an important means of distinguishing species of Micronycteris (see Simmons, 1996b). In brosseti, P3 and P4 are subequal in anteroposterior length, and the height of P3 is very slighly less than that of P4. This is similar to the condition seen in megalotis,microtis,schmidtorum, and homezi. Some variation is known in the former three species, with P3 and P4 subequal in height in some individuals and P3 slightly smaller than P4 in others. Although we did not observe such variation in brosseti (or homezi, known from only one specimen), it would not be unexpected. In contrast to the conditions described above, P3 is noticeably smaller than P4 in both length and height in minuta and sanborni. The degree of development of the posterolingual heel and lingual cusp on P4 varies among and within species of Micronycteris. In brosseti, P4 has a small lingual heel and a poorly developed lingual cusp. The lingual heel of P4 in brosseti is similar in size to those of megalotis,microtis,schmidtorum, and homezi, but the lingual heel is generally smaller in minuta and larger in sanborni. The lingual cusp of P4, which is formed from the raised edge of the tooth, is poorly developed and lacks a sharp point in brosseti,homezi, minuta, and sanborni, and it is entirely lacking in many specimens of schmidtorum.By contrast, this cusp is better developed and tapers to a sharp point in most specimens of 1998 69SIMMONS AND VOSS: PARACOU BATS microtis and megalotis; however, some individuals of both species have a poorly developed cusp, while this cusp may be lacking entirely in others. The first upper molar is noticeably narrower than M2 in brosseti as it is in most other species of Micronycteris. Uniquely, the lingual portion of M1 is expanded so that M1 and M2 are subequal in width in sanborni, which also differs from all other species in having a large gap visible between the posterior edge of the cingulum of M2 and the anterolingual edge of M3 when the toothrow is seen in occlusal view. The lower dentition of brosseti closely resembles that of most other species, with minor differences apparent only in the premolars. In brosseti and schmidtorum, p2 and p4 are subequal in anteroposterior length and in height, while p3 is slightly smaller in both dimensions. In minuta,sanborni, and homezi, p3 is further reduced relative to p2 and p4, which are large and subequal. By contrast, p3 is a larger tooth (approximately the same size as p4) in microtis and megalotis; in these taxa, p2 is often slightly larger than either p3 or p4, although there is some within-species variation in this trait and all three lower premolars are subequal in some individuals of both species. The coronoid process in brosseti is high, and the upper margin of the ascending process has a steep slope (25–30 8 ), similar to the mandibular morphology of megalotis,microtis, and schmidtorum. In contrast, the coronoid process is comparatively low and the upper margin of the ascending ramus has a shallow slope (16–18 8 )inminuta,sanborni, and homezi. F IELD O BSERVATIONS : Of the eight specimens of Micronycteris brosseti that we collected at Paracou, one was taken in a groundlevel mistnet and the remaining seven from a single roost in a hollow tree; both capture sites were in well-drained primary forest. The roost tree, about 1 m in diameter at breast height, had a single entrance hole (about 20 cm in diameter) 1 m above the ground. A mistnet enclosure around the base of the tree captured seven of the eight bats that emerged (between 18:45 and 19:15 hours) on 19 August 1991; the captured animals consisted of four adult males and three adult females. In addition, the mistnet enclosure captured two Phyllostomus elongatus on the outside as they came swooping toward the roost opening, obviously intending to ambush the much smaller emerging bats. We were unable to revisit this roost in subsequent years because the site was cleared for agricultural research in 1992. Micronycteris hirsuta (Peters) V OUCHER M ATERIAL : 3 females (AMNH *267857, *267858; MNHN *1995.800) and 6 males (AMNH *267093, *267894, *267096, 267860; MNHN *1995.801, 1995.802); see table 24 for measurements. I DENTIFICATION :Micronycteris hirsuta is easy to identify owing to its large size (see Simmons [1996b] and tables 23–25) and uniquely high-crowned lower incisors (figured by Genoways and Williams, 1986). Descriptions and comparative measurements of M.hirsuta from the Guianas and elsewhere were provided by Sanborn (1949), Goodwin and Greenhall (1961), Hill (1964), Davis (1976a), Genoways and Williams (1986), Brosset and Charles-Dominique (1990), and Simmons (1996b). No subspecies are currently recognized (Simmons, 1996b). Our specimens of Micronycteris hirsuta conform in all respects to previous descriptions of the species. F IELD O BSERVATIONS : We captured nine individuals of Micronycteris hirsuta at Paracou, three of which were taken in groundlevel mistnets (two in well-drained primary forest and one in swampy primary forest), and the other six at roosts. The two roosts we found were similar: both were hollow trees near streams in primary forest, both trees had sloping trunks, and the central cavity of each tree had a lower entrance 1–2 m above the ground and another smaller aperture several meters higher up the trunk. From one tree (fig. 21) we collected an adult male and an adult female (probably not the entire roosting group); from the other (fig. 29) we collected two adult males, one adult female, and one subadult male (possibly the entire roosting group). The floor of the latter roost was littered with the wings of large orthopterans, presumably the remains of prey car- 70 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 29. Roost of Micronycteris hirsuta in swampy primary forest at Paracou. Like the roost illustrated in figure 21, this is an inclined tree (sloping away from the camera in this view) with a hollow central cavity accessed by a large lower aperture (arrow) and a second smaller opening much higher up the trunk (not visible). The presence of bats in this roost was inferred from a pile of large orthopteran wings found at the base of the lower aperture. In the foreground are cut palm fronds that we used to thatch a mistnet enclosure in which four emerging bats were captured. 1998 71SIMMONS AND VOSS: PARACOU BATS ried back to the roost and eaten by the inhabitants. Micronycteris homezi Pirlot Figures 30–32 V OUCHER M ATERIAL : 1 male (AMNH *267414); see table 25 for measurements. I DENTIFICATION :Micronycteris homezi was originally described by Pirlot (1967), who considered it a subspecies of M.megalotis. Prior to the present study, the only specimens referred to homezi were those of Pirlot’s type series from the Maracaibo Basin of northwestern Venezuela (ca. 2200 km WNW of Paracou). Simmons (1996b: 4) discussed the brief history of this taxon: Micronycteris megalotis homezi Pirlot, 1967, was named based on three specimens collected in the western Venezuelan state of Zulia. These specimens (including the holotype) were destroyed along with the rest of Pirlot’s collection sometime during the 1970s (Pirlot, phone conversation with T. Griffiths). The published description of homezi is incomplete, but several features described by Pirlot (e.g., pale venter, deep notch in ear membrane) suggest that homezi is not related to megalotis. It is not clear from the type description if homezi represents a distinct species or is synonymous with minuta,schmidtorum, or sanborni. Simmons (1996b) found Pirlot’s description of homezi to be inadequate for several reasons, including (1) no description of the dentition, (2) no measurements other than length of the forearm, and (3) no mention of relative size of the calcar. The only truly distinctive character of homezi described by Pirlot was a ‘‘creux cutane´ glabre’’ (smooth cutaneous fossa) on the top of the head between the ears behind the interauricular band. Pirlot (1967: 265) described this peculiar structure as follows: C’est une petite plage subovale, partiellement glabre, limite´e par un bourrelet tre`s net. Les deux versants ante´rieur et poste´rieur de cette plage se terminent dans le sillon transversal qui va d’une oreille a` l’autre. Sur le versant ante´rieur, une range´e de poils relativement longs est implante´e. An accompanying drawing of the dorsum of the head with the ears and interauricularband folded forward (op. cit.: fig. 1) shows little more than the location of the structure in question. Although Pirlot did not explicitly discuss sexual dimorphism, he apparently had both male and female examples of homezi in hand (op. cit.: 262), suggesting that the cutaneous fossa is present in both sexes. Having never seen a structure resembling Pirlot’s cutaneous fossa, Simmons (1996b) inferred that he was just describing the naked patch of skin that lies underneath the interauricular band and the attached hair tuft in all Micronycteris species. The interauricular band is normally folded back against the top of the head so that the posterior (ventral) surface of the band is closely appressed to the skin of the head. The cranial skin covered by the band appears naked (although a sparse coat of fine hairs can be seen under magnification), and it is fully exposed only when the ears are drawn forward. This naked patch varies in size among species of Micronycteris, being largest in species with a high interauricuar band (e.g., minuta,sanborni) and smallest in those with a low interauricular band (e.g., megalotis,microtis). However, this interpretation of Pirlot’s ‘‘creux cutane´ glabre’’ now appears incorrect. One of the specimens that we collected at Paracou—AMNH 267414, an adult male— has a well-defined, hairless fossa on the top of the head between the ears and behind the interauricular band (fig. 30). Smooth and glabrous inside, this pit is surrounded by a rim of skin with coarse projecting hairs, exactly as Pirlot described. Although no histological work has yet been done, this structure appears to be a glandular pouch or a pouch in which glandular secretions might be stored. To the best of our knowledge, this structure is unique among bats (for a summary of known chiropteran glandular structures, see Quay, 1970). Several other characters of our specimen correspond with Pirlot’s (1967) description of homezi, including (1) small size (forearm length 36.5 mm in Paracou specimen, 31.4– 37.4 mm in Pirlot’s series), (2) a pale venter (much lighter than the dorsal fur), and (3) a high interauricular band divided into roughly triangular flaps by a deep median notch. Because our material agrees closely with Pirlot’s description, and because of its clear distinctness from any other named congeneric form, we conclude that AMNH 267414 represents a valid species that should be called Micronycteris homezi. 72 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 30. Live views of Micronycteris homezi (AMNH 267414; male). Note the pale ventral fur, the long hairs on the leading edge of each pinna, and the very deep, V-shaped notch in the interauricular band; the cutaneous fossa is visible through the notch in the interauricular band. The hairs covering the fossa in this view are rooted in the skin surrounding the fossa, and on the underside (posterior surface) of the interauricular band; the fossa itself is hairless. 1998 79SIMMONS AND VOSS: PARACOU BATS ly reidentified as either M.sanborni (Simmons, 1996b) or M.brosseti (see above). Simmons (1996b) provided a summary of measurements of Micronycteris schmidtorum sensu stricto from throughout its known geographic range. No subspecies are currently recognized (Simmons, 1996b). Although our voucher material generally agrees with previous descriptions of Micronycteris schmidtorum, Paracou specimens have pale gray ventral fur rather than the pale buff venters seen in specimens from other localities. Because we did not find any other consistent differences, and because other congeners with pale ventral fur (e.g., M. brosseti and M.minuta) exhibit similar chromatic variability, we conclude that this represents normal intraspecific variation. In addition to characters discussed previously, we found that tibia length was helpful for distinguishing Micronycteris schmidtorum from sympatric species in the hand, at least at Paracou. Whereas both M.schmidtorum and M.homezi have long tibias (15.3– 15.8 mm), the lower leg is consistently shorter (12.8–14.6 mm) in M.brosseti,M.megalotis,M.microtis, and M.minuta. F IELD O BSERVATIONS : Our two examples of Micronycteris schmidtorum were both taken in ground-level mistnets in well-drained primary forest. Mimon bennettii (Gray) Figures 34–37 V OUCHER M ATERIAL : 1 female (AMNH *267109) and 1 male (MNHN *1995.988); see table 26 for measurements. I DENTIFICATION : Although Koopman (1993, 1994) considered Mimon bennettii and M.cozumelae to be conspecific, many other authors have recognized them as distinct species (Dalquest, 1957; Handley, 1960; Carter et al., 1966; Gardner and Patton, 1972; Jones and Carter, 1976; Swanepoel and Genoways, 1979; McCarthy, 1987; McCarthy et al., 1993). Whereas the restricted type locality of M.bennettii is Ipanema in the Brazilian state of Sa˜o Paulo (Hershkovitz, 1951), the type locality of M.cozumelae is Cozumel Island in the Mexican state of Quintana Roo (Goldman, 1914). Currently, the known range of cozumelae extends from southern Mexico to northwestern Colombia, and that of bennettii extends from the Guianas to southeastern Brazil (Koopman, 1994). Dalquest (1957) reported additional specimens of Mimon cozumelae collected in the decades following Goldman’s (1914) original description, and discussed differences between cozumelae and bennettii, which he treated as distinct species. However, as Handley (1960) subsequently noted, Dalquest’s only comparative example of bennettii was a juvenile. Schaldach (1964) subsequently summarized characters supposedly distinguishing these taxa, including size (smaller in bennettii), dorsal pelage color (darker in bennettii, brighter in cozumelae), and length and woolliness of the middorsal hairs (long and not woolly in bennettii, short and woolly in cozumelae). Like Dalquest’s (1957) account, however, Schaldach’s description of the pelage of bennettii seems to have been based on juvenile or subadult specimens. Despite the morphological differences he noted, Schaldach concluded that bennettii and cozumelae are conspecific, a decision based in large part on the geographic gap between their known geographic ranges. Hall (1981) and Koopman (1993, 1994) followed Schaldach (1964) without additional discussion. Contra Schaldach (1964), Hall (1981), and Koopman (1993, 1994), we consider Mimon bennettii and M.cozumelae to represent distinct species based on apparently consistent differences in the following characters: (1) dorsal pelage color (more reddish in adult bennettii, less reddish in cozumelae), (2) color of the wingtip (dark in bennettii, white in cozumelae), (3) shape of the middle upper incisors (tapering to points in bennettii, more spatulate in cozumelae), (4) form of the lower incisors (narrower in bennettii), (5) morphology of m3 (talonid larger and better developed in bennettii), and (6) morphology of the posterior palatal margin (broader with Ushaped mesopterygoid notch in bennettii, narrower with V-shaped notch in cozumelae). 