The Mammals of Paracou, French Guiana: a Neotropical Lowland Rainforest Fauna Part 2

THE MAMMALS OF PARACOU, FRENCH GUIANA: A NEOTROPICAL LOWLAND RAINFOREST FAUNA PART 2. NONVOLANT SPECIES

ROBERT S. VOSS Associate Curator, Division of Vertebrate Zoology (Mammalogy) American Museum of Natural History

DARRIN P. LUNDE Collections Manager, Division of Vertebrate Zoology (Mammalogy) American Museum of Natural History

NANCY B. SIMMONS Associate Curator, Division of Vertebrate Zoology (Mammalogy) American Museum of Natural History

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Number 263, 236 pp., 99 ®gures, 69 tables Issued June 18, 2001 Price: $26.70 a copy

Copyright ᭧ American Museum of Natural History 2001 ISSN 0003-0090 CONTENTS Abstract ...... 3 ReÂsumeÂ ...... 5 Introduction ...... 8 Nonvolant Mammal Inventory Methods ...... 8 Conventional Trapping ...... 8 Arboreal Trapping ...... 12 Pitfall Trapping ...... 12 Hunting and Sight Surveys ...... 12 Interviews ...... 13 Previous Mammalogical Research at Paracou ...... 14 Voucher Preservation ...... 16 Systematic Accounts ...... 16 Marsupialia ...... 17 Xenarthra ...... 64 Primates ...... 66 Carnivora ...... 69 Perissodactyla ...... 72 Artiodactyla ...... 72 Rodentia ...... 73 Analyses of Sampling ...... 164 Sampling Results from Different Methods ...... 167 Estimating Completeness ...... 177 Discussion of the Nonvolant Fauna ...... 179 Taxonomic Composition and Biogeography ...... 180 Species Richness ...... 184 Trophic Guilds and Other Topics ...... 187 General Discussion ...... 192 Species Richness and Inventory Completeness ...... 192 Biogeography ...... 195 Community Structure ...... 207 Suggestions for Future Work ...... 210 Revisionary Taxonomy ...... 210 Mapping Guianan Endemicity ...... 211 Historical Connections with Other Areas of Endemism ...... 211 Causal Explanations for Community Composition ...... 212 Acknowledgments ...... 213 References ...... 214 Appendix 1: Nonvolant Mammals Previously Reported from French Guiana or Surinam, but Not Recorded at Paracou ...... 232 Appendix 2: Species Matrix for 12 Nonvolant Rainforest Mammal Inventories ...... 234

2 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 3

ABSTRACT

This report describes the results of nonvolant mammal inventory ®eldwork at Paracou, a lowland rainforest locality in northern French Guiana, and concludes the faunal analysis in- troduced by our previous monograph on the bats of Paracou (Simmons and Voss, 1998). Working within a 3±km radius over the course of 202 sampling dates from 1991 to 1994, we recorded a total of 64 nonvolant species by conventional trapping, arboreal platform trapping, pitfall trapping, diurnal and nocturnal hunting, and interviews with local residents. Included in this total species count are 12 marsupials, 9 xenarthrans, 6 primates, 10 carnivores, 5 ungulates, and 22 rodents. Systematic research with nonvolant mammal specimens collected as voucher material resulted in the discovery of new taxa, documented range extensions of previously described species, and helped resolve many longstanding taxonomic problems: (1) Gracilinanus emiliae (Thomas), herein reported for the ®rst time from French Guiana, is redescribed and its known geographic distribution documented; based on examination of type material and original descriptions, G. longicaudus Hershkovitz is considered a junior synonym of G. emiliae, but Marmosa agricolai Moojen is not. (2) A new genus is proposed for Gracilinanus kalinowskii Hershkovitz, a taxon previously known only from eastern Peru, in recognition of its trenchant morphological differences from all other known didelphimorph marsupials. (3) Marmosops parvidens (Tate) and M. pinheiroi (Pine), the latter originally described as a subspecies of the former, are distinct species that occur sympatrically at Paracou; based on examination of type material, other taxa hitherto synonymized with M. parvidens are also judged to be valid species, including M. juninensis (Tate) and M. bishopi (Pine). (4) Monodelphis brevicaudata (Erxleben), M. glirina (Wagner), and M. palliolata (Osgood) are all distinct species diagnosable by unique combinations of morphological traits; based on examined specimens, M. brevicaudata (with type locality emended herein as Kartabo, Guyana) appears to be endemic to the Guiana subregion of Amazonia and to include both bicolored and tricolored phenotypes; a neotype from Cayenne, French Guiana, is designated to ®x the application of Viverra touan Shaw as the oldest available name for the tricolored form. (5) Saguinus midas (Linnaeus) and S. niger (E. Geoffroy), currently treated as synonyms or conspeci®c races, are unambiguously diagnosable species that do not appear to be sister taxa; a neotype is designated to conserve current usage of niger E. Geoffroy for the black-handed tamarin of southeastern Amazonia. (6) Two new small species of Neacomys are described from material collected at Paracou; their diagnostic attributes are documented by detailed comparisons with other like-sized congeners from northern South America. (7) Nectomys melanius Thomas is recognized as a species distinct from N. squamipes (Brants) and N. palmipes J. A. Allen and Chapman; however, N. parvipes Petter is not a valid taxon and is herein synonymized with N. melanius. (8) The diagnostic characters of Neusticomys oyapocki (Petter and Dubost), a species previously known only from the holotype, are reevaluated and illustrated from freshly collected material. (9) Oecomys auyantepui Tate and O. paricola (Thomas), previously treated as synonyms, are valid species distinguished by consistent cranial differences and occupy allopatric ranges north and south of the Amazon, respectively. (10) A critical examination of small Oecomys specimens from Paracou and other Guianan localities supports the conclusions of other investigators that O. rutilus Anthony and O. bicolor (Tomes) are unambiguously diagnosable species. (11) Olig- oryzomys fulvescens (Saussure) and O. microtis (J. A. Allen), currently regarded as valid allopatric species occurring north and south of the Amazon, respectively, are dif®cult to diagnose unambiguously and may be conspeci®c; new information is provided about the hitherto ambiguous type locality of the latter taxon. (12) Rhipidomys nitela Thomas is reported from French Guiana for the ®rst time and its previously unpublished diagnostic differences from other congeners are tabulated and discussed. (13) A lectotype is designated for Coendou melanurus (Wagner), and the species is redescribed based on all known specimens in North American and European museums; diagnostic differences between this species and C. insidiosus (Olfers) are illustrated for the ®rst time. (14) A red-rumped agouti (Dasyprocta) is designated as the neotype of Mus aguti Linnaeus to preserve current usage of Dasyprocta prymnolopha (Wagler) for the black-rumped agouti. (15) The diagnostic differences between red and green acouchies (Myoprocta) are discussed and a neotype is designated for Cavia acouchy Erxleben to ®x the application of that name to the red species; other nominal taxa of Myoprocta are identi®ed as red or green acouchies based on examination of type material and original 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

descriptions. (16) The diagnostic morphological traits of Proechimys cuvieri Petter and P. guyannensis (E. Geoffroy) are reevaluated and discussed based on character variation in topotypical (French Guianan) material. Analyses of our sampling results indicate that distinct sets of nonvolant species are effectively sampled by different inventory methods, and that increased sampling effort with any method generally results in more species. Although the rate of discovery of new species always decreases with increasing sample size, none of our graphs of species accumulation indicate that an asymptotic value was reached with any method. Instead, nonparametric statistical extrapolations suggest that the Paracou nonvolant mammal fauna consists of somewhere between 69 and 74 species; by implication, our nonvolant inventory is about 86±93% complete. Most missing species are probably marsupials and rodents, but one or two expected primate species might have been locally extirpated by hunters prior to our ®eldwork. In terms of higher taxonomic composition, the Paracou nonvolant mammal fauna is typical of those found throughout the humid Neotropical lowlands. However, a quantitative analysis of nonvolant faunal similarity at the species level among 12 exemplar rainforest inventories ®rst clusters the Paracou list with others from the Guiana subregion of Amazonia, next with lists from elsewhere in Amazonia, and lastly with Central American lists. Pairwise similarity values likewise show an obvious positive correlation between faunal resemblance and geographic proximity within the Neotropical rainforest biome. At least 24 species (38%) of the Paracou nonvolant fauna are Amazonian endemics, but 18 (28%) are essentially pan-Neotrop- ical in distribution; the remaining 22 species exhibit a variety of distributional patterns that suggest past connections among different sets of currently disjunct rainforested regions. Species richness comparisons among nonvolant faunal inventories are complicated by a variety of familiar problems including inconsistent methodology, presence or absence of certain key habitats, and uneven sampling effort. A conservative interpretation of sampling results from La Selva (Costa Rica), Paracou, and Manu (Peru), however, suggests progressive increases in richness of about 23% from Central America to the Guianas, and of about the same amount from the Guianas to western Amazonia; over the entire gradient (Central America to western Amazonia), the net increase in observed richness is at least 50%. Whereas rodents are consistently the most diverse clade in all well-sampled nonvolant faunas, rankings of other orders by relative richness exhibit considerable site-to-site variation, at least some of which appears to re¯ect real geographic differences in taxonomic diversity rather than sampling artifacts. Nonvolant rainforest mammals are hard to classify into trophic guilds due to behavioral plasticity and incomplete knowledge of relevant natural history. Preliminary guild comparisons among three exemplar faunas, however, suggest that the Paracou nonvolant community is substantially less diverse in arboreal frugivores and more diverse in terrestrial animalivores than are nonvolant communities at some Central American and western Amazonian sites. Subsistence and recreational hunting has clearly affected local populations of some nonvolant mammals at Paracou; whereas popular game species (e.g., large primates) were seldom sighted, density compensation may explain high local densities of certain other taxa (e.g., Potus ¯avus and Cuniculus paca). Patterns of differential habitat use between closely related nonvolant species at Paracou were mostly observed within the terrestrial granivore/frugivore guild. Combining these results with those previously reported for the sympatric bat fauna, we recorded a total of 142 mammalian species at Paracou. By statistical extrapolation from our sampling data, the entire local community perhaps contains 155±168 species; because the known French Guianan rainforest mammal fauna contains at least 167 species for which suitable habitat is present in our study area, such estimates are plausible. By implication, our inventory is perhaps 85±92% complete overall. A synthesis of biogeographic information analyzed in this monograph and by Simmons and Voss (1998) suggests that faunal turnover with increasing geographic distance is much higher for nonvolant mammals than for bats, a necessary consequence of observed group differences in endemicity: whereas many nonvolant rainforest mammals have geographic ranges bounded by obvious topographic or habitat discontinuities (e.g., large rivers, xeromorphic vegetation), most rainforest bats are geographically widespread. Not surprisingly, most of the taxa that usefully de®ne a Guianan center of mammalian endemism are nonvolant species. The geographic limits of Guianan endemism appear to be remarkably similar for mammals, birds, snakes, lizards, and trees, suggesting a common pattern of biotic differentiation. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 5

Overall, the Paracou mammal fauna conforms broadly with previous generalizations about community-wide patterns of diel activity and substrate use by Neotropical rainforest mammals, but appears to diverge signi®cantly from conventional views about trophic structure. Whereas there are many more species of secondary consumers than primary consumers at Paracou, primary consumers appear to outnumber secondary consumers by an equally large margin at some western Amazonian inventory sites. Sampling artifacts perhaps explain some of the community differences observed in such comparisons, but real geographic variation in trophic structure is also apparent.

REÂ SUMEÂ Ce travail deÂcrit les reÂsultats d'un inventaire de la communauteÂ des mammifeÁres non-volants aÁ Paracou, dans un site de foreÃts pluvieuses de basse altitude au Nord de la Guyane francËaise. Cette eÂtude compleÁte et conclut l'analyse faunique de cette localiteÂ, entameÂe par la monographie des chauves-souris de Paracou (Simmons et Voss, 1998). En nous limitant aÁ une surface de 3 km de rayon, eÂtudieÂe durant 202 jours s'eÂtalant de 1991 aÁ 1994, nous avons mis en eÂvidence la preÂsence de 64 mammifeÁres non-volants par la combinaison de diffeÂrentes meÂth- odes d'echantillonnage: pieÂgeage traditionnel, pieÂgeage en canopeÂe, pieÂgeage par seaux en- terreÂs, chasses diurne et nocturne, et entretiens avec des reÂsidents locaux. Cette communauteÂ comprend 12 espeÁces de marsupiaux, 9 xeÂnarthres (eÂdenteÂs), 6 primates, 10 carnivores, 5 onguleÂs, et 22 rongeurs. Les recherches de systeÂmatique compareÂe effectueÂes sur les animaux collecteÂs et preÂserveÂs sous forme de speÂcimens museÂologiques ont permis la deÂcouverte de nouveaux taxons, ont documenteÂ l'extension de la reÂpartition geÂographique d'espeÁces connues par ailleurs, et ont permis d'aborder et de reÂsoudre plusieurs questions taxonomiques longtemps deÂbattues. (1) Le petit opossum Gracilinanus emiliae (Thomas) est pour la premieÁre fois documenteÂ en Guyane; la description de cette espeÁce est eÂlaboreÂe, et sa distribution geÂographique connue est documenteÂe. Par comparaison avec la seÂrie type et les descriptions originelles, G. longicaudus Hershkovitz est consideÂreÂ comme un synonyme posteÂrieur aÁ G. emiliae, mais Marmosa agricolai est une espeÁce diffeÂrente. (2) Un nouveau genre est proposeÂ pour le taxon Gracili- nanus kalinowskii Hershkovitz, qui n'eÂtait auparavant connu que de l'Est du PeÂrou, mais qui diffeÁre de tous les autres marsupiaux didelphimorphes par de singuliers caracteÁres morphologiques. (3) Marmosops parvidens (Tate) et M. pinheiroi (Pine), ce dernier originellement deÂcrit comme une sous-espeÁce du premier, sont deux taxons distincts, qui vivent en sympatrie aÁ Paracou. ApreÁs examen des speÂcimens de reÂfeÂrence (holotype et paratypes), d'autres taxons preÂceÂdemment mis en synonymie avec M. parvidens apparaissent neÂammoins eÃtre des espeÁces distinctes, comme M. juninensis (Tate) et M. bishopi (Pine). (4) Monodelphis brevicaudata (Erxleben), M. glirina (Wagner), et M. palliolata (Osgood) sont trois espeÁces distinctes, reconnaissables par des combinaisons discriminantes de caracteÁres morphologiques. A partir des speÂcimens aÁ disposition, l'espeÁce M. brevicaudata (dont la localiteÂ-type est redeÂ®nie comme Kartabo, au Guyana) apparaõÃt endeÂmique du bouclier guyanais au sein de la reÂgion amazon- ienne, et comporte des formes bi- et tri-coloreÂes. Un neÂotype de Cayenne (Guyane francËaise) est deÂsigneÂ pour repreÂsenter le taxon Viverra touan Shaw, qui est le plus ancien nom disponible pour la forme tricolore. (5) Saguinus midas (Linnaeus) et S. niger (E. Geoffroy), tradition- nellement consideÂreÂs comme synonymes ou comme races conspeÂci®ques, sont reconnaissables de facËon indiscutable, et ne semblent pas eÃtre des espeÁces-soeurs; en conseÂquence de notre analyse, chacun doit eÃtre reconnu comme une espeÁce distincte; un neÂotype est deÂsigneÂ a®n de conserver l'usage reÂpandu du taxon niger E. Geoffroy repreÂsentant les tamarins aux mains noires du sud-est de l'Amazonie. (6) Deux nouvelles petites espeÁces du genre Neacomys sont deÂcrites aÁ partir du mateÂriel collecteÂaÁ Paracou et dans d'autres localiteÂs de la reÂgion des Guyanes; leurs caracteÁres diagnostiques sont documenteÂs par des comparaisons deÂtailleÂes avec d'autres congeÂneÁres de taille similaire vivant en reÂgions septentrionales d'AmeÂrique du Sud. (7) Nectomys melanius Thomas est reconnu comme une espeÁce distincte des taxons N. squamipes (Brants) et N. palmipes J. A. Allen et Chapman. Cependant, N. parvipes Petter n'est pas un taxon valide, et est mis ici en synonymie avec N. melanius. (8) Les caracteÁres diagnostiques de Neusticomys oyapocki (Petter et Dubost), une espeÁce jusqu'ici connue par le seul holotype, sont discuteÂs et illustreÂs aÁ partir de nouveaux speÂcimens reÂcemment collecteÂs. (9) Oecomys 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

auyantepui Tate et O. paricola (Thomas), auparavant mis en synonymie, sont diffeÂrentes es- peÁces reconnaissables par des caracteÁres craniens discriminants; ces taxons ont des reÂpartitions allopatriques, respectivement au Nord et au Sud de l'Amazone. (10) Un examen deÂtailleÂ des petits speÂcimens d'Oecomys provenant de Paracou et d'autres localiteÂs de la reÂgion des Guy- anes con®rme l'interpreÂtation d'autres auteurs comme quoi O. rutilus Anthony et O. bicolor (Tomes) sont des espeÁces distinctes, discreÁtement mais clairement reconnaissables. (11) La diagnose et la distinction d'Oligoryzomys fulvescens (Saussure) et O. microtis (J. A. Allen), actuellement consideÂreÂes comme espeÁces allopatriques distribueÂes respectivement au Nord et au Sud de l'Amazone, sont dif®ciles aÁeÂtablir, et ces deux taxons pourraient eÃtre conspeÂci®ques; de nouvelles donneÂes sont apporteÂes au sujet de la localiteÂ-type d'O. microtis, dont l'emplacement eÂtait jusqu'alors ambigu. (12) Rhipidomys nitela Thomas est signaleÂ pour la premieÁre fois de Guyane francËaise, et les diffeÂrences morphologiques diagnostiques de cette espeÁce sont preÂsenteÂes et discuteÂes par rapport aux autres espeÁces du genre. (13) Un lectotype est deÂsigneÂ pour l'espeÁce Coendou melanurus (Wagner), et la description de ce taxon est deÂtailleÂe aÁ partir des speÂcimens connus dans les museÂes nord-ameÂricains et europeÂens. Des comparaisons et diagnoses par rapport aÁ C. insidiosus (Olfers) sont effectueÂes pour la premieÁre fois. (14) Un speÂcimen d'agouti aÁ croupion rouge (genre Dasyprocta) est deÂsigneÂ comme neÂotype du taxon Mus aguti Linnaeus, a®n de preÂserver l'usage du nom Dasyprocta prymnolopha (Wagler) pour les agoutis aÁ croupion noir. (15) Les diffeÂrences caracteÂristiques des acouchis rouge et olive du genre Myoprocta sont discuteÂes; un speÂcimen collecteÂaÁ Paracou est deÂsigneÂ comme le neÂotype de Cavia acouchy Erxleben, a®n de pouvoir appliquer ce nom aÁ l'acouchi rouge. D'autres taxons du genre Myoprocta sont identi®eÂs comme appartenant aux groupes des acouchis rouges ou olives, aÁ partir du mateÂriel des localiteÂs-type et des descriptions originelles. (16) Les caracteÁres morphologiques permettant de discriminer les espeÁces Proe- chimys cuvieri Petter et P. guyannensis (E. Geoffroy) sont reÂ-eÂvalueÂs et discuteÂs aÁ partir de l'examen de la variation de ceux-ci parmi des speÂcimens topotypiques de Guyane francËaise. Les analyses qualitative et quantitative de nos efforts de collecte indiquent que diffeÂrents ensembles de mammifeÁres non-volants sont mis en eÂvidence par les diffeÂrentes techniques d'observations et de captures. En geÂneÂral, un effort d'eÂchantillonnage accru conduit aÁdeÂtecter un plus grand nombre d'espeÁces, quelle que soit la technique consideÂreÂe. Bien que le taux de deÂcouverte de nouvelles espeÁces deÂcroisse au fur et aÁ mesure des efforts cumuleÂs, l'examen des courbes d'accumulation de taxons indique qu'aucune de nos meÂthodes n'a permis d'atteindre une valeur d'asymptote. Plus preÂciseÂment, des extrapolations utilisant des statistiques non-parameÂtriques suggeÁrent que la communauteÂ des mammifeÁres non-volants de Par- acou renferme de 69 aÁ 74 espeÁces. Cette preÂdiction implique que notre inventaire ne repreÂ- senterait que 86 aÁ 93% des espeÁces potentiellement preÂsentes. La plupart des espeÁces non- deÂtecteÂes sont probablement des marsupiaux et des rongeurs, mais l'un ou l'autre primates absents de nos inventaires ont peut-eÃtre eÂteÂreÂcemment eÂlimineÂs de la zone d'eÂtude par une chasse trop excessive. En ce qui concerne la composition taxonomique des familles et des ordres, la faune des mammifeÁres non-volants de Paracou est bien repreÂsentative de celles connues dans d'autres zones de plaine (foreÃts ombrophiles) des NeÂotropiques. Cependant, une analyse quantitative, au niveau des espeÁces partageÂes, des similariteÂs entre 12 localiteÂs forestieÁres neÂotropicales ouÁ des inventaires comparables ont eÂteÂ effectueÂs, indique que la faune de Paracou se regroupe avec d'autres inventaires de la reÂgion des Guyanes, puis avec des listes fauniques d'autres reÂgions amazoniennes, et en®n avec des peuplements d'AmeÂrique centrale. Ainsi, les indices de similariteÂ d'une matrice d'espeÁces partageÂes montrent une correÂlation positive entre les similariteÂs de faune et la proximiteÂgeÂographique au sein du biome que sont les foreÃts humides neÂotropicales. Au moins 24 espeÁces (soit 38% de notre inventaire) de la faune (aÁ l'exclusion des chiropteÁres) de Paracou sont des endeÂmiques de l'Amazonie, mais 18 (28%) ont une large reÂpartition pan-neÂotropicale; les 22 espeÁces restantes preÂsentent diffeÂrents patrons de distribution qui suggeÁrent d'anciennes connections entre les ensembles forestiers sud-ameÂricains actuellement discontinus. Toutefois, il faut admettre que les comparaisons, en terme de richesses d'espeÁces, entre les diffeÂrents inventaires de mammifeÁres non-volants sont biaiseÂes par divers probleÁme inheÂrents aÁ ces eÂtudes: meÂthodologies non strictement comparables, absence ou preÂsence de certains habitats singuliers, et ineÂgaliteÂs des efforts de collecte. NeÂammoins, une interpreÂtation prudente des donneÂes d'inventaires de trois localiteÂs, La Selva (Costa Rica), Paracou et Manu (PeÂrou), 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 7

suggeÁre un accroissement progressif de la richesse en espeÁces d'environ 23% entre l'AmeÂrique Centrale et les Guyanes, et d'environ la meÃme ampleur entre les Guyanes et l'Amazonie occidentale. Sur l'ensemble du gradient ``AmeÂrique centrale → Guyanes → Amazonie occidentale'', l'augmentation des biodiversiteÂs observeÂes est d'au moins 50%. Alors que les rongeurs repreÂsentent le taxon le plus diversi®eÂ dans toutes les eÂtudes baseÂes sur des inventaires approfondis, l'abondance relative des autres ordres de mammifeÁres non-volants diffeÁre pro- fondeÂment selon les sites examineÂs. Dans certains cas, l'abondance ou la rareteÂ d'un ordre particulier semble re¯eÂter de reÂelles diffeÂrences geÂographiques au travers des foreÃts amazoniennes, et non pas des artefacts d'eÂchantillonnages. Les mammifeÁres non-volants des foreÃts humides sont dif®cilement classables en guildes trophiques du fait d'une part de leur plasticiteÂ comportementale et d'autre part du manque de connaissances quant aÁ leurs traits d'histoire de vie. Nous preÂsentons neÂammoins des comparaisons preÂliminaires, baseÂes sur trois faunes particulieÁrement bien documenteÂes, qui suggeÁrent que la communauteÂ de Paracou est moins riche en frugivores arboricoles et plus diverse en animalivores terrestres par rapport aux communauteÂs d'AmeÂrique centrale et d'Amazonie occidentale. La chasse, qu'elle soit de subsistance ou de loisirs, a clairement modi®eÂ la com- munauteÂ des mammifeÁres de Paracou, eÂradiquant les populations locales de certaines espeÁces. Ainsi, les espeÁces-gibiers les plus rechercheÂes (comme les grands primates) ont eÂteÂ rarement observeÂes, alors que des pheÂnomeÁnes de compensation pourraient expliquer les densiteÂs re- lativement eÂleveÂes d'autres taxons (par exemple, Potos ¯avus et Cuniculus paca). En®n, nos observations suggeÁrent un usage diffeÂrent des habitats par des espeÁces apparenteÂes, ce qui a eÂteÂ noteÂaÁ Paracou principalement au sein de la guilde des mammifeÁres terrestres granivores- frugivores. En reÂunissant ces reÂsultats avec ceux acquis preÂceÂdemment pour la faune des chauves-souris de Paracou, nous avons donc mis en eÂvidence l'existence de 142 espeÁces dans cette localiteÂ de Guyane francËaise. Par des extrapolations statistiques de nos donneÂes d'eÂchantillonnage, nous estimons que la communauteÂ locale renferme probablement de 155 aÁ 168 espeÁces. Com- me la faune des mammifeÁres forestiers de Guyane renferme au moins 167 espeÁces dont l'habitat, tel que nous le connaissons, est repreÂsenteÂaÁ Paracou, notre estimation semble reÂaliste. Ces consideÂrations impliquent que nos inventaires, concernant tant les chauves-souris que les mammifeÁres non-volants, auraient mis en eÂvidence de 85 aÁ 92% des espeÁces preÂsentes. Une syntheÁse des informations biogeÂographiques analyseÂes dans ce travail et dans celui de Simmons et Voss (1998) suggeÁre alors que les changements de faune (renouvellement faunique, ou ``faunal turnover''), en rapport avec l'eÂloignement geÂographique, sont beaucoup plus marqueÂs pour les mammifeÁres non-volants que pour les chauves-souris, une conseÂquence lieÂe aux diffeÂrences d'endeÂmiciteÂ caracteÂrisant ces deux groupes d'animaux. En effet, tandis que de nombreux mammifeÁres forestiers non-volants ont des reÂpartitions geÂographiques circon- scrites par des discontinuiteÂs topographiques ou eÂcologiques (grands ¯euves, veÂgeÂtation xeÂ- romorphe), la plupart des chauves-souris des foreÃts neÂotropicales humides sont largement reÂ- pandues dans l'espace. Il s'ensuit que la plupart des taxons qui pourraient deÂ®nir de facËon claire un endeÂmisme au niveau des Guyanes sont des mammifeÁres non-volants. Nous relevons aussi que les limites geÂographiques de l'endeÂmisme des Guyanes apparaissent treÁs semblables pour divers organismes, mammifeÁres, oiseaux, serpents, leÂzards, et arbres, suggeÂrant un patron commun de diffeÂrentiation pour ce biome. En conclusion, la faune des mammifeÁres de Paracou re¯eÁte bien, en tant qu'exemple concret, les enseignements des ouvrages geÂneÂraux en ce qui concerne les patrons d'activiteÂ journalieÁre et d'utilisation des substrats et des habitats par les mammifeÁres forestiers neÂotropicaux, mais une diffeÂrence d'importance, par rapport aÁ la litteÂrature, concerne la structure trophique. Alors qu'il y a beaucoup plus d'espeÁces de consommateurs secondaires que primaires aÁ Paracou, les consommateurs primaires sont au contraire plus nombreux, par un nombre d'espeÁces tout aussi important, dans certains sites eÂtudieÂs en Amazonie occidentale. Bien qu'il soit possible que des biais d'eÂchantillonnage soient responsables de certaines diffeÂrences observeÂes entre ces communauteÂs, il n'en reste pas moins qu'une importante variation geÂographique existe quant aÁ la structure et aÁ la composition des niches trophiques des mammifeÁres habitant les foreÃts neÂotropicales humides. 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

INTRODUCTION to randomize or standardize the placement of traps. Instead, traps were placed to maximize This report is the second and concluding capture success for the widest range of an- part of our monographic treatment of the ticipated species as suggested by prior ex- mammal fauna of Paracou, a rainforested lo- perience in other habitats and at other rain- cality in the coastal lowlands of northern forested localities. Victor rat traps were French Guiana (®g. 1). In the ®rst part (Sim- sometimes set on the ground (often inside the mons and Voss, 1998) we outlined the over- dark cavities of hollow logs), but we usually all goals of our inventory project, described tied them to lianas, tree trunks, and other our study site, explained the methods we woody supports 0±2 m above the ground to used to sample local bat diversity, summa- capture semiarboreal species (®g. 3, top). rized our systematic research with collected Sherman traps were usually placed on the bat voucher material, and analyzed our bat- ground next to fallen trunks, under large- sampling results. Herein we provide compa- leaved vegetation (palms, heliconias, etc.), rable information about the nonvolant spe- beneath viny tangles and piled branches, or cies, provide a synthetic overview of Paracou in similarly sheltered terrestrial situations mammal diversity, compare our whole-fau- (®g. 3, bottom). nal sampling results with those from other Victor rat traps were usually baited with a inventory sites, and evaluate the prospects ground-up mixture of peanut butter, raisins, and priorities for future mammalian diversity rolled oats, and bacon (in 6:2:2:1 propor- research in the rainforested Neotropical low- tion); although this bait effectively attracts lands. many species of small mammals, it also attracts ants and must be renewed daily in most NONVOLANT MAMMAL INVENTORY situations. Sherman live traps were baited METHODS daily with either cracked corn (sold locally Nonvolant mammals are so variable in as chicken feed) or with commercial bird- size and behavioral traits that capture equip- seed; both baits were effective when dry, but ment or observational methods suitable for cracked corn becomes sticky when wet and some taxa may be entirely ineffective for subsequently spoils unless the trap is washed. others (Voss and Emmons, 1996). Addition- Traplines were checked twice daily, usually ally, whereas some taxa or ecological guilds at dawn and in the late afternoon (between are minimally affected by human activities, 16:00 and 18:30 hours). All traps were re- others can be extirpated by overhunting even baited in the late afternoon. Sherman traps in extensive tracts of uncut forest. To an even were disassembled, thoroughly washed greater extent than in our bat survey (Sim- (without soap), and dried after each capture mons and Voss, 1998), we therefore relied on to remove urine, feces, and old bait. The trea- a combination of methods to inventory non- dles of Victor traps were scraped clean of old volant mammal diversity at Paracou. bait at each rebaiting, but the traps themselves were not washed. We marked all trap locations with brightly colored vinyl survey- CONVENTIONAL TRAPPING or's tape, and all Victor traps were tethered We used standard trapping equipment to nearby stems or roots with a short (ca. 80 (Voss and Emmons, 1996) to sample the lo- cm) length of braided cotton or nylon cord. cal fauna of small marsupials and rodents Most trapped trails traversed both well- near ground level during our 1991 and 1992 drained and swampy primary forest, the two ®eld seasons. Most traplines included both principal habitat types sampled by this meth- Victor rat traps (with push-down wooden bait od. In addition, a few traplines were set for pedals) and folding aluminum Sherman live semiaquatic species by wading small streams traps (measuring 80 ϫ 90 ϫ 230 mm) set at and setting traps along the banks, and some approximately 20-m intervals along existing traps were set in the secondary vegetation trails through our study area (®g. 2). Because bordering our camp clearing and nearby our objective was to sample the fauna as ef- roads. We did not trap in the coastal savannas fectively as possible, no attempt was made north of our study area. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 9

Fig. 1. Map of French Guiana showing the location of our study site at Paracou in relation to other places mentioned in the text. Localities that we could not associate with unique geographic coordinates (e.g., ``RivieÁre Approuague'') are not shown. One degree of latitude or longitude is approximately equal to 110 km. 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 2. Trail through well-drained primary forest sampled by trapping and hunting at Paracou. Most traps were set within a few meters of such narrow paths, which do not materially alter the habitats they traverse. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 11

Fig. 3. Two standard traps as typically deployed to sample nonvolant mammal diversity near ground level. Top, Victor rat trap tied to liana. Bottom, Sherman live trap on the ground. Sixteen species of small marsupials and rodents were taken in mixed Victor/Sherman traplines at Paracou. 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

We periodically used Tomahawk folding a pitfall trapping design suggested by S. M. wire live traps (measuring 145 ϫ 145 ϫ 410 Goodman (personal commun.) to supplement mm) to capture large (Ͼ300 g) terrestrial our conventional trapping in 1993. We used marsupials and rodents (®g. 4). Most Toma- 15-liter plastic buckets as pitfalls, and sunk hawk traps set on the ground were baited these ¯ush with the ground in linear series with coconut or ripe plantain, and a few un- beneath sheet-plastic drift fences (®g. 6). baited large Tomahawks (measuring 250 ϫ Each pitfall trapline consisted of 11 buckets 300 ϫ 810 mm) were set in streams for semi- spaced 5 m apart, with one bucket at either aquatic species. We used small leghold traps end, for a total length of 50 m. Drift fences, and Conibear break-back traps with and consisting of a continuous barrier running the without bait or commercial scent lures in entire length of each trapline, were made of 1991, but not in subsequent ®eld seasons; 50-cm-wide strips of heavyweight (6 mil) most of these traps were set away from es- clear polyethylene stapled to vertical stakes tablished trails, often in burrows or on logs hammered into the ground every 3±4 m. A crossing small streams. In 1992 we also con- broad (7±8 cm) fold of plastic was heaped structed and ®eld-tested two live traps de- with soil to anchor the bottom of the fence, signed for use in armadillo tunnels, but these and a ¯ap was cut in this fold where it passed were unsuccessful and the effort was discon- over each bucket. tinued in subsequent years. We perforated the bottoms of the buckets so that rainwater could drain out, but the per- ARBOREAL TRAPPING forations soon became clogged and we then Because some species of rainforest mar- noticed that undrained buckets containing supials and rodents seldom descend from the 10±15 cm of water were much more effec- canopy or subcanopy, we used a platform- tive traps than dry buckets. Thereafter, we trapping system similar to that described by added water to all of the buckets and tried to Malcolm (1991) to capture arboreal species maintain this level throughout our pitfall- in 1993. Trees were climbed using French trapping effort. Most captured animals pole climbers (see Mori, 1987: ®g. I-3A), of drowned in pitfalls containing water, but which a small pair with a maximum tooth- checking the traplines twice daily was suf®- to-tooth span of 25 cm and a larger pair span- cient to prevent specimen damage due to de- ning 35 cm were useful for climbing boles composition. of different diameters. Wooden platforms provided with eyelets for raising and lower- HUNTING AND SIGHT SURVEYS ing trap assemblies by nylon lines were nailed to tree trunks 7±19 m above the We relied on hunting, sight surveys, and ground in both well-drained and swampy pri- interviews (see below) to census the local mary forest. Each trap assembly consisted of fauna of large mammals (Ն1 kg) that were one Tomahawk folding wire live trap (mea- dif®cult or impossible to capture with our suring 145 ϫ 145 ϫ 410 mm) and one piggy- trapping equipment. However, we seldom backed Sherman folding aluminum live trap hunted or made sight surveys in the daytime (measuring 80 ϫ 90 ϫ 230 mm) attached to because mornings were occupied by check- a wooden frame (®g. 5). Sherman traps were ing traplines, taking down bat nets, or pre- baited with the same oatmeal/peanut/raisin/ paring specimens, and afternoons were usu- bacon mixture described previously, and the ally spent rebaiting traps or setting up bat Tomahawk traps were baited with pieces of nets. Most of our personal observations of coconut and ripe plantain; bait was renewed diurnal mammals at Paracou were therefore as necessary, about every three to four days made haphazardly in the course of other (rel- on average. Tree traps were checked daily atively noisy) activities. Fortunately, local soon after dawn with binoculars. forestry workers were familiar with most species of large diurnal mammals known or PITFALL TRAPPING expected to occur in the area, so interviews Many species of small terrestrial mammals compensated to some extent for the de®cien- are not attracted to baited traps, so we used cies of our diurnal census effort. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 13

Fig. 4. Tomahawk folding wire live trap as typically deployed to supplement Victor- and Sherman- trapping near ground level.

By contrast, we regularly hunted at night, loaded in removable auxilliary barrels (``aux- and we recorded many sightings of nocturnal es'') machined from brass rod stock. Most species encountered by deliberate searching. unvouchered sightings of small marsupials Equipped with notebook, shotgun, and a and rodents could not be con®dently identi- nine-volt headlight with focusable re¯ector, ®ed to species. we walked slowly (about 0.5±1.5 km/hour) along established trails scanning the vegeta- INTERVIEWS tion with the headlight re¯ector adjusted to produce a moderately broad beam. When Small teams of day laborers have been eyeshine was detected, we narrowed the continuously employed in forestry research beam to give brighter illumination; binocu- projects at Paracou since the mid-1980s to lars were sometimes used to identify animals cut trails, clear experimental plots, and to revealed by eyeshine in the canopy or sub- mark and measure tens of thousands of trees. canopy. Most of our nocturnal surveys were Among these workers are members of the made in the early evening (between 19:00 Saramaka and Boni tribes of so-called bush and 24:00 hours), but we also hunted be- negroes, descendants of escaped slaves (ma- tween midnight and dawn on many nights. roons) who ¯ed into the forests of Surinam Some nonvolant species were observed in and French Guiana and re-created traditional the course of bat netting, especially large African communities that have resisted ac- marsupials attracted to the squeals of cap- culturation for centuries (Kahn, 1931; Hu- tured bats. rault, 1961; Price, 1976). Recruited for their Specimens were collected by shooting ability to identify commercially valuable tim- with side-by-side 20-gauge shotguns. We ber, these people have observed (and hunted) used commercial 20-gauge game loads for the Paracou fauna for many years. We peri- large mammals, but small mammals were odically discussed our inventory work with collected with .410 or .22 caliber shot-shells them, comparing our own observations with 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 5. Arboreal platform trap assembly in place at Paracou. Modi®ed from Malcolm's (1991) basic design, each trap assembly consisted of one folding wire Tomahawk trap and a piggy-backed folding aluminum Sherman trap. Both are visible in this photograph, which resembles the perspective of an observer on the ground checking the traps with binoculars. Six species of marsupials and rodents were taken with this equipment, at heights ranging from 7.2 to 16.7 m above the ground. theirs and asking them about species we had asked for details about where and when the not yet seen. Most of our useful interview sighting(s) occurred, and about distinguish- data, however, was obtained from conversa- ing morphological or behavioral characteris- tions with their French-born foreman, Pascal tics; we often used the illustrations in Em- Petronelli, himself a gifted naturalist and a mons (1990) to discuss diagnostic differenc- local resident since 1983. es among related species. For some species, Using distributional data summarized for we were able to examine photographs and/or another study (Voss and Emmons, 1996), we written records (logs of noteworthy sightings compiled a list of the nonvolant mammal on experimental plots). Finally, we asked Mr. species known or expected to occur in Petronelli if any species not mentioned by us French Guiana, and we used this list as the had ever been seen in the area. basis for interviewing Mr. Petronelli in 1993. We focused primarily on large species (Ն1 PREVIOUS MAMMALOGICAL RESEARCH AT kg) that nonmammalogists might reasonably PARACOU be expected to recognize without special effort, but we also asked about a few smaller Whereas information about bat diversity at species identi®able by obvious external char- Paracou (summarized by Simmons and Voss, acters. For each species expected to occur in 1998) derived entirely from our own efforts, the area but not observed by us, we asked if other researchers had previously worked on he had seen it himself, or whether the species the local nonvolant mammal fauna prior to had been seen by the forestry workers he su- our study (Dubost, personal commun.; For- pervised. If the species had been seen, we get, 1991, 1996, 1997; Henry, 1994, 1996, 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 15

Fig. 6. Detail of a pitfall trapline installed beneath a sheet-plastic drift fence at Paracou. Loose soil (excavated to sink the pitfalls) is heaped along a broad bottom fold of plastic to anchor the base of the drift fence; the remaining plastic is stapled to upright stakes for support. Twelve species of small marsupials and rodents were taken by this method at Paracou, in addition to amphibians, squamate reptiles, and arthropods. 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

1999; Forget et al., 1999). Their published SYSTEMATIC ACCOUNTS results and unpublished observations provided supplementary information about local di- A signi®cant difference in the scope of versity, including records of three species taxonomic problems encountered in working that neither we nor our resident interviewees up the bats and the nonvolant mammals from had identi®ed. Of particular importance was Paracou accounts for certain format differ- the very large series of specimens trapped ences between the systematic accounts below over several years by O. Henry, which we and those in Simmons and Voss (1998). were generously allowed to inspect (through Whereas most of the bat genera represented the good of®ces of L. Granjon) for taxa unin the Paracou fauna have been revised tax- represented in our voucher collection. onomically in the last half-century, very few revisions are available to facilitate the iden- VOUCHER PRESERVATION ti®cation of nonvolant taxa. Furthermore, many of the unrevised nonvolant taxa in the We preserved voucher material of nonvo- Paracou fauna were ®rst described by Lin- lant species following standard protocols naeus and other early zoologists, who often (e.g., Nagorsen and Peterson, 1980; Hall, named species based on subsequently mis- 1981). As explained previously (Simmons placed specimens, or on the unvouchered de- and Voss, 1998: 19±21), our primary moti- scriptions of even earlier travellers. Most of vation in preserving voucher material was to these names have been sources of taxonomic obtain enough material from each species to confusion for centuries, and it was found im- assess the taxonomic status of the Paracou possible to use them unambiguously without population with respect to museum samples resolving dif®cult problems of con¯icting us- from other Neotropical localities. Addition- age. To do so, we examined types and other ally, our voucher collection (now divided be- relevant material from many museums, and tween the AMNH and the MNHN) provides we now designate lectotypes or neotypes as a permanent archive of mammalian diversity necessary to ®x the application of problem- at Paracou that can be consulted by other zo- atic names. For some species that lacked ologists, who may con®rm or emend our published syntheses of geographic data, we identi®cations in the light of future taxonom- tried to examine every known specimen in ic revisions. order to map distributions for this report. Nevertheless, we only collected voucher Because taxonomic problems of varying material for small species (few of which could complexity were encountered in working up otherwise be identi®ed with con®dence) and the nonvolant fauna, the organization of for some large mammals judged to be suf®- these accounts differs from species to spe- ciently abundant that removal would not sig- cies. Nevertheless, we tried to maintain some ni®cantly affect the local population. In gen- consistency in the order in which information eral, samples consisting of about 10 adult is presented. Whether or not subheadings are males and 10 adult females are suf®cient to used, we ®rst list the specimens or observa- obtain meaningful estimates of the mean and tions that provide evidence for the species at range of variation in each sex for characters Paracou. Next, we discuss issues of identi®- of taxonomic interest, and our voucher series cation. For species that can be unambiguous- only exceed these minimal counts for a few ly identi®ed with standard references, only a of the commonest marsupials and rodents brief comment to that effect is necessary, but (e.g., Didelphis marsupialis, Philander opos- many species required more extensive treat- sum, Oryzomys megacephalus, Proechimys cuvieri). For open populations of most small nonvolant species, collecting on this scale is ests provides some basis for inference. His estimates of probably trivial by comparison with natural median density (300 individuals per square kilometer) 1 and annual survivorship (36%) suggest that ecologically demographic processes. similar taxa such as P. cuvieri or P. guyannensis could maintain local populations of about 8400 individuals in 1 Demographic data on small nonvolant tropical mam- our study area (28 km2; Simmons and Voss, 1998), of mals are hard to come by, but Fleming's (1971) study which almost 5400 would be expected to die annually of Proechimys semispinosus in Panamanian lowland for- of natural causes. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 17

ment for the reasons just explained. Field ob- CM Carnegie Museum of Natural History servations are summarized last. (Pittsburgh) Where formal taxonomic treatment is re- EBRG EstacioÂn BioloÂgica de Rancho Grande quired, either for the description of new spe- (Maracay) FMNH Field Museum of Natural History cies or to resolve complex issues of usage, (Chicago) we use as many as six subheadings (Type INPA Instituto Nacional de Pesquisas da Material, Distribution, Description, Compar- AmazoÃnia (Manaus) isons, Remarks, Other Specimens Examined) KU University of Kansas Museum of Nat- to organize relevant information. Wherever ural History (Lawrence) possible, we attempted to sequester infor- MHNG MuseÂum d'Histoire Naturelle de Ge- mation about nomenclature under the head- neÁve (Geneva) MHNLS Museo de Historia Natural La Salle ing ``Remarks''. Under ``Other Specimens (Caracas) Examined'' we list the additional material MUSM Museo de Historia Natural de la (besides Paracou vouchers) on which our Universidad Nacional Mayor de San systematic conclusions are based. Marcos (Lima) All linear measurements cited or tabulated MVZ Museum of Vertebrate Zoology Uni- below are in millimeters (mm). External versity of California (Berkeley) measurements and weight of individual spec- NMW Naturhistorisches Museum Wien (Vi- enna) imens are given in the text according to the RMNH Rijksmuseum van Natuurlijke Histo- formula: Length of Head-and-Body (HBL) ϫ rie (Leiden) Length of Tail (LT) ϫ Length of Hindfoot ROM Royal Ontario Museum (Toronto) (HF) ϫ Length of Ear (Ear), followed by UG University of Guyana (Georgetown) weight (Wt) in grams (g) or kilograms (kg). USNM National Museum of Natural History Length of Head-and-Body was obtained by (Washington, D.C.) subtracting Length of Tail (basal ¯exure to UZM Universitets Zoologiske Museum (Copenhagen) ¯eshy tip) from Total Length (nose to ¯eshy V- Institut des Sciences de l'EÂ volution tail-tip); Length of Hindfoot includes the (Montpellier) claws; and Length of Ear was measured from ZMB Museum fuÈr Naturkunde der Hum- the notch (see Husson, 1978: ®g. 1). Unless boldt-UniversitaÈt zu Berlin (Berlin) otherwise noted, external measurements and weights are those recorded by collectors in MARSUPIALIA the ®eld (or were calculated from the collector's measurements as described above for Twelve species of marsupials, all tradition- Head-and-Body Length). Craniodental mea- ally classi®ed in the family Didelphidae,2 are surements are de®ned elsewhere in the text. known to occur at Paracou, and two addi- We tabulate the sample mean plus or minus tional species could be expected to occur lo- one standard deviation, the observed range, cally (see appendix 1). Our terminology for and the sample size (N) for N Ն10 speci- most qualitative aspects of marsupial mor- mens; only the sample mean, the observed phology follows Archer (1976a, 1976b) and range, and the sample size are tabulated for N Ͻ 10 specimens. 2 An alternative classi®cation proposed by Hershkov- Voucher material from Paracou is depos- itz (1992) would assign the Paracou marsupial genera to several different families: Marmosidae (Gracilinanus, ited in the American Museum of Natural His- Marmosa, Marmosops, Metachirus, Micoureus, Mono- tory, New York (AMNH), and in the MuseÂ- delphis), Caluromyidae (Caluromys), and Didelphidae um National d'Histoire Naturelle, Paris (Chironectes, Didelphis, Philander). Morphological sup- (MNHN). Other specimens cited below are port for caluromyid and marmosid monophyly is weak, however, and available molecular data consistently sup- in museums identi®ed by the following ac- port other generic groupings (Patton et al., 1996; Palma ronyms: and Spotorno, 1999; Jansa and Voss, 2000). In the absence of a well-corroborated phylogenetic classi®cation AC MuseÂum d'Anatomie CompareÂe (Par- of New World marsupials, we continue the traditional is) usage of Didelphidae to include the entire didelphi- BMNH Natural History Museum (London) morph crown group. 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Hershkovitz (1992, 1997), but we adhere to However, all of the ``marmosine'' opossums the traditional interpretation of postcanine (species of Gracilinanus, Marmosa, Mar- dental homologies (Flower, 1867; Luckett, mosops, and Micoureus; formerly placed in 1993), wherein the replaced molariform the genus Marmosa sensu Tate, 1933) require tooth of the upper and lower jaw is the de- specimens in the hand for positive identi®- ciduous third premolar (dP3/dp3). Thus, the cation, and some cannot be con®dently dis- adult didelphid dental formula is I 5/4, C 1/ tinguished without cleaned cranial material. 1, P 3/3, M 4/4). We de®ne specimens to be Husson (1978) provided a useful key to Sur- juvenile if dP3 is still in place, subadult if inamese marsupials based on craniodental dP3 has been shed but P3 and/or M4 are still characters, but four species now known to incompletely erupted, and adult if the per- occur in Surinam and/or French Guiana were manent maxillary dentition is complete. omitted: Didelphis albiventris, ``Gracilinan- Our quantitative comparisons of marsupial us'' kalinowskii, Marmosops parvidens, and crania are based on the following measure- Marmosops pinheiroi. Distinguishing mor- ments (®g. 7): phological characteristics of these recently Condylobasal Length (CBL): From the oc- reported members of the eastern Guianan cipital condyles to the anteriormost point of fauna are described and illustrated in the ac- the premaxillae. counts that follow. Maxillary Toothrow (MTR): Crown length, from the anterior margin of the ca- Caluromys philander (Linnaeus) nine to the posterior margin of M4. Molar Length (LM): Crown length of M1± VOUCHER MATERIAL: AMNH 266402, 4, measured on the labial side of the tooth- 266408, 266409, 267330, 267331, 267333± row. 267337; MNHN 1995.884±1995.886, Palatal Breadth (PB): Measured across 1995.894, 1995.902. Total ϭ 15 specimens the labial extremes of the crowns of the last (not including pouch young). molars. IDENTIFICATION: Our material agrees close- Palatal Length (PL): Measured in the mid- ly in external and craniodental characters line from the anteriormost point of the pre- with Husson's (1978) detailed description of maxillae to the end of the palate. topotypic specimens from Surinam, and mea- Nasal Breadth (NB): Measured across the surements of adult Paracou vouchers (table triple-point suture of the nasal, frontal, and 1) broadly overlap those of topotypic adults maxillary bones on each side. (op. cit.: table 1). Least Interorbital Breadth (LIB): Mea- In view of the fact that Caluromys has sured at the narrowest point across the fron- never been revised, it is noteworthy that pub- tals between the orbits. lished measurements of some nominal taxa Least Postorbital Breadth (LPB): Mea- currently regarded as synonyms of C. philan- sured at the narrowest point across the fron- der (see Gardner, 1993) fall outside the tals behind the orbits. known range of variation among Surinamese Zygomatic Breadth (ZB): Greatest breadth and French Guianan specimens. In particular, across the zygomatic arches. specimens from Trinidad and northern Ven- A few other craniodental measurements ezuela formerly referred to C. trinitatis taken for special purposes are either self-ex- Thomas (type locality: ``Botanic Gardens, planatory or are de®ned in the following spe- Trinidad'') are substantially smaller and dif- cies accounts. Because most didelphids ex- fer from typical philander in coloration hibit obvious sexual size dimorphism, we (Thomas, 1894, 1903, 1904; PeÂrez-HernaÂn- summarize morphometric variation separate- dez et al., 1994). Furthermore, we have per- ly by gender. sonally observed conspicuous variation in Most of the large opossums in the Paracou size and pelage traits among populations cur- fauna (Caluromys, Chironectes, Didelphis, rently referred to C. philander from Ama- Metachirus, Philander) are easily distin- zonian Brazil. Although Cabrera (1958) rec- guished, even at a distance, by external char- ognized C. p. philander (including trinitatis acters described by Emmons (1990, 1997). as a synonym), C. p. af®nis (type locality: 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 19

Fig. 7. Dorsal and ventral views of the skull, and occlusal view of the maxillary dentition of Mar- mosa murina, showing the anatomical limits of marsupial measurements de®ned in the text.

Mato Grosso, Brazil), and C. p. dichrurus (or species) limits within what might be (type locality: ``Ypanema'', SaÄo Paulo, Bra- called the Caluromys philander complex is zil) as valid subspecies, the empirical basis too large to review in this faunal report, but for a trinomial nomenclature has never been from the close similarity noted above be- established. The material now available to tween Surinamese and French Guianan ma- evaluate geographic variation and subspecies terial it seems clear that the Paracou popu- 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 1 Measurements (mm) and Weights (g) of Adult Caluromys philander from Paracou

lation is referable either to the nominate sub- young measuring 54 mm crown-rump. With species (if a trinomial nomenclature is war- these exceptions, all of our vouchered and ranted) or to C. philander sensu stricto (if unvouchered records of Caluromys philander additional species are recognized in this are based on solitary individuals; no others group). were trapped or sighted together. FIELD OBSERVATIONS: All of our 15 vouchered records of Caluromys philander at Par- Chironectes minimus (Zimmermann) acou are from specimens trapped or shot at night in trees in primary forest, at both well- VOUCHER MATERIAL: AMNH 266477, drained and swampy sites. In addition, we 266478; MNHN 1998.672. Total ϭ 3 speci- recorded three unvouchered nocturnal obser- mens. vations of this species, two of which were IDENTIFICATION: Our voucher material con- sighted in trees in roadside secondary growth forms closely to most published descriptions and the third on a liana in primary forest. of this widespread and distinctive species Seven (47%) of our vouchers were obtained (e.g., Thomas, 1888; Cabrera, 1919; Krum- by shooting animals sighted in the forest can- biegel, 1940a; Augustiny, 1942; Mondol® opy or subcanopy, and eight others (53%) and Medina, 1957; Husson, 1978; Marshall, were taken in subcanopy platform traps. 1978b; Emmons, 1990, 1997), but a few dis- Measured heights above the ground for the crepancies and supplementary observations trapped specimens ranged from 12 to 16 m, merit comment. Persistent references (op. whereas visually estimated heights of sight- cit.) to brownish and yellowish tints in the ings of free-ranging (untrapped) animals pelage of Chironectes minimus are probably ranged from 3 to 20 m. Figure 8 provides a based on old (faded or stained) museum typical view of the subcanopy habitat of this skins; living animals and fresh skins have species at Paracou. clear black-and-gray dorsal markings and All four of our adult female vouchers were pure white venters, with no trace of other carrying suckling young. The ®rst, captured hues. According to Cabrera (1919), the digits on 10 August 1991, had three nursing young of the manus are webbed to the ends of the measuring 36 mm crown-rump; the second, ®rst phalanges, but our specimens (and those taken on 14 August 1991, had ®ve young illustrated by Augustiny [1942: ®g. 14] and measuring 47 mm crown-rump; the third, Mondol® and Medina [1957: ®g. 1]) have taken on 16 November 1992, had two young unwebbed manual digits. The ``supernumer- measuring 14 mm crown-rump; and the ary facial bristles'' mentioned by Marshall fourth, taken on 28 August 1993, had four (1978b) are the usual superciliary, genal, and 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 21

Fig. 8. Subcanopy habitat of Caluromys philander and other arboreal mammals (e.g., Micoureus demerarae, Potos ¯avus, Oecomys rutilus, O. auyantepui, Rhipidomys nitela) sampled by platform- trapping and by hunting at Paracou.

interramal vibrissae (Brown, 1971), which (1978b) both recognized four subspecies, but are well developed (Augustiny, 1942: ®g. 11) the necessity for a trinomial nomenclature but otherwise unremarkable in C. minimus. for water opossums has yet to be demonstrat- Apparently, the only postcranial vibrissae in ed by any substantive analysis of character this species consist of a prominent tuft of data. In the event that any subspeci®c dis- long carpal hairs at the wrist. tinctions are warranted, the Paracou popula- Our only adult voucher (AMNH 266477), tion would obviously be referable to the an old male, had external measurements of nominate form. 286 ϫ 345 ϫ 63 ϫ 31 mm and weighed 620 FIELD OBSERVATIONS: Water opossums g. Selected craniodental measurements of were commonly seen in all of the four named this specimen fall within the known range of stream systems that have their headwaters in morphometric variation for the species (Mar- our study area; even the smallest and shal- shall, 1978b): condylobasal length, 65.7 mm; lowest creeks were frequented (®g. 9). Be- length of molars, 14.5 mm; palatal breadth, cause this species has bright eyeshine, is 23.5 mm; palatal length, 42.7 mm; least in- boldly marked, and splashes noisily while terorbital breadth, 13.1 mm; least postorbital swimming or wading, it is not dif®cult to ob- breadth, 8.4 mm; zygomatic breadth, 39.0 serve despite its nocturnal habits. Neverthe- mm; length of nasals, 31.8 mm. Both of our less, the occurrence of Chironectes at Para- other vouchers are juveniles. cou was previously unsuspected by the for- REMARKS: Lutra minima Zimmermann estry workers and local hunters whom we in- (1780) was based on Buffon's (1776) de- terviewed. scription of the ``petite loutre d'eau douce de One of our three vouchers was taken in a Cayenne'', so our specimens are practically large (ca. 25 ϫ 30 ϫ 81 cm) wire live trap topotypes. Krumbiegel (1940a) and Marshall set in a small (ca. 2 m wide) shallow (ca. 15 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 9. Aquatic habitat of Chironectes minimus in primary forest at Paracou. Water opossums were encountered almost every night that we worked near such small streams, most of which were less than 2 m wide and 50 cm deep. Streams frequented by C. minimus at Paracou had clear tea-colored (black) water and ¯owed slowly over mostly sandy beds. Other nocturnal vertebrates encountered in the same streams included potential predators such as caimans (Paleosuchus sp.) and anacondas (Eunectes mu- rinus). 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 23

cm deep) stream in primary forest; the trap len, 1901, 1902; Gardner, 1973), and it seems was unbaited, but rows of stakes were driven likely that Husson's large specimens were unto the streambed on either side to funnel just old animals that may have been pre- animals moving downstream into the trap served because of their unusual dimensions. opening. The other two vouchers were col- The essential identity of the Surinamese and lected by shooting. In addition, we recorded French Guianan material we examined sup- 23 unvouchered observations of water opos- ports the conclusions (hitherto undocument- sums, of which 2 were based on juveniles ed by published comparisons) of Thomas trapped in small (145 ϫ 145 ϫ 410 mm) wire (1888) and Allen (1902) that D. karkinopha- live traps set in streams, and 21 were sight- ga Zimmermann and D. cancrivora Gmelin ings of free-ranging individuals. (both based on ``Le Crabier'', an opossum All of our 26 (vouchered and unvouch- described by Buffon from Cayenne) are ju- ered) records of water opossums at Paracou nior synonyms of D. marsupialis. were of animals trapped or sighted while Julien-LaferrieÁre (1991) reported Didel- they waded or swam in primary forest phis albiventris and D. marsupialis as occur- streams at night. Most free-ranging individ- ring syntopically in primary forest at Piste uals alarmed by our presence quickly swam St.-EÂ lie (only 14 km WNW of Paracou), the away, but two animals left the water and dis- ®rst published record of the former species appeared in dense streamside vegetation, and from French Guiana. Catze¯is et al. (1997) another entered a burrow near the water's subsequently reported both species from pri- edge. With the exception of a pair of animals mary forest near Petit Saut (about 28 km apparently engaged in an aggressive inter- SSW of Paracou). In view of these records action, all of our sightings were of solitary of sympatry from nearby localities, our fail- individuals. ure to record the presence of D. albiventris at Paracou merits comment. Didelphis marsupialis Linnaeus The diagnostic morphological characters of Didelphis marsupialis and D. albiventris VOUCHER MATERIAL: AMNH 266456± are suf®ciently striking that collected speci- 266460, 266462±266466, 266468, 266470, mens cannot be misidenti®ed by competent 266471, 266473, 266475, 267367; MNHN researchers alert to the possible presence of 1995.895±1995.901. Total ϭ 23 specimens both species. In the Guiana subregion of (not including pouch young). Amazonia, these taxa are readily distin- IDENTIFICATION: All of our specimens of guished by facial markings (bolder in albi- Didelphis from Paracou are referable to the ventris than in marsupialis; Mondol® and large, black-eared species D. marsupialis, the PeÂrez-HernaÂndez, 1984), ear color (the pin- type locality of which was restricted by nae are usually tipped with white in albiven- Thomas (1911a) to Surinam. Our material tris but are entirely black in adult marsupi- agrees in all qualitative details with Husson's alis; op. cit.), caudal pelage (the furred basal (1978) description of Surinamese topotypes, portion of the tail is conspicuously longer in but three of the four adults measured by Hus- albiventris than in marsupialis; M. D. Engs- son are larger than any collected at Paracou. trom, personal commun.), and size (albiven- To evaluate possible size differences, we bor- tris is smaller; see measurements of the up- rowed Surinamese material for side-by-side per molar row of albiventris tabulated by comparisons with our vouchers; however, Mondol® and PeÂrez-HernaÂndez, 1984). Cor- only four skin-and-skull preparations of adult related molecular characters that may be use- specimens (all females) could be located. No ful for discriminating albiventris and mar- differences in external characters were ob- supialis were recently discussed by Lavergne served between Surinamese and French et al. (1997). Guianan exemplars, and the measurement Although D. albiventris is de®nitely ab- data we obtained (table 2) do not suggest any sent from our voucher material, we did not appreciable morphometric divergence. Con- collect every individual Didelphis that we siderable ontogenetic size variation is appar- encountered at Paracou. Because many ani- ently characteristic of Didelphis species (Al- mals were sighted in dense vegetation where 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 2 Measurements (mm) and Weights (g) of Adult Specimens of Didelphis marsupialis from French Guiana and Surinama

facial markings and other potentially infor- ondary growth, or other disturbed habitats; mative details could not be distinguished, it the other 10 (48%) were taken in primary is therefore possible that this species was forest, at both well-drained and swampy seen but not recognized. Unfortunately, Di- sites. All specimens were shot or trapped at delphis virtually disappeared from our study night, and all unvouchered sightings of Di- area from 1992 to 1994, so we had little op- delphis were also nocturnal. All encountered portunity to collect specimens after we individuals were solitary; no collected fe- learned that D. albiventris could be expected male was carrying suckling young. to occur locally. OTHER SPECIMENS EXAMINED: SurinamÐ Gracilinanus emiliae (Thomas) Brokopondo, Brownsberg Nature Park (CM Figures 11A, 11B, 12A, 17C, 17D, 18C, 18D 52697); Coronie, Totness (CM 52702); Sar- amacca, Bigi Poika (CM 52716, 52724). A single specimen from Paracou (AMNH Listed specimens are adult skin-and-skull 267006) is the ®rst of this widespread but preparations; other Surinamese material ex- rarely collected species to be reported from amined (CM, FMNH) are immature (sub- French Guiana. Because the brief accounts adults, juveniles) or incomplete (skull only/ by Thomas (1909), Tate (1933), Husson skin only) specimens that are not useful for (1978), and Hershkovitz (1992) are inade- taxonomic comparisons. quate for evaluating morphological similari- FIELD OBSERVATIONS: All of our unambig- ties and differences with other congeners, we uous records of Didelphis marsupialis from redescribe the species below. Paracou are based on collected specimens. TYPE MATERIAL: The holotype only, a male Of these, 15 (65%) were trapped and 8 (35%) skin with skull and mandibles (BMNH were shot. Nine specimens (39%) were 9.3.9.10) collected by Emilie Snethlage on trapped or shot in trees or other elevated sub- 13 February 1909 at ``Para'' (ϭ BeleÂm, for- strates at heights ranging from 0.3 to 15 m merly known as ParaÂ), Brazil. The type was above the ground, whereas the remainder described by Thomas (1909: 379) as a ``sub- were trapped or shot on the ground. Of 21 adult'', by Tate (1933: 189) as a ``young specimens accompanied by habitat data, 11 adult'', and by Hershkovitz (1992: 33) as a (52%) were taken in clearings, roadside sec- ``juvenal''. The animal is, in fact, very nearly 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 25

Fig. 10. Known collection localities for Gracilinanus emiliae based on specimens examined. 1, BRAZIL, ParaÂ, BeleÂm; 2, BRAZIL, ParaÂ, Capim; 3, COLOMBIA, Meta, Los Micos; 4, FRENCH GUIANA, Paracou; 5, GUYANA, Upper Takutu-Upper Essequibo, Dadanawa; 6, SURINAM, Marow- ijne, Langamankondre; 7, VENEZUELA, Monagas, 47 km SE MaturõÂn.

adult with P3 and M4 both erupted but still DISTRIBUTION: Specimens that we exam- a little below their adult positions in the ined document the presence of Gracilinanus toothrow. emiliae in eastern Colombia, eastern Vene- Pine (1981: 59) suggested that the type lo- zuela, southern Guyana, northern Surinam, cality should be construed as the state of ParaÂ northern French Guiana, and eastern Brazil rather than the formerly eponymous city, but (®g. 10). Several other published localities Thomas (always scrupulous about type lo- for this species are erroneous or unreliable calities) would surely have noted the lack of (see Remarks, below). de®nite geographic information if he meant EMENDED DESCRIPTION: Very small murine ``Para'' in the sense of a district larger than opossums (table 3) with smooth (not woolly) most European countries. A specimen col- adult pelage; unruf¯ed dorsal fur dull reddish lected at Capim (ca. 90 km ESE of BeleÂm) brown, but basal two-thirds dark gray; ven- provides independent evidence that the spe- tral fur pure white or cream from chin to cies may occur in the environs of the city. groin (the hairs self-colored, not grayish ba- 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 sally). Face marked by mask of black fur ex- TABLE 3 tending from mystacial pad to just behind Measurements (mm) and Weights (g) of outer canthus of eye on each side, and by Subadult and Adult Specimens of narrow midrostral streak of pure orange fur; Gracilinanus emiliae cheeks (below mask) white or cream-colored like throat; facial vibrissae (including long superciliary, mystacial, and genal hairs) mostly black (but some of the ventralmost genals are usually white); vibrissae of chin and throat (submental and interramal hairs) white. Ears not very large (just covering eye when laid forward over face), apparently naked (a sparse pelage of very short hairs is only visible under magni®cation), and very thin; opaque and pale basally (possibly yellow in life, but whitish in fresh alcoholic specimens); translucent and darker (brownish or grayish) distally. Gular glands (indicated by a naked or sparsely haired median patch of skin on the throat) present in all specimens examined, including one juvenile. Wrists, ankles, and dorsal surface of feet covered with short pale (whitish or orange- tinted) hairs. Manus and pes each with six plantar pads (thenar, hypothenar, and four interdigitals); thenar and ®rst interdigital pads of manus separated by at least two rows of minute epidermal tubercles; thenar and ®rst interdigital pads fused on pes of one ¯uid specimen (AMNH 267006), touching but not fused on pes of two others (ROM 35465, 35466); central palmar surface of manus smooth (not densely tubercular); claws of manual digits II±V small, not extending be- rows/cm at middle of tail (counts from three yond ¯eshy apical pads. Scrotal epidermis of ¯uid specimens and two dried skins). Median holotype entirely unpigmented, of another hair of triplet emerging from posterior mar- subadult (AMNH 267006) with dark dorsal gin of each scale about two scale rows long, blotch surrounding suspensory stalk, of one thicker than lateral hairs, but not grossly ¯at- fully adult specimen (ROM 35466) entirely tened or petiolate. dark. At least nine (4±1±4) abdominal-in- Skull (®gs. 11, 12) with slender rostrum, guinal mammae present in one adult female incipiently beaded supraorbital margins, and (ROM 35465). absence of postorbital processes of frontals; Tail much longer than head-and-body (ta- orbits not conspicuously enlarged (delimited ble 3); less than 1 cm furry at base; uniform- posteriorly by well-developed postorbital ly dark (grayish or brownish) without pale processes of jugals); braincase not greatly in- blotches, bands, or countershading. Caudal ¯ated, smooth and unmarked by prominent epidermis covered with very small scales in temporalis scars. Premaxillaries with small annular or spiral series,3 numbering 40±50 (part of Tate's microtarsus group). As noted by Hersh- 3 kovitz (1992), spiral or annular series may occur on the The difference between spiral and annular arrange- same tail, and the perceived arrangement may be subject ments of caudal scales cited by Tate (1933) as a useful to observer bias. With the examples at hand of G. emi- character for diagnosing species groups of Marmosa liae it is dif®cult to determine which, if either, descriptor (sensu lato) is less than obvious within Gracilinanus is appropriate. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 27

Fig. 11. Dorsal and ventral cranial views of (A, B) Gracilinanus emiliae (composite of AMNH 203363, AMNH 267006, and ROM 35466) and of (C, D) Hyladelphys kalinowskii (AMNH 267338). Differences in rostral length, interorbital morphology, size of the orbits, braincase shape, and palatal fenestration are among the taxonomic differences illustrated by these views. 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 12. Lateral views of crania and mandibles of (A) Gracilinanus emiliae and (B) Hyladelphys kalinowskii based on the same specimens as in ®gure 11. Prominent taxonomic differences illustrated in these views include orbital size, development of the postorbital process of the jugal, braincase shape, presence of accessory cusps on upper and lower canines, and relative reduction of the third premolar. but distinct rostral process anterior to inci- Upper incisors 2±5 subequal (not increas- sors; maxillary-premaxillary suture extending in size from front to back); upper canine ing to posteriormost incisor alveolus; palate with small posterior accessory cusp; lower highly fenestrated, with large maxillopalatine canine procumbent and premolariform (with and posteromedial (palatine) vacuities; one ¯attened blade-like apex and small posterior or more small maxillary palatal vacuities pre- accessory cusp); deciduous third premolars sent unilaterally in one specimen (AMNH (dP3/dp3) large and molariform (®gs. 17, 203363), bilaterally in three others (BMNH 18); permanent third upper and lower pre- 9.3.9.10, ROM 35466, RMNH 18231). Ali- molars (P3/p3) slightly smaller than second sphenoid wing of auditory bulla with well- premolars (P2/p2); metacones much larger developed anteromedial process bridging fo- than paracones on M1±3. ramen ovale. COMPARISONS: Gracilinanus emiliae and a 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 29

sympatric taxon originally described by isons of AMNH 203363 (an adult male skin Hershkovitz (1992) as G. kalinowskii are and skull from Capim, Brazil) with the ho- similar in size and coloration (both have red- lotype of emiliae in London and, later, with dish-brown dorsal fur and pure white or the holotype of longicaudus in Chicago, re- cream-colored ventral fur), but they differ in vealed no differences beyond those that many other characters as explained in the fol- might be expected among individuals from a lowing account. Marmosa lepida (not known single local population (for measurements, to occur at Paracou but reported from other see table 3). We therefore regard G. longi- localities in French Guiana and Surinam; ap- caudus as a junior synonym of G. emiliae. pendix 1) is also externally similar to G. em- Eisenberg (1989) suggested that Gracili- iliae. Although adults of M. lepida are much nanus emiliae might be conspeci®c with G. larger than G. emiliae, immature specimens microtarsus from the Atlantic coastal rain- of the former species might be almost im- forest of southeastern Brazil, but it would be possible to distinguish from the latter in the dif®cult to select two more dissimilar con- ®eld. Cranially, adult M. lepida (see Husson, geners for comparison. Among other differ- 1978: pl. 11) can be identi®ed by their (1) ences, Gracilinanus microtarsus is much distinctively longer premaxillary rostral pro- larger and has a proportionately much shorter cesses, (2) prominent postorbital frontal pro- tail (see measurements in Tate, 1933; Hersh- cesses, (3) lack of posteromedial palatal va- kovitz, 1992), the ventral fur is buffy and cuities, (4) lack of anteromedial bullar pro- gray-based, the supraorbital margins are not cesses, and (5) lack of mastoid exposure be- beaded, and the crowns of I2±5 increase in tween the squamosal and parietal bones. size from front to back. Skulls of juvenile M. lepida that lack post- Gardner and Creighton (1989), Hershkov- orbital frontal processes can still be distin- itz (1992), and Gardner (1993) listed Mar- guished from G. emiliae by rostral, palatal, mosa agricolai Moojen (1943) as a synonym and bullar morphology, and by tooth size of Gracilinanus emiliae, but Moojen's illus- (see under Remarks, below). tration (op. cit.: ®g. 1) and description are REMARKS: Originally described as Mar- dif®cult to reconcile with this decision: the mosa emiliae (Thomas, 1909), this species ratio of tail to head-and-body for agricolai is was subsequently referred to Didelphis (sub- only 1.28, the maxillary-premaxillary suture genus Grymaeomys) by Matschie (1916), and (as drawn) does not extend anteriorly to I5, to Marmosa (subgenus Thylamys) by Ca- the frontals are shown without supraorbital brera (1958). The current allocation of emi- beads, the zygomatic arches are widely liae to Gracilinanus follows Gardner and ¯ared, the illustrated bullae appear to lack Creighton (1989). anteromedial processes, the upper canine ap- Gracilinanus longicaudus, named by pears to have a small anterior accessory cusp, Hershkovitz (1992) from a single specimen and the third upper premolar (as drawn) is (FMNH 87924) collected in eastern Colom- larger than the second. The type of agricolai bia, does not differ signi®cantly from G. em- (in Rio de Janeiro) should be reexamined to iliae in any external, cranial, or dental char- evaluate the true status and relationships of acter. According to the original description, this nominal species. ``. . . the combination of small size, long tail, We examined the two Brazilian specimens whitish underparts, incomplete eye ring, and identi®ed by Patterson (1992) as Gracilinan- narrow skull separates [longicaudus] from all us emiliae and found them to be immature other described species [of Gracilinanus]'' examples of Marmosa lepida. Both skulls (op. cit.: 39). However, the black mask of have very long rostral processes of the pre- FMNH 87924 does, in fact, completely en- maxillae, lack posteromedial palatal vacui- circle the eye (a narrow border of black hairs ties, and their rounded alisphenoid bullae is continuous below the lower lid and around lack anteromedial processes; the erupted el- the posterior canthus), and the other charac- ements of the molar dentition are also larger ters cited as diagnostic for longicaudus are than those of any specimens of G. emiliae matched by the type (and other referred spec- but match the homologous teeth of adult M. imens) of emiliae. Our side-by-side compar- lepida from eastern Peru (e.g., AMNH 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