8 Although some overlap exists, bennet8 This list of characters is based on both literature accounts and our examination of voucher specimens. Specimens examined in addition to those from Paracou: Mimon bennettii (Brazil: USNM 123393, 391027); Mimon cozumelae (Mexico: AMNH 144508, 185862–185872). 80 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 34. Dorsal (A) and ventral (B) views of the skull of Mimon bennettii (MNHN 1995.988; male) from Paracou. Dorsal (C) and ventral (D) views of the skull of M.cozumelae (AMNH 265107; male) from Guatemala. Note the species differences in incisor morphology and in the shape of the mesopterygoid notch of the posterior palate. Scale bars 5 10 mm. 1998 81SIMMONS AND VOSS: PARACOU BATS Fig. 35. Lateral views of the skull and lower jaw of (A)Mimon bennettii (MNHN 1995.988; male) and (B)M.cozumelae (AMNH 265107; male). Scale bars 5 10 mm. tii is generally smaller than cozumelae (e.g., forearm length of 50.0–56.6 mm in bennettii, 54.6–60.7 mm in cozumelae; Swanepoel and Genoways, 1979; Hall, 1981; Brosset and Charles-Dominique, 1990). Although these taxa have not been collected in sympatry, it is counterproductive to regard them as subspecies given the magnitude of their mor- 82 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 36. Anterior views of the upper and lower dentition of (A)Mimon bennettii (MNHN 1995.988; male) and (B)M.cozumelae (AMNH 265107; male). Note that the middle upper incisors of bennettii taper to points, whereas these teeth are more spatulate in cozumelae. Scale bars 5 5 mm. Fig. 37. Occlusal views of the crown of the right m3 in (A)Mimon bennettii (MNHN 1995.988; male) and (B)M.cozumelae (AMNH 265107; male). Note that the talonid is larger and better developed in bennettii than it is in cozumelae. Scale bars 5 1 mm. phological divergence; until proven false, the appropriate null hypothesis is that they represent distinct species. Descriptions and measurements of Mimon bennettii sensu stricto were provided by Gray (1838), Dalquest (1957), Husson (1962, 1978), Hill (1964), Swanepoel and Genoways (1979), and Brosset and Charles-Dominique (1990). Our specimens closely resemble others previously reported from Guianas, particularly the bright orange-russet adults described by Brosset and Charles-Dominique (1990). Based on our limited comparisons, recognition of subspecies does not seem warranted. F IELD O BSERVATIONS : One of our two specimens of Mimon bennettii was captured in a ground-level mistnet and the other was taken at a roost; both captures were in creekside primary forest. The roost site was a large tree (ca. 1.5 m in diameter at breast height) with a central cavity extending from ground level to near the crown. We surrounded the only 1998 83SIMMONS AND VOSS: PARACOU BATS roost opening at ground level, a small hole ca. 15 cm in diameter, with a mistnet enclosure that caught one adult male Mimon bennettii and 27 adult male Carollia perspicillata between 18:10 and 18:35 hours on 17 November 1992 (the Mimon bennettii emerged at 18:25 hours); no emerging bats escaped. We did not revisit this roost in subsequent years. Mimon crenulatum (E. Geoffroy) V OUCHER M ATERIAL : 7 females (AMNH *267114, *267115, *267437, *267887, *267889; MNHN *1995.1032, 1995.1033) and 10 males (AMNH *267111, 267113, *267880, 267884, *267885, *267886, 267888; MNHN *1995.1034, *1995.1035, *1995.1036); see table 26 for measurements. I DENTIFICATION : Useful descriptions and measurments of Mimon crenulatum were provided by Handley (1960), Goodwin and Greenhall (1961), Husson (1962, 1978), Hill (1964), Genoways and Williams (1979), Swanepoel and Genoways (1979), Gardner and Patton (1972), Hall (1981), Brosset and Charles-Dominique (1990), and Pedro et al. (1994). Of the five subspecies recognized by Koopman (1994), M.c.crenulatum occurs from Trinidad and eastern Venezuela throughout the Guianas to southern Amazonian Brazil. However, we follow Gardner and Patton (1972) in regarding M.koepckeae (from the highlands of central Peru) as a distinct species contra Koopman (1978, 1993, 1994). Our voucher material conforms with previous descriptions of Mimon crenulatum crenulatum. Measurements of Paracou specimens generally fall within the range of variation previously documented for Guianan populations (Husson, 1962, 1978; Hill, 1964; Genoways and Williams, 1979; Brosset and Charles-Dominique, 1990), although some of our males are slightly smaller in some dimensions than specimens previously reported from the region. F IELD O BSERVATIONS : All of the 17 Mimon crenulatum we captured at Paracou were taken in ground-level mistnets: 9 in welldrained primary forest, 4 in swampy primary forest, and 4 in creekside primary forest. An unusually large proportion (88%) of captures were in the very early evening, before 19:00 hours. 84 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Phylloderma stenops Peters V OUCHER M ATERIAL : 3 females (AMNH *266078, *267441; MNHN *1995.1075) and 5 males (AMNH *266077, *267440, *267890, *267891; MNHN *1995.1076); see table 27 for measurements. I DENTIFICATION : Descriptions and measurements of Phylloderma stenops from the Guianas and elsewhere were provided by Husson (1962, 1978), Hill (1964), Gardner (1976), Genoways and Williams (1979), Swanepoel and Genoways (1979), Williams and Genoways (1980a), Brosset and CharlesDominique (1990), and Anderson (1997). Three subspecies are currently recognized: P. s.septentrionalis (southern Mexico to Costa Rica), P.s.stenops (Panama to southern Brazil), and P.s.boliviensis (central Bolivia) (Barquez and Ojeda, 1979; Koopman, 1994; Anderson, 1997). Our voucher material conforms in all respects to previous descriptions of Phylloderma stenops stenops, and measurements fall within the range of variation reported previously for this taxon. F IELD O BSERVATIONS : All eight of the Phylloderma stenops we captured at Paracou were taken in ground-level mistnets: two in well-drained primary forest, three in swampy primary forest, two in creekside primary forest, and one in closed-canopy secondary growth. Phyllostomus discolor Wagner V OUCHER M ATERIAL : 10 females (AMNH *267116, *267118, *267120, *267121, *267123, *267124; MNHN *1995.1077, *1995.1078, *1995.1079, *1995.1080) and 5 males (AMNH *267117, *267119, *267984, *267986; MNHN *1995.1081); see table 28 for measurements. I DENTIFICATION : Useful descriptions and measurements of Phyllostomus discolor may be found in Goodwin and Greenhall (1961), Husson (1962, 1978), Power and Tamsitt (1973), Taddei (1975a), Swanepoel and Genoways (1979), Brosset and Charles-Dominique (1990), and Anderson (1997). Two subspecies are often recognized: P.d.verrucosus (Mexico to northwestern Peru) and P.d.discolor (South America east of the Andes to northwestern Argentina; Margarita Island; Trinidad) (Koopman, 1994). Power and Tamsitt (1973), however, questioned the recognition of these taxa based on comparisons of specimens from near the supposed contact zone in Colombia. Our Paracou specimens conform closely with previous descriptions of Phyllostomus discolor, and measurements fall within the 1998 85SIMMONS AND VOSS: PARACOU BATS range of variation previously reported from the Guiana region. Our comparisons of published measurements from throughout the known geographic range of P.discolor suggest that Power and Tamsitt (1973) were correct and that there is no current justification for recognizing subspecies. F IELD O BSERVATIONS : We captured 17 Phyllostomus discolor at Paracou: 14 in ground-level mistnets and 3 in elevated nets. Four of the ground-level captures were in well-drained primary forest, eight were in swampy primary forest, and two were in creekside primary forest. One of the elevated mistnet captures was made at 18–21 m over a narrow dirt road and two were made at 10– 13 m above a treefall gap in creekside primary forest. Several individuals netted in the dry season were covered with pollen. Most bats were caught singly, but two adult males and two adult females (possibly members of a foraging flock) were taken within a 10-minute interval at one mistnetting site in swampy primary forest on 24 October 1992, and an- 86 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY other adult female was caught a few hours later at the same place. Phyllostomus elongatus (E. Geoffroy) V OUCHER M ATERIAL : 14 females (AMNH *266051, *266055, *266058, *266062, *266063, *266064, *266067, *266068, *267152, *267897; MNHN *1995.1082, *1995.1083, *1995.1084, *1995.1085) and 12 males (AMNH *266050, *266052, *266053, *266054, *266065, *266066, *266069, *267127; MNHN *1995.1086, *1995.1087, *1995.1088, *1995.1089); see table 28 for measurements. I DENTIFICATION : Descriptions and measurements of Phyllostomus elongatus from the Guianas and elsewhere were provided by Husson (1962, 1978), Hill (1964), Swanepoel and Genoways (1979), Brosset and Charles-Dominique (1990), and Anderson (1997). No subspecies are currently recognized (Koopman, 1994). Our material from Paracou agrees in all respects with previous descriptions of Phyllostomus elongatus. Like other conspecific material from the Guiana region, our specimens fall near the upper end of the known range of size variation for P.elongatus. F IELD O BSERVATIONS : We recorded 162 captures (probably including some recaptures) of Phyllostomus elongatus at Paracou. One hundred fifty-nine captures were in ground-level mistnets, 9 one was in a mistnet suspended 13–16 m above a narrow dirt road, and two were at roosts. Seventy-four of the ground-level mistnet captures were in well-drained primary forest, 45 were in swampy primary forest, 27 were in creekside primary forest, 1 was in a natural treefall opening in primary forest, 7 were in manmade clearings, and 5 were in closed-canopy secondary growth. The two bats taken at roosts were both solitary males; one was collected from a tree cavity with a single large opening about 5 m above the ground (fig. 26), the other from the central shaft of a hol9 This figure includes two bats caught on the outside of the mistnet enclosure described above in the account for Micronycteris brosseti. However, because those captures were incidental to roost sampling, they are excluded from the quantitative analysis of mistnetting results below. low tree with a large lower entrance at ground level and second smaller aperture ca. 15–20 m farther up the trunk (this roost was shared with Saccopteryx bilineata and Carollia perspicillata). Both roosts were in welldrained primary forest. By contrast with the multiple elevatedmistnet captures we recorded for Phyllostomus discolor and P.hastatus (see above and below), the almost complete absence of P. elongatus in nets suspended more than a few meters above the ground is noteworthy. Phyllostomus hastatus (Pallas) V OUCHER M ATERIAL : 11 females (AMNH *266071, *266072, *266073, *266075, *267433, *267901, *267902, *267903; MNHN *1995.1090, *1995.1091, *1995.1092) and 8 males (AMNH *266070, *266074, *267434, *267904, *267907; MNHN *1995.1093, *1995.1094, *1995.1095); see table 28 for measurements. I DENTIFICATION : Descriptions and measurements of Phyllostomus hastatus have appeared in many publications; we consulted those in Goodwin and Greenhall (1961), Husson (1962, 1978), Taddei (1975a), Swanepoel and Genoways (1979), Brosset and Charles-Dominique (1990), and Anderson (1997). Two subspecies of are currently recognized: P.h.panamensis (Honduras to Peru and east to Venezuela) and P.h.hastatus (eastern Venezuela south to northern Bolivia and southeastern Brazil) (Koopman, 1994). These are distinguished principally on the basis of size, with P.h.hastatus being the smaller form. Our voucher material conforms closely with previous descriptions of Phyllostomus hastatus hastatus, with measurements generally falling within the range of variation previously reported from the Guianas. One exception is a particularly small adult male (AMNH 267907), some measurements of which are less than any previously reported for the species (e.g., forearm length of 77.5 mm, compared with 79.0 mm for the smallest male reported by Taddei [1975a]). Comparisons of this diminutive individual with other specimens in our sample, however, failed to reveal any morphological differences other than size. 1998 87SIMMONS AND VOSS: PARACOU BATS F IELD O BSERVATIONS : We recorded 56 captures (possibly including some recaptures) of Phyllostomus hastatus, of which 34 were in ground-level mistnets and 22 were in elevated mistnets. The 34 ground-level captures included 9 in well-drained primary forest, 11 in swampy primary forest, 6 in creekside primary forest, and 8 in manmade clearings (7 were taken in a banana/cacao plantation on a single night). Of the 22 elevated mistnet captures, 17 were in nets suspended 10–23 m above a narrow dirt road, 4 were made at 20–38 m above a treefall gap in well-drained primary forest, and 1 was at 7–10 m above a treefall in creekside primary forest. Tonatia Although keys to the species of Tonatia provided by Genoways and Williams (1984) and Medellı´n and Arita (1989) are useful tools for identification, neither is complete; the former covers only the species known from Surinam, and the latter was published before T.bidens and T.saurophila were recognized as distinct species (see below). All of the five species of Tonatia known from the Guiana subregion of Amazonia are sympatric (and apparently syntopic) at Paracou. Tonatia brasiliense (Peters) V OUCHER M ATERIAL : 3 females (AMNH *267103, *267104; MNHN *1995.1052) and 6 males (AMNH *267101, *267102, *267916, *267917; MNHN *1995.1053, *1995.1054); see table 29 for measurements. I DENTIFICATION : Descriptions and measurements of Tonatia brasiliense from the Guianas and elsewhere can be found in Goodwin (1942), Swanepoel and Genoways (1979), Genoways and Williams (1984), and Brosset and Charles-Dominique (1990). No subspecies are currently recognized (Genoways and Williams, 1984; Koopman, 1994). As the smallest member of its genus, Tonatia brasiliense is most likely to be confused with species of Micronycteris, which are also small gray-brown phyllostomines with large, rounded ears. Genoways and Williams (1984) mentioned several craniodental characters that can be used to distinguish these taxa, but failed to note one obvious external character: whereas all species of Micronycteris have a pair of large dermal pads forming a ‘‘V’’ on the tip of the chin, the tip of the chin in T.brasiliense (and other congeners) has a U-shaped row of tiny dermal papillae (Emmons, 1990, 1997). We also observed that T.brasiliense folds its ears back against the crown of the head when touched, a behavior seen in several other species of Tonatia (see below) but not in Micronycteris. Our voucher material conforms closely with previous descriptions of Tonatia brasiliense, particularly those based on material from the Guianas. Probably because our series contains more individuals than do earlier collections from the region, it exhibits somewhat greater size variability than previously reported. However, measurements of our specimens fall within the range of variation previously documented for the species as a whole. As remarked by Genoways and Williams (1984), Tonatia brasiliense as currently recognized may be composite. Should this prove to be the case, comparisons with published measurements (e.g., those in Goodwin [1942] and Swanepoel and Genoways [1979]) suggest that our specimens, along with others from the Guianas, would be referred to T.brasiliense rather than to other taxa currently treated as synonyms (e.g., minuta,nicaraguae, and venezuelae; Handley [1966], Gardner [1976], Jones and Carter [1976], Genoways and Williams [1984], Medellı´n and Arita [1989], Koopman [1993, 1994]). F IELD O BSERVATIONS : All of the nine specimens of Tonatia brasiliense that we caught at Paracou were taken in ground-level mistnets: six in swampy primary forest and three in creekside primary forest. Tonatia carrikeri (J. A. Allen) V OUCHER M ATERIAL : 1 female (AMNH *267918); see table 29 for measurements. I DENTIFICATION :Tonatia carrikeri has been collected only rarely, so descriptions and measurements of most known specimens have been reported in the literature, for example, by Goodwin (1942), Husson (1962, 1978), Gardner (1976), Swanepoel and Genoways (1979), McCarthy et al. (1983, 1992), Genoways and Williams (1984), and McCar- 88 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY thy and Handley (1987). No subspecies are currently recognized (Koopman, 1994). Our single female specimen of Tonatia carrikeri agrees with previous descriptions of the species in all respects. Like several other congeners, T.carrikeri folds its ears back over the crown of the head when they are touched. F IELD O BSERVATIONS : The single example of Tonatia carrikeri we captured at Paracou was taken in a ground-level mistnet in swampy primary forest dominated by the palm Euterpe oleracea. Tonatia saurophila Koopman and Williams V OUCHER M ATERIAL : 10 females (AMNH *266044, *266045, *266046, *266047, *267429, *267908, *267914; MNHN *1995.1060, *1995.1061, *1995.1062) and 11 males (AMNH *266049, *267099, 1998 95SIMMONS AND VOSS: PARACOU BATS clear.’’ Although the use of trinomial nomenclature for Carollia perspicillata is not currently justified, future morphological or molecular studies may yet show that some subspecific distinctions are warranted. Our specimens of Carollia perspicillata conform in all respects to previous descriptions of the species. F IELD O BSERVATIONS : We recorded 1142 captures (including recaptures) of Carollia perspicillata at Paracou, of which 1048 were in ground-level mistnets, 18 were in elevated mistnets, 75 were at roosts, and 1 was in a harp trap. Of the 1049 ground-level mistnet and harp-trap captures, 252 were in welldrained primary forest, 279 were in swampy primary forest, 110 were in creekside primary forest, 1 was in a treefall opening in primary forest, 325 were in manmade clearings, 9 were in closed-canopy secondary growth, and 73 were over roadside puddles. The 18 captures in elevated mistnets were made between 5 and 13 m above the ground: 7 over a narrow dirt road, 7 over treefalls, and 4 in the subcanopy of swampy primary forest. Of the 11 roosts at which we captured or observed Carollia perspicillata at Paracou, 4 were in culverts under roads, 1 was under a bridge, and 6 were in tree cavities (e.g., figs. 21, 25, 26, 33). We never found C.perspicillata under fallen trees, inside hollow logs, or in foliage of any kind. Other bats observed roosting with C.perspicillata included Rhynchonycteris naso,Saccopteryx bilineata,Micronycteris megalotis,M.microtis,Mimon bennettii,Phyllostomus elongatus,Trachops cirrhosus, and Glossophaga soricina.Carollia perspicillata was the only species that we found roosting in large bachelor groups; for example, 27 males were captured in a mistnet enclosure as they emerged from an opening in the base of a large hollow tree (a roost shared with Mimon bennettii, see above). Another tree-cavity roosting group that we captured in its entirety consisted of two adult males, two lactating adult females, and two juveniles. Rhinophylla pumilio Peters V OUCHER M ATERIAL : 25 females (AMNH *266168, *266171, *266178, *266184, *266186, *266188, *266189, *266193, *266196, *266198, *267159, *267456, *267457, *267458, *267459, *267971; MNHN *1998.623, *1998.624, *1998.625, *1998.626, *1998.627, *1998.628, *1998.629, *1998.630, *1998.631) and 24 males (AMNH *266174, *266175, *266179, *266180, *266181, *266182, *266183, *266185, *266187, *266190, *266191, *266192, *266194, *266197, *267158; MNHN *1998.632, *1998.633, *1998.634, *1998.635, *1998.636, *1998.637, *1998.638, *1998.639, *1998.640); see table 35 for measurements. I DENTIFICATION : Descriptions and comparative measurements of Rhinophylla pumilio that we consulted to identify our material included those in Husson (1962, 1978), Hill (1964), Carter (1966), Swanepoel and Genoways (1979), Williams and Genoways (1980a), and Brosset and Charles-Dominique (1990). No subspecies are currently recognized (Koopman, 1994). Our Paracou specimens conform closely with previous qualitative descriptions of Rhinophylla pumilio in the literature cited above. Likewise, measurements of our series generally fall within the known range of variation for the species, although a few are slightly smaller than any previously reported. Because the small individuals in our collection are similar in all other repects to larger examples, we attribute this minor discrepancy to within-population morphometric variation. Like other collections of R.pumilio from the Guianas, our Paracou vouchers fall at the lower end of the known range of size variation for the species. F IELD O BSERVATIONS : We recorded 128 captures (probably including some recaptures) of Rhinophylla pumilio at Paracou, of which 106 were in ground-level mistnets, 2 were in elevated mistnets, 19 were at roosts, and 1 was in a harp trap near ground level. Of the 107 ground-level mistnet and harptrap captures, 22 were in well-drained primary forest, 44 were in swampy primary forest, 14 were in creekside primary forest, 4 were in treefall openings in primary forest, 19 were in manmade clearings, and 4 were in closed-canopy secondary growth. The two bats captured in elevated mistnets were taken 96 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY 5–10 m above a treefall opening in creekside primary forest. We found eight roosting groups of Rhinophylla pumilio at six unique roost sites (one roost was revisited twice). All of the roosts we found were in foliage between 1.5 and 5 m above the ground. Six roosting groups of R.pumilio occupied ‘‘bifid’’ tents (sensu Kunz et al., 1994) made from the terminal leaflets of fronds of immature understory palms that we provisionally identified as Astrocaryum sciophilum. 11 In all recorded construction details, these tents exactly resembled those in which we also found Artibeus cinereus (see figs. 43–45) and Ectophylla macconnelli (see fig. 47). Although we do not know which (if any) of these bats was actually responsible for making such tents, we once collected (on 6 August 1993) a group of three R.pumilio from a tent previously occupied (on 30 July 1993) by seven E.macconnelli. By contrast, we never ob11 We base this identification on vegetative characters described by Henderson et al. (1995) and de Granville (1997): the plicate and subcoriaceous leaflets, regularly arranged and spreading in one plane, have nonspinous margins, and the long spines on the rachis lack basal wings (see figs. 44, 45). However, none of the plants in which we found bat tents at Paracou were fertile, so diagnostic reproductive characters could not be determined. served other species of bats in tents previously found occupied by R.pumilio. For this reason, and because the delicate anterior dentition of Rhinophylla seems inadequate to the task of chewing through the tough lateral veins of Astrocaryum leaves, we favor the hypothesis suggested by Charles-Dominique (1993) that R.pumilio is a roost parasite that uses tents made by other bats. In addition to finding roosts of Rhinophylla pumilio in bifid Astrocaryum leaf-tents, we disturbed one group of three individuals from an unidentified location in a clump of Phenakospermum guyannense (Strelitziaceae) that contained an ‘‘apical’’ tent (Kunz et al., 1994) similar in construction to those in which we found Uroderma bilobatum at other sites (see fig. 51). On another occasion we found an adult male R.pumilio roosting alone beneath an unmodified leaf of P.guyannense. Of the four Astrocaryum leaf-tents in which we found Rhinophylla pumilio, two were in well-drained primary forest, one was in closed-canopy secondary growth, and one was in primary forest at the edge of a swampy area. The two Phenakospermum roosts were both in swampy primary forest. None of the roosting groups we found at Paracou (table 36) contained more than one adult male. 1998 97SIMMONS AND VOSS: PARACOU BATS STENODERMATINAE Although other Amazonian sites are known to have more than 20 sympatric species of Stenodermatinae, we captured only 15 species of this phyllostomid subfamily at Paracou. Following the taxonomic usage recommended by Peffley et al. ( MS ), the Paracou species represent eight genera: Ametrida,Artibeus (including Dermanura and Koopmania), Chiroderma,Ectophylla (including Mesophylla), Platyrrhinus (formerly Vampyrops), Sturnira,Uroderma, and Vampyressa. Collections from other localities in French Guiana and Surinam include an additional seven stenodermatine species (appendix 1), but for reasons explained below (see Analyses of Sampling: Estimating Completeness) we doubt that any of these normally occur within our study area. Ametrida centurio Gray V OUCHER M ATERIAL : 7 females (AMNH *267973, *267274, *267275, *267276, *267278; MNHN *1995.1037, *1995.1038) and 3 males (AMNH *267279, *267976; MNHN *1995.1039); see table 37 for measurements. I DENTIFICATION : Peterson’s (1965a) revi- 98 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY sion remains the principal reference on Ametrida centurio. Measurements for specimens previously collected in the Guianas can also be found in Husson (1962, 1978), Swanepoel and Genoways (1979), and Brosset and Charles-Dominique (1990). No subspecies are currently recognized (Peterson, 1965a; Jones and Carter, 1976; Koopman, 1994). Paracou specimens of Ametrida centurio conform in all respects to Peterson’s (1965a) account of this species, with measurements falling within the range of variation previously reported in the literature. In particular, morphometric data from our series (table 37) provide compelling confirmation of the remarkable sexual size dimorphism of A.centurio, in which females are substantially larger than males. Measurements of Paracou males and females are completely nonoverlapping in 11 of 13 dimensions, with slight overlap occurring only in hindfoot length and length of ear. F IELD O BSERVATIONS : Of the 10 individuals of Ametrida centurio captured at Paracou, 6 were taken in ground-level mistnets and 4 in elevated mistnets. Of the six ground-level captures, two were in swampy primary forest, one was in creekside primary forest, and three were in manmade clearings. Two individuals were captured between 5 and 20 m over a narrow dirt road, and two were captured 34–37 m above a treefall clearing in well-drained primary forest. Artibeus Reliable species identifications within the genus Artibeus as recognized herein (including Dermanura and Koopmania) requires reference to Handley (1987) and MarquesAguiar (1994). Although Lim and Wilson’s (1993) concept of Artibeus jamaicensis differs from ours, their keys are also useful for distinguishing the larger species of Artibeus (subgenus Artibeus) in northern South America. Artibeus (Artibeus)jamaicensis Leach Figures 38, 39 V OUCHER M ATERIAL : 23 females (AMNH *266322, *266331, *266333, *266334, *266335, *266336, *266337, *266338, *266344, *267998, *268503, *268505, *268508, *268528, *268529; MNHN *1995.1139, *1995.1140, *1995.1141, *1995.1142, *1995.1143, *1995.1144, *1995.1145, *1995.1146) and 12 males (AMNH *266321, *266332, *266341, *266345, *267202, *267999, *268502, *268504; MNHN *1995.1147, *1995.1148, *1995.1149, *1995.1150); see table 38 for measurements. I DENTIFICATION : We follow Handley (1987) and Marques-Aguiar (1994) in regarding Artibeus jamaicensis as the senior synomym of A.planirostris (contra Lim and Wilson [1993] and Koopman [1993, 1994]). Descriptions and comparative measurements can be found in Goodwin and Greenhall (1961), Swanepoel and Genoways (1979), Koepcke and Kraft (1984), Handley (1987), Brosset and Charles-Dominique (1990), Lim and Wilson (1993), and Marques-Aguiar (1994). Although Husson (1962, 1978) provided a detailed account of ‘‘Artibeus lituratus fallax’’ ( 5 A.jamaicensis) from Surinam, his sample may have included specimens of both A.jamaicensis and true A.lituratus. Between 10 and 13 subspecies of Artibeus jamaicensis (including planirostris) are currently recognized (see Hall, 1981; Davis, 1984; Handley, 1987; Marques-Aguiar, 1994; Koopman, 1994), of which 5 occur in South America: A.j.aequatorialis (Pacific slope of the Andes from southern Colombia to northern Peru), A.j.hercules (eastern Ecuador and Peru), A.j.trinitatis (central Colombia and Venezuela north of the Orinoco), A.j.fallax (Venezuela south of the Orinoco throughout the Guianas, thence southward through central Brazil to Bolivia), and A.j.planirostris (eastern Brazil and Paraguay). Our Paracou material conforms closely to published qualitative descriptions of Artibeus jamaicensis fallax, although we observed somewhat more variation in pelage color than previously described for Guianan specimens (see below). Measurements of our material fall within the range of variation previously reported by most authors for A.j.fallax, including specimens referred to A.planirostris by Lim and Wilson (1993). Despite our best efforts to correctly identify large Artibeus in the field, subsequent museum study showed that we initially mis- 1998 99SIMMONS AND VOSS: PARACOU BATS Fig. 38. Dorsal (A) and ventral (B) views of the skull of Artibeus jamaicensis (AMNH 266321; male) from Paracou. Dorsal (C) and ventral (D) views of the skull of A.lituratus (AMNH 267492; male) from Paracou. Note that jamaicensis has a broad interorbital region, small postorbital processes, and three molars, whereas lituratus has a narrower interorbit, better developed postorbital processes, and lacks M3. Scale bars 5 10 mm. 100 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 39. Lateral views of the skull and lower jaw of (A)Artibeus jamaicensis (AMNH 266321; male) and (B)A.lituratus (AMNH 267492; male). Scale bars 5 10 mm. 1998 101SIMMONS AND VOSS: PARACOU BATS dentified 6 specimens (6%) out of 94 adults preserved as vouchers. While we never confused specimens of lituratus and obscurus, we initially misidentified one specimen of jamaicensis as lituratus, two specimens of jamaicensis as obscurus, and three specimens of obscurus as jamaicensis. All of our identification errors involved females. Most of our identification mistakes involved individuals somewhat larger or smaller than the species norm, or those with ambiguous pelage colors and facial markings. For example, most specimens of lituratus have much brighter facial stripes than those typical of jamaicensis, but we found some jamaicensis with bright stripes and some lituratus with pale stripes. We also found intraspecific variation in dorsal pelage color in all three species, with some individuals having blacker or browner fur than that reported as typical in the literature. Unambiguous identifications of our voucher material were ultimately based on the qualitative craniodental characters described by Handley (1987), Brosset and CharlesDominique (1990), and Marques-Aguiar (1994). In particular, Paracou specimens of Artibeus jamaicensis are uniformly distinguished by a broad postorbital region with poorly developed postorbital processes (fig. 38), and M3 is always present. By contrast, the postorbital region is always narrower (table 38) and the postorbital processes are well developed in our specimens of A.lituratus, which uniformly lack M3. Subsequent examination of our measurement data (table 38) showed that whereas mean values for external dimensions of jamaicensis are smaller than those of lituratus, these species have overlapping ranges of morphometric variation at Paracou. Marques-Aguiar (1994) cited two pelage characters as particularly useful for distinguishing Artibeus jamaicensis from A.lituratus: (1) the dorsal surface of the base of the forearm is very sparsely furred (almost naked) in jamaicensis versus densely furred in lituratus; and (2) the ventral fur is frosted (with white or pale gray) in adult jamaicensis, whereas lituratus has completely dark (unfrosted) ventral fur. Examination of our vouchers confirms that these traits are unambiguously diagnostic for jamaicensis and lituratus at Paracou. Unfortunately, we were not aware of these helpful identification criteria in the field. Numerous cranial characters cited by Handley (1989) consistently separate Artibeus jamaicensis and A.obscurus, but these are obviously not useful in the field. Our morphometric data, however, indicate that jamaicensis and obscurus can be separated unambiguously (at least in French Guiana) on the basis of body weight, total length, and forearm length, in all of which dimensions obscurus is substantially smaller. As noted by Handley (1989), jamaicensis has shorter fur than obscurus, but at Paracou this difference is slight (7 mm versus 8–9 mm) and requires careful measurement to be useful for identification. Both taxa, in our judgment, have equally ‘‘soft’’ fur, contra Handley’s observations. Dorsal fur color is typically much darker in obscurus than in jamaicensis, but (as previously noted) we found enough overlapping variation in this character to compromise its usefulness in the field. Handley (1989: 450) also noted that obscurus has ‘‘fewer and smaller ornamental warts on [the] chin,’’ but we did not find this to be consistently true in our material. Most individuals of both species have the same number of chin papillae (nine small papillae arranged in a ‘‘U’’ around a larger central papilla), with considerable variation in papillary size. As noted above, we suspect that Husson’s (1962, 1978) sample of ‘‘Artibeus lituratus fallax’’ was a composite of individuals properly referred to jamaicensis and lituratus. This conclusion is based on two observations. First, Husson (1962, 1978) reported that M3 was present in 26 of his 34 specimens, and absent in 7 specimens. By contrast, we found M3 to be uniformly present in jamaicensis and uniformly absent in lituratus, a pattern that was also observed by Brosset and Charles-Dominique (1990). Second, Husson (1962: table XX) provided measurements of 10 individuals, 9 of which fall within the range of variation that we observed for jamaicensis, but 1 of which does not. The latter individual (a female from the Stuttgart museum, SNM 686.1) was reported to have a postorbital breadth of 6.3 mm, a value falling well below the range of varia- 102 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY tion in our series of jamaicensis, but agreeing perfectly with our measurement data for lituratus. F IELD O BSERVATIONS : We recorded 73 captures (possibly including some recaptures) of Artibeus jamaicensis at Paracou, of which 71 were in ground-level mistnets and 2 were in elevated mistnets. The 71 ground-level captures included 18 in well-drained primary forest, 38 in swampy primary forest, and 15 in creekside primary forest. The two elevated net captures were made 5–10 m above the ground in the subcanopy of swampy primary forest. Artibeus (Artibeus)lituratus (Olfers) Figures 38, 39 V OUCHER M ATERIAL : 15 females (AMNH *267204, *267495, *267496, *267497, *268506, *268507, *268509, *268510, *268512, *268513; MNHN *1995.1151, *1995.1152, *1995.1153, *1995.1154, *1995.1155) and 10 males (AMNH *266346, *267206, *267492, *268511, *268514, *268515; MNHN *1995.1156, *1995.1157, *1995.1158, *1995.1159); see table 38 for measurements. I DENTIFICATION : Characters useful for separating Artibeus lituratus from A.jamaicensis at Paracou are discussed in the preceding species account. Descriptions and comparative measurements of A.lituratus can also be found in Goodwin and Greenhall (1961), Hill (1964), Tamsitt and Valdivieso (1966), Swanepoel and Genoways (1979), Davis (1984), Koepcke and Kraft (1984), Handley (1987), Brosset and Charles-Dominique (1990), Lim and Wilson (1993), and Marques-Aguiar (1994). As noted earlier, Husson’s (1962, 1978) account of ‘‘Artibeus lituratus fallax’’ was apparently based on a composite series of A.lituratus and A.jamaicensis. Two subspecies of Artibeus lituratus are generally recognized: A.l.palmarum (Central America to northwestern Peru, northern Colombia, northern Venezuela, Trinidad to St. Vincent) and A.l.lituratus (throughout Amazonia to northern Argentina) (Davis, 1984; Koopman, 1994; Marques-Aguiar, 1994). Artibeus intermedius, considered a distinct species by many authors (e.g., Davis, 1984; Koopman, 1993, 1994), may represent a subspecies of A.lituratus endemic to Mexico and Central America (Marques-Aguiar, 1994). Our voucher material conforms closely with most previous qualitative descriptions of Artibeus lituratus (except Husson’s) in the literature cited above. In particular, measurements of our specimens fall within the range of variation previously reported for A.l.lituratus in the Guianas and elsewhere in northern South America. F IELD O BSERVATIONS : We recorded 53 captures (possibly including some recaptures) of Artibeus lituratus at Paracou, of which 40 were in ground-level mistnets and 12 were in elevated mistnets; in addition, 1 individual was shot at night. Of the 40 ground-level mistnet captures, 17 were in well-drained primary forest, 8 were in swampy primary forest, 3 were in creekside primary forest, 3 were in manmade clearings, and 9 were over roadside puddles. Of the 12 individuals captured in elevated nets, 11 were taken between 10 and 23 m above a narrow dirt road, and 1 was taken at 5–8 m in the subcanopy of swampy primary forest. The shot bat was found hanging from a palm frond about 10 m above the ground in well-drained primary forest. Artibeus (Artibeus)obscurus (Schinz) V OUCHER M ATERIAL : 27 females (AMNH *266271, *266273, *266279, *266281, *266287, *267997, *268000, *268501, *268516, *268518, *268520, 268522, *268523, *268524, *268525, *268526, *268527; MNHN *1995.1160, *1995.1161, *1995.1162, *1995.1163, *1995.1164, *1995.1165, *1995.1166, *1995.1167, *1995.1168, *1995.1169) and 10 males (AMNH *266272, *266286, *266288, *267208, *267210, *268517, 268519; MNHN *1995.1170, *1995.1171, *1995.1172); see table 38 for measurements. I DENTIFICATION : We follow Handley (1989) in using the name Artibeus obscurus instead of A.fuliginosus for the smallest and darkest member of the subgenus Artibeus found throughout the wet South American lowlands east of the