78001, 98656). Because immature Marmosa laea) in reference to the predominant vege- lepida can be confused with Gracilinanus tation of the Amazonian lowlands (e.g., by emiliae in size and external appearance, lit- Ducke and Black, 1953); and ␦⑀␭␾␷Â ␵ erature records of the latter (e.g., AÂ vila-Pires, (womb), a traditional suf®x for New World 1964) should be regarded as suspect until di- marsupial genera. agnostic characters can be recon®rmed from specimens. Hyladelphys kalinowskii (Hershkovitz) OTHER SPECIMENS EXAMINED: BrazilÐ Figures 11C, 11D, 12B, 15A, 16A, 17A, 17B, 18A, ParaÂ, BeleÂm (BMNH 9.3.9.10, type), Capim 18B, 19B (AMNH 203363). ColombiaÐMeta, Los Micos (FMNH 87924, type of Gracilinanus Three specimens from Paracou (AMNH longicaudus). GuyanaÐUpper Takutu-Up- 267338, 267339; MNHN 1995.903) and one per Essequibo, 12 km E Dadanawa (ROM from Guyana (kindly loaned to us for iden- 35465, 35466), no other locality data (ROM ti®cation by M. D. Engstrom) represent ex- 33807). SurinamÐMarowijne, Langaman- traordinary range extensions of this distinc- kondre (RMNH 18231). VenezuelaÐMon- tive species, originally described as Gracili- agas, 47 km SE MaturõÂn (USNM 385066). nanus kalinowskii by Hershkovitz (1992) on FIELD OBSERVATIONS: Our single specimen the basis of two Peruvian specimens. We pro- from Paracou was shot at 22:18 hours on 21 vide an emended description with compari- October 1992 as it perched about 4 m above sons and remarks based on this new material the ground in dense secondary growth along and our reexamination of Hershkovitz's type a dirt road through well-drained forest (®g. series. 13). TYPE MATERIAL: The holotype, an adult female skin and skull (FMNH 89991), collect- Hyladelphys, new genus ed on 9 July 1958 by Celestino Kalinowski at Hacienda Cadena (890 m elevation), Mar- DIAGNOSIS: Very small didelphids distin- capata, Departamento Cuzco, Peru; and the guished from all other family members by paratype, also an adult female skin and skull the following combination of traits: four (FMNH 65754), collected in October 1948 mammae in two abdominal-inguinal pairs; by J. M. Schunke at Chanchamayo (1100 m dorsolateral margins of frontals beaded and elevation), Departamento JunõÂn, Peru. strongly convergent anteriorly, without post- DISTRIBUTION: Known only from Amazo- orbital processes; premaxillae short, without nian Peru, southern Guyana, and northern rostral process anterior to incisor row; pos- French Guiana (®g. 14). teromedial (palatine) palatal vacuities absent; EMENDED DESCRIPTION: Very small in all tympanic wing of alisphenoid without a well- external and craniodental dimensions (table developed anteromedial strut forming sec- 4). Body pelage smooth (not woolly); dorsal ondary foramen ovale; I2±5 not increasing in fur unpatterned dull reddish brown, but basal size from front to back, their crowns asym- two-thirds of hairs dark gray; ventral fur pure mmetrical and nonoverlapping; upper canine white or cream from chin to groin (the hairs without anterior or posterior accessory cusps; self-colored, not grayish basally). Face bold- P2 much larger than P3; deciduous premolars ly marked by broad mask of black fur ex- (dP3/dp3) very small and nonmolariform; tending from mystacial pad to base of ear on molars not highly carnassialized (paracone each side, and by prominent median streak smaller than metacone on M1±2, but para- of very pale orange fur extending from be- cone and metacone subequal on M3); lower tween eyes to rhinarium; cheeks (below canine not premolariform. mask) white like throat. Facial vibrissae (in- TYPE SPECIES: Gracilinanus kalinowskii cluding long superciliary, mystacial, and ge- Hershkovitz, 1992. nal hairs) mostly black (but a few of the ven- CONTENT: Hyladelphys currently contains tralmost genals are white); vibrissae of chin only the type species. and throat (submental and interramal hairs) ETYMOLOGY: From ‘␷Â␭␩ (wood or forest), white. Ears very large (covering eyes and ex- sometimes used in its Latinized form (hy- tending to mystacial pads when laid forward 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 31

Fig. 13. Secondary growth bordering a narrow dirt road through our study area at Paracou. Our only specimen of Gracilinanus emiliae was taken in this habitat, but primary forest occurs only a few meters away and it is not known which successional stage is typical of this rarely collected species. More commonly encountered inhabitants of secondary growth at Paracou include Didelphis marsupialis, Marmosa murina, Micoureus demerarae, Philander opossum, Dasypus novemcinctus, Saguinas midas, Potos ¯avus, Neacomys paracou, Oligoryzomys fulvescens, Oryzomys megacephalus, and Proechimys cuvieri. 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 14. Known collection localities of Hyladelphys kalinowskii based on specimens examined. 1, FRENCH GUIANA, Paracou; 2, Guyana, East Berbice-Corentyne, New River Falls; 3, PERU, Cuzco, Hacienda Cadena; 4, PERU, JunõÂn, Chanchamayo; 5, PERU, Loreto, Nuevo San Juan on RõÂo GaÂlvez. over face), apparently naked (a sparse pelage Wrists, ankles, and dorsal surface of me- of tiny hairs is visible only under magni®- tapodials covered with short, orange fur; ma- cation), and paper-thin; opaque and bright nus and pes each with six plantar pads (the- yellow-orange basally in fresh specimens nar, hypothenar, and four interdigitals), the (this color fading to white after a few months thenar and ®rst interdigital touching but not in alcohol), translucent and brownish distal- fused on either manus or pes; central palmar ly. Eyes very large. Throat glands apparently surface smooth (not densely tubercular); absent.4 claws of manual digits II±V large, extending beyond ¯eshy apical pads. Scrotum (of one 4 Throat glands in murine opossums are indicated by adult male, AMNH 267338) pigmented, a midventral gular patch of pigmented hair or bare skin (Tate, 1933). Although Hershkovitz (1992: 11) remarked fully examined the throat of FMNH 65754 under a dis- that ``neither marker is present in the two females of G. secting microscope and found no external trace of glan- kalinowskii'', the paratype (FMNH 65754) was subse- dular development. No discolored fur or bare patches quently described as resembling the holotype externally are present on the throats of any of the four ¯uid spec- ``but with gular gland evident'' (op. cit.: 38). We care- imens we examined from the Guianas. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 33

TABLE 4 Measurements (mm) and Weights (g) of Adult Specimens of Hyladelphys kalinowskii

blue. Mammae 2±0±2 ϭ 4, in two abdomi- Skull (®gs. 11, 12) with distinctively short, nal-inguinal pairs (without an unpaired me- blunt rostrum; beaded and anteriorly conver- dian teat; ®g. 15).5 gent supraorbital margins; no postorbital Tail not very long and without furry base; frontal processes; very large orbits (their pos- dark (brownish) above and below, but indis- terior limits indicated by weakly developed tinctly banded by absence of pigment over postorbital processes of the jugals); and lat- vertebral articulations. Caudal epidermis erally in¯ated braincase. Premaxillae without scaly beneath sparse covering of ®ne hairs; rostral process anterior to incisors; maxil- caudal scales in annular or spiral series, num- lary-premaxillary suture not extending to bering about 25±35 rows/cm at middle of tail posteriormost incisor alveolus; maxillopala- (counts from four ¯uid specimens). Median tine vacuities narrow, discontinuous in some hair of triplet emerging from posterior mar- specimens; posteromedial (palatine) and gin of each scale usually about three scale maxillary vacuities absent.6 Alisphenoid rows long, thicker than lateral hairs but not wing of auditory bulla without a well-devel- grossly ¯attened or petiolate. oped anteromedial process bridging the foramen ovale (see Remarks, below). 5 Hershkovitz (1992: 37) gave the mammary formula Upper incisors 2±5 subequal (not increas- of kalinowskii as 7±1±7 ϭ 15. This notation corresponds ing in size from front to back), asymmetrical to what he described earlier (op. cit.: 11) as the ``. . . prime formula for Gracilinanus . . . of which the pec- (the crowns with anterior but no posterior toral is 2±0±2 ϭ 4, the abdominal-inguinal, 5±1±5 ϭ 11. styles), and nonoverlapping (®g. 16); upper The functional teat formula may be the same or less, and lower canines without accessory cusps; depending on the number of attached young. Unused lower canine erect and not premolariform; nipples are more-or-less resorbed'' (italics original). Contrary to this last assertion, however, unused teats re- deciduous third premolars (dP3/dp3) ex- main everted and are not ``resorbed'' in adult female tremely reduced, not molariform (®gs. 17, marsupials (Bresslau, 1920; Tyndale-Biscoe and Ren- 18); permanent upper and lower third pre- free, 1987). Although unused teats are smaller than ac- tively lactating teats, accurate mammary counts can be 6 Hershkovitz (1992: 37±38) described the postero- obtained even from anestrous adults. All of the ®ve adult medial vacuities of Hyladelphys kalinowskii as ``small'', females of Hyladelphys kalinowskii that we examined but these fenestrae are absent in the holotype and the have four well-developed mammae in two abdominal- posterior palate of the paratype is destroyed. None of inguinal pairs; no other mammary loci were apparent, the specimens from Guyana and French Guiana have even with careful searching under high magni®cation. posteromedial vacuities. 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 16. Details of upper incisor morphology of (A) Hyladelphys kalinowskii (AMNH 267338) and (B) Marmosops pinheiroi (AMNH 267341). In all examined specimens of Hyladelphys, the crowns of I2±5 are narrow, asymmetrical, and widely spaced, with no consistent size increase from front to back. By contrast, the crowns of I2± 5inMarmosops are broad, symmetrical, overlapping or closely approximated rhomboids of posteriorly increasing size.

imens (with extracted skulls) from Guyana and French Guiana agree in all qualitative characters with the Peruvian holotype despite several inconsistencies between Hershkov- Fig. 15. Mammary morphology of adult fe- itz's (1992) original description and that giv- male specimens of Hyladelphys kalinowskii (A, en above (see footnotes 4±6). The Peruvian AMNH 267339) and Marmosops parvidens (B, specimens are slightly larger than French AMNH 267344). Only two pairs of inguinal-ab- Guianan material (table 4), but the apparent dominal teats are present in H. kalinowskii, whose size difference (perhaps best indexed by mo- mammary formula can be written as 2±0±2 ϭ 4. lar measurements) is no more than might be By contrast, this specimen of M. parvidens has four pairs of inguinal-abdominal teats plus an un- expected of conspeci®c samples collected al- paired median teat (4±1±4 ϭ 9). most 3000 km apart. The alisphenoid bullae of the holotype and of all four specimens from the Guianas lack anteromedial process- molars (P3/p3) much smaller than second es, but weakly developed processes (which premolars (P2/p2); molars not highly carnas- are not fused to the ¯oor of the braincase as sialized (metacone larger than paracone on they normally are in Gracilinanus, Marmo- M1 and M2, but metacone and paracone sub- sops, and Thylamys) are present on the an- equal on M3). teromedial aspect of both alisphenoid bullae VARIATION: The four ¯uid-preserved spec- of the paratype. We interpret this as intraspe- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 35

Fig. 17. Lateral and occlusal views of the left upper postcanine dentition of juvenile Hyladelphys kalinowskii (A, B, MUSM 11031) and Gracilinanus emiliae (C, D, USNM 385066) showing taxonomic differences in size and shape of the deciduous third premolar (dP3, arrows; see text). The fourth upper molar is unerupted in this example of H. kalinoswkii and incompletely erupted in G. emiliae.

ci®c variation because the paratype is not hibit odd-numbered mammary counts due to morphologically remarkable in any other re- the presence of an unpaired median teat (®g. spect. 15B); by contrast, caenolestids, microbioth- COMPARISONS: Hyladelphys kalinowskii eriids, and Old World marsupials have even- appears to differ from all other Recent di- numbered mammary counts because their delphids in at least two qualitative characters. teats are bilaterally paired (Bresslau, 1920; As far as is known, well-preserved adult fe- Osgood, 1921, 1943; Tate, 1933, 1947, males of other didelphid taxa uniformly ex- 1948). All reports of even-numbered didel- 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 18. Lateral and occlusal views of the left lower postcanine dentitions of juvenile Hyladelphys kalinowskii (A, B, MUSM 11031) and Gracilinanus emiliae (C, D, USNM 385066) showing taxonomic differences in the size and shape of the deciduous third premolar (dp3, arrows; see text). phid mammary counts that we investigated mary con®gurations in two Hyladelphys were either ®eld observations of lactating skins from Peru and two ¯uid-preserved teats only (e.g., Hershkovitz, 1997: table 4), specimens from French Guiana suggest that or were observations from specimens with these are not rare variants or preservational midventral incisions that might have de- artifacts. Instead, this taxon appears to be stroyed the median nipple (e.g., Marshall's genuinely divergent from other didelphids in [1978a] count of four mammae from an a character that displays considerable higher- FMNH skin of Glironia venusta; W. T. Stan- taxonomic stability among marsupials. ley, personal commun.). By contrast, our ob- Hyladelphys also differs markedly from servations of identical (2±0±2 ϭ 4) mam- other didelphids in the morphology of its de- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 37

TABLE 5 Morphology and Size of Milk Premolars among Didelphids and Other Marsupials

ciduous dentition. Whereas previous descrip- by having much smaller caudal scales (Ͼ 40 tions of didelphid milk premolars have con- rows/cm), longer and narrower rostrums, less sistently reported these teeth as large and pronounced interorbital constrictions, smaller molariform (Flower, 1867; Thomas, 1888; orbits, less in¯ated braincases, much more Bensley, 1903; Tate, 1948; Archer, 1976b), highly fenestrated palates, secondary foram- there is in fact some substantial variation in ina ovale formed by anteromedial struts of the size and morphology of dP3/dp3 within the alisphenoid tympanic wing, more carnas- the family (table 5). However, Hyladelphys sialized molars, and premolariform lower ca- is a conspicuous outlier: its milk teeth are nines. Individual species of Gracilinanus dif- much smaller than those of other confamili- fer from H. kalinowskii in other respects, but als, and they are uniquely nonmolariform in only G. emiliae merits explicit comparison occlusal structure (®gs. 17, 18). Instead, dP3/ here. dp3 in Hyladelphys fall well within the range Gracilinanus emiliae occurs sympatrically of milk-premolar morphologies seen in some with H. kalinowskii at Paracou, and because Old World marsupial groups (Tate, 1947, these species are similar in size and colora- 1948; Archer, 1976b; see illustrations in tion (both have reddish dorsal fur and self- Luckett, 1993, 1994). colored whitish venters), they might be con- From super®cially similar ``marmosines'', fused in the ®eld. However, numerous exter- Hyladelphys differs in additional characters. nal characters permit unambiguous discrim- Species of Gracilinanus, among which H. ination. Gracilinanus emiliae differs from H. kalinowskii was previously classi®ed, differ kalinowskii in facial markings (its black fa- 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 cial mask does not extend to the base of the is misleading in several details. The bulging ear, and the streak of orange fur between its eyes of our ¯uid-preserved specimens are eyes is darker and narrower) and has smaller proportionately about twice as large as those ears, smaller manual claws (not extending in the drawing. The pinnae in the illustration beyond the ¯eshy digital pads), 4±1±4 ϭ 9 are shown bristling with short hairs, but the mammae, and a much longer tail with small- auricular pelage is actually microscopic and er scales and no trace of lighter banding over the unmagni®ed ears appear quite naked; the the vertebral articulations. Another external ears in life are also proportionately much character that might distinguish G. emiliae larger than drawn. The portrait does not and H. kalinowskii is the presence/absence of show the genal vibrissae, but these long gular glands, but the variability of this trait black hairs are conspicuous against the short within species is not well documented and, white fur of the cheeks in all of the speci- with so few examples available for compar- mens at hand. The mystacial vibrissae are de- ison, its diagnostic value is uncertain. The picted as ®ne, inconspicuous hairs that ex- skull of G. emiliae differs from that of H. tend only to the outer canthus of the eye, but kalinowskii in the cranial characters listed these robust whiskers actually extend nearly above for Gracilinanus and by the presence to the tips of the pinnae when laid back along of a posterior accessory cusp on the upper the side of the head. The facial markings in canine (a trait not consistently exhibited by the portrait also lack the vivid de®nition other congeners). characteristic of this species: the mask is in- Members of other ``marmosine'' genera tensely black in fresh specimens and is bold- differ consistently from Hyladelphys by their ly accentuated by a broad streak of very pale larger size and size-correlated proportions orange fur down the midline of the rostrum. (relatively longer rostrums, smaller orbits, We examined the western Ecuadorean less in¯ated braincases) and in the following specimen (KU 135097) that Hershkovitz qualitative contrasts (in addition to the mam- (1992: 42) identi®ed as ``Gracilinanus sp. mary and milk-dentitional traits mentioned (new species)'' and that he subsequently (op. previously): incrassate tails (Thylamys), cit.: 45) conjectured ``. . . is most nearly like grossly enlarged central hairs of each caudal- adult Gracilinanus kalinowskii ...''. The scale triplet (Marmosops), distinct postorbital animal in question is a juvenile male (not a frontal processes (Marmosa, Micoureus), female as originally reported) preserved in secondary foramina ovale (Marmosops, Thy- ¯uid with an extracted skull, of which the lamys), I2±5 conspicuously increasing in size ®rst upper molar (crown length ϭ 2.26 mm) from front to back (Marmosops and some suggests an adult size far larger than that of Marmosa), P3/p3 larger than or subequal to Hyladelphys or any known species of Gra- P2/p2 (Thylamys), more carnassialized mo- cilinanus. The large size of its caudal scales, lars (all genera), and premolariform c1 (Mar- absence of posteromedial palatal vacuities, mosops). Insofar as can be inferred from po- and absence of an anteromedial alisphenoid lyprotodont character polarities (e.g., as hy- strut bridging the foramen ovale are addi- pothesized by Archer, 1976a, 1976b; Creigh- tional traits that cannot be reconciled with ton, 1984; Reig et al., 1987; Wroe, 1997), Gardner and Creighton's (1989) diagnosis of Hyladelphys shows no clear pattern of syn- Gracilinanus. From Hyladelphys, KU apomorphic resemblances with any other di- 135097 differs conspicuously by its well-de- delphid taxon. By the same token, evidence veloped rostral process of the premaxillae, of a closer relationship to nondidelphid well-developed postorbital jugal process, up- clades is also weak. In effect, our assignment per incisor morphology (I1±5 have large, of Hyladelphys to the family Didelphidae is overlapping, symmetrically rhomboidal based primarily on zoogeography and on crowns that increase in size from front to morphological traits that are currently inter- back), fully molariform dP3, and highly car- preted as marsupial plesiomorphies. nassialized molars. Based on these and other REMARKS: A published portrait of the head attributes, we refer this specimen to the geof Hyladelphys kalinowskii reconstructed nus Marmosa (sensu Gardner and Creighton, from dried skins (Hershkovitz, 1992: ®g. 14) 1989), within which it most closely resem- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 39

Fig. 19. Cross-sections and dorsal views of the interorbital regions of some Paracou marsupials: (A) Marmosops pinheiroi (AMNH 267341), (B) Hyladelphys kalinowskii (AMNH 267339), (C) Marmosa murina (AMNH 267368). Whereas the interorbital region of Marmosops parvidens is ``hourglass- shaped'' in dorsal view and has more-or-less rounded supraorbital margins, the interorbital region is ``convergent'' in Hyladelphys kalinowskii and has well-developed supraorbital beads. Adult specimens of Marmosa murina are characterized by large, triangular postorbital processes. All scale bars ϭ 5 mm.

bles M. mimetra Thomas (1921) and other which was restricted by Thomas (1911a) to nominal taxa currently synonymized (Gard- Surinam. According to Tate (1933: 95), two ner, 1993) with M. robinsoni. subspecies are parapatrically distributed in OTHER SPECIMENS EXAMINED: GuyanaÐ Surinam and French Guiana: supposedly, East Berbice-Corentyne, New River Falls Marmosa murina murina ranges ``along the (ROM 34271). PeruÐCuzco, Hacienda Cad- narrow coastal strip between the sea and the ena (FMNH 8991 [type]); JunõÂn, Chancha- heavy rainforest'', whereas ``in the rainforest mayo (FMNH 65754 [paratype]); Loreto, it is replaced by the smaller darker [M. m.] RõÂo GaÂlvez, Nuevo San Juan (MUSM muscula [(Cabanis)].'' Husson (1978), how- 11031). ever, recognized only the former taxon in his FIELD OBSERVATIONS: Our ®rst example of Surinamese material, and all of the Surina- this species from Paracou (MNHN 1995.903) mese specimens that we examined appear to was shot as it perched on a palm frond about represent a single recognizable formÐclose- 1 m above the ground in swampy primary ly conforming to Husson's description and forest at 18:35 hours on 25 October 1992. resembling our Paracou vouchers in qualita- The other two specimens (AMNH 267338, tive characters and measurements (table 6)Ð 267339) were taken from the same pitfall that we assume to represent typical M. mu- trapline, near a small stream in well-drained rina. Unfortunately, possible character dif- primary forest, on 21 August 1993. ferences between the types of murina and muscula are now dif®cult to evaluate. Marmosa murina (Linnaeus) Thomas (1892) identi®ed two syntypes of Figures 7, 19C, 21B Didelphys murina Linnaeus (1758) among VOUCHER MATERIAL: AMNH 266416, the specimens of the Lidth de Jeude collec- 266417, 267368, 267816; MNHN 1995.904. tion purchased by the BMNH in 1867. One Total ϭ 5 specimens. of these (BMNH 67.4.12.542) was designat- IDENTIFICATION: Our vouchers conform ed as the lectotype by Husson (1978), but his closely in all essential details of external and choice was unfortunate: the lectotype is a ¯u- craniodental morphology with Tate's (1933) id-preserved adult female from which the and Husson's (1978) authoritative descrip- skull has been extracted and apparently lost. tions of this species, the type locality of The skull of the adult male paralectotype 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 6 Measurements (mm) and Weights (g) of Adult Marmosa murina from French Guiana and Surinam

(BMNH 67.4.12.541), however, has mea- zones of gray-based fur. Whereas the dorsal surements that fall within the range of vari- color difference is consistent with Tate's di- ation that we observed among recently col- agnosis of muscula versus murina, the ven- lected specimens from Surinam (table 6). tral color difference is not. The only quali- Tate (1933) described the pelage color of tative cranial character cited by Tate as di- BMNH 67.4.12.541 in considerable detail, agnostic of muscula, the absence of dorsal but neither this specimen nor the lectotype grooves along the supraorbital ridges, is ap- are really suitable for subspeci®c color com- parently useless for de®ning this taxon (as parisons because their pigments may have represented by Tate's own identi®cations) in- faded after more than two centuries in alco- asmuch as all the fully adult skulls from the hol. Kartabo series (AMNH 42908, 48135, The holotype of Marmosa murina muscula 99983, 142807) have grooved supraorbital (type locality: ``Caraiben Niederlassung Ar- ridges. Given that (1) the other character dif- rai am obern Pomeroon'' [Cabanis, 1848: ferences between murina and muscula cited 778], Pomeroon-Supenaam, Guyana) is a ju- by Tate are inde®nite, (2) that the relevant venile specimen (with dP3 in place and M4 types have discolored pelage and are incom- unerupted) that consists of the skin and skull mensurate in age, and (3) that the material of an animal originally preserved in ¯uid. we examined from Surinam and French Gui- Because of its immaturity, and because the ana shows no appreciable divergence in size pelage is now faded from preservative and or coat color between coastal and interior discolored by age, the type of muscula is populations, it does not seem useful to rec- likewise unsuitable for subspeci®c compari- ognize these subspecies as valid at the pre- sons. A small series of skins from Kartabo, sent time. Guyana, that Tate (1933) identi®ed as M. m. We also examined the types of other nom- muscula, however, are dorsally somewhat inal taxa from the Guiana subregion of Ama- darker than our Paracou vouchers, but have zonia currently treated as subjective syno- whiter venters with less extensive lateral nyms of Marmosa murina, including klagesi 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 41

Allen (1900), chloe Thomas (1907), rorai- Marmosa murina at Paracou are based on mae Tate (1931), and duidae Tate (1931). All specimens collected in secondary growth and of these are chie¯y distinguishable by pelage other manmade habitats: two were shot at charactersÐsubtle differences in fur color, night as they perched 1±2 m above the length, and textureÐthat perhaps vary clin- ground in dense roadside vegetation, one was ally with environmental conditions as sug- caught by hand at night in a garden, and two gested by Tate himself (1933, 1939), who were found killed and partially eaten (prob- ranked them as no more than subspeci®cally ably by domestic cats) near houses in clear- distinct. By contrast, the conspeci®city of ings. some taxa from outside the Guiana subregion that are currently referred to M. murina (e.g., Marmosops Matschie by Gardner, 1993) is more problematic. For example, M. quichua Thomas (1899a) from Thirty-®ve voucher specimens from Para- western Amazonia is craniodentally distinc- cou are referable to Marmosops parvidens in tive (Tate, 1933; personal obs.) and no jus- the sense of Pine (1981), but close exami- ti®cation for its synonymy with M. murina nation of pelage characters and correlated has apparently been published. In view of the morphometric variation in our sample indi- lack of any critical review of the species-lev- cates that two species are present. The fol- el taxonomy of Marmosa (sensu stricto) lowing paragraphs summarize the evidence since Tate's 1933 monograph, that work supporting this conclusion. should still be considered the primary au- All of the specimens in question are small thority for species limits until compelling ev- opossums (21±33 g adult weight) with dark idence is provided for alternative classi®ca- facial masks, dull reddish-brown or grayish- tions. brown dorsal fur, small fore- and hindfeet OTHER SPECIMENS EXAMINED: French Gui- (each with diminutive claws and six separate anaÐArataye (MNHN 1981.172, 1981.173, plantar tubercles), and long tails. The scrotal 1982.597, 1986.125), Cayenne (MNHN sacs of males are entirely white. Female 1986.1024, 1986.1025), Les Nouragues (MNHN 1998.307), Piste St.-EÂ lie (MNHN specimens lack any trace of a pouch and ap- 1981.417±1981.419, 1981.421, 1981.422, pear to have have 3±1±3 ϭ 7 or 4±1±4 ϭ 9 1982.598), SauÈl (MNHN 1982.596, inguinal-abdominal mammae (®g. 15B). Oth- 1986.484). GuyanaÐ``Demerara River 29 er distinctive attributes include: the grossly miles above Georgetown'' (BMNH 7.6.20.16 enlarged central hair of each caudal-scale [holotype of chloe]); Cuyuni-Mazaruni, Kar- triplet; a prominent, spoon-shaped, ¯eshy tu- tabo (AMNH 42907, 42908, 48135, 99983, bercle supported internally by bone on the 142807); Pomeroon-Supenaam, ``Caraiben- lateral aspect of the wrist of males (®g. 20); Niederlassung Arrai am obern Pomeroon'' smoothly rounded supraorbital margins with- (ZMB 2331 [holotype of muscula]). Suri- out distinct beads or processes (®g. 19A); namÐBrokopondo, Finisanti (FMNH conical alisphenoid bullae with prominent 95315±95319, 95321±95326, 95328); Ma- anteromedial processes (®g. 21); absence of rowijne, Oelemarie (CM 76729); Para, Zan- posteromedial (palatine) palatal vacuities derij (CM 68346, 68353); Saramacca, Ra- (®g. 22); upper canines with distinct anterior leigh Falls (CM 68354, 68355, 68356); Su- and posterior accessory cusps; and lower ca- riname, Lelydorpplan (FMNH 95329, 95330, nines that resemble the lower premolars in 95332). VenezuelaÐAmazonas, Mt. Duida shape and size, forming a more-or-less un- (AMNH 76984 [holotype of duidae]); BolõÂ- differentiated series of four subequal teeth var, Ciudad BolõÂvar (AMNH 16121 [holo- (®g. 23). Some of these traits are common to type of klagesi]), Arabupu (AMNH 75703 many ``marmosines'', others are perhaps di- [holotype of roraimae]). Without locality agnostic of the genus Marmosops, and a few dataÐ(BMNH 67.4.12.541 [paralectotype may de®ne a distinctive group of species of murina], 67.4.12.542 [lectotype of muri- closely related to M. parvidens. For present na]). purposes, however, these attributes serve to FIELD OBSERVATIONS: All of our records of distinguish examples of parvidens-like Mar- 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 20. Sexually dimorphic wrist morphology of members of the Marmosops parvidens species group. Adult males (left) possess an externally obvious lateral carpal tubercle that is supported internally by an enlarged pisiform bone (arrows); females (right) do not show this character (see Lunde and Schutt, 1999, for further details). Marmosops parvidens and M. pinheiroi, both members of the Paracou fauna, have similar wrist morphologies, but males of other taxa referable to the M. parvidens species group (e.g., M. juninensis, see text) differ in the size and shape of the lateral carpal tubercle. Illustrated specimens are examples of Marmosops pinheiroi from Paracou: top and bottom left, AMNH 267346; top and bottom right, AMNH 267342. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 43

Fig. 21. Detail of auditory region in (A) Marmosops pinheiroi (AMNH 267346) and (B) Marmosa murina (AMNH 267368). In Marmosops, the tympanic wing of the alisphenoid (twa) produces an anteromedial strut (st) that partially encloses the mandibular branch of the trigeminal nerve (V3, reconstructed course shown by heavy arrow); the nerve then emerges from a secondary foramen ovale. In Marmosa, the extracranial course of the nerve is unenclosed by the alisphenoid because the anteromedial strut is absent; there is no secondary foramen ovale in this species. Other abbreviations: als, alisphenoid; bo, basioccipital; bs, basisphenoid; et, ectotympanic; fo, foramen ovale; per, periotic.

mosops from all of the other marsupial taxa tral coloration was classi®ed as ``white'' if that we collected at Paracou. the self-colored fur and the tips of gray-based Within this series, the most conspicuous hairs lacked any pigmentation, or ``cream'' if external variation involves fur color. We the self-colored fur and gray-based hair tips scored three pelage characters for statistical were pale yellowish. Ventral pattern was analysis. Dorsal coloration was classi®ed as classed as ``narrow'' if self-colored (pure ``fuscous'' (dusky grayish-brown) or ``red- white or cream) fur was con®ned to the mid- dish'' (a subtly warmer tone) according to line (sometimes as a discontinuous streak) by the predominant hue of the unruf¯ed fur. Al- a broad zone of gray-based fur on each side, though this chromatic contrast disappears in ``broad'' if almost the entire ventrum was material stored for years in alcohol, dried self-colored, or ``intermediate'' for speci- skins and fresh ¯uid specimens (recently re- moved from formalin) were readily assigned ever, retain their original colors: AMNH 266423, to one or the other of these two states.7 Ven- 267007, and 267352 have ``fuscous'' dorsal fur and ``white'' underparts; AMNH 266426 has ``reddish'' dorsal fur and ``cream'' underparts; and AMNH 267817 has 7 At the time of writing (1997), specimen assignments ``reddish'' dorsal fur and ``white'' underparts. The extent to dorsal and ventral color classes (recorded in 1993) are of gray-based ventral fur (another pelage character we dif®cult to verify in ¯uid-preserved material (the bulk scored for statistical analysis) is not affected by ageing of our vouchers) due to fading. Five dried skins, how- or preservatives. 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 22. Dorsal and ventral cranial views of (A, B) Marmosops parvidens (AMNH 267359, male) and (C, D) M. pinheiroi (AMNH 267345, male). Note species differences in size and in the relative breadth of the rostrum. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 45

Fig. 23. Lateral views of crania and mandibles of (A) Marmosops parvidens and (B) M. pinheiroi based on the same specimens as in ®gure 22. Note the smaller upper canine of M. parvidens. The premolariform lower canines (forming a continuous series of four subequal teeth with p1±3) distinguish both species from other sympatric marmosines. mens with substantial amounts of both self- ventral fur narrowly con®ned to the midline colored and gray-based ventral fur. or bordered by extensive lateral zones of Chi-squared tests of independence provide gray-based fur; by contrast, many animals no evidence for age or sex effects on the ex- with reddish dorsal fur have almost entirely pression of these pelage traits. In fact, the self-colored ventral fur, and none has self- largest age-sex class in our sample (16 adult colored fur narrowly con®ned to the midline. males) includes individuals with fuscous and Similarly, almost all animals with fuscous reddish dorsal fur, specimens with white and dorsal fur have white (or white-tipped) ven- cream-colored ventral fur, and examples of tral fur, whereas about one-third of the spec- all three conditions of ventral pattern. How- imens with reddish dorsal fur have cream (or ever, pelage characters are not independently cream-tipped) underparts. distributed inter se (table 7). Most animals We used one-way ANOVAs to test for with fuscous dorsal fur have self-colored morphometric divergence between individu- 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 7 difference between the two groups of speci- Chi-Square Tests for Independence of Pelage mens previously sorted by chromatic and Charactersa among Marmosops Specimens morphometric traits. In specimens with fus- from Paracou cous dorsal fur, taller canines, and broader skulls, the lacrimal bone forms a prominent part of the anteroventral margin of the orbit; the lacrimal foramina then perforate the orbital margin, where they are exposed in lateral view (®g. 25B). By contrast, the lacrimal bone is not a prominent part of the anteroventral orbital margin in specimens with reddish dorsal fur, smaller teeth, and narrower skulls; in these, the lacrimal foramina are always located within the orbit, where they are more-or-less concealed from lateral view (®g. 25A). We interpret the correlation between pelage characters, morphometric differences, and lacrimal morphology described above as evidence that two species of Marmosops are sympatric at Paracou. Based on our examination of relevant type material (see below), we identify the reddish individuals with shorter canines, narrower skulls, and reduced lacrimals as M. parvidens, and the fuscous individuals with taller canines, broader skulls, and prominent lacrimals as M. pin- als with red and fuscous dorsal fur among heiroi. A single Paracou specimen (AMNH adult males (the only age-sex class in our 267358), consisting of the skeleton only of a sample large enough for such analyses) and juvenile animal with a smashed skull, is not found highly signi®cant differences (p Ͻ assignable with certainty to either species. 0.01) in length of molars, palatal breadth, zygomatic breadth, height of the upper canine Marmosops parvidens (Tate) (measured from the posterior accessory cusp Figures 15B, 22A, 22B, 23A, 25A, 26A to the tip of unworn teeth), and nasal breadth. Other external and craniodental measure- VOUCHER MATERIAL: AMNH 266425, ments showed no signi®cant divergence. Not 266426, 267344, 267347, 267348, 267350, surprisingly, principal components analysis 267353, 267359, 267361, 267817; MNHN of the log-transformed craniodental measure- 1995.927±1995.930, 1995.933. Total ϭ 15 ment data provides clear separation of dorsal specimens. fur color classes aligned with the ®rst eigen- IDENTIFICATION: Specimens of Marmosops vector (®g. 24), the coef®cients of which (ta- parvidens can be distinguished from sym- ble 8) re¯ect essentially the same differences patric examples of M. pinheiroi at Paracou as those indicated by univariate statistics: by their warmer (more reddish) dorsal fur, fuscous adult males have taller canines, broader extent of self-colored fur (which is broader nasals, and wider zygomas than do never discontinuous between chin and anus), reddish adult males, with smaller (but still smaller teeth (especially upper canines), nar- substantial) differences in molar size and pal- rower skull, and by the reduced lacrimal con- atal breadth. Similar morphometric contrasts tribution to the anteroventral orbital margin between dorsal fur color classes can be seen (see above). Because the difference in dorsal among the few adult females in our sample, pelage color is subtle, and because there is and among immatures. some overlapping variation in the extent of Close inspection of skulls revealed another self-colored ventral fur, unambiguous species 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 47

Fig. 24. Specimen scores of adult male Marmosops parvidens (⅙) and M. pinheiroi (●) on the ®rst two principal components extracted from the covariance matrix of eight log-transformed craniodental measurements (see text). Variable coef®cients and eigenvalues (scaled as percentages of the total variance in these data) are provided in table 8.

TABLE 8 identi®cations require cleaned cranial mate- Principal Components Analysis of rial. In our voucher series, height of canine Craniodental Measurement Variation among (HC) affords the clearest discrimination (ta- 16 Adult Male Marmosops from Paracoua bles 9, 10), but toothwear and sexual dimorphism must be taken into account in sorting specimens by this criterion. Lacrimal morphology (®g. 25) is perhaps the most reliable cranial character for identifying juveniles with incompletely erupted (and therefore un- measurable) canines. Although a revision of what may be called the parvidens group of Marmosops is beyond the scope of this faunal report, we note that our conclusion that two species assignable to this complex are sympatric at Paracou is consistent with recent molecular evidence that M. parvidens sensu Pine (1981) is composite (Mustrangi and Patton, 1997; Patton et al., 2000). In addition to recognizing M. pinheiroi as a valid species, we note that M. juninensis (another ``subspecies'' of M. parvi- 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 25. Detail of the anterior orbital region in (A) Marmosops parvidens and (B) M. pinheiroi based on the same specimens as in ®gures 22 and 23. In M. parvidens the lacrimal bone does not form a large part of the orbital margin, and the lacrimal foramina are concealed from lateral view inside the orbit. In M. pinheiroi the lacrimal bone forms a larger part of the orbital margin, and the lacrimal foramina (lf) are laterally exposed. dens sensu Pine, 1981) is equally distinctive from M. parvidens in lacking any trace of an and also merits speci®c recognition. Origi- anterior accessory cusp on the upper canine. nally described by Tate (1931) on the basis Specimens from Bolivia (e.g., AMNH of a single specimen (AMNH 63864), M. 268938) and Peru (e.g., AMNH 67243) with juninensis (now additionally represented by darker (more saturated) dorsal fur than the AMNH 230014±230016, all collected near type may nevertheless be provisionally re- Tarma in Depto. JunõÂn, Peru) can be unam- ferred to M. bishopi based on upper canine biguously distinguished from M. parvidens morphology. (sensu stricto) and M. pinheiroi by the size A specimen from northern Venezuela and shape of the male carpal tubercle (small- (USNM 371299) that Pine (1981) referred to er and not spoon-shaped), by the consistent Marmosops parvidens parvidens lacks an an- presence of posteromedial palatal vacuities, terior accessory upper canine cusp (like M. and by the absence of distinct accessory bishopi) and exhibits other differences from cusps on the upper canine (®g. 26). Like par- typical examples of M. parvidens as recog- videns, but unlike pinheiroi, the lacrimal nized in this report. The mystacial vibrissae bone in juninensis does not form part of the of USNM 371299 appear to be very short, anteroventral orbital margin, so the lacrimal probably not extending much if at all beyond foramen lies inside the orbit. Unlike any the base of the ear in life, whereas the mys- specimens of parvidens, however, the ventral tacial hairs reach at least to the posterior mar- pelage of juninensis is entirely gray-based. gins of the pinnae in typical examples of M. We examined the type of Marmosops par- parvidens. The dorsal fur of USNM 371299 videns bishopi (USNM 393535), the pelage is longer than in typical M. parvidens (about of which Pine (1981) described as colored 9 mm middorsally versus about 7 mm), is essentially like that of M. p. parvidens, but ¯uf®er in texture, and has a faintly marbled paler. In our opinion, this specimen repre- appearance that is not characteristic of other sents another distinct species that differs specimens that we refer to this species. Taken 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 49

TABLE 9 was caught by hand, one was caught in a Measurements (mm) and Weights (g) of Adult Victor trap, one in a Sherman trap, and the Specimens of Marmosops parvidens rest (seven) were captured in pitfalls; all were collected at night. Nine specimens were taken on the ground, ®ve were found perching in understory vegetation (usually on vertical stems) 0.2±1.5 m above the ground, and one was trapped on a liana 1.8 m above the ground. Seven specimens were collected in well-drained primary forest, ®ve in creekside primary forest, one in swampy primary forest, one in primary forest of unspeci®ed character, and one in secondary growth. See the following account for habitat comparisons between Marmosops parvidens and M. pinheiroi.