Andes. Artibeus obscurus is best identified by reference to Handley (1987, 1998 103SIMMONS AND VOSS: PARACOU BATS 1989), Lim and Wilson (1993), and MarquesAguiar (1994). Additional descriptions and comparative measurements can also be found (under the name A.fuliginosus) in Swanepoel and Genoways (1979), Koepcke and Kraft (1984), and Brosset and Charles-Dominique (1990). No subspecies are currently recognized (Handley, 1987, 1989; MarquesAguiar, 1994). Although our voucher material generally agrees with previous qualitative descriptions of Artibeus obscurus, three of our specimens lack M3 completely, a polymorphism noted by Handley (1989) and Marques-Aguiar (1994), but not by Handley (1987) or Lim and Wilson (1993). Measurements of the Paracou series likewise fall within the known range of variation for this species with the exception of our largest specimens, which have slightly longer forearms than previously reported. F IELD O BSERVATIONS : We recorded 117 captures (probably including some recaptures) of Artibeus obscurus at Paracou, of which 104 were in ground-level mistnets, 7 were in elevated mistnets, and 6 were at 104 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY roosts. Of the 104 ground-level mistnet captures, 42 were in well-drained primary forest, 37 were in swampy primary forest, 19 were in creekside primary forest, 4 were in manmade clearings, 1 was in closed-canopy secondary growth, and 1 was over a roadside puddle. Of the seven elevated mistnet captures, five were made between 7 and 20 m above a narrow dirt road, one was 10 m above a treefall in well-drained primary forest, and one was 5–8 m above the ground in the subcanopy of swampy primary forest. We found three roosting groups of Artibeus obscurus under exfoliating pieces of bark 6–7 m above the ground on the trunks of grignon trees, Ocotea rubra (Lauraceae), in well-drained primary forest (fig. 40). One of these groups consisted of an adult female with a nursing juvenile; another consisted of an adult female, a nursing juvenile, and an escaped adult of unknown sex; and the third was a solitary near-term pregnant female. We also found a solitary adult male Artibeus obscurus roosting beneath an unmodified leaf of Phenakospermum guyannense (Strelitziaceae) about 4 m above the ground at the edge of a clearing in secondary vegetation. Artibeus (Dermanura)cinereus (Gervais) Figures 41, 42 V OUCHER M ATERIAL : 9 females (AMNH *266259, *266265, *266266, *266270, *267499, *267991; MNHN *1995.1108, *1995.1109, *1995.1110) and 16 males (AMNH *266260, *266261, *266290, *266291, *266302, *266306, *266307, *266320, *267196, *267978, *267980; MNHN *1995.1111, *1995.1112, *1995.1113, *1995.1114, *1995.1115); see table 39 for measurements. I DENTIFICATION : The most recent revision of the smaller species of Artibeus (subgenus Dermanura) is Handley’s (1987), whose meticulous species comparisons and key were our primary resource for identification. Additional comparative measurements of A.cinereus and A.gnomus from French Guiana provided by Brosset and Charles-Dominique (1990) were also helpful. Although both species probably occur in Surinam and may have been mixed in earlier collections, Husson’s (1962, 1978) account of A.cinereus seems to have been based entirely upon specimens properly referred to that species. Koopman (1994) listed seven subspecies of A.cinereus, however all but two of these appear to represent other species (see Handley, 1987, and below). Our voucher material, one of the largest series of Artibeus cinereus reported from a single locality, conforms in all respects to Handley’s (1987) description of the species. Although our specimens generally fall within the range of morphometric variation previously reported in the literature, a few individuals of both sexes are slightly smaller in some dimensions than those documented from the Guianas by Husson (1962, 1978) and Brosset and Charles-Dominique (1990). As with the larger species of Artibeus (subgenus Artibeus), we found that we could not always distinguish species of Dermanura in the field. Most individuals of cinereus are larger than most specimens of gnomus (table 39), and most cinereus have cream-colored ear margins and tragus while those structures are bright yellow in most gnomus. However, we captured some individuals that we initially misidentified using these characters. Of 67 collected specimens of Dermanura,65 (97%) were correctly identified in the field; two individuals (3%) were initially identified incorrectly, both cases involving specimens of A.cinereus (one male and one female) misidentified as A.gnomus. We had to clean the skulls of these and several other specimens that fell in the zone of size overlap between A.cinereus and A.gnomus (see table 39) in order to determine their correct identifications. We are unaware of any external characters that allow unequivocal identification of Artibeus cinereus and A.gnomus where these taxa occur in sympatry, but the number of lower molars appears to be reliable in our Paracou sample: all individuals with two lower molars present on both sides proved to be A.cinereus upon subsequent examination, and all individuals with three lower molars on both sides were A.gnomus (one individual of cinereus had two lower molars on one side and three on the other). Although the number of lower molars is perhaps the best single character for distinguishing these taxa in the field, it is not very useful for nonde- 1998 111SIMMONS AND VOSS: PARACOU BATS Fig. 45. Detail of the leaf-tent roost of Artibeus cinereus shown in figure 43. This is the lower surface of the leaf, which is conspicuously whitish by contrast with the dark green upper surface (facing page). The long black spines, grouped in clusters all along the midrib, had been chewed to short stubs for the distalmost 15 cm or so, near the apex of the tent where the bats were hanging (arrow). 112 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY *1995.1137, *1995.1138); see table 39 for measurements. I DENTIFICATION : The original description of Artibeus gnomus by Handley (1987) is still the basic reference for this species, although the measurements tabulated by Brosset and Charles-Dominique (1990) for their French Guianan material are also helpful. The few problems we encountered in distinguishing A.gnomus from A.cinereus were discussed in the preceeding species account. No subspecies of Artibeus gnomus are recognized (Handley, 1987). Our voucher material corresponds closely with Handley’s (1987) description of Artibeus gnomus, and measurements of our specimens generally fall within the range of variation previously reported for the species. One exception is a particularly large male (AMNH 267987), whose measurements in several dimensions exceed any previously reported in the literature (e.g., length of maxillary toothrow, 6.08 mm; breadth across molars, 8.30 mm; zygomatic breadth, 11.47 mm). Except in size, however, this specimen agrees morphologically with the remainder of our specimens, and we conclude that it simply represents an unusually large example. Unpublished measurement data from other localities (in Venezuela, Guyana, Peru, and Brazil; Handley, personal commun.) indicate that AMNH 267987 is not unique in this respect, and that the normal range of size variation in A.gnomus is somewhat greater than that originally reported by Handley (1987). F IELD O BSERVATIONS : We recorded 59 captures (possibly including some recaptures) of Artibeus gnomus at Paracou: 51 in groundlevel mistnets, 4 in elevated mistnets, and another 4 at roosts. Of the 51 ground-level mistnet captures, 6 were in well-drained primary forest, 17 were in swampy primary forest, 5 were in creekside primary forest, 22 were in manmade clearings, and 1 was in closed-canopy secondary growth. It is noteworthy that of the 22 ground-level mistnet captures in manmade clearings, 18 were taken in nets erected around small fruiting trees, Vismia sp. (Clusiaceae), growing along a narrow road through well-drained primary forest; none were made in large clearings far removed from primary forest. Of the elevated mistnet captures, two were made between 4 and 21 m above a narrow dirt road and two were made at 7–10 m in the subcanopy of swampy primary forest. We found four roosting groups of Artibeus gnomus, all of which occupied leaf-tents. Two large roosting groups (one with 4–8 individuals, the other with perhaps 6–10) each inhabited a single modified leaf of the giant herb Phenakospermum guyannense (Strelitziaceae); both roosts were hard to see, about 3 m above the ground in the densely cluttered understory of swampy primary forest (fig. 46). We captured (and preserved as vouchers) only a single adult female from each of these roosting groups, the age and sex composition of which are otherwise unknown. The other two roosts, inhabited by solitary adult males, were ‘‘apical’’ tents (sensu Kunz et al., 1994) made from the spade-shaped leaves of epiphytic Philodendron sp. (Araceae). One of these roosts was about 5 m above the ground in well-drained disturbed forest, the other about 10 m above the ground in well-drained primary forest. We found many epiphytic Philodendron plants with similarly modified leaves throughout our study area, but only these two examples were occupied by bats (wasp nests were found under the modified leaves of a few plants). Although Artibeus gnomus and A.cinereus are morphologically similar, our data suggest they are ecologically divergent. Consistent with Brosset and Charles-Dominique’s (1990) generalizations concerning habitat use in these species, an analysis of ground-level capture frequency data from Paracou (table 40) suggests that A.gnomus is more abundant in primary forest than A.cinereus, which apparently favors the early-successional vegetation of modified habitats. Whether or not A.cinereus consistently inhabits bifid palm-leaf tents whereas A.gnomus uses differently designed tents made from the leaves of large herbs is unknown, but this hypothetical difference could easily be tested by following the movements of positively identified bats fitted with radio transmitters. Because A.gnomus was only recently recognized as taxonomically distinct from A.cinereus, and because they are hard 1998 113SIMMONS AND VOSS: PARACOU BATS Fig. 46. Roost of Artibeus gnomus in swampy primary forest at Paracou. As many as eight individuals may have inhabited the dark apex of this leaf-tent (arrow), but it was impossible to obtain a clear view to make an accurate count. Phenakospermum guyannense (Strelitziaceae) is moderately common in swampy primary forest and secondary vegetation at Paracou, where tent-roosts made from its large leaves are also inhabited by Uroderma bilobatum (fig. 51). This tent was made by cuts in the lateral veins and interstitial tissue along both sides of the midrib causing the sides of the leaf to droop (as in the ‘‘boat’’ tents described and illustrated by Kunz et al., 1994), and by a deep cut through the midrib, causing the apex of the leaf to hang straight down (a characteristic of ‘‘apical’’ tents described by the same authors). A similar roost inhabited by another large group of A.gnomus was found several kilometers from this site. 114 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY (but not impossible) to distinguish in the field, unvouchered ecological observations reported in the literature for these species should be regarded with caution. Artibeus (Koopmania)concolor Peters V OUCHER M ATERIAL : 17 females (AMNH *266267, *266269, *267192, *267194, *267476, *267478, *267479, *267487, *267488, *267981, *267982, *267983; MNHN *1995.1116, *1995.1117, *1995.1118, *1995.1119, *1995.1120) and 7 males (AMNH *267193, *267195, *267477, *267502; MNHN *1995.1121, *1995.1122, *1995.1123); see table 41 for measurements. I DENTIFICATION : Features useful for identifying Artibeus concolor were summarized by Handley (1987) and Acosta and Owen (1993); additional descriptions and measurements can be found in Husson (1962, 1978), Hill (1964), Barriga-Bonilla (1965), Linares (1969), Genoways and Williams (1979), Swanepoel and Genoways (1979), and Brosset and Charles-Dominique (1990). In contrast to the situation with most other species of Artibeus, no taxonomic problems are apparently associated with A.concolor,so identification is relatively straightforward. No subspecies are currently recognized (Handley, 1987; Acosta and Owen, 1993; Koopman, 1994). Our voucher material, apparently the largest series available from any single locality, conforms closely with previous qualitative and morphometric descriptions of Artibeus concolor. In particular, measurements of our specimens confirm Brosset and CharlesDominique’s (1990) report of considerable size variation within French Guianan populations of this species. The size range among our specimens is even greater than they reported, yet we found no evidence that the Paracou sample includes more than one taxon. 