Marmosops pinheiroi (Pine) Figures 16B, 19A, 20, 21A, 22C, 22D, 23B, 25B

VOUCHER MATERIAL: AMNH 266423, 266424, 267007, 267008, 267341±267343, 267345, 267346, 267349, 267351, 267352, 267354, 267357; MNHN 1995.925, 1995.926, 1995.931, 1995.932, 1995.934. Total ϭ 19 specimens. IDENTIFICATION: See the preceding account for diagnostic comparisons with Marmosops parvidens. together, these differences suggest that In addition to the holotype and paratypes USNM 371299 may represent an undescri- of Marmosops pinheiroi, we examined the bed taxon, but more material should be ex- type series of M. parvidens woodalli (USNM amined to evaluate this conjecture. 393529±393532, 393534, 545543), a subspe- With a single exception (AMNH 97333, cies that Pine (1981) described from the vi- see below), all of the specimens that we iden- cinity of BeleÂm, Brazil. Although these spec- tify as Marmosops parvidens sensu stricto imens average paler dorsally than examples are from the Guiana subregion of Amazonia of M. pinheiroi from north of the Amazon, (®g. 27). they are otherwise similar in pelage and OTHER SPECIMENS EXAMINED: BrazilÐ craniodental characters, and we provisionally Amazonas, Boca Rio Piratucu (AMNH regard them as conspeci®c. An adult male 93970), 80 km N Manaus (USNM 579985± specimen from the right bank of the lower 579990); ParaÂ, Ilha do Taiuna on lower Rio Rio Xingu (USNM 549294), however, may Tocantins (AMNH 97333). French Gui- represent an undescribed taxon. Although anaÐArataye (USNM 548439). GuyanaÐ most of its qualitative traits match those of Demerara-Mahaica, Hyde Park (FMNH M. pinheiroi, it is much paler dorsally and 18545 [holotype]); Upper Takutu-Upper Es- has substantially smaller upper canines (HC sequibo, Karanambo (ROM 97938). ϭ 1.06 mm) for its sex than any example of FIELD OBSERVATIONS: Because unvouch- that species as recognized by us. ered sightings of Marmosops could not be OTHER SPECIMENS EXAMINED: BrazilÐ unambiguously identi®ed to species, all of AmapaÂ, Serra do Navio (USNM 461459 [ho- our de®nite records of M. parvidens at Par- lotype], 461460, 461462±461465); ParaÂ, Be- acou are from collected specimens. Five leÂm (USNM 545543), Utinga (USNM specimens of M. parvidens were shot, one 393529±393532, 393534). GuyanaÐPota- 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 10 Measurements (mm) and Weights (g) of Adult Specimens of Marmosops pinheiroi

ro-Siparuni, Iwokrama Reserve (ROM est''), we sometimes took both species in 108920). VenezuelaÐBolõÂvar, Auyan-tepui close proximity. One line of pitfall traps that (AMNH 130521, 130568, 130570), Churi-te- traversed 50 m of apparently homogeneous pui (AMNH 176352, 176353), 85 km SE El primary forest along a small stream, for ex- Dorado (USNM 385046). ample, captured three parvidens and four FIELD OBSERVATIONS: All of our de®nite pinheiroi between 29 July and 13 August records of Marmosops pinheiroi at Paracou 1993. Nevertheless, it is noteworthy that are from collected specimens. Four speci- whereas we rarely took parvidens in swampy mens were shot, three were captured in Vic- forest or secondary growth, over half of our tor traps, and the remainder (12 specimens) pinheiroi specimens were collected in those were caught in pitfalls; all were taken at habitats. Both species appear to occur only night. Thirteen specimens were trapped or in the forest understory: no Marmosops were shot on the ground (one in a hollow log), but sighted or trapped at heights greater than six were taken 0.3±1.5 m above the ground about 2 m above the ground. on vertical stems and lianas. Four specimens None of the adult female Marmosops we were collected in well-drained primary for- collected were carrying suckling young. est, ®ve in creekside primary forest, six in swampy primary forest, and four in second- Metachirus nudicaudatus (E. Geoffroy) ary growth. Marmosops parvidens and M. pinheiroi VOUCHER MATERIAL: AMNH 266435, clearly overlap in habitats at Paracou. Al- 266439, 266440, 266449, 266450, 266452, though our ecological classi®cation is coarse 266453, 266455, 267009, 267010, 267362, and doubtless obscures many subtle differ- 267365; MNHN 1995.905±1995.910. Total ences among capture sites with the same de- ϭ 18 (not including suckling young). scriptor (e.g., ``well-drained primary for- IDENTIFICATION: Our voucher material is al- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 51

Fig. 26. Details of palatal and upper canine morphology of Marmosops parvidens (A, AMNH 267359) and M. juninensis (B, AMNH 230016). Among other differences between these distinctive species (see text), the posterior palate is more extensively fenestrated in M. juninensis than in M. parvidens, and M. juninensis lacks the distinct accessory cusps that are always present on the unworn upper canines of M. parvidens.

most topotypical of this species, which was role in any future attempt to resolve this un- originally described from a specimen col- satisfactory state of affairs. lected at Cayenne (Julien-LaferrieÁre, 1994). REMARKS: We agree with Julien-LaferrieÁre The Paracou series agrees closely in quali- (1994) that the name Didelphis nudicaudata tative external characters with the description is available from Geoffroy's (1803) catalog given by Husson (1978), and most cranio- for the reasons explained by Hershkovitz dental measurements of the type fall within (1955) and Holthuis (1963). the range of metric variation in our voucher OTHER SPECIMENS EXAMINED: French Gui- collection (table 11). anaÐCayenne (MNHN 1990.420 [holo- Although Metachirus has long been type]). thought to contain but a single valid species FIELD OBSERVATIONS: We recorded 22 ob- (Tate, 1939; Cabrera, 1958; Gardner, 1993), servations of Metachirus nudicaudatus at this historical consensus is challenged by re- Paracou, of which 18 are based on collected cently analyzed mtDNA sequences that sug- specimens and 4 are unvouchered sightings. gest deep evolutionary divergence among Eighteen records (82%) are of animals shot samples from different Amazonian subre- or sighted on the ground, but one specimen gions (Patton et al., 2000). In the absence of (4%) was trapped on the ground in a Toma- any revisionary analysis of morphological hawk live trap, and three specimens (14%, specimens, however, it is unclear how mito- all juveniles) were taken in Victor snap-traps chondrial haplotypes might correspond with tied to lianas 0.5±1.3 m above the ground. named taxa. Inevitably, nominotypical ma- All of our records are from animals shot, terial from French Guiana will play a pivotal sighted, or trapped at night. Fourteen indi- 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 27. Known collection localities for Marmosops parvidens and M. pinheiroi based on specimens examined. 1, BRAZIL, AmapaÂ, Serra do Navio; 2, BRAZIL, Amazonas, Boca Rio Piratucu; 3, BRAZIL, Amazonas, 80 km N Manaus; 4, BRAZIL, ParaÂ, BeleÂm and Utinga; 5, BRAZIL, ParaÂ, Ilha do Taiuna; 6, FRENCH GUIANA, Arataye; 7, FRENCH GUIANA, Paracou; 8, GUYANA, Demerara-Mahaica, Hyde Park; 9, GUYANA, Potaro-Siparuni, Iwokrama Reserve; 10, Upper Takutu-Upper Essequibo, Karanambo; 11, VENEZUELA, BolõÂvar, Auyan-tepui; 12, VENEZUELA, BolõÂvar, Churi-tepui; 13, VENEZUELA, BolõÂvar, 85 km SE El Dorado. One collection locality for Marmosops pinheiroi (BeleÂm/ Utinga, in the Brazilian state of ParaÂ) lies just outside the right margin of this map. viduals (64%) were shot, sighted, or trapped Micoureus demerarae (Thomas) in well-drained primary forest, but one (4%) was encountered in swampy primary forest, VOUCHER MATERIAL: AMNH 266428, four (18%) in primary forest of unspeci®ed 266429, 266431±266434, 267370, 267371, character, and three (14%) in secondary veg- 267818; MNHN 1995.911±1995.914. Total etation. With the exception of females with ϭ 13 specimens. nursing young, all shot, sighted, or trapped IDENTIFICATION: Although our voucher ma- animals were solitary. terial corresponds closely to Tate's (1933) One female shot on 7 July 1991 had seven and Husson's (1978) descriptions of this tax- nursing young measuring 19 mm crown- on, the coloration of the ventral pelage and rump, and another shot on 17 August 1991 of the tail are variable among Paracou spec- had eight nursing young measuring 29 mm imens and merits comment. Most of the ven- crown-rump. tral surface is covered by gray-based fur that 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 53

TABLE 11 terial, the best external characters for dis- Measurements (mm) and Weights (g) of Adult crimination are tail color (most, but not all, Metachirus nudicaudatus from French Guiana examples of M. demerarae have tails blotched or tipped with white, whereas M. murina has consistently all-dark tails), fur texture (longer and woolly in M. demerarae, close and smooth in M. murina), the extent of fur at the base of the tail (conspicuously greater in M. demerarae than in M. murina), and size of the manual claws (extending beyond the ¯eshy apical pads in M. demerarae but not in M. murina). REMARKS: Originally described as a subspecies of Marmosa cinerea by Thomas (1905), demerarae was treated as a distinct species of the cinerea group in Tate's (1933) monographic revision of Marmosa (sensu lato). Cabrera (1958), however, considered demerarae to be a subspecies of cinerea, cit- ing doubts that Tate (1939: 164, footnote 2) expressed about his own prior classi®cation. The current allocation of Tate's cinerea group to Micoureus follows Gardner and Creighton (1989). Among the many nominal taxa now synonymized with Micoureus de- is heavily washed with buff, but self-colored merarae (sensu Gardner, 1993) are several (pure buff) hairs cover the groin and throat, that Tate (1933) recognized as full species, and a few specimens have narrow streaks of all or some of which may yet prove to be pure buff fur along the midline of the chest valid (Patton et al., 2000). With the extensive or abdomen. The tails of most Paracou spec- series of specimens now available to evaluate imens have white tips or are mottled with the taxonomy of this widespread complex, large white spots distally, but two adults the group is ripe for modern revisionary (AMNH 267370, 267818) have entirely dark treatment. tails. OTHER SPECIMENS EXAMINED: GuyanaÐ The holotype (BMNH 5.11.1.25) and other ``R. Demerara'' (BMNH 7.6.20.14); Cuyuni- adult specimens from Guyana that we mea- Mazaruni, Kartabo (AMNH 42887, 64156); sured for comparison exhibit broad morpho- Potaro-Siparuni, 5 km SE Surama (ROM metric overlap with our vouchers (table 12). 103146), Iwokrama Reserve (ROM 104708); Additionally, fresh Guyanese skins (e.g., Upper Demerara-Berbice, Comackka on De- ROM 103370, 104708) are indistinguishable merara River (BMNH 5.11.1.25 [holotype]), in coloration from our Paracou material. Tropenbos (ROM 103370); Upper Takutu- Thus, although several valid taxa may even- Upper Essequibo, Achamere Wan (ROM tually be recognized among the many names 34514), Ireng Valley (BMNH 3.4.6.10), Kui- currently synonymized with Micoureus de- taro River 40 mi E Dadanawa (ROM 35453), merarae (see below), the Paracou population Weri More (ROM 33201). can be con®dently assigned to this species, FIELD OBSERVATIONS: All of our de®nite and to the nominate race if a trinomial no- records of Micoureus demerarae at Paracou menclature is warranted. are based on collected specimens; of these, Juveniles of Micoureus demerarae some- eight were shot, three were caught in Victor what resemble Marmosa murina in size and traps, and two were caught in elevated plat- external appearance, and the two species form traps. Most (12) of our specimens were might therefore be confused in the ®eld, even taken 1±17 m above the ground on lianas or with specimens in hand. Based on our main trees, but one juvenile was found climbing 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 12 Measurements (mm) and Weights (g) of Adult Micoureus demerarae

among dead branches at ground level in a previous research with many of the same treefall. Five specimens were collected in specimens that we studied. His suggestions well-drained primary forest, one in swampy prompted us to reexamine some of our earlier primary forest, and seven in more-or-less dis- ideas about character variation in the M. turbed habitats (roadside secondary growth brevicaudata complex, and the following ac- and selectively logged forest). count therefore re¯ects his critical input. Our single adult female specimen, collect- Among the many named forms of red- ed on 12 August, was not carrying suckling ¯anked Monodelphis currently synonymized young. with M. brevicaudata (see Gardner, 1993) are several readily diagnosable taxa that we pro- Monodelphis brevicaudata (Erxleben) visionally recognize as full species. As un- Figures 29±31 derstood by us, M. brevicaudata is restricted VOUCHER MATERIAL: AMNH 267000. To- to the Guiana subregion of Amazonia (®g. tal ϭ 1 specimen. 28) and is distinguished from other species IDENTIFICATION: The genus Monodelphis of the brevicaudata complex, all of which has never been revised, and many aspects of are allopatric, by the extension of body fur the currently accepted species-level taxono- onto the proximal one-third or more of the my of these short-tailed opossums (summa- caudal dorsum; the ventral surface of the tail rized by Gardner, 1993) remain untested by is just furred at the base (®g. 29). By con- substantive analyses of specimen data. To de- trast, only the basal one-sixth or less of the termine the correct identi®cation of our sin- tail is furred, to about the same extent above gle Paracou voucher, we examined compar- and below, in M. palliolata (which occurs ative series, types, and original descriptions west of the Orinoco in northern Venezuela of all relevant taxa. Our resulting systematic and northeastern Colombia), M. glirina conclusions broadly overlapped those of the (south of the Amazon and west of the Xin- late C. O. Handley, Jr., who generously gu), and in an unnamed form (a subspecies shared with us the unpublished results of his of M. brevicaudata in the view of C. O. Han- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 55

Fig. 28. Known collection localities for Monodelphis brevicaudata based on specimens examined. 1, BRAZIL, AmapaÂ, Serra do Navio; 2, BRAZIL, Amazonas, Faro; 3, BRAZIL, Amazonas, 80 km N Manaus; 4, BRAZIL, Amazonas, Sto. Antonio de Amatary; 5, BRAZIL, ParaÂ, Cachoeira Porteira; 6, FRENCH GUIANA, Arataye; 7, FRENCH GUIANA, Cacao; 8, FRENCH GUIANA, Cayenne and Montjoly; 9, FRENCH GUIANA, Paracou; 10, FRENCH GUIANA, St.-EugeÁne; 11, FRENCH GUI- ANA, Sophie; 12, FRENCH GUIANA, Tamanoir; 13, GUYANA, Cuyuni-Mazaruni, Bartica Grove; 14, GUYANA, Cuyuni-Mazaruni, First Falls on Cuyuni River; 15, GUYANA, Cuyuni-Mazaruni, Kartabo; 16, GUYANA, Cuyuni-Mazaruni, Kamakusa; 17, GUYANA, Essequibo Islands-West Demerara, Buck Hall; 18, GUYANA, Potaro-Siparuni, Anundabaru; 19, GUYANA, Potaro-Siparuni, Minnehaha Creek; 20, GUYANA, Potaro-Siparuni, Potaro; 21, GUYANA, Upper Demerara-Berbice, Dubulay Ranch; 22, SURINAM, Brokopondo, Brownsberg; 23, SURINAM, Brokopondo, Finisanti; 24, SURINAM, Marow- ijne, Albina; 25, SURINAM, Marowijne, Langamankondre; 26, SURINAM, Marowijne, 10 km N and 24 km W Moengo; 27, SURINAM, Marowijne, Oelemarie; 28, SURINAM, Marowijne, Paloemeu Air- strip; 29, SURINAM, Nickerie, Avanavero; 30, SURINAM, Nickerie, Kayserberg Airstrip; 31, SURI- NAM, Nickerie, King Frederick William Falls; 32, SURINAM, Saramacca, La Poule; 33, SURINAM, Saramacca, Raleigh Falls; 34, SURINAM, Suriname, Paramaribo; 35, SURINAM, Suriname, Jarikaba near Uitkijk; 36, VENEZUELA, Amazonas, Boca Rio Ocamo; 37, Amazonas, Esmeralda and Mt. Duida; 38, VENEZUELA, Amazonas, Serra de Neblina; 39, VENEZUELA, BolõÂvar, Arabupu; 40, VENEZUE- LA, BolõÂvar, Auyantepui; 41, VENEZUELA, BolõÂvar, Caicara; 42, VENEZUELA, BolõÂvar, Ciudad BolõÂvar; 43, VENEZUELA, BolõÂvar, 65 km SSE El Dorado; 44, VENEZUELA, BolõÂvar, Reserva For- estal Imataca. 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 29. Distribution of body pelage on tails of Monodelphis brevicaudata (A, AMNH 93972) and M. palliolata (B, AMNH 69942). In brevicaudata, body fur extends onto the caudal dorsum for one- third or more of the length of the tail, but the caudal ventrum is only furred at the base. In palliolata and other taxa formerly synonymized with brevicaudata, the base of the tail is furred to about the same extent above and below. Both views about ϫ1.2. dley, Jr., personal commun.) distributed south considerable geographic variation in pelage of the Amazon and east of the Xingu. In ad- color, the taxonomic signi®cance of which is dition, whereas M. palliolata and M. glirina dif®cult to evaluate. Possibly, pelage color have orange ventral fur (not sharply differ- may re¯ect species-level divergence that is entiated from the color of the ¯anks) and a not apparent in craniodental characters, but broad cap of grizzled-grayish fur extending the material currently available is insuf®cient across the crown of the head between the to determine whether chromatic variation is eyes, fully adult specimens of M. brevicau- clinal or discontinuous. The following obser- data have whitish, cream, or buffy ventral fur vations are intended to establish which of the (sharply differentiated from the reddish above names is applicable to the coat-color ¯anks) and a narrower cap of grizzled-gray- phenotype represented by our Paracou ish fur that (when present) is con®ned mid- voucher, not to revise the nomenclature of dorsally by a broad band of red above each the entire brevicaudata complex, a task be- eye. Monodelphis palliolata and M. glirina yond the scope of this report. are externally similar, but differ in size (up- Erxleben's (1777) description of Didelphis per molar row Յ7.9 mm in palliolata, Յ7.9 brevicaudata and Schreber's (1778) descrip- mm in glirina). The unnamed form south of tion of Didelphys brachyuros were both the Amazon and east of the Xingu resembles based on Seba's (1734: 50) description and geographically adjacent populations of M. illustration (op. cit.: pl. xxxi, ®g. 6) of ''M- brevicaudata (from AmapaÂ and the eastern uris sylvestris Americani faemina''. Seba's Guianas) in size and coloration but lacks the original specimen (BMNH 67.4.12.540; see dorsal extension of body fur onto the tail. Thomas, 1892) is therefore the type of both Thus restricted, Monodelphis brevicaudata brevicaudata and brachyuros, so these nom- includes the following nominal taxa: brevi- inal taxa are objective synonyms (for the pri- caudata Erxleben (1777), brachyuros Schre- ority of Erxleben's name see Thomas, 1888: ber (1778), touan Shaw (1800), tricolor 356). We examined BMNH 67.4.12.540, an Geoffroy (1803), hunteri Waterhouse (1841), adult female preserved in ¯uid with an ex- orinoci Thomas (1899b), and dorsalis Allen tracted skull. Despite more than two centu- (1904). The material we examined exhibits ries in preservative, this specimen is in re- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 57

markably good condition and the apparently of intermediates between the bicolored and unfaded pelage is distinctly bicolored (red- tricolored phenotypes, all of the Surinamese dish dorsally and abruptly paler ventrally), specimens we examined (including Husson's exactly as described by Seba, Erxleben, and material, obtained on loan from the RMNH) Schreber; there is no middorsal stripe of griz- are tricolored. zled-brownish, -grayish, or -blackish fur. Al- The oldest available name for any tricol- though Matschie (1916) restricted the type ored form of Monodelphis brevicaudata is locality of Monodelphis brevicaudata to Su- Viverra touan Shaw (1800), which was rinam, bicolored specimens resembling the based on Buffon's (1789) description of ``Le type are only known from the inter¯uvial re- Touan'' from Cayenne.9 That the name touan gion between the lower Caroni-Orinoco and properly applies to the tricolored phenotype the lower Mazaruni-Essequibo in northeast- is unambiguously supported both by Buf- ern Venezuela and northwestern Guyana (®g. fon's and Shaw's explicit mention of a black- 28). Because Matschie's restriction was ob- ish middorsal stripe extending from the ros- viously erroneous, we hereby emend the type trum to the base of the tail, and by the fact locality to the vicinity of Kartabo, Cuyuni- that only tricolored animals are known from Mazaruni District, Guyana (locality 15 in ®g. French Guiana. The appropriate trinomial, if 28), from which a well-preserved bicolored one is needed, for our Paracou voucher is specimen (AMNH 48133) closely resem- therefore M. b. touan. Unfortunately, the ap- bling the type was collected by William Bee- plication of the name touan has been a per- be in 1919.8 sistent source of confusion in the literature. By contrast with the limited geographic Thomas (1888) and Cabrera (1919) regarded distribution of bicolored animals, tricolored touan as a synonym of brevicaudata, but Ca- specimens of Monodelphis brevicaudata brera (1958) listed touan (without comment) have been collected throughout French Gui- as a distinct species that included such di- ana, Surinam, Guyana, Guianan Venezuela vergent forms as emiliae, paulensis, and rub- (south and east of the Orinoco), and Guianan ida as subspecies. Current usage (Gardner, Brazil (north of the Amazon and east of the 1993) recognizes M. emiliae, M. rubida, and Rio Negro). In fully adult examples of this M. sorex (including paulensis) as full species coat-color phenotype, a broad middorsal that can be distinguished from M. brevicau- stripe of grizzled-brownish, -grayish, or data by trenchant craniodental characters -blackish fur is sharply set off from the (e.g., those described by Pine and Handley, clear (ungrizzled) reddish ¯anks, which are 1984). separated by a similarly abrupt transition In a recent treatment of Venezuelan Mon- from the pale (whitish, cream, or buffy, but odelphis, Linares (1998) inexplicably re- sometimes partly gray-based) ventral fur. versed the application of touan and brevi- Whereas tricolored skins from Brazil and the caudata by assigning the former name to the Guianas are often brightly colored (with bicolored phenotype and the latter name to blackish or grayish middorsal stripes and tricolored animals (including M. palliolata). cream or whitish venters), most tricolored According to Linares, touan and brevicau- Venezuelan skins have brownish middorsal data are distinguished by craniodental char- stripes and buffy venters that exhibit less acters in addition to pelage color pattern, and chromatic contrast with the reddish ¯anks. occur sympatrically in northeastern Venezue- Despite Husson's (1978) remark that his Sur- la, the Guianas, and in the Brazilian state of inamese specimens represented a full range ParaÂ. However, specimens referable to touan and brevicaudata are not craniodentally dif- 8 Kyk-over-al, a small island opposite Kartabo at the con¯uence of the Cuyuni and Mazaruni Rivers, was the 9 Didelphis tricolor E. Geoffroy is another name pro- seat of government of the Dutch colony of Essequibo posed for ``Le Touan'', but it is based on a different from the early 1600s until 1740 (Beebe, 1925). The en- specimen in the Paris museum than that illustrated and virons of Kartabo (Kartabu Point on recent maps; described by Buffon (Geoffroy, 1803). The type of tri- 6Њ23ЈN, 58Њ41ЈW) are therefore a plausible source of color (MNHN 1990.421), like Buffon's lost specimen, some of the South American material assembled by Al- was apparently collected by M. de la Borde of Cayenne bert Seba (b.1665, d.1736; Engel, 1937) in Amsterdam. (Julien-LaferrieÁre, 1994). 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 30. Dorsal, ventral, and lateral cranial views of FMNH 21720, neotype of Viverra touan Shaw (ϭ Monodelphis brevicaudata). All views approximately ϫ2. ferentiated in our experience, nor have we The neotype, consisting of the skull (®g. 30) seen evidence that two species assignable to and the tricolored skin (®g. 31) of an adult the brevicaudata complex are sympatric any- male, was collected by S. Klages at Cayenne, where. Linares also resurrected Cabrera's French Guiana, on 26 February 1917. (1958) hypothesis that touan and emiliae are Despite the conspicuous geographic vari- conspeci®c, but our observations support ation in pelage color within Monodelphis Pine and Handley's (1984) conclusion that brevicaudata noted above, we are not per- emiliae is a distinctive species with no spe- suaded of the necessity for a formal trinomial cial similarity to touan or to other members nomenclature. Although the brevicaudata of the brevicaudata complex. and touan phenotypes are clearly distinct in In order to de®nitively resolve these con- Guyana, some bicolored Venezuelan speci- ¯icting usages, we hereby designate FMNH mens from NE BolõÂvar (e.g., EBRG 17536, 21720 as the neotype of Viverra touan Shaw. USNM 385004, 385005) have indistinctly 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 59

Fig. 31. Dorsal, lateral, and ventral views of the skin of FMNH 21720, neotype of Viverra touan Shaw (ϭ Monodelphis brevicaudata). All views approximately life-size. grizzled middorsal fur, somewhat resembling as evidence that bicolored and tricolored the middorsal pigmentation of drab-tricolor populations intergrade clinally in Guianan skins from Amazonas and southeastern Bo- Venezuela. Furthermore, although our sam- lõÂvar, an observation that could be interpreted ples of measurable adults are too small for 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 13 Measurements (mm) of Adult Monodelphis brevicaudataa

con®dent statistical inference (table 13), no 93972±93974, 94161, 94221), 80 km N Ma- morphometric differences are apparent be- naus (USNM 579976±579979), Santo Anto- tween bicolored and tricolored animals to nio da Amatary (AMNH 92879); ParaÂ, suggest that these are anything more than lo- Cachoeira Porteira (USNM 546209± cal coat-color variants. Larger series of spec- 546219), ``Serra do Tumucumaque'' (USNM imens, especially from western Guyana and 392044±392049). French GuianaÐArataye eastern Venezuela, together with molecular (USNM 578009), Cacao (MNHN 1981.168, data would be helpful in any future effort to 1981.412, 1981.414±1981.416, 1982.599), evaluate the taxonomic signi®cance of pelage Cayenne (FMNH 21720 [neotype of touan], color variation in this species. MNHN 1990.421 [holotype of tricolor]), BICOLORED SPECIMENS EXAMINED: Guya- Montjoly (MNHN 1994.122), St.-EugeÁne naÐCuyuni-Mazaruni, First Falls on the Cu- (MNHN 1995.205, 1995.3216), Sophie yuni River (BMNH 34.6.30.65), Kartabo (MNHN 1966.1, 1966.2), Tamanoir (FMNH (AMNH 48133); Essequibo Islands-West 21793), Trois Sauts (MNHN 1981.413). Demerara, Buck Hall (BMNH 17.7.7.1). GuyanaÐ``River Supinaam'' (BMNH VenezuelaÐBolõÂvar, 65 km SSE El Dorado 10.9.29.26, 10.9.29.27); Cuyuni-Mazaruni, (USNM 385005), 56 km SE El Manteco Bartica Grove (BMNH 10.11.10.20), Ka- (USNM 385004), Reserva Forestal Imataca makusa (AMNH 140465, 140466); Potaro- (EBRG 17536). Without locality dataÐ Siparuni, Anundabaru (AMNH 75830, (BMNH 67.4.12.540 [holotype of brevicau- 75831), Minnehaha Creek (AMNH 36317), data], 88.1.31.1 [holotype of hunteri]). Potaro (BMNH 12.6.9.9); Upper Demerara- TRICOLORED SPECIMENS EXAMINED: Bra- Berbice, Dubulay Ranch (AMNH 267744, zilÐAmapaÂ, Serra do Navio (USNM 268060, 268061). SurinamÐBrokopondo, 392050, 392051, 393424, 393428, 393430, Brownsberg (CM 52729, RMNH 23403, 393435, 393436, 393438, 393439, 393441, 23404), Finisanti (FMNH 95338); Marowi- 461434, 461435); Amazonas, Faro (AMNH jne, 3 km SW Albina (CM 76730), Langa- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 61

Fig. 32. Capture site of Monodelphis brevicaudata at Paracou. Our single voucher specimen was trapped in the dark cave-like space (white arrow) beneath this fallen tree in well-drained primary forest. Such cavities, carpeted and sheltered by decaying wood, are typical microhabitats for Monodelphis species but not for other Neotropical rainforest marsupials. All of the other nonvolant mammals captured under fallen trees at Paracou were rodents, including Neacomys paracou, Oryzomys megacephalus, O. yunganus, and Proechimys cuvieri.

mankondre (RMNH 18227), 10 km N and 24 75681±75687), Auyantepui (AMNH 130516, km W Moengo (CM 52730), Oelemarie (CM 130560±130565, 130573±130576, 130727), 76731, 76732), Paloemeu Airstrip (FMNH Caicara (BMNH 98.12.1.22 [holotype of or- 94018, 94019); Nickerie, Avanavero (CM inoci]), Ciudad BolõÂvar (AMNH 16124± 68358), Kayserberg Airstrip (CM 68359), 16126 [type series of dorsalis]. King Frederick William Falls (FMNH FIELD OBSERVATIONS: The single example 48416); Saramacca, La Poule (FMNH of Monodelphis brevicaudata that we col- 95339), Raleigh Falls (CM 63510, 63511, lected at Paracou is a partially eaten speci- 68361); Suriname, Cultuurtuin near Para- men caught in a Victor trap set under a fallen maribo (RMNH 18076), Jarikaba near Uit- tree (®g. 32) in well-drained primary forest. kijk (RMNH 20672), Plantage De Morgen- Several additional specimens were previous- stond near Paramaribo (RMNH 17223), ly taken by O. Henry in the course of his Plantation Clevia near Paramaribo (RMNH multiyear trapping study (G. Dubost, person- 21654). VenezuelaÐAmazonas, Boca RõÂo al commun.), but we have not examined his Ocamo (AMNH 78093±78095), Esmeralda material. (AMNH 77281, 77282, 77287, 77288), Mt. Duida (AMNH 77283±77285, 77289, Philander opossum (Linnaeus) 77290±77296), ``RõÂo Casiquiare'' (AMNH 77286, 78096±78100), Serra de Neblina VOUCHER MATERIAL: AMNH 266379± (AMNH 244469); BolõÂvar, Arabupu (AMNH 266381, 266383±266387, 266389±266391, 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 14 Measurements (mm) and Weights (g) of Adult Philander opossum from French Guiana and Surinama

266394, 266395, 266398, 266400, 266994, back in all adult and subadult Paracou skins, 266995, 266997±266999, 267014, 267328; even the aforementioned example with dark- MNHN 1995.915±1995.924. Total ϭ 32 er middorsal fur. (3) Husson (1978: 26) de- specimens. scribed the ears as ``whitish with a broad IDENTIFICATION: The species-level taxono- black rim'', but the ears of our Paracou spec- my of Philander was recently reviewed by imens are only whitish at the baseÐmost of Hershkovitz (1997) and by Patton and da Sil- the pinna is black. (4) Husson's juvenile va (1997), whose taxonomic conclusions dif- specimens were described as darker than fer considerably. Both studies, however, re- adults and with less distinct facial markings, ferred Guianan populations of gray four-eyed but three juvenile skins from Paracou opossums to P. opossum, the type locality of (AMNH 266385, 266391, 266997) are com- which (as restricted by Matschie, 1916) is parable to those of adults and subadults ex- Paramaribo, Surinam. Our Paracou vouchers cept in pelage texture (the juvenile fur is agree closely with Husson's (1978) descrip- softer). tion of topotypic specimens of P. opossum Although most of the Surinamese material in most details, but a few discrepancies merit we borrowed for side-by-side comparison are comment. (1) Whereas Husson described the darker middorsally than on the sides, as Hus- pelage of his topotypes as usually distinctly son described, a few (e.g., CM 52735) are darker middorsally than on the sides, only uniformly gray like our Paracou vouchers. one skin from Paracou (AMNH 266995) has The other external differences implied by fur that is slightly darker middorsally than on Husson's description are not apparent in the the sides; the rest of the adult and subadult Surinamese material we examined.10 Cranial skins from Paracou are a uniform grizzled- measurements of adult females from Paracou gray over the entire dorsum. (2) Husson (1978: 25) described the fur of the head as 10 Some color variation among the series at hand may ``of the same dark blackish-brown colour as be artifactual. For example, the blackish color of the crown of the head probably fades with age (becoming the median part of the back or even slightly dusky-brown on old skins), and dark-tipped dorsal hairs darker'', but the blackish fur of the head con- may be more concentrated in the midline of ¯at skins trasts conspicuously with the grizzled-gray than on round skins. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 63

and Surinam overlap broadly (table 14), but currently treated as subspecies of P. opossum our few adult males seem exceptionally (e.g., canus and fuscogriseus) might also be large. We did not see any consistent quali- recognized as full species. However, many tative craniodental differences between Sur- nominal taxa were not represented in Patton inamese and Paracou specimens. and da Silva's molecular analyses, and sev- Patton and da Silva (1997: 98) character- eral important issues of species-level synon- ized the phenotype of Philander opossum as ymy remain unresolved. Therefore, the geo- ``uniformly pale-gray'', a description that graphic limits of P. opossum are still uncertain. better ®ts our Paracou material than it does REMARKS: We agree with Hershkovitz the Surinamese topotypes we examined. (1976, 1981) that Philander, not Metachi- However, Patton and da Silva's context for rops, is the correct name for pouched four- taxonomic comparisons included the almost eyed opossums (contra Husson [1978] and entirely black P. mcilhennyi, as well as the other authors). Linnaeus's (1758) original de- black-striped P. andersoni, both of which scription of Didelphis opossum was based on contrast strikingly with predominantly gray- an adult male and an adult female described ish animals that those authors referred to P. and ®gured by Seba (1734). Hershkovitz opossum and P. frenata. (1976: 297) designated Seba's female as the According to Hershkovitz (1997: 39), both lectotype, but the specimen itself was appar- gray and ``brown'' color phases occur ently not then known to have survived the throughout the range of P. opossum. We ex- breakup and dispersion of Seba's museum amined the Surinamese specimens that he (for the history of which, see Boeseman, cited as coat-color exemplars, however, and 1970). Hershkovitz (1997) subsequently stat- did not observe any consistent differences ed that the lectotype is still preserved as an between gray and ``brown'' individuals (e.g., alcoholic specimen in the Rijksmuseum van between FMNH 95312 and 95313, both col- Naturlijke Historie in Leiden.11 According to lected by H. A. Beatty in the Wilhelmina the current RMNH curator of mammals (C. Mountains). Yellowish or brownish tints do Smeenk, personal commun.), the lectotype is occur in some Philander skins, but whether RMNH 25421, clearly recognizable as the these represent true coat-color variants rather adult female with three pouch young illus- than preservational artifacts (e.g., staining by trated and described by Seba. sebaceous secretions or subcutaneous fat) is OTHER SPECIMENS EXAMINED: SurinamÐ dif®cult to determine with the material at hand. Coronie, Totness (CM 52731, 52733); Ma- Both Patton and da Silva (1997) and rowijne, Moengo (CM 52734), Perica (CM Hershkovitz (1997) referred French Guianan 76736); Nickerie, Avanavero (CM 68363), populations of Philander opossum to the Kayserberg Airstrip (FMNH 93168), Sipali- nominate form P. o. opossum, a usage con- wini Airstrip (CM 63517, 76739), Wilhemi- sistent with the results of our comparisons. na Mountains (FMNH 95312, 95313); Para, According to these authors, the nominate Zanderij (CM 68365, 76742); Saramacca, race occurs throughout the Guianas and the Bigi Poika (CM 52735), La Poule (FMNH eastern Amazon basin of Brazil, and differs 95309); Suriname, Clevia (FMNH 95310), from other subspecies principally by geo- Lelydorpplan (FMNH 95308), Plantation graphic variation in size and pelage color. Clevia (CM 76743, 76744), Powakka (CM However, whereas Patton and da Silva rec- 52741, 52742). ognized four species of Philander (see FIELD OBSERVATIONS: In addition to data above), Hershkovitz recognized only P. accompanying our 32 voucher specimens, we opossum (including frenata among other recorded 18 unvouchered observations of subspecies) and P. andersoni (including mcilhennyi as a subspecies). We follow Pat- 11 Hershkovitz (1997) gave no catalog number for the ton and da Silva in recognizing the gray four- lectotype, and the caption to his (op. cit.) ®gure 21 con- eyed opossum of the Brazilian Atlantic forest fusingly refers to the ``Male and female lectotypes [sic] of Didelphis opossum Linnaeus (1758)''. The caption is as a distinct species, P. frenata, and we in- an obvious lapsus that does not affect the validity of his terpret their phylogenetic analysis of mtDNA earlier (1976) selection of the female as lectotype. Seba's sequences as evidence that additional taxa male specimen is therefore a paralectotype. 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Philander opossum at Paracou, for a total of Choloepus didactylus (Linnaeus) 50 documented records of this species. Of our vouchers, 21 (66%) were shot, 5 (16%) VOUCHER MATERIAL: AMNH 265952, were taken in Conibear traps, 3 (9%) were MNHN 1995.952. Total ϭ 2 specimens. taken in Sherman live traps, 1 was taken in IDENTIFICATION: Although the species-level a Tomahawk wire live trap, and 1 was taken taxonomy of two-toed sloths, genus Chol- in a Victor snap trap, and 1 was caught by oepus, has never been formally revised, two hand. One juvenile captured in a pitfall trap species are consistently recognized in mod- and another taken in a Sherman trap were ern synoptic treatments of edentate classi®- Â released. Of all 50 records, 31 (62%) were cation (e.g., Wetzel and Avila-Pires, 1980; of individuals shot, trapped, or sighted on the Wetzel, 1982, 1985); of these, only C. di- ground, whereas 19 (38%) were of individ- dactylus is known to occur in the Guiana uals shot, trapped, caught by hand, or sighted subregion of Amazonia. Our voucher mate- in trees (or on other elevated substrates such rial conforms with the brief description of as inclined trunks, lianas, etc.) from 1.5 to 6 topotypic specimens of C. didactylus from m above the ground. With the exception of Surinam provided by Husson (1978), and females carrying nursing young, all P. opos- with the cranial diagnosis of this species pro- sum that we encountered were solitary, and vided by Wetzel (1985). However, the fur on all were collected or sighted at night. Of 49 the throat of our adult female specimen records accompanied by habitat data, 32 (AMNH 265952) is distinctly paler and (65%) were of individuals encountered in shorter than the pectoral fur, a pelage trait primary forest, usually near streams or in that Wetzel regarded as diagnostic of C. hoff- swamps, but occasionally at well-drained manni (from western Amazonia and Central sites; 17 observations (35%), however, were America). The throat fur of our juvenile male based on individuals encountered in roadside (MNHN 1995.952) is likewise paler than the secondary growth or other more-or-less dis- chest fur, but the contrast is less marked than turbed habitats. in the adult female. For comparison with One female, collected on 9 July 1991, had Surinamese topotypes measured by Husson two pouch young measuring 49 mm crown- (1978: table 38), external measurements of rump; another, collected on 24 October 1992, our adult female voucher were 698 ϫ 12 ϫ had three pouch young measuring 24 mm; 164 ϫ 29 mm, and it weighed 7.3 kg. Se- and a third, collected on 11 August 1993, had lected cranial measurements of this specimen four pouch young measuring 27 mm. are: condylobasal length (``greatest length of skull''), 119.1 mm; zygomatic breadth, 73.0 XENARTHRA mm; interorbital constriction, 36.3 mm; alveolar length of maxillary toothrow, 44.5 Nine xenarthran species have been de®- mm. The ratio of the minimal to the maximal nitely recorded from Paracou and it is un- interpterygoid width (after Wetzel, 1985) in likely that any others occur locally (see ap- this specimen is 0.49. pendix 1). Because most xenarthran species FIELD OBSERVATIONS: Two-toed sloths are are easily recognized by external characters probably common at Paracou. P. Petronelli (Husson, 1978; Emmons, 1990, 1997), we (personal commun., 1993) estimated that the collected few vouchers. species is sighted by forestry workers about ®ve or six times a year in our study area, a Bradypus tridactylus Linnaeus reasonably high frequency given its cryptic Three-toed sloths are possibly common at appearance and inconspicuous habits. Of ®ve Paracou, but they are seldom seen and we sightings by inventory personnel from 1991 did not encounter any in the course of our to 1994, four were nocturnal and one was ®eldwork. P. Petronelli (personal commun., diurnal: (1) RSV found a lactating female 1993) estimated that the species is sighted and a juvenile male (the vouchers described about once or twice a year at Paracou, much above) hanging together motionless in sub- less frequently than two-toed sloths (see be- canopy vegetation (ca. 20 m above the low). ground) in well-drained primary forest at 20: 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 65