1998 115SIMMONS AND VOSS: PARACOU BATS F IELD O BSERVATIONS : We caught 24 Artibeus concolor at Paracou, of which 22 were taken in ground-level mistnets and 2 in elevated mistnets. Of the 22 ground-level captures, 3 were in well-drained primary forest, 7 were in swampy primary forest, 3 were in creekside primary forest, 7 were in manmade clearings, and 2 were over roadside puddles. The two elevated mistnet captures were made between 10 and 21 m over a narrow dirt road. Chiroderma trinitatum Goodwin V OUCHER M ATERIAL : 8 females (AMNH *266255, *267189, *267473, *268531, *269117; MNHN *1995.1191, *1995.1192, *1995.1193) and 4 males (AMNH *266256, *268532, *269118; MNHN *1995.1194); see table 42 for measurements. I DENTIFICATION : Descriptions and measurements of Chiroderma trinitatum from the Guianas and elsewhere can be found in Goodwin (1958), Goodwin and Greenhall (1961), Ojasti and Linares (1971), Gardner (1976), Genoways and Williams (1979), Swanepoel and Genoways (1979), and Brosset and Charles-Dominique (1990). Although subspecies of C.trinitatum have been recognized by some authors (e.g., Jones and Carter, 1976), increased sampling throughout the range of this species has demonstrated more within-population variation than previously suspected (see discussion in Williams and Genoways, 1980a). Pending a thorough systematic review, no trinomial nomenclature seems warranted (Koopman, 1994). Our Paracou specimens conform in all respects to previous qualitative and morphometric descriptions of Chiroderma trinitatum. F IELD O BSERVATIONS : We caught 13 Chiroderma trinitatum at Paracou, of which 11 were taken in ground-level mistnets and 2 in elevated mistnets. Four of the ground-level captures were made in swampy primary forest and the other seven in manmade clearings. The elevated mistnet captures were made at 17–20 m above a narrow dirt road. Chiroderma villosum Peters V OUCHER M ATERIAL : 3 females (AMNH *267191, *267474, *268536) and 5 males (AMNH *267190, *267475, *268535; MNHN *1995.1195, *1995.1196); see table 42 for measurements. I DENTIFICATION : Descriptions and comparative measurements of Chiroderma villosum can be found in Goodwin and Greenhall 116 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY (1961), Husson (1962, 1978), Hill (1964), Genoways and Williams (1979), Swanepoel and Genoways (1979), Hall (1981), Brosset and Charles-Dominique (1990), and Anderson (1997). Two subspecies are currently recognized: C.v.jesupi (Mexico to northern Colombia) and C.v.villosum (tropical South America east of the Andes from Colombia to southeastern Brazil, including Trinidad and Tobago) (Koopman, 1994). Although our Paracou series of Chiroderma villosum agrees in qualitative and quantitative characters with previous descriptions of the species as a whole, craniodental measurement comparisons indicate that our specimens are substantially smaller than most of those previously reported from the Guianas. For example, the observed range in length of the maxillary toothrow is 7.96–8.66 mm at Paracou versus 8.6–10.2 mm at other Guianan localities from which measured specimens are reported in the literature cited above. An apparently individual anomaly is represented by one of our male specimens, AMNH 268535, which has only one pair of upper and lower incisors; all of our remaining specimens have two pairs in both jaws, the normal formula for Chiroderma (see Koopman, 1994). F IELD O BSERVATIONS : We caught eight Chiroderma villosum at Paracou, of which four were taken in ground-level mistnets and four in elevated nets. Two ground-level captures were made in well-drained primary forest, one in swampy primary forest, and one in creekside primary forest. Two specimens were captured 34–37 m above a treefall opening in well-drained primary forest and two others were netted 17–20 m above a narrow dirt road. Ectophylla Phylogenetic relationships between Ectophylla,Mesophylla, and Vampyressa have been the subject of much debate in the literature. Most workers have retained all three as distinct genera (e.g., Hall, 1981; Koopman, 1993, 1994), but some authors have considered Mesophylla to be a junior synomym of Vampyressa (e.g., Owen, 1987) or of Ectophylla (e.g., Goodwin and Greenhall, 1962; Handley, 1976). Recently, Peffley et al. ( MS ) found strong support for a sistergroup relationship between Ectophylla and Mesophylla, both of which are monotypic, and argued that Ectophylla alba H. Allen (1892) and Mesophylla macconnelli Thomas (1901a) should be placed in a single genus to reflect this relationship. We therefore use the older generic name for both species, and provide a formal diagnosis for Ectophylla as so defined. E MENDED D IAGNOSIS OF E CTOPHYLLA: Size small (weight less than 11 g and forearm length less than 35 mm); dorsal and ventral fur pale buff, grayish white, or white; no white facial stripes or middorsal stripe; skin of noseleaf, ears, and thumb bright yellow; ventral border of narial horseshoe defined by a free flap of skin; uropatagium short, naked, translucent; length of calcar less than onehalf length of hindfoot; dental formula I 2/2, C 2/2, P 2/2, M 2/2–3 3 2 5 28–30; rostrum approximately three-fourths the length of the braincase; rostrum not inflated and without a deep depression or long nasal emargination; interpterygoid space not extended by a deep palatal emargination; inner upper incisors elongate, unworn crown height more than twice that of outer incisors; inner upper incisors not deeply bifid; m1 without posterolingual cusp (crown resembles that of last premolar); lingual cusps of m2 vestigial or absent. Ectophylla macconnelli (Thomas) V OUCHER M ATERIAL : 9 females (AMNH *267281, *267537, *267538, *267556, *267558, *267559; MNHN *1995.1181, *1995.1182, *1995.1183) and 4 males (AMNH *267557, *267562, *268539; MNHN *1995.1184); see table 43 for measurements. I DENTIFICATION : We consulted descriptions and measurements of Ectophylla macconnelli provided by Goodwin and Greenhall (1962), Swanepoel and Genoways (1979), Williams and Genoways (1980a), Brosset and CharlesDominique (1990), and Kunz and Pena (1992). Two subspecies are currently recognized, of which the nominate form occurs throughout most of the humid Neotropical lowlands, including the Guianas (Koopman, 1994). 1998 117SIMMONS AND VOSS: PARACOU BATS Our Paracou series conforms in all respects to previous descriptions of Ectophylla macconnelli except that of Brosset and Charles-Dominique (1990), who reported unusually high values for length of the maxillary toothrow (6.7–7.4 mm) in their French Guianan material. By contrast, the observed range for this measurement in our series (5.61–6.09 mm) falls within the range previously reported by other authors (5.5–6.6 mm). Because the other external and craniodental measurements reported by Brosset and Charles-Dominique are not aberrant, we presume that their maxillary toothrow measurements were erroneously reported. F IELD O BSERVATIONS : We caught 13 Ectophylla macconnelli at Paracou, of which only 1 was taken in a mistnet; the remaining 12 were caught at roosts. Our single mistnet capture was at ground level in swampy primary forest. We found three roosting groups of Ectophylla macconnelli, all of which inhabited leaf-tents (fig. 47) made from the bifid terminal leaflets of fronds of young understory palms provisionally identified as Astrocaryum sciophilum. In all construction details that we noted, these tents exactly resembled those described above in the accounts for Rhinophylla pumilio and Artibeus cinereus. All three roosts were in well-drained primary forest. One roosting group of seven bats, collected in its entirety, consisted of two adult males and five adult females. Another entire group of three consisted of one adult male and two adult females. Our roost observations, together with those reported by Foster (1992) and CharlesDominique (1993), suggest that Ectophylla macconnelli regularly inhabits tents manufactured from the leaves of Astrocaryum palms. 12 Because the undersurfaces of Astrocaryum leaves are characteristically whitish (Henderson et al., 1995), we agree with Hingston (1932) that the unsually pale fur of 12 Emmons (1990) was apparently the first to report that Ectophylla macconnelli inhabits palm-leaf tents, but she did not identify the host plant. Roosts have been reported in the foliage of other palms (and aroids), but multiple independent observations of occupied bifid tents in Astrocaryum spp. from opposite ends of Amazonia suggest that this genus is favored by Ectophylla macconnelli. Emmons’ (1990, 1997) suggestion that E. macconnelli sometimes inhabits hollow trees was based on Handley’s (1976: 30) report of one individual ‘‘found roosting in a tree.’’ However, the original field record for the specimen in question (USNM 405185) notes ‘‘shot in tree after frightened up‘‘—the bat having presumably been dislodged from an unobserved roost in the undergrowth. All other published accounts (including Beebe [1925], Hingston [1932], Charles-Dominique [1993], and Kunz et al. [1994] in addition to references cited by Foster [1992]) explicitly identify foliage as the roosting substrate of E.macconnelli. 118 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 47. Roost of Ectophylla macconnelli in well-drained primary forest at Paracou. Made from the bifid terminal leaflet of a young palm (Astrocaryum cf. sciophilum), this shelter contained three E. macconnelli (one adult male and two adult females) hanging in a tight cluster from the midrib about 1.5 m above the ground (arrow). Note the much sparser undergrowth at this primary-forest site than that surrounding an otherwise similar roost in secondary vegetation (figure 43). 1998 119SIMMONS AND VOSS: PARACOU BATS E.macconnelli is correlated with the background color of its typical roosts, perhaps making the bats less conspicuous to diurnal predators. Platyrrhinus helleri (Peters) V OUCHER M ATERIAL : 9 females (AMNH *266254, *267182, *267551, *267554, *267555, *268540; MNHN *1995.1187, *1995.1188, *1995.1189) and 5 males (AMNH *267179, *267550, *268541, *268542; MNHN *1995.1190); see table 44 for measurements. I DENTIFICATION : A key to the species of Platyrrhinus was provided by Ferrell and Wilson (1991), and Anderson (1996) helpfully reviewed characters distinguishing P. helleri from P.brachycephalus (see also Rouk and Carter, 1972). Other useful descriptions and measurements of P.helleri can be found in Sanborn (1955), Goodwin and Greenhall (1961), Husson (1962, 1978), Gardner and Carter (1972), Rouk and Carter (1972), Swanepoel and Genoways (1979), Brosset and Charles-Dominique (1990), and Anderson (1997). Two subspecies of P.helleri are currently recognized, of which P.h. incarum ranges throughout Amazonia, including the Guianas (Koopman, 1994). Our Paracou specimens conform closely with previous descriptions of Platyrrhinus helleri. Although some populations (e.g., in Bolivia; Anderson, 1996) exhibit variation in the number of accessory cusps present on the anterior margin of the second lower premolar, all of our vouchers have only a single accessory cusp in this location. Measurements of the Paracou series (table 44) fall within the range of variation previously reported for P.h.incarum. F IELD O BSERVATIONS : We captured 14 individuals of Platyrrhinus helleri at Paracou, of which 11 were taken in ground-level mistnets and 3 in elevated mistnets. Four groundlevel captures were in well-drained primary forest, one in swampy primary forest, two in creekside primary forest, two in manmade clearings, and two over roadside puddles. The three elevated net captures were made between 6 and 21 m above a narrow dirt road. Sturnira lilium (E. Geoffroy) Figures 48–50 V OUCHER M ATERIAL : 19 females (AMNH *266210, *266226, *266231, *266234, *266235, *266236, *267170, *268543, *268546, *268549, *268552, *268553; MNHN *1995.1197, *1995.1198, *1995.1199, *1995.1200, *1998.600, *1998.601, *1998- .602) and 35 males (AMNH *266199, *266200, *266201, *266203, *266205, 120 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 48. Dorsal (A) and ventral (B) views of the skull of Sturnira lilium (AMNH 266232; male) from Paracou. Dorsal (C) and ventral (D) views of the skull of S.tildae (AMNH 267461; male) from Paracou. Note the species differences in incisor morphology, size and shape of the molars, and relative breadth of the mesopterygoid fossa and mastoid region. Scale bars 5 10 mm. 1998 127SIMMONS AND VOSS: PARACOU BATS Fig. 51. Leaf-tent roost of the type inhabited by Uroderma bilobatum, in a wet opening in swampy primary forest at Paracou. This simple conical shelter, made by cutting the midrib of a Phenakospermum guyannense frond (arrow), was unoccupied when we found it, but an identical tent nearby contained three bats. Because it is difficult to approach such roosts without alarming the inhabitants, we found many empty tents that might have contained bats only moments before. 128 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY ant readings are current in the literature. Thus, Williams and Genoways (1980a) stated that V.brocki lacks a middorsal stripe, citing Peterson as their authority, whereas other authors have evidently interpreted Peterson to mean that a stripe is present in this species (e.g., Lewis and Wilson, 1987). Peterson (1968: 1) himself clearly stated that the holotype of brocki is ‘‘greyish brown above with a faint indistinct dorsal whitish stripe.’’ A diagrammatic character summary accompanying the original description (op. cit.: fig. 6), however, does not distinguish between the faint middorsal stripe of brocki and the much brighter markings of two other congeners (nymphaea and bidens). Our Paracou vouchers have a middorsal stripe, but it is extremely faint and is visible only when the dry dorsal pelage is carefully brushed. The stripe is hard to see under any conditions because it is only slightly paler than the surrounding fur, and it becomes quite invisible when the fur is wet (precluding observation in alcohol-preserved specimens). Because the dorsal stripe is hard to see, Vampyressa brocki externally resembles V. pusilla, with which it might be confused in the field. The latter species entirely lacks a dorsal stripe and shares several traits with brocki, including small size, two pairs of lower incisors, and absence of m3 (Peterson, 1968). Characters that unambiguously distinguish brocki and pusilla include a suite of craniodental features (op. cit.), of which the most useful include: (1) form of the inner upper incisors (tapering to a point in brocki, bilobed in pusilla), (2) shape of the anterior lower premolar (low-crowned and bladelike in brocki, with higher crown and spearlike anterior cusp in pusilla), and (3) shape of the nasal aperture (with straight ventral border in brocki, V-shaped ventral border in pusilla). The latter character is visible only on cleaned skulls, but the distinction is unambiguous and provides a means of unequivocally identifying old individuals with worn or broken teeth. F IELD O BSERVATIONS : Of the seven Vampyressa brocki we captured at Paracou, four were taken in ground-level mistnets and three in elevated nets. One of the groundlevel captures was in well-drained primary forest, two were in creekside primary forest, and one was in a manmade clearing. The elevated net captures were made 17–21 m over a narrow dirt road. FURIPTERIDAE We captured a single furipterid species at Paracou, the only member of its family known to occur in rainforest habitats. 1998 129SIMMONS AND VOSS: PARACOU BATS Furipterus horrens (F. Cuvier) V OUCHER M ATERIAL : 9 females (AMNH *265975, *265979, 265980, *267213, *267214, *268572, 268573; MNHN *1995.870, *1995.871) and 4 males (AMNH *267212, *267507; MNHN *1995.872, 1995.873); see table 49 for measurements. I DENTIFICATION : Husson (1962, 1978) provided a detailed description and measurements of Furipterus horrens, and additional morphometric data were published by Brosset and Charles-Dominique (1990). No subspecies are recognized (Koopman, 1994). Although our voucher material does not differ in any respect from Husson’s (1962, 1978) careful qualitative description of Furipterus horrens, measurements of the Paracou series document a greater range of size variation than that previously reported from the Guianas. Our measurement data (table 49) additionally suggest that the species may be sexually dimorphic (females averaging slightly larger than males in most dimensions), an observation that tends to corroborate Uieda et al.’s (1980) report of sexual dimorphism in a northeastern Brazilian population. In her otherwise excellent external descriptions of Furipterus horrens, Emmons stated that the thumb has ‘‘no claw’’ (1990: 82) or ‘‘almost no claw’’ (1997: 91). The thumb, small and enclosed in the wing membrane, bears a tiny but distinct claw in all the specimens we examined. F IELD O BSERVATIONS : We collected 13 specimens of Furipterus horrens at Paracou, of which 12 were taken at roosts and 1 was shot as it flew back and forth along a regular beat about a meter above the ground in welldrained primary forest at night. The nine roosting groups we found consisted of one to two individuals, either solitary adults (of both sexes) or lactating females with nursing juveniles. In no case did we find more than one adult occupying a roost. All of the roosts we found were in or under fallen trees in various stages of decomposition. One roost was a small cavity in the broken end of a rotting log (fig. 52), but most were on the undersides of trunks or in dark chambers between buttresses (fig. 17). Four roosts were in well-drained primary forest, one was in swampy primary forest, two were in primary forest of unrecorded character, one was in selectively logged forest, and one was in closed-canopy secondary growth. Small, dark, solitary bats that flew away unidentified from refugia in or under woody debris on many occasions throughout the course of our fieldwork at Paracou were probably F. horrens. Although we never caught this species in mistnets, our impression was that roosts of F.horrens could be found by careful searching almost anywhere in the forest. THYROPTERIDAE Although we caught only one thyropterid species at Paracou, a second is known from 130 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 52. Roost of Furipterus horrens in well-drained primary forest at Paracou. An adult female with her nursing offspring inhabited this dark but shallow cavity in the broken end of a rotting log, about 40 cm above the ground (arrow). Such inconspicuous refugia occur everywhere in the forest and we probably overlooked many F.horrens roosts even along frequently traveled trails through our study area. French Guiana and might also be expected to occur in our study area (appendix 1). Thyroptera tricolor Spix V OUCHER M ATERIAL : 16 females (AMNH *266348, *266356, *266358, 266359, *266364, *267215, *267216, *267217, *267218, *268576; MNHN *1995.874, *1995.875, *1995.876, *1995.877, 1995.878, *1995.879) and 12 males (AMNH *266352, 266355, 266357, *266361, *266363, 266365, *268574, 268577; MNHN 1995.880, *1995.881, 1995.882, 1995.883); see table 50 for measurements. I DENTIFICATION : We consulted the descriptions and measurements of Thypotera tricolor provided by Husson (1962, 1978), Brosset and Charles-Dominique (1990), and Pine (1993) to confirm the identification of our material. Although three subspecies were recognized by Wilson and Findley (1977) and Koopman (1994), the morphological justification for a trinomial classification is not clear. Furthermore, Pine (1993) suggested that some published observations of geographic variation within T.tricolor may have been based on material that was not correctly identified to species. Pending a thorough review of the problem, it currently seems pointless to employ subspecific nomenclature. Our material from Paracou conforms with published descriptions of Thyroptera tricolor in all respects. Although Pine (1993) noted some variation in the number of lappets on the calcar in some populations of this species, all of our specimens have two lappets on the calcar. The free portion of the tail is relatively long in all our fluid-preserved material, which conforms to Pine’s (1993) ob- 1998 131SIMMONS AND VOSS: PARACOU BATS servations. We note, however, that the free portion of the tail appears quite short in our skins, an artifact that resulted from pinning the specimens to dry with the uropatagium maximally extended. The size range documented by our voucher specimens (table 50) is somewhat greater than that previously reported from Surinamese and French Guianan populations of Thyroptera tricolor by Husson (1962, 1978) and Brosset and Charles-Dominique (1990). In addition, the Paracou measurement data suggest some slight sexual dimorphism, with females exceeding males in average body weight and total length. F IELD O BSERVATIONS : We recorded 40 captures (possibly including some recaptures) of Thyroptera tricolor at Paracou, of which 3 were in ground-level mistnets, 1 was in an elevated net, and 36 were at roosts. One ground-level mistnet capture was in welldrained primary forest, one was in swampy 132 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 53. Roost of Thyroptera tricolor in the half-unrolled new leaf of a small Heliconia growing in swampy primary forest at Paracou. The adhesive suckers of the roosting bats are visible as dark spots through the translucent tissue of the leaf; the bats themselves (an adult male and three adult females) form a dark mass inside their tubular shelter (arrow). 1998 133SIMMONS AND VOSS: PARACOU BATS Fig. 54. Roost of Thyroptera tricolor in a scrolled dead leaf of Phenakospermum guyannense (arrow) in secondary vegetation at Paracou. We found two other roosts of T.tricolor in dead leaves like this one, which is hanging downward from its broken petiole. This roost contained about four bats, one of which was collected as a voucher. 134 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY primary forest, and one was in a manmade clearing. Our single elevated mistnet capture was made at 7–8 m above the ground in the subcanopy of swampy primary forest. We found 12 roosting groups of Thyroptera tricolor, all of them in foliage (table 51). Most (nine) roosts were in the erect, halfunrolled new leaves of heliconias, Heliconia sp. (Heliconiaceae) (fig. 53), or Phenakospermum guyannense (Strelitziaceae), but three roosting groups occupied scrolled dead leaves hanging from large Phenakospermum plants (fig. 54). Of the five Heliconia roosts we found, one was in well-drained primary forest, two were in swampy primary forest, and two were in creekside primary forest. Five Phenakospermum roosts were in secondary vegetation (especially along the margins of a small savanna enclave) and two were in wet glades surrounded by swampy primary forest. All of the leaves that we found used by T.tricolor as roosts were shaded; none was in direct sunlight. The number of bats per roost varied from one to six with a well-defined mode of four. We did not record more than a single adult male in any roost. VESPERTILIONIDAE We captured five vespertilionid species at Paracou belonging to the genera Eptesicus, Lasiurus, and Myotis. Of the six other vespertilionids known from French Guiana and Surinam, four could plausibly be expected to occur in our study area also (appendix 1). Eptesicus chiriquinus Thomas Figures 55, 56 V OUCHER M ATERIAL : 2 females (AMNH *267531, *268584) and 4 males (AMNH *267234, *267530; MNHN *1995.961, 1995.962); see table 52 for measurements. I DENTIFICATION : In his revision of the South American species of Eptesicus, Davis (1966) recognized an andinus group of species with long blackish fur. Several taxa were included in this complex: andinus Allen (1914), chiriquinus Thomas (1920b), inca Thomas (1920b), montosus Thomas (1920b) and chiralensis Anthony (1926). Of these, Davis considered inca and chiriquinus to be strict junior synonyms of E.andinus, the larger of the two species he recognized; the smaller species, E.montosus, included chiralensis as a valid subspecies. Koopman (1978) disagreed, claiming that the long-haired forms of Neotropical Eptesicus are restricted to cool highland areas, whereas the short-haired forms occur in the warm lowlands. By his interpretation, the long-haired taxa are local variants of shorthaired species that have adapted to cooler conditions at higher elevations. Accordingly, Koopman (1978, 1993, 1994) treated the members of Davis’ (1966) andinus group as subspecies of nomenclaturally older shorthaired species based on size, assigning andinus to E.brasiliensis, and montosus and chiralensis to E.furinalis. However, Koopman’s adaptive scenario concerning hair length is refuted by the occurrence of longhaired Eptesicus—clearly referable to Davis’ andinus group—in lowland areas of Brazil, Panama, Venezuela, and French Guiana. 13 In addition, the sympatric occurrence in northeastern Venezuela of E.brasiliensis,E.furinalis, and a third species that Ochoa et al. (1993) identified as E.andinus definitely indicates that Eptesicus species with different pelage types can coexist at the same elevation. Specimen data cited by Davis (1966), together with other collections subsequently reported in the literature (e.g., by Handley, 1976), provide compelling evidence that two species referable to the andinus group are sympatric at several South American localities. However, Davis did not personally examine any of the relevant holotypes in this complex, so his decisions about synonymies 13 We examined dark, long-furred specimens of Eptesicus, clearly referable to Davis’ andinus group, from lowland areas of Brazil (Amazonas, Rio Madeira, Santo Antonio do Guajara´ [ca. 25 m]: AMNH 92251, 93787), French Guiana (Paracou [ca. 30 m]: AMNH 267234, 267530, 267531, 268584; MNHN 1995.961, 1995.962), and Panama (San Blas, Armila, Quebrada Venado [sea level]: USNM 335411). In addition, unambiguous descriptions of andinus-group specimens have been reported in the literature from a lowland site in Venezuela (Bolı´var, Imataca Forest Reserve [180 m]; Ochoa et al., 1993) and another in French Guiana (Piste Saint-E ´lie [ca. 45 m]; Brosset and Charles-Dominique, 1990). 