15 hours on 30 July 1991. (2) RSV found a counts and cranial measurements follow large adult of undetermined sex descending Wetzel and Mondol®'s conventions). a tree head-®rst (probably to defecate) in FIELD OBSERVATIONS: Our single voucher well-drained primary forest at ca. 23:00 of Dasypus kappleri and two other individ- hours on 6 August 1991; the animal was uals (unambiguously identi®ed but not col- about 2 m above the ground and remained lected) were all encountered at night, forag- motionless throughout the encounter (about ing singly on the ground in primary forest. 10 min), even when touched; its eyeshine Several other armadillo encounters recorded was very faint. (3) RSV saw another solitary in our ®eldnotes might have been of this spe- adult of unknown sex sleeping suspended cies, but positive identi®cation requires a from a liana ca. 4 m above the ground in old clear and reasonably close view (to accurate- secondary growth at 02:00 hours on 31 July ly judge size or to see the enlarged scutes on 1993. (4) L. H. Emmons saw a solitary ani- the knee), and many animals were only seen mal of unknown sex climbing rapidly about ¯eetingly or at a distance. Because we doubt- 20 m above the ground in well-drained pri- ed that the forestry workers and local hunters mary forest at 10:15 hours on 24 September with whom we spoke reliably distinguished 1994. (5) L. H. Emmons found an adult of this armadillo from the smaller but otherwise unknown sex hanging motionless under a externally similar D. novemcinctus, we did branch about 20 m above a stream in primary not collect second-hand information about forest at 20:21 hours on 5 October 1994. Dasypus species.

Cabassous unicinctus (Linnaeus) Dasypus novemcinctus Linnaeus

The only record of this elusive armadillo VOUCHER MATERIAL: AMNH 266483, at Paracou is an infant specimen (with un- 267012; MNHN 1995.953. Total ϭ 3 speci- opened eyes) excavated by a bulldozer when mens. a tract of well-drained primary forest was be- IDENTIFICATION: Our three vouchers con- ing cleared for a new experimental planta- form in all qualitative external and cranial tion; forestry workers brought the animal to characters with the descriptions of this spe- P. Petronelli (personal commun., 1993), who cies by Husson (1978) and Wetzel and Mon- showed us the photographs he had taken of dol® (1979). For comparison with Wetzel it. and Mondol®'s summary of quantitative data from Dasypus novemcinctus (op. cit.: table Dasypus kappleri Krauss 1), the external measurements of our only adult specimen (AMNH 267012, female) VOUCHER MATERIAL: AMNH 267011. To- were 506 ϫ 380 ϫ 95 ϫ 51 mm, and it tal ϭ 1 specimen. weighed 4.8 kg. The shell of AMNH 267012 IDENTIFICATION: Our single voucher con- was not preserved, but the other two speci- forms exactly in qualitative external and cra- mens (both subadults with unfused basicranial characters to Husson's (1978) and Wetz- nial sutures and incompletely erupted denti- el and Mondol®'s (1979) descriptions of this tions) each have nine movable bands, of species, the type locality of which is in Su- which the fourth has 57 (MNHN 1995.953) rinam. For comparison with quantitative data or 60 (AMNH 266483) scutes. The condy- summarized by those authors, the external lonasal length of the skull of AMNH 267012 measurements of our adult male voucher is 102.4 mm, the zygomatic width 44.8 mm, were 565 ϫ 405 ϫ 125 ϫ 53 mm, and it and the mastoidal width 29.0 mm; this spec- weighed 10.6 kg. The carapace of this specimen has seven paired maxillary teeth and imen has eight movable bands, of which the eight paired mandibular teeth. fourth has 58 scutes; the condylonasal length FIELD OBSERVATIONS: We heard armadillos of the skull is 128.6 mm, the zygomatic crashing through the undergrowth or caught width 53.2 mm, and the mastoidal width 35.3 brief glimpses of them as they ¯ed almost mm; there are eight paired maxillary teeth every night; most were probably Dasypus and eight paired mandibular teeth (meristic novemcinctus (usually the commonest rain- 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 forest armadillo throughout its extensive geo- residence at Paracou, these scant observa- graphic range; Emmons, 1990, 1997), but it tions suggest that this large and conspicuous was impossible to make certain identi®ca- diurnal species is very uncommon. tions in such cases. Our three vouchers were all shot at night in primary forest, at both Tamandua tetradactyla (Linnaeus) well-drained and swampy sites. Numerous unambiguous sightings (many of which were Although we encountered this species only not recorded) indicate that this species is in 1992, when what was perhaps a single in- common in all local habitats including pri- dividual was heard twice (tearing apart dead mary forest and secondary growth. All of our wood or termite nests in trees at night) and observations were of solitary individuals. Al- sighted once (walking down a road at 07:20 though most sightings were nocturnal, we hours) during a two-day interval, tamanduas occasionally encountered nine-banded arma- are apparently not uncommon locally. P. Pe- dillos in the late afternoon, usually an hour tronelli (personal commun., 1993) estimated or less before dusk. that they are typically seen about ®ve or six times a year by local forestry workers. Priodontes maximus (Kerr) Husson (1978) referred the tamanduas of the Guiana subregion to Tamandua longicau- We only saw the giant armadillo once, on data (Wagner), but we follow Wetzel's the night of 27 October 1994, when A. L. (1975) revision of Tamandua in regarding Pef¯ey and RSV encountered a large (ca. 1 longicaudata as a junior synonym of tetra- m HBL) adult of unknown sex excavating a dactyla. Wetzel's revision should be consult- mound of dead wood beside a rotting log in ed for diagnostic characters, as well as for a primary forest near a small stream at ca. 20: discussion of the considerable geographic 00 hours. Alarmed by our headlights, it ¯ed and nongeographic variation in coat color of uphill toward a large treefall, where we were this species. Although Wetzel's map (op. cit.: unable to follow in the dense undergrowth. ®g. 1) indicates that only partially vested, un- Bushnegro forestry workers have sometimes vested, or melanistic specimens are known pointed out burrows and excavations said to from French Guiana, our single sighting (by be made by this species to P. Petronelli (per- DPL in broad daylight at a distance of ca. 10 sonal commun., 1993); according to them, m) was of a yellow individual with a dis- collared peccaries (Pecari tajacu) use the tinctly blackish vest. abandoned burrows of giant armadillos as nocturnal retreats. Only one individual is PRIMATES known to have been shot by local hunters, about ten years before our inventory ®eld- Six species of primates are de®nitely work began. known to occur at Paracou, or to have occurred there in the recent past, and a seventh Cyclopes didactylus (Linnaeus) species could be expected (see appendix 1). The only records of the pygmy anteater at Unfortunately, the local primate fauna has Paracou are two sightings by P. Petronelli been decimated by uncontrolled hunting, a (personal commun., 1993), each of a single process that accelerated following comple- animal of unkown sex crossing a dirt road tion of the new asphalt highway from Kour- through our study area. ou to Sinnamary in 1992. Even at the begin- ing of our ®eldwork in 1991, however, spider monkeys (Ateles paniscus), sakis (Pithecia Myrmecophaga tridactyla Linnaeus pithecia), and capuchins (Cebus sp.) were Four sightings of giant anteaters by local rare. By 1994 (our last ®eld season), only forestry workers were reported to us (P. Pe- howlers (Alouatta seniculus) and tamarins tronelli, personal commun., 1993). One ani- (Saguinus midas) were commonly heard or mal was seen in savanna vegetation, the oth- seen in the vicinity of our camp, although er three in the forest; all were seen in the capuchins could still be found a few kilo- daytime. Covering about ten years of human meters away. Squirrel monkeys (Saimiri sci- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 67

ureus) have probably always been local va- eastern Amazonia should be called S. midas grants, not regular residents. niger (E. Geoffroy). Apparently, the only Thus, we never had the opportunity to cen- published justi®cation for treating these un- sus an intact primate community at Paracou, equivocally diagnosable taxa as conspeci®c and some of the observations cited below is the following (Hershkovitz, 1977: 207): were necessarily recorded at second hand. Fortunately, all Guianan primates are easily The color of the cheiridia dictates the proffered hypothesis of racial differentiation. The pheomelanic or identi®ed by obvious external characters eumelanic cheiridia can be derived directly from the (Emmons, 1990, 1997), so the likelihood of primitive agouti colored cheiridia . . . and either of mistakes in the second-hand identi®cations the saturate patterns can switch to the other. Further- we cite is remote. more, presence of callitrichids on the Ilha de MarajoÂ . . . discounts the probability of one race arising from the stock of another. It remains to be determined if Saguinus midas (Linnaeus) tamarins with agouti cheiridia still persist on any of the innumerable islands of the lower Amazon. VOUCHER MATERIAL: AMNH 266481, 266482; MNHN 1998.699. Total ϭ 3 speci- Apparently, Hershkovitz judged the chro- mens. matic differences between midas (sensu stric- IDENTIFICATION: Our three vouchers corre- to) and niger to be evolutionarily labile and spond exactly with Husson's (1978) detailed predicted the existence of an extinct (or un- description of this species, which was based discovered) form that was (or is) intermedi- on topotypic material from Surinam. In par- ate in coloration and geography. However, ticular, the diagnostic external markings of the constancy of tamarin markings on op- golden-handed tamarinsÐbright orange posite sides of the Amazon suggests that the (sometimes reddish or yellow) hands and feet character transformation in question is not that contrast with the blackish limbsÐare evolutionarily labile, nor have populations conspicuous in our vouchers as they are in with intermediate phenotypes yet been re- all specimens referred to this taxon through- ported from any Amazonian islands. out the Guiana subregion of Amazonia According to Hershkovitz, only the col- (Hershkovitz, 1977). For comparison with oration of the hands and feet distinguishes published measurement data (Husson, 1957, midas from niger, but his monograph con- 1978; Napier, 1976; Hershkovitz, 1977), se- tains no explicit comparison of these taxa in lected external and craniodental dimensions nonpelage characters. By contrast, subse- (mm) of our two adult female vouchers are: quent research has shown that midas and ni- head-and-body length 258, 285; length of tail ger have divergent dental measurements (Na- 425, 440; length of hindfoot 77, 80; ear 36, tori and Hanihara, 1992) and ␤2-microglob- 37; condylobasal length 38.9, 41.2; orbital ulin DNA sequences (Canavez et al., 1999). breadth 28.7, 29.1; postorbital constriction Indeed, parsimony analysis of the ␤2-microg- 24.3, 25.6; zygomatic breadth 33.3, 34.8; lobulin data provides compelling evidence maxillary toothrow (crown length C±M2) that midas is more closely related to another 12.8, 12.9. species that occurs north of the Amazon, S. The black-handed tamarins that occur bicolor (Spix), than it is to niger on the op- south of the Amazon and east of the Xingu posite bank (Canavez et al., 1999). (including Ilha de MarajoÂ) were long consid- In view of (1) the diagnosability of gold- ered to be a distinct species from the golden- en- and black-handed tamarins by bold and handed tamarins of the Guiana subregion consistent pelage markings, (2) the existence (e.g., by Elliot, 1912; Cruz Lima, 1945; Hill, of correlated divergence in nonpelage char- 1957; Cabrera, 1961). Hershkovitz (1977), acters, and (3) clear indications from phylo- however, treated golden- and black-handed genetic analysis that these taxa are not sister tamarins as no more than subspeci®cally dis- taxa, the currently accepted use of midas and tinct; according to this authority, the correct niger as subspecies (Hershkovitz, 1977) or name for the golden-handed Guianan tama- synonyms (Groves, 1993) is not defensible. rins is Saguinus midas midas (Linnaeus), Instead, we recognize the golden-handed whereas the black-handed tamarins of south- tamarins of the Guiana subregion, Saguinus 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 midas, as a species distinct from the black- groups were seen crossing dirt roads on the handed tamarins of southeastern Amazonia ground when the gap between trees on either historically known by authors as S. niger, S. side was too wide to leap. Unlike other pri- tamarin, or S. ursulus (see below). mate species at Paracou, tamarins did not no- REMARKS: Whereas most early authors ticeably decline in density from 1991 to used the epithets tamarin Link or ursulus 1994, probably because they are not locally Hoffmannsegg for the black-handed tamarin hunted for meat. of southeastern Amazonia, Hershkovitz (1977) argued that the oldest applicable Alouatta seniculus (Linnaeus) name for the zoological taxon in question is niger E. Geoffroy. Confusingly, the holotype We heard howlers night and day through- of niger was clearly stated to have come out the course of our ®eldwork at Paracou. from Cayenne (Geoffroy, 1803), far from the Despite the audible evidence of their contin- known range of black-handed tamarins. Al- uous presence, however, we seldom caught though Hershkovitz reassigned the type lo- more than ¯eeting glimpses of this locally cality to BeleÂm, he did not examine Geof- persecuted species. The few groups surviv- froy's specimen, the identity of which is ob- ing in our study area never closely ap- viously problematic. Unfortunately, the ho- proached our camp, and they seldom ven- lotype of niger (No. XXIV in Geoffroy's tured within the radius of our daily activities catalog) is lost: it was not listed in Rode's except to visit fruiting ®g trees and other (1938) catalog of MNHN primate types, and transient resources. it is not part of the current Paris museum collection (M. Tranier, personal commun.). Ateles paniscus (Linnaeus) Whether the original specimen of Geoffroy's Spider monkeys, always vulnerable to lo- niger was a melanistic individual of the gold- cal extirpation by hunters because of their en-handed species collected at Cayenne or loud vocalizations, large size, highly prized was a mislabelled example of the black- meat, and low reproductive rate, have been handed species is now impossible to deter- scarce at Paracou for the last decade or more. mine. Nevertheless, the black-handed species P. Petronelli (personal commun., 1993) is now widely and consistently known by the guessed that he had seen solitary individuals, epithet niger E. Geoffroy, a usage that should never groups, on only ®ve or six occasions be preserved in the interest of taxonomic sta- since the early 1980s. We recorded only a bility. For that purpose, we hereby designate single encounter with this species, when DPL as neotype of Sagouin niger E. Geoffroy, saw a solitary individual in the canopy of 1803, an adult male specimen represented by well-drained primary forest near a fruiting ®g a well-preserved skin and skull in the Amer- tree on 2 July 1991; no noncaptive spider ican Museum of Natural History, AMNH monkeys were seen or heard by us in later 96500, collected by A. M. Olalla on 2 No- years. vember 1931 at CametaÂ on the Rio Tocan- tins, ParaÂ, Brazil, from which locality a large Cebus apella (Linnaeus) series of topotypes is also available. OTHER SPECIMENS EXAMINED: GuyanaÐ One or two species of Cebus were com- Cuyuni-Mazaruni, Kartabo (AMNH 65159, mon at Paracou in the early 1980s (P. Petro- 142936). nelli, personal commun., 1993), but capu- FIELD OBSERVATIONS: This is the common- chins are now rarely seen or heard in the est primate species at Paracou. We saw area. The older forestry workers are familiar groups of tamarins daily, in both swampy with both a large and a small species, pre- and well-drained primary forest and in road- sumably C. apella and C. olivaceus (see ap- side secondary growth. In primary forest, pendix 1), but only C. apella has been de®- tamarins were invariably sighted in the can- nitely identi®ed by sight at Paracou (P.-M. opy or subcanopy, but we often saw them Forget, personal commun., 1994). We heard descend to within a few meters of the ground Cebus on several occasions in 1994 when in roadside secondary growth; occasionally, our inventory activities extended ca. 3 km to 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 69

the NNW of camp (the limit of our sampling Speothos venaticus (Lund) radius), but we caught no more than a brief We did not encounter bush dogs during glimpse of the animals as they ¯ed in the our 1991±1994 ®eldwork at Paracou, but P. distance and were not able to determine the Petronelli (personal commun., 1993) told us species. he had seen them twice in previous years: once as a group of four individuals, and an- Pithecia pithecia (Linnaeus) other time as a group of three. A visiting photographer also saw a pack of seven bush Although saki monkeys may once have dogs chase a paca across a road through the been common at Paracou, P. Petronelli (per- forest near our camp. All of these sightings sonal commun., 1993) recalled only one were diurnal. sighting of this species prior to our inventory: a pair that he observed on 31 January Herpailurus yaguarondi (LaceÂpeÁde) 1991. Subsequently, we recorded three observations of Pithecia pithecia in our ®eld- We did not see jaguarundis at Paracou, but notes: (1) DPL saw a solitary individual on P. Petronelli (personal commun., 1993) re- 3 July 1991; (2) RSV saw what might have ported three sightings, all of solitary individ- been the same animal in the subcanopy of uals on the ground in the daytime: one was well-drained primary forest at 07:15 hours crossing a road, one was near a stream in on 4 July 1991; and (3) Roland W. Kays saw primary forest, and another was in secondary two individuals leaping from tree to tree as growth near the edge of a small patch of sa- he was walking along a dirt road at 16:30 vanna vegetation. hours on 6 August 1993. Silent and retiring by disposition, sakis may still linger in re- Leopardus pardalis (Linnaeus) mote and seldom-visited parts of our study Although ocelots are probably not uncom- area, but their numbers have certainly been mon at Paracou, they are rarely seen. We saw much reduced by hunting. none, but local hunters have killed at least three in the last decade, one of which had the Saimiri sciureus (Linnaeus) distinctively banded quills of Coendou prehensilis embedded in its neck and shoulders Squirrel monkeys have apparently been (P. Petronelli, personal commun., 1993). sighted only once at Paracou, when P. Petro- nelli (personal commun., 1993) encountered Leopardus wiedii (Schinz) a group of 10±12 individuals on the edge of Small spotted cats are often seen at night a small patch of savanna vegetation in the by local forestry workers (P. Petronelli, per- early 1980s. This species is probably a local sonal commun., 1993), and we recorded sev- vagrant that seldom strays far from the tan- eral ¯eeting encounters in our ®eldnotes, but gled growth at the forested margins of nearby the identi®cation of such observations is savannas and rivers. problematic. We have only two de®nite records of margays from our study area. (1) In CARNIVORA 1993 we examined and measured an adult female, shot by a local hunter, that measured Ten species of carnivores are de®nitely 563 ϫ 427 ϫ 124 ϫ 54 mm and weighed known to occur at Paracou, and it is doubtful 3.4 kg; the fur of the nape of the neck was that any others occur in our study area except reversed on this specimen, which likewise as rare vagrants (see appendix 1). Because corresponded in other external characters to all rainforest carnivores can be con®dently the descriptions provided by Pocock (1941), identi®ed by external characters (Emmons, Emmons (1990, 1997), and Oliveira (1998). 1990, 1997), and because many are uncom- (2) On 19 September 1994, L. H. Emmons mon and/or elusive, most of the information observed an emaciated male at a distance of reported below was obtained by interviewing only 6 m from 19:20 to 19:35 hours; the an- local forestry personnel. imal was encountered near a stream in pri- 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 mary forest, and half of its face was bristling Nasua nasua (Linnaeus) with the large white quills of Coendou prehensilis. VOUCHER MATERIAL: AMNH 267605; MNHN 1995.959. Total ϭ 2 specimens. IDENTIFICATION: Our two vouchers, both Panthera onca (Linnaeus) juveniles, conform closely in pelage charac- In the course of ten years' residence at ters with the description of Surinamese spec- Paracou, P. Petronelli (personal commun., imens identi®ed by Husson (1978) as Nasua 1993) told us that he had seen jaguars four nasua vittata Tschudi (1845), the type local- times: twice in the daytime and twice at ity of which is in Guyana. Unfortunately, the night, all in primary forest. Although we saw appropriate trinomial designation for coatis the distant eyeshine of large cats on several from the Guiana subregion of Amazonia, if occasions, our only de®nite record of this indeed a subspeci®c classi®cation is neces- species is based on the pugs of a young adult sary, remains to be convincingly determined. that L. H. Emmons observed along a dirt Tate (1939) identi®ed the lowland coatis road on 13 October 1994. of the Guiana subregion as Nasua phaeocephala J. A. Allen (1904)Ðapparently over- Puma concolor (Linnaeus) looking the availability of Tschudi's older nameÐand proposed a new name, dichro- We did not see this species at Paracou in matica, for the montane population on Au- the course of our ®eldwork, but P. Petronelli yantepui in Venezuela. Subsequently, Ca- (personal commun., 1993) told us he had en- brera (1958) referred all named forms of countered pumas locally on four occasions, South American coatis to N. nasua, of which twice at night and twice in the daytime. One 11 subspecies were recognized as valid; in of the diurnal sightings was of an animal his classi®cation, phaeocephala and dichro- asleep on a high tree limb; another individual matica were both treated as subjective syn- was found asleep between the buttresses of a onyms of N. n. vittata. According to Hersh- big tree. kovitz (1959), however, the oldest valid name for a Guianan coati is Viverra quasje Eira barbara (Linnaeus) Gmelin (1788), a name said to be based pri- Although tayras are common at Paracou, marily on a Surinamese specimen described where they are sighted on average about 10± and illustrated by Seba (1734). Although 12 times a year by forestry workers (P. Pe- Seba's specimen (BMNH 67.4.12.447), the tronelli, personal commun., 1993), we saw presumptive type of quasje, was reported to only three in the course of our ®eldwork: be extant by Thomas (1892), we have found RSV encountered a solitary individual in no evidence that it has been examined by any well-drained primary forest at 14:45 hours subsequent author. on 9 July 1991, DPL saw one crossing a dirt Decker's (1991) revision of Nasua did not road in the daytime on 6 November 1992, explicitly recognize any valid subspecies of and Nancy A. Voss sighted another in pri- N. nasua, but Gompper and Decker (1998) mary forest on an unrecorded date in 1994. listed ten, including N. n. vittata (with phaeocephala and dichromatica as synonyms). Confusingly, Gompper and Decker listed Galictis vittata (Schreber) quasje as a synonym of N. n. nasua, thus The only de®nite record of grisons within implying that two valid taxa of coatis occur the limits of our study area is a sighting by in the Guianas. Without having undertaken a P. Petronelli (personal commun., 1993), who specimen-based revision of coati taxonomy, observed a pair travelling together on the we are unable to evaluate the possible sig- ground in primary forest in 1990. A large ni®cance of any geographic variation in Na- adult male that we found dead on a road near sua nasua that might occur in the Guiana Sinnamary (ca. 12 km NNW of Paracou) in subregion of Amazonia. However, it is clear 1992 measured 553 ϫ 155 ϫ 92 ϫ 32 mm from the literature reviewed above that if the and weighed 3.8 kg. French Guianan and Surinamese populations 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 71

TABLE 15 Measurements (mm) and Weights (kg) of Adult Potos ¯avus from Paracoua

(as represented by our vouchers and Hus- Potos ¯avus (Schreber) son's material) are distinct from the nominate form, the correct name for them may be VOUCHER MATERIAL: AMNH 265956, quasje Gmelin, not vittata Tschudi. Clearly, 265958, 265959, 266597±266599, 267048, resolving the application of Gmelin's name 267050, 267051, 267053, 267607, 267608; and reexamining Seba's original specimen MNHN 1995.954±1995.958. Total ϭ 17 should be a priority in any future revisionary specimens. study. IDENTIFICATION: Our voucher material cor- FIELD OBSERVATIONS: Coatis are uncom- responds exactly in qualitative characters mon at Paracou. P. Petronelli (personal com- with Husson's (1978) description of topotyp- mun., 1993) told us that he had seen coatis ic specimens from Surinam. External and only twice in ten years, once in an experi- craniodental dimensions of Paracou speci- mental plot of disturbed forest (a group of mens (table 15) likewise overlap those of two animals), and the second time in primary topotypes measured by Husson (op. cit.: ta- forest near a stream (a group of seven or ble 43). eight); both sightings were in daytime. We Although partial revisions of Potos ¯avus encountered coatis only three times in the by Thomas (1902), Kortlucke (1973), and course of our inventory ®eldwork. (1) On 15 HernaÂndez-Camacho (1977) each recognized August 1993, R. W. Kays sighted a group several subspecies as valid, there has been no estimated to consist of about 20 individuals geographically comprehensive study of kin- in a large tree at night; the tree was in well- kajou taxonomy to date. The Paracou popu- drained primary forest, and the animals (in- lation is presumably referrable to P. f. ¯avus, cluding our two vouchers) were perching but the necessity for a trinomial classi®cation among the branches, about 20 m above the remains to be convincingly demonstrated. ground. (2) L. H. Emmons sighted a group FIELD OBSERVATIONS: Kinkajous are by far of unknown size in swampy primary forest the commonest carnivore at Paracou. We at 09:25 hours on 13 October 1994. (3) A. heard them squealing and crashing about in L. Pef¯ey encountered a solitary individual the canopy virtually every night throughout of unknown sex in swampy primary forest at the course of our 1991±1994 ®eldwork. All 09:00 hours on 18 October 1994. of our vouchers were shot at night in trees, 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 at estimated heights ranging from 10 to 30 cana is primarily active at night; occasional m above the ground; recorded habitats in- daytime sightings are probably of individuals clude well-drained and swampy primary for- disturbed near their resting places. Although est and roadside secondary growth. Although M. americana is believed to be locally more many apparently solitary individuals were abundant than M. gouazoubira (below), we encountered, kinkajous were also encoun- nevertheless obtained only two unambiguous tered foraging or travelling in pairs and larg- sightings of red brockets in the course of our er groups. Collected specimens accompanied 1991±1994 inventory: (1) RSV observed a by information about group size include (1) solitary individual stealthily retreating an apparently solitary subadult female, (2) an through the undergrowth of swampy primary apparently solitary lactating adult female, (3) forest in the mid-afternoon of 25 June 1991; a nonlactating and nonpregnant adult female (2) DPL saw a solitary individual in well- accompanied by at least one other individual, drained primary forest at 10:10 hours on 23 (4) a juvenile male from a group of three July 1993. Many unidenti®ed deer whose individuals, (5) an apparently solitary sub- eyeshine was detected in the forest under- adult male, (6) an apparently solitary adult growth at night, or whose alarmed reactions male, and (7) an adult male accompanied by (snorts and foot-stamps) were heard in the at least one other individual. None of the six darkness, could have belonged to this spe- adult females we collected (in August and cies, or to the next. November) were pregnant. Mazama gouazoubira (G. Fischer) PERISSODACTYLA VOUCHER MATERIAL: AMNH 265960, Only one perissodactyl species occurs at 265961; MNHN 1995.960. Total ϭ 3 speci- Paracou, and no others are known from any mens. Amazonian locality. Tapir dung, spoor, and IDENTIFICATION: Both of our adult exam- the animal itself are unmistakable. ples, one male (MNHN 1995.960) and one female (AMNH 265961), agree closely with Tapirus terrestris (Linnaeus) Husson's (1978) qualitative description of Although feces and tracks of this species Surinamese specimens that he identi®ed as are said to be commonly encountered in the Mazama gouazoubira nemorivaga. Measure- more remote and swampy parts of our study ments of our vouchers (table 16) likewise area, the animal itself is seldom seen. During correspond closely with morphometric data ten years' residence at Paracou, P. Petronelli from M. g. nemorivaga summarized by Hus- (personal commun., 1993) encountered only son (1978: table 61) and Bisbal (1991: table one tapir, near a headwater stream of Crique II). In qualitative cranial traits, our two adult Paracou in the daytime. Local hunters, how- skulls conform with MedellõÂn et al.`s (1998) ever, are known to have killed at least two characterization of northern South American during the same interval. populations of M. gouazoubira, except that the mesopterygoid fossae of both specimens ARTIODACTYLA have broadly U-shaped (not V-shaped) anterior margins. Four rainforest artiodactyl species are Our vouchers are practically topotypes of known to occur at Paracou, and no others are Cervus nemorivagus F. Cuvier, the descrip- expected (see appendix 1). All are unmistak- tion of which was based primarily on speci- able by external characters (Emmons, 1990, mens from Cayenne (Allen, 1915b; contra 1997) if a suf®ciently good view is obtained. Miranda-Ribeiro, 1919). Although this taxon is currently considered a subspecies of Ma- Mazama americana (Erxleben) zama gouazoubira (e.g., by Czernay [1987] Red brockets are apparently common at and Grubb [1993], presumably following Paracou, local hunters having killed dozens AÂ vila-Pires [1959]), no revisionary study in the last decade. According to P. Petronelli based on extensive specimen data has, in (personal commun., 1993), Mazama ameri- fact, shown that the small grayish brockets 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 73

TABLE 16 sonnel, about 10±12 times per year on av- Measurements (mm) and Weights (kg) of erage. Adult Mazama gouazoubira from Paracou Pecari tajacu (Linnaeus) Collared peccaries were once common at Paracou, but although we often saw tracks and wallows (usually in parts of the forest distant from our normal inventory activities), we rarely saw the animals themselves. Ac- cording to P. Petronelli (personal commun., 1993), this species is typically encountered locally in groups of 6±7 animals, always in the daytime, about 6±8 times per year on average.

Tayassu pecari (Link) Groups of about 20 white-lipped peccaries have been sighted at Paracou on at least three occasions, all in the daytime (P. Petronelli, personal commun., 1993). Bushnegro forestry workers say that such groups visit the area of the Guianas and those of Paraguay (the about every four years. Although we did not type locality of gouazoubira) are really con- see the animals themselves in the course of speci®c. The character differences that Tate our 1991±1994 inventory ®eldwork, L. H. (1939) observed between specimens that he Emmons observed fresh tracks of this species referred to M. nemorivaga and M. simplici- in primary forest on 12 October 1994. cornis Illiger (ϭ M. gouazoubira) should be carefully evaluated in any future taxonomic RODENTIA analysis of these deer. Rodents constitute the most diverse group REMARKS: The speci®c epithet of the gray of nonvolant mammals at Paracou, where we brocket was originally spelled ``gouazoupi- documented the occurrence of 22 species in ra'' as noted by Grubb (1993). However, six families: Sciuridae (2 species), Muridae gouazoubira is the spelling that has been fol- (11 species), Erethizontidae (2 species), Da- lowed almost universally for many years, and syproctidae (2 species), Cuniculidae (1 spe- we agree with Gardner (1999) that this usage cies), and Echimyidae (4 species). An addi- should be maintained in the interest of no- tional 11 species are known from other lo- menclatural stability. calities in French Guiana or Surinam (appen- FIELD OBSERVATIONS: Our three vouchers dix 1), and some of these could also be were all shot at night in well-drained primary expected to occur in our study area. Two spe- forest. The still-nursing fawn (its stomach cies are described as new below. containing milk only) and the lactating adult Predictably, most Paracou rodent identi®- female were collected together, whereas the cations proved to involve signi®cant taxo- young adult male (with antlers in velvet) was nomic problems, the resolution of which oc- apparently solitary. In addition, L. H. Em- cupies the bulk of the following accounts. mons observed what was probably a single We de®ne specimens to be adult if the per- individual on four different nights from 21 manent dentition is fully erupted, subadult if September to 10 October 1993 in well- the molar dentition is completely erupted but drained primary forest. the permanent premolars are not, and juve- According to P. Petronelli (personal com- nile if one or more molars are incompletely mun., 1993), gray brockets are less common erupted. Our quantitative comparisons of ro- at Paracou than are red brockets, but they are dent crania and dentitions are based on the nevertheless often sighted by forestry per- following measurements (®gs. 33, 34). 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 33. Limits of 12 rodent craniodental measurements de®ned in the text.

Condylo-incisive Length (CIL): From the greater some tables, where the alveolar equivalent was curvature of one upper incisor to the articular measured). surface of the occipital condyle on the same Maxillary Toothrow (MTR): Crown length, from side. P4 to M3 (except as noted in some tables, Length of Diastema (LD): From the crown of the where the alveolar equivalent was measured). ®rst cheektooth to the lesser curvature of the Length of Molars (LM): Crown length from M1 incisor on the same side (except as noted in to M3. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 75

Fig. 34. Limits of four craniodental dimensions as measured for caviomorph rodents.