1998 135SIMMONS AND VOSS: PARACOU BATS Fig. 55. Dorsal (A) and ventral (B) views of the skull of the holotype of Eptesicus andinus (AMNH 33807; male) from Colombia. Dorsal (C) and ventral (D) views of the skull of E.chiriquinus (AMNH 267234; male) from Paracou. E.chiriquinus has a consistently larger and more heavily built skull than does E.andinus. Also note the flattened, triangular bony plate at the intersection of the sagittal and nuchal crests in andinus, a structure never present in chiriquinus. Scale bars 5 10 mm. 136 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 56. Lateral views of the skull and lower jaw of (A) the holotype of Eptesicus andinus (AMNH 33807; male) and (B)E.chiriquinus from Paracou (AMNH 267234; male). Note the species difference in development of the sagittal crest. Scale bars 5 10 mm. are problematic. Because Koopman’s (1978) review of the situation was obviously not an improvement, a fresh appraisal of the systematics of Neotropical Eptesicus is necessary. To determine the correct identification of several blackish, long-haired specimens of Eptesicus captured at Paracou, we examined the holotypes of E.andinus and E.chiralensis as well as 136 other specimens of Neo- 1998 143SIMMONS AND VOSS: PARACOU BATS Fig. 57. Dorsal (A) and ventral (B) views of the skull of Myotis nigricans (AMNH 267228; female) from Paracou. Dorsal (C) and ventral (D) views of the skull of M.riparius (AMNH 268591; female) from Paracou. Note differences in structure of the anterior braincase, palate, and auditory region. Scale bars 5 5 mm. 144 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 58. Lateral views of the skull and lower jaw of (A)Myotis nigricans (AMNH 267228; female) and (B)M.riparius (AMNH 268591; female). Scale bars 5 5 mm. ica based on postorbital breadth and braincase breadth, both of which are greater in M. albescens where these taxa occur in sympatry or near-sympatry (Husson, 1962, 1978; LaVal, 1973; Baud and Menu, 1993). Husson (1962, 1978) noted that Surinamese specimens of M.albescens have a braincase breadth . 6.8 mm and a postorbital breadth . 3.6 mm, whereas these dimensions are smaller in M.nigricans from Surinam. None of our specimens from Paracou has a braincase breadth . 6.75 mm, and only one specimen has a postorbital breadth . 3.50 mm (table 57). In our sample of 29 individuals, 1998 145SIMMONS AND VOSS: PARACOU BATS the specimen with the largest postorbital breadth (MNHN 1995.943, postorbital breadth 3.65 mm) has a braincase breadth identical to the sample mean (6.41 mm). Based on these data and comparisons of external morphology, we are confident that none of the Myotis collected at Paracou represents M.albescens.Myotis albescens has yet to be reported in French Guiana despite its occurrence in Surinam and northeastern Brazil. However, we note that several specimens identified as M.nigricans by Brosset and Charles-Dominique were reported to have postorbital breadth $ 3.7 mm, and it is possible that one or more of these specimens may represent M.albescens. Although M.albescens generally has a distinctively ‘‘frosted’’ pelage, LaVal (1973) noted that some specimens lack this feature and externally resemble M.nigricans. Our voucher material of Myotis nigricans falls within the lower range of size variation previously reported for this species (e.g., by LaVal, 1973; Wilson and LaVal, 1974). With the exception of postorbital breadth (noted above), the Paracou specimens are similar in all dimensions to specimens reported from other localities in French Guiana and Surinam (Husson, 1962, 1978; Brosset and Charles-Dominique, 1990). Most of our specimens have long, silky, brownish dorsal fur. However, several have fur that is brownish but woollier in texture, and one individual (a male) has almost black fur. The ventral fur is always slightly lighter in color than the dorsal fur, appearing frosted in many specimens. F IELD O BSERVATIONS : We made 29 vouchered captures of Myotis nigricans at Paracou, 25 of which were in ground-level mistnets and 4 of which were in elevated nets. Of the 25 ground-level captures, 3 were in creekside primary forest, 9 were in manmade clearings, and 13 were over roadside puddles. The elevated net captures were made between 4 and 13 m above a narrow dirt road. See the following species account for habitat comparisons with Myotis riparius and for information about captures of unidentified Myotis. Myotis riparius Handley Figures 57, 58 V OUCHER M ATERIAL : 6 females (AMNH *267224, *267524, *268591, *268592; MNHN *1995.946, *1995.947) and 6 males (AMNH *266366, *266376, *267523, *268589; MNHN *1995.948, *1995.949); see table 57 for measurements. 146 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY I DENTIFICATION : As noted above, identification of species of Neotropical Myotis requires reference to LaVal (1973). Descriptions and measurements of Myotis riparius can also be found in Handley (1960) and Brosset and Charles-Dominique (1990). No subspecies of M.riparius are currently recognized (Koopman, 1994). Our specimens of Myotis riparius conform to the description provided by LaVal (1973). A sagittal crest is present in all individuals, although it is poorly developed in two specimens; P3 is less than one-fourth the height of P4 in all individuals with both teeth (P3 is missing in one individual); and P3 is shifted to the inside of the toothrow in 64% of our specimens. Most of our specimens of M. riparius have woolly brown dorsal fur and slightly lighter ventral fur. However, some individuals have fur that is silkier in texture, and one specimen (clearly adult based on epiphyseal fusion) is gray-brown with slightly darker underparts. Measurements of our material resemble those reported by Brosset and Charles-Dominique (1990) for French Guianan M.riparius, although our larger series exhibits somewhat more size variation (table 57). F IELD O BSERVATIONS : We made 12 vouchered captures of Myotis riparius at Paracou: 11 individuals were taken in ground-level mistnets and 1 was shot as it flew back and forth in a regular beat about 5 m above a narrow dirt road at night. Of the 11 groundlevel mistnet captures, 5 were in well-drained primary forest, 4 were in swampy primary forest, 1 was in creekside primary forest, and 1 was in a manmade clearing. The observed habitat difference in groundlevel mistnet capture frequencies between Myotis nigricans and M.riparius is noteworthy despite the rather small numbers of vouchered captures available for comparison (table 58). Apparently, M.nigricans favors clearings whereas M.riparius is more commonly found beneath the primary forest canopy. To our knowledge, ecological differences between sympatric populations of these species have not previously been reported in the literature, possibly due to the difficulty of field identification. In addition to the vouchered captures of Myotis nigricans and M.riparius reported above, we recorded five unvouchered captures of Myotis in ground-level mistnets over roadside puddles. These bats were identified in the field as representing either M.nigricans or M.riparius, but were released without adequate confirmation of identification. MOLOSSIDAE We captured nine molossid species at Paracou, including members of the genera Eumops,Molossops,Molossus, and Promops. Four of these species have not been reported previously from French Guiana, and another, originally described from the department, has long been lost in synonymy. Another eight molossids known from elsewhere in French Guiana or Surinam might also occur in our study area (appendix 1). Eumops auripendulus (Shaw) V OUCHER M ATERIAL : 2 females (AMNH *267537, *268594) and 1 male (MNHN *1995.950); see table 59 for measurements. I DENTIFICATION : Eger (1977) provided species diagnoses and a key in her revision of the genus Eumops, which has not been superceded by any comparably comprehensive study. Husson (1962, 1978) gave a detailed description and comparative measurements 1998 147SIMMONS AND VOSS: PARACOU BATS of E.auripendulus from Surinam, and additional information can be found in Sanborn (1932) under the account for E.abrasus (now considered a junior synonym of auripendulus; see Goodwin, 1960; Husson, 1962, 1978; Eger, 1974, 1977). Two subspecies of E.auripendulus are currently recognized with the following continental distributions (Eger, 1974, 1977; Koopman, 1994): E.a. auripendulus (southern Mexico to Amazonia, including the Guianas) and E.a.major (northern Argentina, southern Paraguay, and eastern Brazil). These taxa are distinguished on the basis of size, with major being the larger of the two. Our voucher material conforms in all respects to Eger’s (1977) description of Eumops auripendulus auripendulus. The Paracou specimens are also comparable in size to those reported from Surinam by Husson (1962, 1978), although one of our females has a shorter maxillary toothrow and narrower skull than either of the females whose measurements he published (op. cit.: table 32). F IELD O BSERVATIONS : Our three specimens of Eumops auripendulus from Paracou were captured in mistnets suspended 17--23 m above a narrow dirt road through the forest. Eumops hansae Sanborn V OUCHER M ATERIAL : 1 female (AMNH *267538); see table 59 for measurements. I DENTIFICATION : Diagnostic characters and measurements of Eumops hansae have been discussed by Gardner et al. (1970), Eger (1977), and Brosset and Charles-Dominique (1990). No subspecies are currently recognized (Eger, 1977; Koopman, 1994). Our Paracou specimen conforms to published descriptions of Eumops hansae and falls within the range of size variation previously reported by Gardner et al. (1970), Eger (1977), and Brosset and Charles-Dominique (1990). F IELD O BSERVATIONS : Our single specimen of Eumops hansae was caught in a mistnet suspended 10–13 m over a narrow dirt road. Molossops Both of the species of Molossops that we captured at Paracou belong to the subgenus Cynomops, the contents of which have never been subjected to modern revisionary treatment. Table 60 summarizes the diagnostic traits of the four species that we recognize based on the literature and our examination of representative museum specimens. All of 148 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY 1998 149SIMMONS AND VOSS: PARACOU BATS the substantive taxonomic difficulties we encountered concern the three smaller species, an extended commentary on which is provided below in the account for M.paranus. Molossops (Cynomops)abrasus (Temminck) V OUCHER M ATERIAL : 1 female (AMNH *267534); see table 62 for measurements. I DENTIFICATION :Molossops abrasus is easily identified by size and other morphological traits from other members of the subgenus Cynomops (tables 60, 61). Four subspecies are currently recognized (Williams and Genoways, 1980a; Koopman, 1994): M.a.mastivus (Venezuela and the Guianas), M.a.brachymeles (eastern Peru), M.a.abrasus (eastern Brazil), and M.a.cerastes (Paraguay and northern Argentina). These taxa are distinguished principally on the basis of size, with mastivus being the largest (Goodwin, 1958; Taddei et al., 1976; Williams and Genoways, 1980a, 1980b). Our Paracou voucher, apparently the first record from French Guiana, conforms to previous qualitative descriptions of Molossops abrasus and falls within the range of size variation previously reported for the species (Goodwin, 1958; Taddei et al., 1976; Williams and Genoways, 1980a, 1980b; Koopman, 1994). Our specimen, a female, is somewhat smaller than a male identified as M.a.mastivus that Williams and Genoways (1980a) reported from Surinam, but this difference is of the same magnitude as the sexual dimorphism that Taddei et al. (1976) documented in a collection of M.abrasus from eastern Brazil. F IELD O BSERVATIONS : We caught our only specimen of Molossops abrasus in a mistnet suspended 18–21 m above a narrow dirt road. Molossops (Cynomops)paranus (Thomas) Figures 59, 60 V OUCHER M ATERIAL : 1 male (AMNH *267535); see table 62 for measurements. I DENTIFICATION :Molossops paranus, originally described as Molossus planirostris paranus by Thomas (1901b), continues to be recognized as a subspecies of Molossops planirostris by authors (e.g, Koopman, 1978, 1993, 1994). Although Handley (1976) reported specimens identified as Molossops paranus,M.planirostris, and M.greenhalli from the Venezuelan state of Bolı´var, he did not comment on the criteria he used to distinguish these taxa. Partial diagnoses and descriptions of the species and subspecies of Cynomops occur throughout the literature, but significant inconsistencies exist among published accounts. Conflicting descriptions 150 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY of ventral coloration, presence or absence of white hair bases, and degree of development of hair patches on the wings make unambiguous identification difficult. To address this problem, we consulted original taxonomic descriptions, literature accounts of additional material, and examined representative specimens of all of the smaller species to develop working diagnoses for identification. 16 16 The specimens we examined in addition to those from Paracou are: Molossops greenhalli (Mexico: USNM 511543, 523453; Panama: USNM 310264– 310268, 310270–310275, 368108, 396481, 449875; Venezuela: USNM 387745, 517509; Trinidad: AMNH 175326 [holotype], 176285, 207071); Molossops paranus (Panama: AMNH 183868, USNM 317627; Colombia: ROM 41479; Venezuela: USNM 387744; Guyana: ROM 32426, 57337–57338, 57375; Brazil: AMNH 79744–79745); Molossops planirostris (Panama: AMNH 183161, 183863; Venezuela: AMNH 17096– 17097; Brazil: AMNH 37043–37049, 37050–37052, 79725, 79727, 79731, 79733, 93879–93886, 92971, 92753–92755, 94630–94653, 236221; Paraguay: 234455–234459) 1998 151SIMMONS AND VOSS: PARACOU BATS Fig. 59. Dorsal (A) and ventral (B) views of the skull of Molossops paranus (AMNH 267535; male) from Paracou. Dorsal (C) and ventral (D) views of the skull of the holotype of M.greenhalli (AMNH 175326; male) from Trinidad. Note species differences in size and cranial proportions. Scale bars 5 10 mm. 152 NO. 237BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY Fig. 60. Lateral views of the skull and lower jaw of (A)Molossops paranus (AMNH 267535; male) and (B) the holotype M.greenhalli (AMNH 175326; male). Scale bars 5 10 mm.