Breadth of M1 (BM1): Greatest crown breadth of gin of the infraorbital foramen to the postero- the ®rst maxillary molar. lateral corner of the zygomatic arch. Length of Incisive Foramen (LIF): Greatest anterior-posterior dimension of one incisive fora- A few other measurements taken for spe- men. cial purposes are de®ned as necessary in the Breadth of Incisive Foramina (BIF): Greatest text and tables that follow. Because sexual transverse dimension across both incisive for- dimorphism is an insigni®cant source of amina. measurement variation in most rodents (e.g., Breadth of Palatal Bridge (BPB): Measured be- see Straney [1978] and references cited by tween the protocones of the right and left ®rst Voss [1988: 362]), we do not summarize maxillary molars (ϭ ``Anterior Palatal morphometric data separately by gender. Breadth'' of Voss and Angermann, 1997). Most of the larger rodents in the Paracou Breadth of Zygomatic Plate (BZP): Least distance between anterior and posterior edges of the zy- fauna (all sciurids, erethizontids, dasyproc- gomatic plate. tids, and Cuniculus paca) can be identi®ed Length of Rostrum (LR): From the tip of one nasal at a distance by obvious external characters bone to the posterior margin of the zygomatic (Emmons, 1990, 1997), but most of the notch on the same side. smaller rodents cannot be con®dently iden- Length of Nasals (LN): Greatest anterior-posterior ti®ed without specimens in hand, and some dimension of one nasal bone. closely related species cannot be unambigu- Least Interorbital Breadth (LIB): Least distance ously distinguished except from cleaned cra- across the frontal bones between the orbital fos- nial material. In lieu of keys, we provide tab- sae. ular summaries of diagnostic traits to facili- Breadth of Braincase (BB): Greatest transverse dimension across the braincase above and tate identi®cations in some speciose genera. slightly behind the squamosal zygomatic processes. SCIURIDAE Zygomatic Breadth (ZB): Greatest transverse dimension across the squamosal zygomatic pro- The two squirrels that we found at Paracou cesses (ϭ ``Posterior Zygomatic Breadth'' of are the only species known to occur in Voss and Angermann, 1997). French Guiana, Surinam, and AmapaÂ; there- Zygomatic Length (ZL): From the posterior mar- fore, no future additions to the local sciurid 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 fauna are expected. Pending a revision of the TABLE 17 complex nomenclature of these animals (M. Measurements (mm) and Weights of Adult de Vivo, in prep.), we follow the usages rec- Sciurus aestuans from Paracou ommended by Husson (1978), who examined type material that we have not seen.

Sciurillus pusillus (E. Geoffroy)

VOUCHER MATERIAL: AMNH 269119. To- tal ϭ 1 specimen. IDENTIFICATION: Our voucher is almost topotypical of this species (originally described from specimens collected at Cayenne) and exhibits the reddish head, black ear tips, and white postauricular patches said to distinguish S. p. pusillus from other nominal taxa of South American pygmy squirrels (An- thony and Tate, 1935; Husson, 1978). The external dimensions of AMNH 269119, an adult female, were 109 ϫ 74 ϫ 28 ϫ 14 mm; including two embryos in utero, this specimen weighed 51 g. Because ¯uid-preserved ular patches'', but most do not; there is no material of Sciurillus is rare in museum col- trace of a postauricular patch on any speci- lections, we did not extract the skull of men from Paracou. (2) Husson described the AMNH 269119 for measurement. ventral coloration as ``usually pale reddish REMARKS: For the availability of names brown, sharply separated from the colour of from Geoffroy's (1803) catalog (rejected by the dorsal surface, at least in about the mid- Wilson and Reeder, 1993: 831), see Hersh- dle of the body, but considerably less so in kovitz (1955) and Holthuis (1963). the anterior and posterior parts''. By contrast, FIELD OBSERVATIONS: Our single voucher Paracou skins have clear (self-colored) or- was shot by L. H. Emmons at 11:45 hours ange fur on the chest, fading to buff or cream on 21 September 1994 as it fed on something on the throat; some clear orange fur extends growing on (or concealed beneath) the bark posteriorly along the ventral midline onto the of a large Inga sp. (Mimosoideae) at a height abdomen, but most of the abdominal fur is of about 18 m in well-drained primary forest. gray-based, appearing grizzled like the ¯anks In addition, we recorded ¯eeting diurnal ob- although much paler. (3) Some Surinamese servations of this species on ®ve dates from specimens have substantially smaller mea- 1991 to 1993; all of these sightings were of surements (op. cit.: table 62) than our vouch- solitary individuals in trees in well-drained ers (table 17), but it is possible that Husson primary forest. included subadults in his sample. This species (together with other members Sciurus aestuans Linnaeus of the so-called aestuans group of Sciurus) VOUCHER MATERIAL: AMNH 266485± was referred to the genus Guerlinguetus 266488, 266492, 266493, 267013, 267565; Gray by Allen (1915a), Tate (1939), Moojen MNHN 1995.989±1995.991. Total ϭ 11 (1942), Moore (1959), and others, but most specimens. recent authors have followed Cabrera (1961) IDENTIFICATION: Our voucher material in treating Guerlinguetus as a subgenus of agrees closely in most details with Husson's Sciurus. Cabrera cited no published analysis (1978: 386±387) description of topotypic of character data to support his opinion, how- specimens from Surinam, but several points ever, and it seems probable that renewed of comparison merit comment. (1) According morphological and molecular studies of Neo- to Husson, a few Surinamese examples have tropical squirrels will advocate a return to the ``very inconspicuous buffy yellow postauric- older generic usage. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 77

Although the currently accepted synony- vide illustrations and supplementary infor- my (Hoffmann et al., 1993) for Sciurus aes- mation to facilitate ®eld determinations of tuans implies that this species is distributed problematic taxa. Our anatomical terminol- throughout eastern Amazonia to southeastern ogy for muroid morphological characters fol- Brazil (including such forms as alphonsei lows that referenced or de®ned by Reig Thomas, garbei Pinto, henseli Miranda-Ri- (1977), Voss (1988, 1993), Carleton and beiro, ingrami Thomas, poaiae Moojen, and Musser (1989), Voss and Carleton (1993), roberti Thomas; see Cabrera [1961] for bib- and Musser et al. (1998). liographic citations and type localities), we follow the last substantive specimen-based Neacomys Thomas revisionary treatment of Amazonian squirrels (Moojen, 1942) in restricting S. aestuans to The Neotropical spiny mice of the genus the Guiana subregion of Amazonia. In the Neacomys have never been revised, and Amazonian lowlands of southeastern Vene- many aspects of their species-level taxonomy zuela (geographically part of the Guiana sub- have long been problematic. The identi®ca- region), however, S. aestuans is replaced by tion of spiny mice from the Guiana subregion a different species that is usually identi®ed of Amazonia is a case in point: although (e.g., by Tate, 1939; Handley, 1976; Linares, these have traditionally been identi®ed as N. 1998) as S. gilvigularis Wagner. guianae (e.g., by Anthony, 1921a; Tate, FIELD OBSERVATIONS: Sciurus aestuans is 1939; Carvahlo, 1962; Husson, 1978; Gen- one of only three common diurnal rodents at oways et al., 1981; Guillotin, 1982; Mal- Paracou. Most (9) of our 11 voucher speci- colm, 1990; Voss and Emmons, 1996), the mens were shot in the daytime, and we re- diagnostic morphological characters and geo- corded an additional 11 unvouchered day- graphic range of this species are not docu- time sightings in our ®eldnotes; two trapped mented in the literature. In his original de- specimens were found at or near dawn, but scription, Thomas (1905: 310) compared N. might have been captured the preceding af- guianae only with N. spinosus (Thomas, ternoon. With the exception of the latter, 1882), stating that the new species was very which were taken near ground level in Con- similar but ``conspicuously smaller''. How- ibear and leghold traps set on tree trunks ever, size does not distinguish guianae from over a small stream, all of our observations such other diminutive forms as tenuipes of this species at Paracou were of animals Thomas (1900), pusillus Allen (1912), and perched in trees at heights of 3±30 m above pictus Goldman (1912). Musser and Carleton the ground. Most individuals were solitary, (1993) listed N. guianae, N. pictus, N. spi- but an adult male and an adult female were nosus, and N. tenuipes (including pusillus)as collected together on 13 August 1991. Hab- valid species, but the recent description of itat data accompanying specimens or sight two additional species from western Ama- records include 16 observations in primary zonia (N. minutus and N. musseri), together forest at both well-drained and swampy sites, with sequence comparisons showing high and 5 in more-or-less disturbed habitats. levels of genetic differentiation among several undescribed mtDNA clades of spiny MURIDAE mice, suggests that the genus is much more diverse than previously recognized (Patton et All of the genera of rainforest murids al., 2000). known to occur in the Guiana subregion of In order to identify our Paracou vouchers, Amazonia are documented by vouchers col- we examined original descriptions of all lected at Paracou, including species of Nea- nominal taxa of Neacomys, and we examined comys, Nectomys, Neusticomys, Oecomys, holotypes or paratypes of all the smaller Oligoryzomys, Oryzomys, and Rhipidomys. named forms (guianae, minutus, musseri, Although most Guianan murids can be pro- pictus, pusillus, tenuipes). We tried to locate visionally identi®ed to genus in the ®eld by every Guianan Neacomys specimen currently external characters described by Husson housed in North American and European (1978) and Emmons (1990, 1997), we pro- museums, and we measured representative 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 series to document morphometric variation M1 and M2 always well developed and fused within and among species. The results of our to mesostyle on labial cingulum); M1/m1 comparisons indicate that at least three dis- without accessory roots. tinct species are present in the Guiana subregion of Amazonia, of which two are new Neacomys dubosti, new species and occur sympatrically at Paracou. Because Figures 36, 37, 38B, 39A, 39C, 43 the very brief diagnoses of Neacomys provided by Thomas (1900), Gyldenstolpe TYPE MATERIAL AND TYPE LOCALITY: The (1932), and Ellerman (1941) are now insuf- holotype, AMNH 267569, an adult female ®cient as a basis for taxonomic inference, we preserved as a ¯uid specimen with the skull rediagnose the genus here. extracted and cleaned, was collected at Par- EMENDED DIAGNOSIS OF NEACOMYS: Small acou by R. W. Kays (original number: RWK oryzomyines (sensu Voss and Carleton, 9) on 7 August 1993. No other material is 1993: 31) with coarsely grizzled yellowish-, known from the type locality, but all of the reddish-, or grayish-brown dorsal fur con- additional specimens we examined from taining short, grooved spines in addition to French Guiana and AmapaÂ are hereby des- conventional guard hairs and underfur; ven- ignated as paratypes. tral fur similar in composition to dorsal fur, DISTRIBUTION AND SYMPATRY: Based on but shorter and always contrastingly colored; specimens we examined, Neacomys dubosti pinnae small, dark, and sparsely haired; occurs in French Guiana, AmapaÂ (Brazil), mammae eight in inguinal, abdominal, post- and southeastern Surinam (®g. 35). In Suri- axial, and pectoral pairs; hindfoot with outer nam, N. dubosti has been collected sympat- digits (I and V) much shorter than three mid- rically with N. guianae (at the Sipaliwini dle digits (claw of dI not extending beyond Airstrip, Nickerie District), and it occurs middle of ®rst phalange of dII, claw of dV sympatrically with another new species in not extending beyond ®rst interphalangeal French Guiana and AmapaÂ (see the next ac- joint of dIV); claws of pedal digits II±V pro- count, below). vided with ungual tufts of long whitish or ETYMOLOGY: The speci®c epithet honors silvery hairs that exceed the claws in length; GeÂrard Dubost for his many contributions to tail sparsely haired (appearing naked except knowledge of mammalian ecology and nat- under magni®cation) with prominent epider- ural history in the lowland rainforests of mal scales in annular series, sometimes with French Guiana and Gabon. We are also grate- a thin terminal pencil but never with a con- ful for his original suggestion of Paracou as spicuous tuft of long hairs at tip. Skull with the site for our mammal inventory, and for prominently beaded supraorbital margins; in- his subsequent support and advice through- terparietal large; palate long and wide, with out the course of our ®eldwork there. prominent and often complex posterolateral DIAGNOSIS: A small species of Neacomys pits ¯anking anterolateral margins of mesop- (measurements in table 18) distinguished terygoid fossa; parapterygoid fossae shallow from other diminutive congeners by its short, (never deeply excavated above level of pal- usually unicolored tail; moderately short ros- ate); alisphenoid strut absent (except as rare, trum ¯anked by relatively shallow zygomatic usually unilateral variant); carotid circulation notches; broad and strongly convergent in- includes large stapedial artery (pattern 1 or terorbital region with highly developed, 2 of Voss, 1988); tegmen tympani not over- shelf-like supraorbital beads; broad and dis- lapping posterior margin of squamosal (pos- tinctly in¯ated braincase; short, convex-sided terior suspensory process of squamosal ab- incisive foramina; carotid circulation pattern sent); subsquamosal fenestra sometimes 1; M1 with undivided anterocone; mesoloph small but always present and usually patent. of M1 with more-or-less symmetrical con- Upper incisors small, narrow, and opistho- nections to protocone and hypocone; persis- dont (never orthodont or proodont); lower in- tently tubercular molar cusps; and a distinc- cisor root contained in prominent capsular tive range of craniometric variation. process on lateral surface of mandible; mo- MORPHOLOGICAL DESCRIPTION: Dorsal pel- lars small and pentalophodont (mesolophs on age coarsely grizzled tawny- or reddish- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 79

Fig. 35. Known collection localities of Neacomys dubosti based on specimens examined. 1, BRA- ZIL, AmapaÂ, Serra do Navio; 2, FRENCH GUIANA, Cacao; 3, FRENCH GUIANA, Camopi; 4, FRENCH GUIANA, Iracoubo; 5, FRENCH GUIANA, Paracou; 6, FRENCH GUIANA, Piste St.-EÂ lie; 7, FRENCH GUIANA, SauÈl; 8, FRENCH GUIANA, St.-EugeÁne; 9, FRENCH GUIANA, Trois Sauts; 10, SURINAM, Marowijne, Oelemarie; 11, SURINAM, Nickerie, Sipaliwini Airstrip. brown, somewhat paler along sides due to hairs over central metapodials sometimes in- middorsal concentration of dark-tipped distinctly darker than those on digits and out- spines; ventral fur abruptly paler, sometimes er metapodials; claw of pedal digit I extend- pure white from chin to anus (e.g., CM ing about one-half length of phalange 1 of 76840, MNHN 1983.412, USNM 46182), adjacent digit II; claw of pedal digit V ex- but more commonly suffused to a greater or tending to but not beyond end of ®rst pha- lesser extent with buff or orange; ventral lange of adjacent digit IV; small but distinct hairs usually pale to roots, very rarely with hypothenar (lateral tarsal) plantar pad present indistinctly gray bases (e.g., USNM 461571, on hindfoot of one ¯uid specimen (this trait 461590); broad lateral line of clear buff or is dif®cult to score reliably on dried skins). orange separating dorsal and ventral pelage Tail about as long as combined length of present in all specimens examined. Supercil- head-and-body; almost always unicolored iary, genal, and some mystacial vibrissae ex- (dark above and below), but occasionally in- tend behind pinnae when laid back against distinctly paler ventrally at base (e.g., CM head. Dorsal surface of manus and pes cov- 76842, MNHN 1972.641); with small caudal ered by short pale fur in most specimens, but scales (21 rows/cm near base of tail in one 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 18 Measurements (mm) and Weights (g) of Neacomys dubosti

¯uid specimen) forming relatively narrow and patent; auditory bullae usually ¯ask- annulations. shaped, tapering gradually from tympanic Skull with moderately short, tapering ros- ring to unconstricted bony eustacian tubes. trum ¯anked by relatively shallow zygomatic First maxillary molar with undivided an- notches; interorbital region broad, with terocone; anteroloph of M1 seldom distinct, strongly convergent lateral margins; supra- usually fused labially with anterocone (an- orbital beads highly developed, projecting as tero¯exus usually distinguishable only as small shelves over posterior orbits and con- persistent internal fossette); mesoloph of M1 tinuing onto braincase as low temporal straight and slender, projecting labially from crests; braincase in¯ated, conspicuously symmetrically Y-shaped junction with me- domed, and very broad behind squamosal dian mure, without disproportionate connec- zygomatic processes. Incisive foramina relation to hypocone; principal labial cusps tively short (averaging about 57% of diaste- (paracone, metacone) slightly reduced in size mal length), usually with distinctly convex relative to lingual cusps (protocone, hypo- lateral margins; zygomatic plate relatively cone); principal cusps persisting as distinctly narrow; carotid circulation with well-devel- tubercular elements with moderate wear. oped supraorbital ramus of stapedial artery KARYOTYPES: Two specimens from AmapaÂ (occupying squamosal-alisphenoid groove (MNHN 1972.640, 1972.641) karyotyped by and sphenofrontal foramen; pattern 1 of M. Tranier had diploid counts of 2Nϭ62 Voss, 1988); subsquamosal fenestra smaller chromosomes (as recorded on skin labels). than postglenoid foramen but always distinct VARIATION: The three geographic samples 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 81

at hand, from Surinam, French Guiana, and 1983.403), Iracoubo (MNHN 1983.409), AmapaÂ (table 18), are very similar in most Piste St.-EÂ lie km 16 (MNHN 1986.876, qualitative and quantitative characters. In- 1986.877), SauÈl (MNHN 1983.405± stead, most of the variation in the material 1983.407, 1983.422, 1983.423, 1983.425), we examined (e.g., as noted parenthetically St.-EugeÁne (MNHN 1995.3226±1995.3229, in the preceeding description) occurs as in- 1998.1835, 1998.1839), Trois-Sauts (MNHN dividual differences within local populations. 1982.629, 1982.630, 1983.410, 1983.412, However, resemblances are strongest be- 1983.414±1983.416). SurinamÐMarowi- tween French Guianan specimens and a large jne, Oelemarie (CM 76835±76837, 76839± series from the Serra do Navio in AmapaÂ, 76843); Nickerie, Sipaliwini Airstrip (CM Brazil. By contrast, our few Surinamese ex- 76846). amples have slightly narrower molars and in- FIELD OBSERVATIONS: Our single specimen terorbital regions, and their supraorbital of Neacomys dubosti from Paracou was tak- beads appear somewhat less developed as en in a pitfall trap in creekside primary for- projecting shelves. est. COMPARISONS: Neacomys dubosti could potentially be confused with two previously Neacomys paracou, new species described congeners from northern South Figures 36, 37, 39B, 39D, 42B, 43 AmericaÐN. tenuipes and N. guianaeÐin addition to N. paracou, another new species TYPE MATERIAL AND TYPE LOCALITY: The described below. Selected qualitative con- holotype, MNHN 1995.1020, an adult male trasts among these four taxa are summarized preserved as a complete skeleton, was col- in table 19, descriptive univariate statistics lected at Paracou on 23 August 1993 by Ro- for measurements of representative series are land W. Kays (original number: RWK 30). provided in table 20, and the results of mul- All of the additional specimens we examined tivariate morphometric analyses are repre- from Paracou (see Specimens Examined, be- sented in ®gure 40 and table 21. More de- low) are hereby designated as paratypes. tailed, character-by-character comparisons DISTRIBUTION AND SYMPATRY: Specimens are deferred to the next account. Patton et al that we refer to Neacomys paracou are from (2000) recently described additional species French Guiana, Surinam, Guyana, eastern of small-bodied Neacomys from western Bra- Venezuela (BolõÂvar state), and Guianan Bra- zil, but those bear no close resemblance to zil (north of the Amazon and east of the Rio either N. dubosti or N. paracou and so are Negro) (®g. 41). Based on these records it not treated in these accounts. would be reasonable to expect that the spe- REMARKS: At least some of the specimens cies occurs throughout the Guiana subregion previously reported in the literature as Nea- of Amazonia, but we have not seen any ma- comys guianae by Carvalho (1962), Geno- terial from the Amazonas federal territory of ways et al. (1981), and Guillotin (1982) are Venezuela. Neacomys paracou has been col- probably referable to N. dubosti, but those lected sympatrically with N. guianae in Guy- authors did not provide the museum catalog ana (at Kartabo), and with N. dubosti in Su- numbers of relevant voucher material and we rinam (Oelemarie), French Guiana (Cacao, are therefore unable to associate con®dent Paracou, SauÈl, St.-EugeÁne), and AmapaÂ (Ser- species identi®cations with their observa- ra do Navio). tions. ETYMOLOGY: The species is named for our OTHER SPECIMENS EXAMINED: BrazilÐ study area, treated as a noun standing in ap- AmapaÂ, Serra do Navio (USNM 461563± position to the generic name. 461569, 461571, 461572, 461574±461576, DIAGNOSIS: A small species of Neacomys 461579±461582, 461584, 461588, 461590± (measurements in table 22) distinguished 461595, 461601, 461604, 461612); no other from other like-sized congeners by its very locality data (MNHN 1972.640, 1972.641). short outer pedal digits; short, unicolored French GuianaÐCacao (MNHN 1983.426, tail; short rostrum ¯anked by relatively deep 1986.534, 1986.535, 1986.537, 1986.538, zygomatic notches; broad and usually strong- 1986.541±1986.545), Camopi (MNHN ly convergent interorbital region with well- 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 19 Diagnostic Qualitative Comparisons among Four Species of Neacomys

developed and often shelf-like supraorbital curving from and disproportionately con- beads; narrow, unin¯ated braincase; long, nected to hypocone; principal molar cusps parallel-sided incisive foramina; carotid cir- quickly worn to enamel loops, not persis- culation pattern 1; M1 with narrow, undivid- tently tubercular; and a distinctive range of ed anterocone; mesoloph of M1 stout, often morphometric variation. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 83

TABLE 20 Comparative Measurements (mm) and Weights (g) of Representative Series of Four Species of Neacomys

MORPHOLOGICAL DESCRIPTION: Dorsal pel- not extending much if at all behind pinnae age coarsely grizzled tawny- or reddish- on properly made-up skins. Dorsal surface of brown, somewhat paler along sides due to manus and pes covered with short pale fur, middorsal concentration of dark-tipped often with indistinctly darker markings over spines; ventral fur abruptly paler, often pure central metapodials; claw of pedal digit I ex- white from chin to anus, but sometimes with tending less than one-half length of phalange orange pectoral markings (e.g., AMNH 1 of adjacent digit II; claw of pedal digit V 266548), or broadly suffused with orange extending no more than three-fourths length (e.g., CM 76845); ventral hairs pale to roots of phalange 1 of adjacent digit IV; hypoth- in most specimens (rarely with indistinctly enar (lateral metatarsal) plantar pad small but gray bases between the fore- and hindlegs; distinct in some specimens, indistinct or ab- e.g., AMNH 266545); broad lateral line of sent in others. Tail about as long as, or a little clear buff or orange separating dorsal and shorter than, combined length of head-and- ventral pelage in many specimens, but lateral body; unicolored (dark above and below), line narrow or absent in others (e.g., AMNH rarely indistinctly paler ventrally at base 266542, CM 76844, MNHN 1986.285). Su- (e.g., AMNH 266548); with large caudal perciliary and genal vibrissae extending be- scales (15±18 rows/cm near the base of the hind pinnae when laid back alongside head, tail in nine ¯uid specimens) forming coarse but mystacial vibrissae consistently shorter, and conspicuous annulations. 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 36. Dorsal and ventral cranial views of four species of Neacomys. From left to right: N. tenuipes (FMNH 71778); N. guianae (CM 76850); N. dubosti (AMNH 267569, holotype); N. paracou (MNHN 1995.1020, holotype). All views about ϫ3.

Skull with very short rostrum ¯anked by matic plate relatively broad; carotid circula- moderately deep zygomatic notches; inter- tion with well-developed supraorbital ramus orbital region broad and usually strongly of stapedial artery (occupying squamosal-ali- convergent; supraorbital beads well devel- sphenoid groove and sphenofrontal foramen; oped, often projecting as small shelves over pattern 1 of Voss, 1988); subsquamosal fe- posterior orbits and continuing onto brain- nestra often very small and sometimes oc- case as low temporal crests; braincase rela- cluded by internal ¯ange of petrosal; audi- tively narrow in most specimens and not con- tory bullae usually globular, with spherical spicuously in¯ated. Incisive foramina rela- tympanic capsules and abruptly constricted tively long (averaging about 65% of diaste- bony eustacian tubes. mal length) and narrow, usually with First maxillary molar typically with very more-or-less parallel lateral margins; zygo- narrow, undivided anterocone; anteroloph of 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 85

TABLE 21 Results of Linear Discriminant Function Analysis of Craniodental Measurement Data from Four Neacomys Species Samplesa

M1 usually indistinct (fused with anterocone) even on unworn teeth; mesoloph of M1 very prominent (mesoloph/mesostyle complex sometimes rivalling paracone and/or metacone in size) and disproportionately connected to hypocone by median mure, arising anterolabially from that cusp in an uninter- upted curve on most unworn teeth; principal labial cusps (paracone, metacone) distinctly smaller than lingual cusps (protocone, hypocone); all principal cusps quickly worn down to enamel loops, not persisting as distinctly tubercular elements in most adult dentitions. KARYOTYPES: Two specimens of Neacomys paracou karyotyped by M. Tranier from ``Cayenne, Rte. de Cacao'', French Guiana (MNHN 1983.419, 1983.420) and another Fig. 37. Lateral cranial and mandibular views from SauÈl (MNHN 1983.418) had diploid of four species of Neacomys. From top to bottom: counts of 2Nϭ56 chromosomes (recorded on N. tenuipes (FMNH 71778); N. guianae (CM skin tags). The same diploid counts were ob- 76850); N. dubosti (AMNH 267569, holotype); N. tained by E. Bach from chromosomal prep- paracou (MNHN 1995.1020, holotype). All views arations of two Venezuelan specimens about ϫ3. 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 38. Dorsal views and cross-sections of the interorbital region in Neacomys tenuipes (A, FMNH 71778) and N. dubosti (B, AMNH 267569, holotype). In most examined specimens of N. tenuipes, the interorbital region is relatively narrow and unin¯ated, with weakly convergent supraorbital margins; the supraorbital beads are usually strongly re¯ected and seldom form projecting shelves. By contrast, the interorbital region of N. dubosti is typically broader and more in¯ated, the supraorbital margins are more strongly convergent, and the supraorbital beads are more frequently developed as projecting shelves.

(AMNH 257270, MNHLS 8064) that were Brazilian states of Amazonas and ParaÂ; mea- part of the series from San Ignacio YuruanõÂ surements not tabulated) have less well-de- originally misidenti®ed by Voss (1991: table veloped supraorbital beads than most topo- 23) as N. tenuipes. typical specimens but do not appear to be VARIATION: Samples that we refer to Nea- morphometrically divergent or remarkable in comys paracou are remarkably similar in other qualitative respects. morphological characters across a very large COMPARISONS: Neacomys paracou requires geographic range. The most metrically di- close comparisons with three other small vergent series consists of three Venezuelan species from northern South America, N. du- examples (from San Ignacio YuruanõÂ in east- bosti, N. guianae, and N. tenuipes. Of these, ern BolõÂvar state; table 22), which have lon- dubosti, guianae, and paracou occur in the ger hindfeet, slightly larger molars, and Guiana subregion of Amazonia, where they slightly longer rostrums than most Paracou have been collected sympatrically in all pair- specimens; broad overlap between these wise combinations (but never all three to- samples in most measured dimensions (together). Neacomys tenuipes does not occur in gether with the lack of other distinguishing the Guiana subregion, but its nomenclatur- characters), however, suggest that they are ally crucial status as the oldest named species not speci®cally distinct. Specimens from two of small spiny mice compels us to include it other geographically outlying samples (in the in this comparative analysis. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 87

Fig. 39. Left maxillary toothrows (A, B) and right mandibular toothrows (C, D)ofNeacomys dubosti (A, C; USNM 461580) and N. paracou (B, D; USNM 461609).

Although all small species of spiny mice method (total length and tail length [dorsal are similar in most external characters, the ¯exure to ¯eshy tip] measured in the ®eld; morphology of the hindfoot and the tail can head-and-body length calculated by subtrac- be used in combination to provide tentative tion), the tail is consistently longer than the ®eld identi®cations. In tenuipes and dubosti, head-and-body by a substantial amount (the the outer pedal digits (dI and dV) are rela- ratio LT/HBL averaging about 115%) in ten- tively long: the claw of dI extends almost or uipes, but in the other three species the tail fully half the length of the ®rst phalange of is about the same length as the head-and- dII, and the claw of dV extends almost or body, on average. Unfortunately, the ratio of fully to the end of the ®rst phalange of dIV. tail to head-and-body cannot be compared By contrast, the claw of dI does not extend meaningfully among specimens measured by much beyond the base of the ®rst phalange different protocols, and many specimens of of dII in guianae and paracou, and the claw Neacomys are captured with bobbed tails. of dV in these two species does not extend Nevertheless, the contrast in tail length is vi- more than about half the length of the ®rst sually obvious when comparing series of phalange of dIV. These proportional differ- skins of tenuipes with those of dubosti, gui- ences are easiest to see in fresh material, or anae, and paracou. in ¯uids, where the digits can be straightened The tail is distinctly bicolored (dark above, and freely manipulated; in carelessly made- pale below), at least near the base, in most up skins (with twisted or bent toes), however, specimens of tenuipes, but most specimens digital proportions can be hard to evaluate. of dubosti and paracou have unicolored (all In specimens measured by the American dark) tails. Unfortunately, the few available 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 40. Results of pairwise principal components analyses of representative samples of Neacomys dubosti (dub, N ϭ 29), N. guianae (gui, N ϭ 6), N. paracou (par, N ϭ 37), and N. tenuipes (ten, N ϭ 36). For each of the six analyses illustrated, principal components were extracted from the covariance matrix of log-transformed craniodental measurements of specimens with complete data in the series identi®ed by footnotes to table 20. The ®rst principal component is represented by the horizontal axis in each panel, the second principal component by the vertical axis; species samples are depicted as 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 89

Fig. 41. Known collection localities of Neacomys paracou based on specimens examined. 1, BRA- ZIL, AmapaÂ, Serra do Navio; 2, BRAZIL, Amazonas, 80 km N Manaus; 3, BRAZIL, ParaÂ, Cachoeira Porteira; 4, FRENCH GUIANA, Arataye; 5, FRENCH GUIANA, Cacao; 6, FRENCH GUIANA, Cay- enne; 7, FRENCH GUIANA, Mont St.-Michel; 8, FRENCH GUIANA, Paracou; 9, FRENCH GUIANA, SauÈl; 10, FRENCH GUIANA, St.-EugeÁne; 11, GUYANA, Barima-Waini, Baramita; 12, GUYANA, Cuyuni-Mazaruni, Kartabo; 13, GUYANA, Potaro-Siparuni, Kurupukari; 14, GUYANA, Upper Takutu- Upper Essequibo, Nappi Creek; 15, SURINAM, Brokopondo, 18.5 km W Afobakka; 16, SURINAM, Brokopondo, Locksie Hattie; 17, SURINAM, Marowijne, Oelemarie; 18, SURINAM, Marowijne, Perica; 19, VENEZUELA, BolõÂvar, San Ignacio YuruanõÂ. skins of guianae are too variable to charac- colored at the base, four other specimens terize the species with con®dence for this have indistinctly bicolored or unicolored trait: whereas the type and two Surinamese tails. specimens have tails that are distinctly bi- In visual comparisons of dried skins, the

← minimum convex polygons enclosing all referred specimens. Together, the ®rst and second principal components accounted for 65% of the total variance in analyses A±D, 70% in analysis E, and 69% in analysis F. Species scores did not differ signi®cantly on subsequent components (after the second) in analyses A, B, D, E and F, but the scores of N. dubosti and N. tenuipes differed signi®cantly (p K0.01 by 1-way ANOVA) on the fourth component (not shown) in analysis C. 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 22 Measurements (mm) and Weights (g) of Neacomys paracou

caudal scales appear to be larger and to form necessarily omitted from this faunal report, coarser annulations in paracou than in the the scatter plots in ®gure 40 depict patterns other three species, but this difference is hard of multivariate divergence revealed by six to quantify because tails are stretched to pairwise principal components ordinations, varying degrees when skins are stuffed. Al- and the matrix in table 21 summarizes the though we counted the number of scale rows outcome of a linear discriminant function per centimeter near the base of the tail on analysis with all species treated simulta- ¯uid specimens, only a few ¯uids were avail- neously. Both methods indicate that these able for most species. Nevertheless, our data taxa are craniometrically distinct in all pair- suggest that this character might be useful for wise combinations with the exception of du- ®eld identi®cations: whereas nine adult ¯uid bosti and tenuipes, which have partially over- specimens of paracou from the type locality lapping multivariate distributions. The fol- had 15±18 (mean ϭ 16) scale rows/cm, the lowing are the principal points of quantita- ¯uid holotype of dubosti had 21 rows/cm. tive and qualitative cranial difference based We were not able to examine any ¯uid spec- on our visual and analytic comparisons. imens of tenuipes or guianae. When samples of skulls are lined up in We assessed species differences in cranial comparative series, each species has a dis- morphology by visual comparisons supple- tinctive dorsal gestalt as a consequence of mented by measurements of representative taxonomic variation in four anatomically ad- samples (table 20). Although most statistical jacent and visually juxtaposed structures: (1) details of our morphometric analyses are The rostrum varies in absolute and relative 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 91

length, being longest on average in tenuipes and shortest in paracou; the rostrum is of intermediate length in guianae and dubosti. (2) The zygomatic notches (dorsal emargi- nations of the maxillary bone ¯anking the base of the rostrum) vary in depth as a correlate of variation in the width of the zygomatic plate; the zygomatic notches are deep- est and the zygomatic plates widest in paracou, whereas tenuipes, guianae, and dubosti have shallower zygomatic notches and cor- respondingly narrower zygomatic plates. (3) The interorbital region is relatively narrow, and the supraorbital beads are relatively weakly developed (seldom produced as shelf-like projections over the posterior orbits) in tenuipes and guianae; in these species, the modal interobital morphology could be described as weakly convergent (®g. 38A). By contrast, dubosti and paracou have relatively broader interorbits, and the supraorbital beads are more frequently developed as projecting shelves; their modal interorbital morphology is strongly convergent (®g. 38B). (4) The braincase is relatively broader and more in¯ated in dubosti than in any of the other three species. Taxonomic variation in other quantitative and qualitative cranial traits also contributes to species recognition. The incisive foramina of paracou are longer in relation to the diastema (LIF averaging about 65% of LD) than those of tenuipes, guianae, and dubosti (in which this proportion averages about 56± 57%), and subtle taxonomic differences in the shape of these diastemal perforations are Fig. 42. Ventral views of the left auditory bul- also present. Thus, the foramina are relative- la in Neacomys guianae (A, CM 76850) and N. ly narrow in proportion to their length and paracou (B, MNHN 1995.1020, holotype). In N. usually have subparallel lateral margins in guianae the capsular part of the auditory bulla paracou, whereas the foramina are relatively (cap) narrows gradually to merge with the bony broader with more convex or anteriorly con- eustacian tube (et), and the bulla is more-or-less vergent lateral margins in the other species. ¯ask-shaped. By contrast, the bulla is abruptly constricted anteromedially in N. paracou, result- Although the shape of the auditory bullae ing in a sharper transition between a more glob- exhibits individual variation within most ular tympanic capsule and a narrower eustacian population samples, the bullae of paracou tube. are more consistently globular in form, each consisting of a roughly spherical tympanic capsule that is usually abruptly constricted tively broader eustacian tube. Insuf®cient in anteromedially to form narrow bony eusta- itself for species diagnosis, this character is cian tubes (®g. 42B). By contrast, the bullae nevertheless useful for corroborating identi- of tenuipes, guianae (®g. 42A), and dubosti ®cations when used in conjunction with other are usually ¯ask-shaped, each tapering grad- traits. ually from the tympanic annulus to a rela- The morphology of the ®rst maxillary mo- 92 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 43. Occlusal morphology of the left ®rst maxillary molar (M1) of Neacomys species. Left to right: N. tenuipes (USNM 499555), N. guianae (CM 76849), N. dubosti (USNM 461580), and N. paracou (USNM 461609). lar also differs signi®cantly among the four dency, that is more marked in some speci- species (®g. 43). In tenuipes, M1 is more-or- mens than in others, for the labial cusps less rectangular in outline because the anter- (paracone and metacone) to be reduced in ocone is almost as broad as the paracone- size relative to their lingual counterparts protocone cusp-pair behind it. In many spec- (protocone and hypocone), resulting in a less imens with unworn dentitions (especially symmetrical occlusal design. In addition to from the central Andean cordillera of Colom- these shape differences, the toothrow is ab- bia; e.g., FMNH 70126, USNM 499555), the solutely shorter in guianae than in either ten- anterocone is deeply divided into anterolin- uipes or dubosti. gual and anterolabial conules by an antero- The typical morphology of M1 in paracou median ¯exus, and the anteroloph is large differs in several respects from that seen in and distinct. The occlusal organization of the the other three species. The tooth is visibly tooth is strikingly symmetrical, with sube- narrower in relation to its length, on average, qual lingual and labial cusps that remain per- and the undivided anterocone is usually sistently tubercular with moderate wear. The much narrower than the protocone-paracone mesoloph is a slender crest of enamel, per- cusp-pair behind it. In most specimens, this pendicular to the long axis of the tooth, that tooth exhibits a striking departure from bi- forms a Y-shaped junction with the median lateral symmetry, with the labial cusps being mure and lacks a disproportionate connection much reduced in size relative to their lingual to either of the two principal lingual cusps counterparts, and with an enlarged mesoloph (protocone and hypocone). that runs obliquely and disproportionately The modal morphology of M1 in guianae from the hypocone to the labial cingulum. In and dubosti is essentially similar to that seen addition, the principal cusps are not persis- in tenuipes, but differs in certain details. tently tubercular because they are quickly Thus, the anterocone is undivided and usu- worn down to enamel loops; thus, even mod- ally distinctly narrower than the protocone- erately worn dentitions are essentially ¯at- paracone cusp-pair, giving the tooth a less crowned, lacking any signi®cant occlusal re- rectangular and more egg-shaped outline, lief. and the anteroloph is seldom distinct (the an- REMARKS: Specimens that we examined tero¯exus usually persisting, if at all, only as and determined to be Neacomys paracou in- a small internal fossette). There is also a ten- clude at least some of the material previously 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 93

identi®ed as N. guianae by Anthony (1921a), etation, 10 (35%) were trapped in well- Husson (1978), Guillotin (1982), Malcolm drained primary forest, and 5 (17%) were (1990), and Voss and Emmons (1996: appen- trapped in swampy primary forest. All spec- dices 4 and 5). It is probable that other lit- imens were collected at or near ground level. erature records of N. guianae are also based Of the 18 Sherman- or Victor-trapped spec- partly or entirely on specimens of N. para- imens, most were taken in dense under- cou, which appears to be the commonest and growth near woody shelter: under logs (6 most widespread of the three Neacomys spe- specimens), on top of logs (3), among stilt cies now known from the Guiana subregion roots (3), beside logs (2), under piled branch- of Amazonia. es (2), inside a hollow log (1), and at the base SPECIMENS EXAMINED: BrazilÐAmapaÂ, of a tree (1). Serra do Navio (USNM 461570, 461577, 461578, 461583, 461585, 461586, 461596, Nectomys melanius Thomas 461597±461600, 461603, 461605, 461608, Figures 44, 45B, 46B, 47B 461609); Amazonas, 80 km N Manaus (USNM 580008±580011); ParaÂ, Cachoeira VOUCHER MATERIAL: MNHN 1998.680, Porteira (USNM 546277±546281). French 1998.681. Total ϭ 2 specimens. GuianaÐArataye (MNHN 1986.284± IDENTIFICATION: The genus Nectomys was 1986.286, 1986.870±1986.875), Cacao last revised by Hershkovitz (1944), who rec- (MNHN 1986.536, 1986.539), Cayenne ognized all of the material he examined as (MNHN 1983.419, 1983.420), Mont St.- belonging to one or the other of two poly- Michel (MNHN 1983.411), Paracou (AMNH typic species assigned to different subgenera, 266542, 266544±266546, 266548±266550, Nectomys (Nectomys) squamipes (Brants) 266552±266557, 267570, 267572, 267574± and N. (Sigmodontomys) alfari (J. A. Allen). 267577; MNHN 1995.1013±1995.1022 [type Current usage (summarized by Musser and series]), SauÈl (MNHN 1983.405, 1983.418, Carleton, 1993), however, recognizes Sig- 1983.421, 1983.424), St.-EugeÁne (MNHN modontomys and Nectomys as full genera, the 1998.1834, 1998.1836±1998.1838). Guya- former with two valid species (S. alfari and naÐ``River Supinaam'' (BMNH 10.5.4.22); S. aphrastus) and the latter with three (N. Barima-Waini, Baramita (ROM 100947); palmipes, N. parvipes, and N. squamipes). Cuyuni-Mazaruni, Kartabo (AMNH 42893, Only Nectomys (sensu stricto) is known to 64146, 64147, 142821, 245037); Potaro-Si- occur in Amazonia, Sigmodontomys being paruni, Kurupukari in Iwokrama Reserve restricted to trans-Andean and Venezuelan (BMNH 1997.44, 1997.46); Upper Takutu- coastal rainforests (Voss and Emmons, 1996: Upper Essequibo, Nappi Creek in Kanuku table 1). Mountains (ROM 31760). SurinamÐBro- Nectomys squamipes melanius was origi- kopondo, 18.5 km W Afobakka (CM 54016), nally described by Thomas (1910: 185±186), Locksie Hattie on Saramacca River (FMNH who considered it ``the Guianan representa- 95642, 95643); Marowijne, Oelemarie (CM tive of the Brazilian water rat, N. squamipes, 76838, 76844), Perica (CM 76845). Vene- but . . . distinguishable by its darker dorsal zuelaÐBolõÂvar, San Ignacio YuruanõÂ color and smaller skull and teeth.'' Thomas's (AMNH 257269±257271). account was based on a small series of spec- FIELD OBSERVATIONS: All of our inventory imens from Guyana and Surinam, but addi- records of Neacomys paracou are based on tional material identi®ed as melanius was collected specimens (N ϭ 29), of which 17 subsequently described by Hershkovitz (62%) were taken in Sherman traps, 8 (28%) (1944) and Husson (1978). Whereas both were taken in pitfalls, 3 (10%) were taken in Hershkovitz and Husson followed Thomas in Victor rat traps, and 1 was shot. Most trapped treating melanius as a valid subspecies of N. specimens were found at dawn, but a single squamipes (Brants, 1827), Tate (1939) syn- specimen was found in the late afternoon in onymized melanius with N. s. palmipes, a a pitfall that had been checked earlier on the taxon that was originally described (as a full same day. Fourteen specimens (48% of the species) by Allen and Chapman (1893) from total) were shot or trapped in secondary veg- Trinidad. Petter (1979) subsequently de- 94 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 44. Outlines of nasal bones in typical examples of Nectomys palmipes (top row) and N. melanius (bottom row). Whereas the nasals are abruptly constricted behind the premaxillary-frontal suture in N. palmipes, these bones are more-or-less evenly tapered throughout their length in N. melanius. Although atypical morphologies occur uncommonly in both species, this character is useful for taxonomic identi®cations when used in combination with other cranial traits (table 24; ®gs. 45±47). scribed N. parvipes from French Guiana and by Thomas (1910), Hershkovitz (1944), and compared it with sympatrically collected ma- Husson (1978). The French Guianan holo- terial that he identi®ed as N. s. melanius. The type of Nectomys parvipes, raised in the lab- identi®cation of our voucher material there- oratory from a wild-caught nestling (Petter, fore involves each of the three species of 1979), appears to be no more than an unusu- Nectomys regarded as valid by Musser and ally small individual (table 23), perhaps Carleton (1993). stunted by an inadequate diet or other captive All of the French Guianan material we ex- conditions. We examined this specimen amined agrees closely with the descriptions (MNHN 1979.345) and determined that none of Nectomys squamipes melanius provided of its qualitative characters diverge from the 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 95

examined, the same phenotype apparently extends westward into the Venezuelan state of Amazonas, and southward into the Bra- zilian state of ParaÂ on the north bank of the Amazon. We provisionally recognize Nectomys melanius as a distinct species based on geo- graphical, morphological, and cytogenetic comparisons with both of the taxa that have previously been considered to be senior synonyms (table 24). From N. palmipes, its near- est neighbor, melanius differs conspicuously in diploid chromosome counts (2N ϭ 16±17 versus 2N ϭ 52±56; references in footnotes to table 24) and in several morphological traits that can be used to identify museum specimens unaccompanied by karyotypes. (1) Whereas the lateral margins of the nasal bones taper gradually from front to back in melanius without a sharp change in angle at the premaxillary-frontal suture, the nasals of Fig. 45. Morphology of the occipital region palmipes are more abruptly constricted be- in Nectomys palmipes (A, AMNH 4658 [holo- hind the premaxillae (®g. 44) in most of the type]) and N. melanius (B, MNHN 1981.1296). In specimens we examined.12 (2) The interpa- N. palmipes, the interparietal (ip) is deep (antero- rietal bone is a shallow and wide element in posterior dimension) relative to its width (trans- melanius (®g. 45B) versus deeper and nar- verse dimension), and the exoccipital (exo) oc- rower in palmipes (®g. 45A), a difference cupies most of the dorsolateral surface of the oc- that is correlated with a marked dorsolateral ciput. By contrast, the interparietal of N. melanius expansion of the exoccipital in the latter spe- is relatively much wider than deep, and the dor- cies. (3) The nasolacrimal capsules on the solateral exposure of the exoccipital is corre- spondingly reduced. sides of the rostrum are mostly exposed to lateral view in melanius (®g. 46B), but these structures are partially concealed by the zy- range of variation exhibited by other speci- gomatic plate in palmipes (®g. 46A). (4) The mens of French Guianan Nectomys with sub- tegmen tympani is usually inconspicuous in stantially larger measurements. Although the melanius (®g. 47B), but a large anterior pro- range of variation in molar length (LM) in cess of the tegmen tympani is always present our French Guianan series is considerable in palmipes (®g. 47A). (5.9±7.1 mm), there is no hint of bimodality Based on museum specimens that we sort- in the frequency distribution of this measure- ed by these four morphological characters, ment, and there is no correlated variability in Nectomys palmipes occurs throughout the is- other characters to suggest that our sample is land of Trinidad, where the type (AMNH composite. French Guianan skins are not 5928/4658) was collected at Princestown. quite as dark, on average, as topotypical me- This species also occurs on the adjacent Ve- lanius from Guyana, but this appears to be nezuelan mainland, from which we examined the only point of external difference. Al- several specimens including the type of Nec- though measurements of Guyanese exem- tomys squamipes tatei Hershkovitz (1948a). plars suggest that there may be a modest Collected at San Antonio in the Venezuelan east-to-west increase in average molar size in this taxon (table 23), our side-by-side com- 12 In addition to the Trinidadian specimens listed in parisons of French Guianan and Guyanese table 23 (footnote e), we examined AMNH 69899 (the specimens revealed no qualitative cranio- holotype of tatei), AMNH 142608, and USNM 415009 dental differences. Based on specimens we from northeastern Venezuela. 96 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 46. Lateral views of the rostrum in Nectomys palmipes (A, AMNH 174241) and N. melanius (B, USNM 560824). In N. palmipes, the nasolacrimal capsule (nlc) is mostly obscured in lateral view by the very broad zygomatic plate (zp), but the nasolacrimal capsule is mostly exposed anterior to the narrower zygomatic plate of N. melanius.

Fig. 47. Lateral view of the ear region in Nectomys palmipes (A, AMNH 235065) and N. melanius (B, USNM 406062). A large anterior process of the tegmen tympani (tt) that touches or overlaps the squamosal (sq) anterodorsal to the auditory bulla and anteroventral to the postglenoid foramen (pgf)is always present in N. palmipes, but the tegmen tympani is usually less exposed to lateral view in N. melanius. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 97

TABLE 23 Measurements (mm) and Weights (g) of Nectomys melanius and N. palmipes

state of Monagas, the type of tatei (AMNH Nectomys melanius closely resembles N. 69899) is craniodentally indistinguishable squamipes in all of the morphological traits from Trinidadian material, as are two addi- by which N. palmipes differs from both (tational specimens from Monagas (AMNH ble 24), but we are persuaded by the karyo- 142608, USNM 415009); all were collected typic data and breeding experiments reported within the mapped distribution of karyotyped by Bonvincino et al. (1996), which suggest individuals with 2N ϭ 16±17 chromosomes that the water rats of southeastern Brazil rep- (in the states of Anzoategui, Delta Amacuro, resent a distinct species from Amazonian Monagas, and Sucre; Barros et al., 1992). populations. Based on the restricted type lo- However, a single specimen with the same cality of squamipes (SaÄo SebastiaÄo, SaÄo Pau- morphological characters (AMNH 16964) is lo state; see Hershkovitz, 1944), this would from El Llagual (ca. 7Њ25ЈN, 65Њ10ЈW) in the appear to be the correct name for Bonvincino northern part of BolõÂvar state. Because ma- et al.'s southeastern Brazilian taxon. We terial from southern BolõÂvar (e.g., AMNH agree with Patton et al. (2000) that aquaticus 75634, 75635, 130733, 130784) is unambig- Lund (type locality: near Lagoa Santa, Minas uously referable to melanius, specimens of Gerais) and olivaceus Hershkovitz (type lo- water rats from intermediate localities in that cality: Therezopolis, Rio de Janeiro) are state should be examined carefully to deter- probable synonyms of N. squamipes, and that mine whether melanius and palmipes (the this species probably extends into northern latter including tatei as a subjective junior Argentina. The observations of Peters (1861) synonym) are parapatrically or sympatrically and Hershkovitz (1944), together with our distributed in eastern Venezuela. own examination of southeastern Brazilian 98 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 24 Geographic, Morphological, and Karyotypic Comparisons among Three Species of Nectomys

material suggest that the hindfeet of N. ern Amazonian populations of Nectomys, we squamipes usually have six plantar tubercles, are unable to rule out this possibility. whereas the hindfeet of most specimens of REMARKS: It seems probable that Nectomys N. melanius lack a distinct hypothenar (lat- rattus, originally described by Pelzeln (1883) eral tarsal) pad. based on a single immature specimen col- Patton et al. (2000) used the name apicalis lected at Marabitanas (0Њ58ЈN, 66Њ51ЈW) on Peters (1861) for western Amazonian popu- the upper Rio Negro in the Brazilian state of lations of Nectomys with low diploid num- Amazonas, is a senior synonym of melanius. bers (2N ϭ 38±42 chromosomes), large teeth Although we have not seen Pelzeln's prob- (LM ϭ 7.0±7.4 mm), and deep-narrow inter- lematic type,14 the geographic proximity of parietals.13 Based on the difference in diploid numbers alone, it seems unlikely that api- 14 Thomas (1897: 497) remarked that the immature calis and melanius intergrade in western type of Hesperomys rattus Pelzeln ``is clearly a Necto- mys'', but he did not explicitly state that he had seen the Amazonia (contra Hershkovitz, 1944: 51± specimen. Tate (1939) apparently accepted Thomas's as- 52), but without having seen the type of ap- signment of rattus to Nectomys, but Hershkovitz (1944: icalis and without undertaking an extensive 30) objected on the grounds that Pelzeln's original de- analysis of morphological variation in west- scription was not informative about generic characters: ``Until the type itself, if still extant, can be examined, Hesperomys rattus, if at all identi®able, cannot be iden- 13 Barros et al. (1992) previously used the junior name ti®ed as a Nectomys.'' Although we have not seen the garleppi Thomas (1899) for this form, and it is possible type, we did consult Thomas's manuscript notes (in the that other nominal taxa may also be synonyms. library of the Natural History Museum, London), which 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 99

Marabitanas to Venezuelan localities from Neusticomys oyapocki (Dubost and Petter) which we have examined specimens refer- Figures 48, 49B, 49D able to melanius tends to support this synonymy. Clearly, a comprehensive revision of VOUCHER MATERIAL: AMNH 267597; Nectomys based on ®rst-hand examination of MNHN 1995.992. Total ϭ 2 specimens. relevant types and a critical analysis of mor- IDENTIFICATION: Our two vouchers and an- phological variation among the many hun- other specimen subsequently collected at dreds of museum specimens now available nearby St.-EugeÁne provide an opportunity to reevaluate the characters of this obscure tax- for study will be essential for resolving this on. Previously known from a single speci- and other taxonomic enigmas. In the mean- men (MNHN 1977.775) from Trois Sauts in time, N. melanius is the oldest available southeastern French Guiana, Daptomys oya- name that we can con®dently apply to the pocki was initially diagnosed only by the ab- material at hand. sence of upper and lower third molars, and OTHER SPECIMENS EXAMINED: BrazilÐ by the small size of its remaining cheekteeth ParaÂ, Cachoeira Porteira (USNM 546290, (Dubost and Petter, 1978). Other distinctive 546291). French GuianaÐArataye (MNHN attributes of the holotype were reported by 1981.162), Awara (MNHN 1986.271), Cacao Voss (1988), who treated Daptomys as a ju- (MNHN 1979.345 [holotype of parvipes], nior synonym of Neusticomys. The new ma- 1981.1303, 1981.1304, 1986.272, 1986.273), terial closely resembles the type and corrob- Cayenne (MNHN 1970.224, 1981.1298, orates the status of N. oyapocki as a valid 1981.1299, 1986.274, 1986.275), Rorota species. (MNHN 1981.1305), Ouanary (MNHN The Paracou examples are both males. The Â 1981.1297), Piste St.-Elie (MNHN smaller specimen (MNHN 1995.992) we 1981.184), SauÈl (MNHN 1980.407, judge to be subadult because of its uniformly 1981.1296, 1986.270). GuyanaÐCuyuni- blackish pelage, undescended testes, conspic- Mazaruni, Kartabo (AMNH 42332, 42333, uous metapodial epiphyses, and open basi- 42882, 42885, 42891, 64140), Oko Moun- cranial sutures; although its molar dentition tains (USNM 46216); Upper Demerara-Ber- (see below) is completely erupted, the animal bice, Rockstone (AMNH 34651). Venezue- is obviously immature. The larger specimen laÐAmazonas, AcananÄa (USNM 406237), (AMNH 267597; ®gs. 48, 49B, 49D) appears Boca Mavaca (USNM 374662, 374664, to be a young adult; its pelage is also dark, 374665, 406062, 406063, 406233), Cerro but the color is distinctly brownish and ®nely Neblina Base Camp (USNM 560824), Es- ticked with tawny-banded hairs, the testes are meralda (AMNH 77303), Mt. Duida (AMNH scrotal, the metapodial epiphyses are incon- 77306), RõÂo Casiquiare (AMNH 78080), San spicuous (but perhaps not completely ab- Carlos de RõÂo Negro (USNM 560650); Bo- sorbed), and the basicranial sutures are lõÂvar, Auyantepui (AMNH 130733, 130784), closed (although not completely fused). In Mt. Roraima (AMNH 75634, 75635). fact, AMNH 267597 seems to be nearly the FIELD OBSERVATIONS: Both of our vouch- same age as the holotype and compares with ers of Nectomys melanius were collected by it closely in external and cranial dimensions O. Henry, whose ®eld notes indicate that one (table 25). The specimen from St.-EugeÁne is was trapped ``preÁs de la crique'' on 4 No- a fully adult male, with fur like the larger vember 1989, and the other ``sur la piste'' on Paracou specimen, well-worn molars, and 20 April 1990. fused basicranial sutures. Like the holotype, all of the three new include a bound volume of observations about the spec- specimens of Neusticomys oyapocki lack M3/ imens he examined in European museums. Page 73 re- m3, and the remaining molars are small by cords his description of Pelzeln's type of Hesperomys comparison with their homologs in the only rattus (in the NMW) as ``a young . . . Nectomys ... the other Guianan congener, N. venezuelae (®g. toe webbing quite visible''. Unfortunately, no other diagnostic characters were given, and we are not aware 49). As noted by Voss (1988), loss of the that anyone else has subsequently examined this speci- third molar is accompanied by morphological men. changes in the second, now the most poste- 100 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 48. Dorsal, ventral, and lateral views of the skull of Neusticomys oyapocki (AMNH 267597). Scale bar ϭ 10 mm. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 101

Fig. 49. Occlusal views of upper (A, B) and lower (C, D) molars of Neusticomys venezuelae (A, C [AMNH 69907, holotype]) and N. oyapocki (B, D [AMNH 267597]). Absence of upper and lower third molars conspicuously distinguishes N. oyapocki from N. venezuelae; the remaining cheekteeth of N. oyapocki are also absolutely smaller than their homologues in N. venezuelae. Additionally, the posterior lobe (metacone/hypocone cusp-pair) of M2 in N. oyapocki is greatly reduced relative to the anterior lobe (paracone/protocone) of that tooth, whereas the anterior and posterior lobes of M2 are subequal in N. venezuelae.

rior element in the toothrow. In all specimens riano [1991]). (3) A small orbicular apoph- of N. oyapocki, the hypocone/metacone lobe ysis of the malleus is present on the type (and of M2 is greatly reduced by comparison with only known fully adult specimen) of N. pe- that of N. venezuelae, and there is no trace ruviensis, but this structure is absent in both of a posteroloph. N. oyapocki and N. venezuelae; the character Neusticomys oyapocki can be distin- has not been described or illustrated for N. guished from other lowland congeners (for- mussoi. Table 26 summarizes these and other merly classi®ed as Daptomys; ®g. 50) by ad- relevant comparisons. ditional characters: (1) The ears and feet of Two peculiarities of the holotype of Neus- N. mussoi and N. peruviensis are cream-col- ticomys oyapocki noted by Voss (1988) are ored and contrast with the brownish dorsal apparently not diagnostic for the species. The body pelage, but the ears and feet of N. oya- type lacks masseteric tubercles, the bony pro- pocki and N. venezuelae are dark brown and cesses from which M. masseter super®cialis do not contrast with the dorsal fur. (2) In N. venezuelae and N. mussoi, the posterior edge originates in ichthyomyines, but a distinct of the inferior zygomatic root (zygomatic masseteric tubercle is present at the base of plate) lies above or just anterior to the anter- the inferior zygomatic root on each side of ocone of M1; in fully adult examples of N. the skull in the larger Paracou specimen oyapocki and N. peruviensis, however, the (AMNH 267597) and in MNHN 1995.3234 posterior edge of the inferior zygomatic root (from St.-EugeÁne). The type of N. oyapocki is located well anterior to the toothrow (®g. also appears to have an unusually narrow in- 48; also see illustrations in Musser and Gard- terorbital constriction by comparison with ner [1974], Voss [1988], and Ochoa and So- both of the other conspeci®c adults at hand, 102 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 25 Measurements (mm) and Weights (g) of Neusticomys oyapocki and N. venezuelae

and with specimens of N. venezuelae (table Oecomys Thomas 25). OTHER SPECIMENS EXAMINED: French Gui- For most of the last three decades, Neo- anaÐSt.-EugeÁne (MNHN 1995.3234), Trois tropical mammalogists have identi®ed small Sauts (MNHN 1977.775 [holotype]). specimens of Oecomys (formerly considered FIELD OBSERVATIONS: Both of our speci- a subgenus of Oryzomys)asO. bicolor and mens of Neusticomys oyapocki from Paracou large specimens as O. concolor following were taken in pitfall traplines in primary for- Hershkovitz (1960). Unpublished revisionary est. The ®rst example (AMNH 267597) was research, however, suggests that at least 13 collected on 14 August 1993 about 5 m from valid species of Oecomys are represented a small (ca. 1.4 m wide), shallow (ca. 20 cm among the many nominal taxa that Hersh- deep), clear, sandy-bottomed stream; the hab- kovitz lumped into bicolor and concolor (see itat at this site is perhaps best characterized Musser and Carleton, 1993). Four morpho- as moist creekside forest on level sandy soil logically diagnosable species are known to (®g. 51). The second animal (MNHN occur in French Guiana (table 27), of which 1995.992) was taken on 8 September 1993 two are represented among our vouchers. at approximately the same distance from a Pending the publication of a comprehensive slightly smaller stream, but on sloping, well- revision of this dif®cult genus, we offer pre- drained ground. The remains of small crabs liminary descriptions and diagnostic compar- (®g. 52) found along other streams in our isons of both Paracou species to document study area suggest that this species is not un- our identi®cations. common locally, but intensive trapping with Species of Oecomys are small to medium- Victors and Tomahawks set at streamside and sized murids, ranging in average adult body baited with crabs produced no additional weight from about 20 to 60 g. The dorsal fur specimens. is soft (not spiny) and, in adults with fresh 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 103

Fig. 50. Known collection localities of lowland species of Neusticomys based on specimens examined and the literature (see citations in footnotes to table 26). 1a, FRENCH GUIANA, Paracou; 1b, FRENCH GUIANA, St.-EugeÁne; 2, FRENCH GUIANA, Trois Sauts; 3, GUYANA, Cuyuni-Mazaruni, Kartabo; 4, PERU, Ucayali, Balta; 5, PERU, Madre de Dios, Pakitza; 6, VENEZUELA, Amazonas, Cerro Duida; 7, VENEZUELA, BolõÂvar, Los Pijiguaos; 8, VENEZUELA, BolõÂvar, San Ignacio YuruanõÂ; 9, VENEZUELA, Sucre, RõÂo NeverõÂ; 10, VENEZUELA, TaÂchira, 14 km SE Pregonero. pelage, usually some shade of reddish brown. or amber in dried skins) or lightly pigmented The ventral fur can be either self-colored (grayish), but apparently never blackish. (pure white) or gray-based with a super®cial Structurally, the hindfeet are short and broad, wash of white, buff, or orange. The mystacial with large plantar pads and semi-opposable vibrissae are long, extending well behind the ®fth digits (®g. 53B). Tails are unicolored in pinnae when laid back alongside the head. most species (dark above and below), and The dorsal pelage of the hindfeet is some- they are usually longer than the combined times indistinctly darker over the metatarsals length of head-and-body; a terminal tuft of than on the digits, but sharply de®ned meta- long hairs is present in some, but not all spe- tarsal spots or bands are absent; the plantar cies. surface of the hindfoot is either unpigmented Amazonian species of Oryzomys are some- (whitish in preservative, pink in life, brown times misidenti®ed in the ®eld as Oecomys 104 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 26 Geographic and Morphological Comparisons among Lowland Species of Neusticomys

(and vice versa), but differ externally by their Auyantepui in Estado BolõÂvar, Venezuela. much shorter mystacial vibrissae (not extend- Hitherto regarded as a junior synonym of O. ing behind the pinnae), distinctive hindfeet concolor (by Hershkovitz, 1960), O. trinita- (®g. 53A), and shorter tails that are often bi- tis (by Cabrera, 1961), or O. paricola (by colored (at least basally) and never have ter- Musser and Carleton, 1993), auyantepui is minal tufts of long hairs (see the account of unambiguously diagnosable from other Oryzomys below for more detailed descrip- named forms of Oecomys and merits recog- tions of external traits). nition as a distinct species. Distinguishing Species of Rhipidomys resemble Oecomys traits include its predominantly gray-based externally in possessing long mystacial vi- ventral fur, a distinctly tufted tail, lack of brissae; short-broad hindfeet with unpig- broadly shelved supraorbital margins and mented soles, large plantar pads, and semi- postorbital processes, a primitive carotid ar- opposable ®fth digits; and long, usually un- terial circulation, presence of an alisphenoid icolored, tufted tails. However, the hindfeet strut, a large postglenoid foramen, complete of Rhipidomys are distinctive, with darker closure of the subsquamosal fenestra, and a metatarsal markings, larger plantar pads, and distinctive range of morphometric variation relatively longer ®fth pedal digits (®g. 53C). (see table 27 and below). All of the speci- In addition, whereas female Oecomys have mens we refer to O. auyantepui are from the eight mammae, female Rhipidomys have Guiana subregion of Amazonia (®g. 54). only six. More detailed comparisons between In the hand, Oecomys auyantepui is an at- like-sized species of Oecomys and Rhipido- tractive mouse with soft reddish-brown fur mys that might be confused in the ®eld are that is much brighter in mature adults with provided in the account that follows. fresh glossy pelage than in juveniles, subadults, or specimens with obviously worn, Oecomys auyantepui Tate dull coats. The small ears are covered with a Figures 53B, 55, 56, 57A, 62B short but macroscopically visible pelage that VOUCHER MATERIAL: AMNH 266560, is colored essentially like that of the head 266564, 267593, 267595, 267596; MNHN and nape (not contrastingly darker). The ven- 1995.1027, 1995.1028. Total ϭ 7 specimens. tral fur, super®cially whitish, cream-colored, IDENTIFICATION: Oecomys auyantepui was or pale buff, is sharply set off from the red- originally described by Tate (1939) from two dish-brown fur of the sides and back. The specimens collected at 1100 m elevation on ventral fur is predominantly gray-based in 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 105

Fig. 51. Pitfall trapline that captured Neusticomys oyapocki in creekside primary forest at Paracou. This trapline roughly paralleled the small stream illustrated in ®gure 9, which was only about 10 m to the right of this view. 106 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

of average length relative to the diastema (neither very short nor very long by oryzomyine standards), with wide and more-or- less evenly convex lateral margins. The roof of the mesopterygoid fossa is completely bony, with no trace of sphenopalatine perforations in most of the specimens at hand. A large stapedial foramen on the medial surface of the bulla, a translucent squamosal- alisphenoid groove on the internal surface of the braincase, and a distinct sphenofrontal foramen in the rear of the orbit indicate that the pattern of carotid arterial supply is primitive (pattern 1 of Voss, 1988). A robust diagonal strut of the alisphenoid bone separates the foramen ovale accessorius from the buccinator-masticatory foramen on both sides of the skull in most specimens, and the subsquamosal fenestra is invariably absent (®g. 57A; see table 29). Fig. 52. Dorsal and ventral views of one of Morphometrically, series of Oecomys au- several pseudothelphusid crab carapaces found yantepui that we measured from Guyana, near small streams at Paracou. Patterns of break- French Guiana, and Brazil (AmapaÂ) are re- age along the posterior and ventrolateral margins markably similar (table 28). In fact, the most of this example match those observed on the car- divergent specimen we examined is the Ve- apaces of like-sized pseudothelphusids eaten by nezuelan type (AMNH 131156), which is captive ichthyomyines (Voss et al., 1982: ®g. 3). larger than most of the other material at hand Such remains provided the ®rst evidence that ichthyomyines were resident in our study area, an but does not differ in any qualitative external inference subsequently con®rmed by trapped or craniodental character. Additional Vene- specimens of Neusticomys oyapocki. Both views zuelan material would be useful to determine about 2ϫ. whether the type represents a western population characterized by large size, or is mere- most of the specimens at hand, but the fur of ly an unusually large individual. the chin and throat is usually self-colored (all Oecomys auyantepui is intermediate in pale), and a few specimens have self-colored size to other congeneric species that occur in fur extending caudally along the ventral mid- the Guiana subregion of Amazonia, two of line to the groin. The hindfeet are either cov- which (O. bicolor and O. rutilus) are sub- ered uniformly with pale buffy hairs or the stantially smaller, and four of which (O. con- metatarsus is indistinctly darker than the color, O. rex, O. roberti, O. trinitatis) are toes, but a distinct metatarsal band of black- larger. Additionally, other Guianan subregion ish fur is apparently never present. Undam- species differ from auyantepui by having en- aged tails are uniformly dark (almost black- tirely self-colored (pure white) ventral fur ish in some specimens) with a terminal tuft (bicolor, rutilus), untufted tails (concolor, of hairs that are distinctly longer (6±10 mm) rex, roberti, trinitatis), more-or-less bicol- than the short (Ͻ2 mm) hairs on the proximal ored tails (trinitatis), broadly shelved supra- part of that organ. orbital margins (rex), a derived pattern of ca- The skull (®gs. 55, 56) is unremarkable in rotid arterial supply (concolor), con¯uent ac- general aspect, with the short rostrum, shal- cessory oval and buccinator-masticatory fo- low zygomatic notches, convergent interor- ramina (most bicolor specimens), and/or bit, beaded supraorbital margins, wide-long patent subsquamosal fenestrae (bicolor, con- palate, and small bullae characteristic of this color, rutilus, trinitatis). Instead, Oecomys oryzomyine genus. The incisive foramina are auyantepui is morphologically most similar 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 107

TABLE 27 Diagnostic Morphological Comparisons among French Guianan Species of Oecomys (Tabulated data include information from specimens collected throughout the Guiana subregion of Amazonia.)

to O. paricola, a species that we do not rec- of all museum material currently identi®ed as ognize as occurring in the Guiana subregion. O. paricola, we did not include specimens so Oecomys paricola (Thomas, 1904) was determined from other inter¯uvial regions, originally described on the basis of a very some of which exhibit characters not shown young female specimen (BMNH 4.7.4.63) by the type. collected at ``IgarapeÂ-Assu'' (ϭ IgarapeÂAcËu Oecomys paricola is about the same size at 1Њ07ЈS, 47Њ37ЈW; Paynter and Traylor, as O. auyantepui, and these two species are 1991) near BeleÂm. To evaluate the hypothe- perhaps indistinguishable in pelage color and sis that O. paricola and O. auyantepui are external morphology. Cranially, however, synonyms (Musser and Carleton, 1993), we two qualitative characters (table 29, ®g. 57) examined the type of paricola and 21 addi- permit unambiguous identi®cation. (1) A tional specimens collected in the same inter- well-developed alisphenoid strut almost al- ¯uvial region (south of the Amazon and east ways separates the buccinator-masticatory of the Rio Tocantins).15 Because it was not and accessory oval foramina in auyantepui, our intention to evaluate the taxonomic status but an alisphenoid strut is missing and these foramina are consistently con¯uent in pari- 15 We examined specimens of Oecomys paricola from cola. (2) Whereas the subsquamosal fenestra the following localities: BrazilÐParaÂ, Belem (USNM is consistently absent in auyantepui, a dis- 393819, 393842, 461386, 545232, 545233), Capim (AMNH 188963, 203391, 203392, 203397; USNM tinct subsquamosal fenestra (separated from 461385, 461388±461395, 461400), IgarapeÂAcËu the postglenoid foramen by the hamular pro- (BMNH 4.7.4.63, 4.7.4.64, 4.7.4 108). cess of the squamosal bone) is always pre- 108 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 53. Plantar views of left hindfeet of Oryzomys megacephalus (A, AMNH 266527), Oecomys auyantepui (B, AMNH 267595), and Rhipidomys nitela (C, AMNH 267583) illustrating diagnostic intergeneric differences. The hindfeet of Amazonian species of Oryzomys are proportionately narrow; the three central digits (II, III, and IV) are conspicuously longer than the outer digits (I and V); digit V is not opposable; the plantar epithelium is at least partially pigmented (grayish in fresh material); the plantar pads are small, hard, and more-or-less conical; and the skin between the pads is studded with small, scale-like tubercles. By contrast, the hindfeet of Oecomys and Rhipidomys are proportionately broader; digits I and V are longer relative to II±IV; digit V is semi-opposable; the plantar epithelium is usually entirely unpigmented (pinkish in fresh material); the plantar pads are larger, softer, cushion-like structures; and the skin between the plantar pads is smooth or creased but never tuberculate. Despite differences in hindfoot size and other features between the illustrated examples of O. auyantepui and R. nitela, their respective genera do not seem to be consistently distinguishable by pedal morphology. sent in paricola. Although their close resem- 1983.395), St.-EugeÁne (1994.124, blance in other respects suggest that auyan- 1995.3235, 1998.1844), Trois Sauts (MNHN tepui and paricola are likely to be sister taxa, 1983.398). GuyanaÐCuyuni-Mazaruni, we recognize them as valid (diagnosable) Kartabo (AMNH 64135); Potaro-Siparuni, 5 species allopatrically distributed north and km SE Surama (ROM 102944, 103051, south of the Amazon. 103052, 103244, 103288); Upper Demerara- OTHER SPECIMENS EXAMINED: BrazilÐ Berbice, 18 km SW Kwakwani (AMNH AmapaÂ, Serra do Navio (USNM 393820, 269829, 269830), Tropenbos (ROM 103433, 393821, 394239±394243, 394246±394249, 103502). VenezuelaÐBolõÂvar, Auyantepui 461521); Amazonas, 80 km N Manaus (AMNH 131108, 131156 [holotype]). (USNM 579996±580001). French Gui- FIELD OBSERVATIONS: All of our de®nite anaÐArataye (MNHN 1986.865; USNM records of Oecomys auyantepui from Para- 578015, 578019), Iracoubo (1983.394, cou are based on collected specimens. Of 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 109

Fig. 54. Known collection localities for Oecomys auyantepui based on specimens examined. 1, BRAZIL, AmapaÂ, Serra do Navio; 2, BRAZIL, Amazonas, 80 km N Manaus; 3, FRENCH GUIANA, Arataye; 4, FRENCH GUIANA, Iracoubo; 5, FRENCH GUIANA, Paracou; 6, FRENCH GUIANA, St.- EugeÁne; 7, FRENCH GUIANA, Trois Sauts; 8, GUYANA, Cuyuni-Mazaruni, Kartabo; 9, GUYANA, Potaro-Siparuni, 5 km SE Surama; 10, GUYANA, Upper Demerara-Berbice, 18 km SE Kwakwani; 11, GUYANA, Upper Demerara-Berbice, Tropenbos; 12, VENEZUELA, BolõÂvar, Auyan-tepui.

these, one was taken on the ground in a Sher- 1995.1023±1995.1026. Total ϭ 12 speci- man trap, three were taken in Victor snap mens. traps tied to lianas 1.2±1.5 m above the IDENTIFICATION: Oecomys rutilus was orig- ground (®g. 58), and three were taken in ar- inally described by Anthony (1921b) based boreal platform traps 7.2±10.5 m above the on a single specimen collected by W. Beebe ground. Three specimens were trapped in at Kartabo, Cuyuni-Mazaruni District, Guyana. creekside primary forest, two in well-drained Anthony (p. 4) characterized rutilus as ``A primary forest, and two in swampy primary small, brightly colored species, with very short forest. All specimens were found in the traps tail and clear white under parts'', and he re- at or soon after dawn. marked that it was quite distinct from another species originally described from Guyana, Oecomys rutilus Anthony Oecomys nitedulus Thomas (1910): Figures 55, 56 VOUCHERS: AMNH 266561, 267584, Compared with Oecomys nitedulus, collected at the 267586, 267588±267591, 269121; MNHN same place, rutilus is somewhat smaller super®cially, 110 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 55. Dorsal and ventral cranial views of four species of Oecomys from French Guiana. Left to right: Oecomys rex (DIM 58); O. auyantepui (USNM 578015); O. bicolor (USNM 394252); O. rutilus (AMNH 267589). All views about ϫ2.

much brighter in color, with longer, softer, pelage, same Kartabo series identi®ed by Anthony as shorter tail and conspicuously smaller skull. nitedulus.'' Despite Anthony's explicit statement that Oecomys rutilus was listed as a valid spe- rutilus and nitedulus were valid sympatric cies by Musser and Carleton (1993), but no species, Hershkovitz (1960) listed both account of its diagnostic characteristics has names as synonyms of Oecomys bicolor, a yet been published to supplement Anthony's taxon originally described (Tomes, 1860) preliminary observations. To identify our from eastern Ecuador. Hershkovitz (p. 539) Paracou material, we examined every avail- admitted that he had not seen Anthony's type able specimen of small Oecomys from the of rutilus, but remarked that ``Judged by the Guiana subregion of Amazonia, including original description, it is a subadult of the the types of rutilus and nitedulus. In the fol- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 111

lowing account we provisionally accept the hypothesis that nitedulus and bicolor are conspeci®c (Hershkovitz, 1960; Musser and Carleton, 1993), but we note that this synonymy remains untested by published analyses of character data, and that we have not made a careful study of typical bicolor (from eastern Ecuador). Instead, our taxonomic comparisons are based exclusively on Guian- an material.16 The frequency distribution of the length of the upper molar row (LM) for small Oeco- mys collected in the Guiana subregion of Amazonia (N ϭ 66) is distinctly bimodal (®g. 59), with one peak in the interval 3.2± 3.4 mm (including the type of rutilus with LM ϭ 3.26 mm) and another in the interval 3.6±3.9 mm (including the type of nitedulus with LM ϭ 3.79 mm). Although this univariate graph does not of itself show any morphometric discontinuity that could be used to sort specimens into discrete size classes (all frequency intervals between 3.0 and 4.0 mm are occupied), the bimodal pattern clearly indicates that our sample is heterogeneous. Fortunately, variation in other characters is correlated with molar toothrow length and provides unambiguous evidence for species diagnosis. In fact, the specimens we measured of the smaller species, O. rutilus, have molar toothrows shorter than 3.5 mm, whereas our measured specimens of O. bicolor have toothrows longer than 3.5 mm. With larger samples of both species, however, it seems inevitable that the observed ranges of variation in this dimension will eventually be found to overlap. Oecomys rutilus and Guianan specimens of O. bicolor (ϭ nitedulus) are similar in external appearance: both are small, usually reddish mice with dark, tufted tails and pure white venters (table 27). Although rutilus averages smaller than bicolor in all standard

Fig. 56. Lateral cranial and mandibular views 16 We examined specimens of Oecomys bicolor from of four species of Oecomys from French Guiana. the following localities: BrazilÐAmapaÂ, Serra do Navio Top to Bottom: Oecomys rex (DIM 58); O. au- (USNM 393822±393830, 393833±393839, 393843, yantepui (USNM 578015); O. bicolor (USNM 394251±394253, 461522). French GuianaÐLes No- uragues (AMNH 269823; V-882, Ϫ895, Ϫ896, Ϫ898, 394252); O. rutilus (AMNH 267589). All views Ϫ912), RivieÁre Approuague (MNHN 1983.393). Guy- about ϫ2.5. anaÐCuyuni-Mazaruni, Kartabo (AMNH 41910, 60641, 64130, 64132, 64133); Potaro-Siparuni, 5km SE Surama (ROM 102970); ``Essequibo River 13 mi from mouth'' (BMNH 6.4.8.31 [holotype of nitedulus]; ``R. Supinaam'' (BMNH 10.9.29.18). 112 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 57. Ventrolateral cranial views of Oecomys auyantepui (A, USNM 578015) and O. paricola (B, USNM 461389). In O. auyantepui, a bony strut of the alisphenoid (bsa) separates the buccinator- masticatory foramen (bmf) from the foramen ovale accessorius (foa), whereas in O. paricola the alisphenoid strut is absent and the two foramina are con¯uent (bmf ؉ foa). Additionally, O. auyantepui consistently lacks the subsquamosal foramen (ssf), a prominent perforation in the posterolateral wall of the braincase that is consistently present in O. paricola and many other congeners.

TABLE 28 Measurements (mm) and Weights (g) of Adult Oecomys auyantepuia 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 113

TABLE 29 external dimensions (table 30), there is suf- Comparisons of State Frequencies for Two ®cient morphometric overlap between them Qualitative Craniodental Characters in that no measurement is diagnostic. Most of Oecomys auyantepui and O. paricolaa the material at hand is preserved in ¯uid, so it is dif®cult to evaluate Anthony's statement that these species differ in fur color; however, no consistent differences were apparent in the few dried skins we examined. The species difference in tail length mentioned by Anthony (presumably meant to be considered in relation to the head-and-body) is also dif®cult to assess because we have not measured any specimens of bicolor ourselves (minor differences in measurement methodology can produce substantial artifactual divergence in computed ratios). Instead, two other characters are useful for ®eld identi®- cation. As observed by Anthony, the dorsal fur is longer in Oecomys rutilus than it is in O. bicolor, and the two species also differ in fur texture. In rutilus, the fur averages about 6±

Fig. 58. Tangled lianas in the understory of swampy primary forest at Paracou, capture site for Oecomys auyantepui (trapped 1.2 m above the ground, white arrow) and a typical substrate for other arboreal small mammals that occasionally descend to near ground level (e.g., Micoureus demerarae and Oecomys rutilus). 114 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 59. Frequency histogram of upper molar toothrow length (LM) for 66 specimens of small Oecomys collected in the Guiana subregion of Amazonia. This sample includes material that we identify as O. rutilus (including Paracou vouchers and other conspeci®c specimens listed in the text) and O. bicolor (see text footnote 16).

8 mm middorsally near the rump, and it feels is visually conspicuous and is useful for sort- soft and lax when ruf¯ed because the unre- ing skins in combination with other traits. sistant hairs return slowly to their normal Skulls of Oecomys bicolor and O. rutilus (unruf¯ed) condition. By contrast, the dorsal differ in size (®gs. 55, 56; table 30) and in fur of bicolor is only 4±5 mm long in most the usual size-correlated proportions, but the specimens, and because bicolor is the larger incisive foramina are notably longer relative species this absolute difference makes the to the diastema in bicolor than in rutilus, a pelage appear relatively much shorter. Also, difference that is not attributable to standard the short fur of bicolor feels ``crisp'' to the patterns of muroid craniodental allometry. touch because the stiffer hairs are more re- Although bicolor and rutilus are similar in sistant to ruf¯ing. most qualitative osteological traits, they dif- The tuft of hairs at the tail-tip is signi®- fer signi®cantly in the frequency of occur- cantly longer in Oecomys rutilus, averaging rence of the alisphenoid strut, an ossi®cation 7.7 Ϯ 1.4 mm (observed range: 5±11 mm, N that occurs bilaterally in most rutilus, but is ϭ 19). In O. bicolor, this tuft measured 4±5 bilaterally absent in most bicolor (table 31). mm in all nine specimens we examined with Otherwise, these species are craniodentally intact tail tips. Since the long hairs of the tail similar, both having primitive carotid circu- tuft are often exposed to bending and com- lations (pattern 1 of Voss, 1988), beaded but pression in museum trays, it is probable that unshelved supraorbital margins, and consis- some of the tufts we measured were broken tently large postglenoid foramina and sub- short, so the mean tuft length determined squamosal fenestrae. from fresh specimens with undamaged tails Although Oecomys rutilus and O. bicolor might be larger for both species than the val- are widely distributed in the Guiana subre- ues reported here. Nonethless, the difference gion of Amazonia, they have been collected 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 115

TABLE 30 Measurements (mm) and Weights (g) of Adult Oecomys rutilus and O. bicolora

sympatrically only at Kartabo (Cuyuni-Ma- 42910), a mature adult female (not a subadult zaruni District, Guyana) and Les Nouragues as conjectured by Hershkovitz, 1960), are all (French Guiana). Musser and Carleton within the range of variation exhibited by (1993) gave the range of rutilus as restricted French Guianan specimens (table 30). A few to Guyana, Surinam, and French Guiana, but measurements of Brazilian and Venezuelan material from San Ignacio YuruanõÂ (Vene- specimens exceed the observed range of var- zuela) and from 80 km N Manaus (Brazil) iation for homologous measurements of previously misidenti®ed as bicolor (e.g., by French Guianan material, but the discrepan- Voss, 1991: table 23; Voss and Emmons, cies are small in all cases. 1996: appendix 7) extend the range of this OTHER SPECIMENS EXAMINED: BrazilÐ species east and south of the Guianas proper Amazonas, 80 km N Manaus (USNM (®g. 60). Although we have not examined 579992±579995). French GuianaÐCacao any material of rutilus from outside the Gui- (MNHN 1983.400), Les Nouragues (AMNH ana subregion of Amazonia, the small unnamed Oecomys that Patton et al. (2000) re- 269822, V-889, Ϫ892, Ϫ899, Ϫ900, Ϫ906, ported from the Rio JuruaÂ appears to be sim- Ϫ913), St.-EugeÁne (MNHN 1995.3236, ilar in some respects and merits close com- 1995.3237, 1998.1845, 1998.1846). Guya- parison in any future revisionary study. naÐCuyuni-Mazaruni, Kartabo (AMNH The material we examined of Oecomys ru- 42910 [holotype], 142820); Upper Demer- tilus is remarkably uniform with little indi- ara-Berbice, Dubulay Ranch (AMNH cation of signi®cant variation among samples 267745), 18 mi SW Kwakwani (AMNH collected at widely separated localities. Thus, 269828), Tropenbos (ROM 103482). Suri- measurements of the Guyanese type (AMNH namÐSuriname, Carolinakreek (FMNH 116 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 60. Known collection localities for Oecomys rutilus based on specimens examined. 1, BRAZIL, Amazonas, 80 km N Manaus; 2, French Guiana, Cacao; 3, FRENCH GUIANA, Les Nouragues; 4, FRENCH GUIANA, Paracou; 5, FRENCH GUIANA, St.-EugeÁne; 6, GUYANA, Cuyuni-Mazaruni, Kartabo; 7, GUYANA, Upper Demerara-Berbice, Dubulay Ranch; 8, GUYANA, Upper Demerara- Berbice, 18 mi SW Kwakwani; 9, GUYANA, Upper Demerara-Berbice, Tropenbos; 10, SURINAM, Suriname, Carolinakreek; 11, VENEZUELA, BolõÂvar, San Ignacio YuruanõÂ.

95591). VenezuelaÐBolõÂvar, San Ignacio Six specimens (50%) were taken in well- YuruanõÂ (AMNH 257268, USNM 448576). drained primary forest, 4 (33%) in swampy FIELD OBSERVATIONS: Although we saw primary forest, 1 (8%) in creekside primary small reddish mice racing along branches or forest, and 1 in roadside secondary growth. lianas at night on many occasions, all of our Both shot specimens were taken at night, and de®nite records of Oecomys rutilus at Para- all of the other specimens were found in the cou are based on collected specimens. Of our traps at dawn. 12 vouchers, 7 (58%) were taken in pitfall traps, 2 (17%) were shot at heights of 4±5 m Oligoryzomys fulvescens (Saussure) in trees, 1 (8%) was taken in a Victor snap- trap tied to a liana 2 m above the ground, 1 VOUCHER MATERIAL: AMNH 267022, was taken in a Sherman trap placed on a li- 267023; MNHN 1998.673. Total ϭ 3 speci- ana 30 cm above the ground, and 1 was taken mens. in a platform trap 15.9 m above the ground. IDENTIFICATION: Although the genus Oli- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 117

TABLE 31 (near BeleÂm) were identi®ed as comprising a Comparison of Character-State Frequencies distinct species, O. microtis (J. A. Allen). for Occurrence of the Alisphenoid Strut in The taxonomic status of Oligoryzomys Oecomys bicolor and O. rutilusa populations in French Guiana (®rst reported as O. delicatus by Charles-Dominique, 1993) and in the Brazilian state of AmapaÂ (tenta- tively identi®ed as O. navus by Carvalho, 1962) has yet to be critically evaluated. Be- cause these regions constitute a geographic hiatus between the known ranges of O. fulvescens and O. microtis (as delimited by Carleton and Musser, 1989; see above), we compared our Paracou vouchers with typical material of both species. Craniodental measurements of our three Paracou specimens (table 32) fall almost entirely within the range of morphometric variation observed by Carleton and Musser (1995) for a large sample of typical Oligo- ryzomys fulvescens (topotypes and other goryzomys (formerly a subgenus of Oryzo- specimens from Veracruz, Mexico); the few mys; see Carleton and Musser, 1989) has exceptions are Paracou values (of BM1, BIF, never been comprehensively revised, several BZP, and BB) that only exceed the observed publications have at least partially clari®ed range in homologous dimensions of typical the species-level systematics of Oligoryzo- fulvescens by 0.1 mm. Our vouchers are larg- mys in certain regions, notably Paraguay er than average fulvescens from Mexico, but (Myers and Carleton, 1981), Bolivia (Olds Carleton and Musser (1995) documented a and Anderson, 1987), and Central America southward cline of increasing size among (Carleton and Musser, 1995). Unfortunately, their Central American samples; in cranio- the Oligoryzomys of northern South America dental measurements, our vouchers more have received no revisionary attention to closely resemble Costa Rican and Panaman- date. In a preliminary review of the genus, ian populations that Carleton and Musser re- however, Carleton and Musser (1989) hy- ferred to O. f. costaricensis than they do typ- pothesized that a single widespread polytypic ical (Mexican) material. speciesÐO. fulvescensÐextends from Mex- The only noteworthy morphometric con- ico throughout most of Central America, trast between our Paracou vouchers and typ- thence southward into Colombia and north- ical Oligoryzomys fulvescens appears to be ern Ecuador and eastward throughout Vene- the ratio of tail length to head-and-body zuela, Guyana, and Surinam. Included as length. That ratio is about 1.16 in our two subjective synonyms in a subsequent syn- specimens with intact tails, whereas the ratio opsis of O. fulvescens (see Carleton and of mean tail length to mean head-and-body Musser, 1995) were costaricensis J. A. Allen, length calculated from Carleton and Musser's delicatus J. A. Allen and Chapman, navus (1995) data for Mexican fulvescens is 1.31. Bangs, messorius Thomas, tenuipes J. A. Al- The same ratio for Costa Rican and Pana- len, munchiquensis J. A. Allen, lenis Gold- manian samples of fulvescens measured by man, mayensis Goldman, engraciae Osgood, those authors ranges from 1.27 to 1.44. In and paci®cus Hooper. As de®ned geograph- side-by-side comparisons, Mexican and Cen- ically by Carleton and Musser (1989), O. ful- tral American skins of fulvescens appear vis- vescens was not known to occur east of Su- ibly longer-tailed than our Paracou vouchers. rinam or south of the Amazon. Instead, pop- Subtle qualitative cranial differences are ulations of Oligoryzomys extending along the present between our three specimens and the entire south bank of the Amazon from near Mexican and Costa Rican exemplars of ful- the headwaters of that river to its mouth vescens with which we compared them. For 118 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 32 Sex, Measurements (mm), and Weights (g) of Oligoryzomys Vouchers from Paracou with Comparative Data from Samples of Typical O. fulvescens and O. microtis (All measurements from adult specimens.)

example, the frontal sinuses appear slightly veolus, position of the incisive foramina rel- more in¯ated in the Mexican and Central ative to the toothrows, pelage color and tex- American samples, producing a noticeably ture), our vouchers appear to be indistin- larger swelling behind the lacrimal bone in guishable from Mexican and Central Amer- the front of the orbit than that seen in our ican examples of fulvescens. French Guianan material. The rostrum also Confusingly, the Paracou material is also appears relatively longer and more slender, morphometrically similar to typical Oligo- and the upper incisors perhaps less strongly ryzomys microtis as represented by Allen's opisthodont in Mexican and Central Ameri- (1916a) original series (and several topo- can specimens than in our vouchers. These types, table 32) collected by Leo E. Miller are not conspicuous contrasts, however, and on the ``Lower Rio Solimoens (®fty miles their possible taxonomic signi®cance is dif- above mouth)'', a locality that can now be ®cult to assess without a careful study of the restricted with some con®dence to a speci®c many hundreds of museum specimens that site on the north bank of the river.17 Apparent are now available from dozens of geographically intermediate localities. In other quali- 17 According to Allen's (1916b) gazetteer of the Roo- tative characters, including those that have sevelt Brazilian Expedition, Miller collected at this locality from 16 to 30 April 1914. A letter from Miller to previously been used to diagnose species of F. M. Chapman (in the AMNH Ornithology archives) Oligoryzomys (e.g., the dorsal projection of written in Manaus on 24 April 1914, however, states that the capsular process of the lower incisor al- he had just ``. . . spent a week up on the Solimoens, 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 119

mensural differences between these very based on our side-by side comparisons of small samples (3 and 10 specimens, respec- skins. Small differences in ear size are hard tively) are unimpressive, especially when the to evaluate without measurements taken in considerable distance between collection lo- the ¯esh, which Leo E. Miller (the collector calities (roughly 1300 km) is taken into ac- of Allen's specimens) did not record, and our count. Length of the molar toothrow is the visual comparisons of dried ears between only nonoverlapping measurement, but the typical material of microtis and fulvescens re- mean sample difference (0.2 mm) is trivial. vealed no obvious size contrast. The reput- The few specimens at hand from immediate- edly diagnostic short tail of microtis is also ly south of the Amazon (e.g., AMNH 95983, hard to assess. Although Miller's ®eld mea- 188964) have toothrow measurements that surements of the type (AMNH 37091) indi- do overlap those of our vouchers, so very cate that its tail was slightly shorter than the large samples will probably be required to head-and-body,18 three paratypes (AMNH demonstate any signi®cant divergence in this 37088, 37089, 37097) have an average ratio dimension between Oligoryzomys popula- of tail to head-and-body of 1.20. Because tions on opposite banks of the river, if any Miller did not indicate whether his measure- such difference in fact exists. ments were of complete or bobbed tails Allen's (1916a) original description of Oli- (AMNH 37096 has an obviously bobbed-tail goryzomys microtis emphasized the diagnos- measurement of 20 mm with no accompa- tic value of pale coloration, relatively small nying notation to that effect), and because ears, and short tail for distinguishing this most of his skins no longer have intact tail- species from other congeners. In coloration, tips (the result of bending and compression however, Allen's material of microtis appears in museum trays), it is now impossible to to be indistinguihable from typical fulvescens evaluate the ratio of tail to head-and-body length in Allen's series. Other specimens of with a [G]erman friend, and got some nice things in the microtis (sensu Carleton and Musser, 1989), mammal line . . . but I hope to go to another fazenda however, are not very short-tailed: one spec- within a day or two.'' Miller therefore visited (or in- imen from the south bank of the lower Am- tended to visit) two localities near Manaus between 16 and 30 April. Perplexingly, Miller's skin tags give only azon (AMNH 203400) has a tail:body ratio ``Solimoens'' as the locality where he worked in this of 1.17, whereas another (AMNH 188964) interval, and his ®eld catalog (AMNH Mammalogy ar- has an improbably large ratio of 1.57. Given chives) contains no additional geographic information. the usual methodological inaccuracies asso- Fortunately, this problem did not escape the notice of mammalogists who were still in time to obtain at least ciated with muroid tail measurements (How- a partial ®rst-hand clari®cation: in a letter dated 4 June ell, 1924), any taxonomic inferences based 1945, G. H. H. Tate wrote Miller (then retired and living on this dimension in the small samples at in Connecticut) asking him to provide additional geo- hand would be premature. graphic details about the types of Proechimys kermiti We found no obvious qualitative charac- (collected 20 April 1914) and Oligoryzomys microtis (collected 29 April 1914). Miller replied in a letter dated ters to distinguish Oligoryzomys fulvescens 6 June 1945 that ``After a short time in Manaos, I went (as represented by our Mexican and Central down the Rio Negro and up the Solimoes about 50 or American exemplars) from typical O. micro- 60 miles to make the collections on the North bank of tis, and in view of the negligible morpho- the river. The location is approximately between 3Њ and 4Њ S Lat., and Long. 61Њ. I think this will locate the spot metric differences indicated by table 32 we closely enough for your purposes.'' (italics ours; corre- are unable to con®dently assign our Paracou spondence in AMNH Mammalogy archives). Because material to one or the other taxon. Although Miller did not mention collecting at separate localities on the 20th and the 29th, the second sojourn from Ma- naus mentioned in his letter to Chapman of 24 April 18 Miller's original measurements, recorded in pencil 1914 may have been a return trip to his previously pro- on a small ®eld label tied to the right hindfoot of the ductive site. Based on Miller's reply to Tate, Moojen type skin, were correctly reported by Allen (1916a) as (1948) published the restricted type locality of P. kermiti 183 mm (total length), 90 mm (tail), and 20 mm (hind- as ManacaparuÂ, a settlement located 70 km WSW of foot). However, Olds and Anderson (1987: 271) stated Manaus on the north bank of the lower SolimoÄes at that ``. . . the tail of the holotype is more than half the 3Њ08ЈS, 60Њ01ЈW (Paynter and Traylor, 1991). We as- total length, a discrepancy noted by Goodwin (1953).'' sume that the type of O. microtis was collected at the In fact, Goodwin (p. 299) did not note any discrepancy same place. and reported the same external measurements as Allen. 120 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 an obvious implication of the preceding dis- cept in juveniles), but the ventral hair bases cussion is that fulvescens and microtis are are always dark gray. The mystacial vibrissae conspeci®c, it is also possible that multivar- are short, not extending beyond the tips of iate analyses of larger samples could detect the pinnae when laid back alongside the morphological discontinuities that are not head. The dorsal surface of the hindfoot is now apparent; karyological and biochemical unmarked by dark metatarsal spots or bands, comparisons among key geographic popula- and the naked plantar surface is pigmented tions could likewise contribute to an assess- (grayish in life and in ¯uid-preserved mate- ment of these and other phenotypically sim- rial, blackish on dried skins). The hindfoot ilar nominal taxa of Oligoryzomys as valid (®g. 53A) appears narrow because the three species. Lacking the time to undertake such central pedal digits (II, III, and IV) are much a critical revisionary study for this faunal re- longer than the outer digits (I and V); the port, we simply apply the older name to our ®fth digit is not semi-opposable (its claw not Paracou material. extending to the end of the ®rst phalange of OTHER SPECIMENS EXAMINED: BrazilÐ dIV). Tails are usually at least partially bi- Amazonas, near ManacaparuÂ on north bank colored (at least near the base), average about of lower Rio SolimoÄes (AMNH 37088, as long as (never much shorter or longer 37089, 37091 [holotype of microtis], 37092± than) the combined length of head-and-body, 37097, 37157); ParaÂ, Capim (AMNH and appear quite naked (a sparse caudal pel- 188964, 203400), PoÃrto de Moz (AMNH age is visible only under magni®cation). 95983), Vilarinho do Monte (AMNH 95984± Contrasting external characters of Oecomys 95986, 95997). Costa RicaÐPuntarenas, species, which are sometimes misidenti®ed CanÄas Gordas (AMNH 142440±142458, as Oryzomys (and vice versa) by ®eldwork- 142490±142495, 142500). MexicoÐVera- ers, are given in the introductory account for cruz, Jalapa (AMNH 12536/10846±12541/ that genus (above). 10851, 12543/10853±12549/10959, 12583± Three species, Oryzomys macconnelli, O. 12585). megacephalus, and O. yunganus, occur sym- FIELD OBSERVATIONS: All of our de®nite patrically at Paracou as they probably do records of Oligoryzomys fulvescens at Para- throughout the Guiana subregion of Ama- cou are based on collected specimens. All zonia (see range maps in Musser et al., three of our vouchers were caught by hand 1998). Whereas adult specimens of O. mac- at night in roadside secondary growth (®g. connelli are easily recognized by external 13), where two were encountered running on characters, accurate ®eld identi®cations of O. the ground through sparse weeds and the megacephalus and O. yunganus are more dif- third was climbing among the dried leaves of ®cult. Juvenile Oryzomys, which lack the di- a felled tree. agnostic external dimensions and coloration of adults, cannot be reliably identi®ed in the Oryzomys Baird ®eld.

The Amazonian species of Oryzomys were Oryzomys macconnelli Thomas recently revised by Musser et al. (1998), whose species-level taxonomy is followed VOUCHER MATERIAL: MNHN 1998.674± herein. As a group, Amazonian species of 1998.676. Total ϭ 3 specimens. Oryzomys are easily distinguished from other IDENTIFICATION: Our three Paracou vouch- sympatric rodents by external characters. ers (collected by O. Henry, see below), to- These are medium-sized murids, ranging in gether with additional material that we ex- average adult body weight from about 40 to amined from other French Guianan localities, 80 g. The dorsal fur is soft (not spiny) and agree with Musser et al.'s (1998: 225±232) varies in color (depending on species, age, description of Oryzomys macconnelli, an and stage of molt) from rich reddish or tawny identi®cation that we con®rmed by direct hues to drab brown. The ventral fur is su- comparison with Thomas's (1910) type series per®cially whitish or whitish gray, contrast- from the Supenaam River, Guyana. Although ing abruptly in color with the dorsal fur (ex- the French Guianan material averages small- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 121

er than the type series in many dimensions cies with which it is sympatric in French (table 33), most French Guianan specimens Guiana, a contrast that is obvious in cranial are young adults (with lightly worn molars) comparisons (Musser et al., 1998) but can whereas the type series is composed of older also be seen in living specimens. specimens (with more advanced toothwear). OTHER SPECIMENS EXAMINED: French Gui- In the absence of other noteworthy differ- anaÐArataye (MNHN 1983.371±1983.373, ences between the two series, most of the 1986.276), St.-EugeÁne (MNHN 1994.126, observed measurement divergence could be 1994.127, 1995.208, 1998.1842±1998.1844), attributed to sample age composition. The SauÈl (MNHN 1983.365, 1983.367). Guya- longer molar rows of the Guyanese series, naÐ``River Supinaam'' (BMNH 10.5.4.29± however, cannot be attributed to advanced 10.5.4.34 [type series]). VenezuelaÐBolõÂ- age, and the observation of even longer var, San Ignacio YuruanõÂ (AMNH 257236± toothrows in some Venezuelan series (LM 257238; MHNLS 7831, 7836, 7880, 8075, averages 5.2 mm in 12 specimens that we 8076, 8088; USNM 448584±448586). measured from Estado BolõÂvar) suggests that FIELD OBSERVATIONS: Our only records of a real east-to-west size gradient may exist Oryzomys macconnelli at Paracou are based among O. macconnelli populations from the on three specimens trapped by O. Henry, Guiana subregion of Amazonia. Western none of which were accompanied by ecolog- Amazonian samples are even more divergent ical information. morphometrically (Musser et al., 1998: ®g. 106), and available karyotypes from western Oryzomys megacephalus Fischer Amazonia differ dramatically from those of Figure 53A the single population sampled for chromosomes in the Guianan subregion (op. cit.: ®g. VOUCHER MATERIAL: AMNH 266494, 105). In the event that western Amazonian 266497, 266498, 266501, 266502, 266504, populations currently referred to O. maccon- 266508, 266514, 266515, 266518, 266521, nelli merit formal taxonomic recognition, the 266523, 266525, 266527±266530, 266533, name O. mureliae J. A. Allen, based on a 266535, 266538, 266539, 266541, 267018, specimen collected in eastern Colombia, is 267566; MNHN 1995.999±1995.1010. Total available (Musser et al., 1998: 278±280). The ϭ 36 specimens. Paracou population, however, is unambigu- IDENTIFICATION: At Paracou, and apparent- ously assignable to O. macconnelli, or to the ly throughout most of Amazonia, Oryzomys nominate race if a trinomial nomenclature is megacephalus (formerly O. capito, see be- adopted. low) occurs sympatrically with another mor- Based on the samples at hand from Para- phologically similar species, O. yunganus cou and other localities in French Guiana, (see range maps in Musser et al., 1998). Both adult specimens of Oryzomys macconnelli are drab-colored Oryzomys with brownish can be readily distinguished in the ®eld from dorsal fur, whitish-gray ventral fur, and in- O. megacephalus and O. yunganus by their distinctly bicolored or unicolored-dark tails larger external dimensions (especially hind- that average a little shorter than heads-and- foot length: tables 33, 34), brighter pelage bodies. Comparisons of external measure- colors (redder dorsally and whiter ventrally ments from our voucher material suggest that versus drab brown dorsally and grayer ven- French Guianan O. megacephalus have trally in megacephalus and yunganus), lon- slightly longer tails than sympatric O. yun- ger dorsal fur (12±15 mm versus Ͻ10 mm in ganus, but the difference in average values megacephalus and yunganus), sharply bicol- for the ratio LT/HBL is small (0.94 versus ored tails that are slightly longer than heads- 0.89) and insuf®cient for ®eld identi®cation and-bodies (versus indistinctly bicolored or because of overlapping variation (the ob- unicolored-dark and shorter in megacephalus served range in this ratio among our vouch- and yunganus), and six plantar pads on the ers is 0.78±1.03 for megacephalus, 0.69± hindfoot (versus ®ve in most yunganus, see 1.13 for yunganus). Our measurement data below). Oryzomys macconnelli also has a suggest no appreciable species difference in longer rostrum than either of the other spe- absolute or relative size of the hindfoot. On 122 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 33 vative approach to identifying these species Measurements (mm) and Weights (g) of in the ®eld in French Guiana (classifying an- Oryzomys macconnelli from French Guiana imals with six distinct plantar pads on both and the Type Localitya hindfeet as megacephalus and those with only ®ve pads on both feet as yunganus) would leave an estimated 3% of specimens undetermined (i.e., those with asymmetrical numbers of pads, plus those with indistinct hypothenars on both feet) and would result in an expected error rate of about 5% (mostly yunganus misidenti®ed as megacephalus). The only truly reliable basis for identifying megacephalus and yunganus is cleaned cranial material, from which diagnostic molar characters can be determined. All of the Paracou vouchers we identify as megacephalus have (1) the long para¯exus and single fossette on M2, and (2) the long hypo- ¯exid, diagonal median murid, and lack of fossetid on m2 described and illustrated by Musser et al. (1998). By contrast, specimens we identify as yunganus have (1) a short para¯exus and two fossettes on M2, and (2) a short hypo¯exid separated from a lingual fossetid by a less oblique median murid on m2 (op. cit.). Our multivariate statistical analyses of cranial measurement data from Paracou vouchers identi®ed by these molar traits reveal that specimens of megacephalus have, on average, wider incisive foramina and narrower zygomatic plates than like- average, megacephalus has somewhat bright- sized examples of yunganus. However, mea- er adult dorsal pelage than yunganus, but the surement variation is too extensive for cra- difference is subtle and not useful for ®eld nial proportions to be used as the sole basis identi®cation. for species identi®cation at this locality. Al- The only external character potentially though molar measurements differ signi®- useful for distinguishing Oryzomys megace- cantly between sympatric samples of mega- phalus from O. yunganus in the ®eld is the cephalus and yunganus from some parts of number of plantar pads on the hindfoot (for Amazonia (Musser et al., 1998: table 8), illustrations, see Musser et al., 1998: ®g. 17). these species do not diverge in dental dimen- Almost all specimens of megacephalus have sions at Paracou. six plantar pads: thenar, hypothenar, and four Husson (1978) and most other recent au- interdigitals. By contrast, most French thors have called this species Oryzomys cap- Guianan yunganus have ®ve pads because ito (Olfers), but Tate (1939) referred Guianan the hypothenar is absent (table 35). One Par- populations to O. laticeps (Lund). Both cap- acou specimen of megacephalus (AMNH ito Olfers and megacephalus Fischer are 266527), however, lacks the hypothenar based on Azara's (1801) description of the completely on one foot and has only an in- ``Rat Seconde ou Rat aÁ Grosse TeÃte'' from distinct hypothenar on the other. Similarly, a Paraguay. Musser et al. (1998) designated a few yunganus from Paracou (e.g., AMNH Paraguayan neotype for megacephalus and 266495, MNHN 1995.994) have distinct hy- explained why this name should replace cap- pothenars on one or both hindfeet. Based on ito (a junior objective synonym). Musser and our tabulation of trait frequencies, a conser- his colleagues also diagnosed Oryzomys la- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 123

TABLE 34 Measurements (mm) and Weights (g) of Oryzomys megacephalus and O. yunganus from Paracoua

ticeps as a distinct species restricted to the Amazonia and extends southward into the Atlantic rainforest region of southeastern ParanaÂ basin of eastern Paraguay. Morpho- Brazil. metric, karyotypic, and molecular data sum- As understood by Musser et al. (1998), marized by Musser and his colleagues, how- Oryzomys megacephalus occurs throughout ever, strongly suggest that western Amazo- nian populations (characterized by large size, diploid counts of 52 chromosomes, and dis- TABLE 35 tinctive mtDNA haplotypes) are genetically Frequency of Occurrence of the Hypothenar Pad on the Hindfeet of Oryzomys and evolutionarily distinct from Paraguayan megacephalus and O. yunganus and eastern Amazonian populations (charac- from French Guianaa terized by small body size, diploid counts of 54 chromosomes, and different mtDNA sequences). Although Musser et al. (1998) recognized this dichotomy, they emphasized the dif®culty of identifying geographically intermediate samples as belonging to either the eastern or western clades by morphological criteria and provisionally regarded all of their megacephalus-like Amazonian material as conspeci®c. Patton et al. (2000) subsequently recognized the western Amazonian form as a distinct species, O. perenensis J. A. Allen, a decision with which we concur. In fact, Oryzomys megacephalus may be composite even in the restricted sense of Pat- ton et al. (2000) because their cytochrome-b 124 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 sequence analyses (op. cit.: ®g. 97) suggest TABLE 36 that samples of the small 2N ϭ 54 taxon Comparison of Capture-Habitat Frequencies from north and south of the Amazon form between Oryzomys megacephalus and O. reciprocally monophyletic groups. In the yunganus Trapped in Primary Forest at a event that these geographic moieties should Paracou prove to be diagnosably different by additional criteria, the oldest available name for the northern form (to which our Paracou sample is presumably referable) is velutinus J. A. Allen and Chapman (1893), based on a holotype collected at Princestown, Trinidad. OTHER SPECIMENS EXAMINED: French Gui- anaÐArataye (MNHN 1986.287±1986.293, 1986.295, 1986.296, 1986.298±1986.301, 1986.314, 1986.878±1986.880, 1986.882), Cacao (MNHN 1980.275, 1983.370, 1986.278, 1986.279, 1986.316, 1986.493± 1986.499, 1986.501, 1986.506, 1986.511, 1986.514, 1986.531, 1986.533), Camopi (MNHN 1982.600, 1982.620), Cayenne primary forest of unspeci®ed character, and (MNHN 1970.225, 1986.317, 1986.319, 4 (11%) in secondary vegetation. See table 1986.952, 1986.953), ``Marais de Kaw'' 36 and the following account for capture- (MNHN 1986.1106, 1986.1107), Rorota habitat comparisons with O. yunganus. (MNHN 1986.954, 1986.955), Piste St.-EÂ lie km 16 (MNHN 1986.884±1986.886), SauÈl Oryzomys yunganus Thomas (MNHN 1981.181, 1983.368, 1983.369, 1986.489, 1986.492, 1986.518, 1986.521, VOUCHER MATERIAL: AMNH 266495, 1986.525), Sauts de l'Itany (MNHN 266496, 266503, 266510, 266511, 266513, 1962.1024±1962.1028), Trois Sauts (MNHN 266516, 266517, 266520, 266532, 267017, 1981.150, 1982.616±1982.619, 1982.623± 267567; MNHN 1995.993±1995.998. Total 1982.627, 1986.283, 1990.908±1990.910). ϭ 18 specimens. FIELD OBSERVATIONS: All of our de®nite IDENTIFICATION: See the account above for records of Oryzomys megacephalus at Para- morphological comparisons with Oryzomys cou are based on collected specimens. Of our megacephalus, the only species with which 36 vouchers, 23 (64%) were taken in Sher- O. yunganus could plausibly be confused. man or Victor traps set on the ground, 7 Long unrecognized as a member of the (19%) were taken in Sherman or Victor traps Guianan fauna, O. yunganus was recently re- tied to lianas 0.3±1.2 m above the ground, 5 vised by Musser et al. (1998), who docu- (14%) were shot on the ground, and 1 (3%) mented the extensive Amazonian distribution was taken in a pitfall. Microhabitat notes ac- of this species by mapping all known collec- companying 26 specimens shot or trapped on tion localities (op. cit.: ®g. 14). As noted by the ground record 18 captures under or be- Musser and his colleagues, the geographic side logs, 4 captures under the roots or but- samples they refered to O. yunganus exhibit tresses of fallen trees, 2 captures under tan- considerable divergence in body size. Espe- gled branches of fallen trees, 1 capture at the cially notable are specimens from Guyana, base of a buttressed tree, and 1 capture at the Surinam, French Guiana, and eastern Ama- entrance to a hollow log. All of the shot zonian Brazil, which are diminutive by com- specimens were encountered at night, and all parison with specimens from Venezuela, Co- of the other specimens were found in traps lombia, Ecuador, Peru, and western Brazil. at or near dawn. Twenty-two specimens For example, the observed range of variation (61%) were taken in well-drained primary in crown length of the upper molar series forest, 3 (8%) in swampy primary forest, 2 (LM) among our vouchers (4.3±4.6 mm; ta- (6%) in creekside primary forest, 5 (14%) in ble 34) does not overlap with the observed 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 125

range of variation among 52 specimens (in- specimens were found in the traps at or near cluding the type) that we measured from Co- dawn. Seven captures (44%) were in well- lombia, Ecuador, Peru, and Bolivia (4.8±5.6 drained primary forest, another 7 were in mm). Additionally, French Guianan samples swampy primary forest, 1 (6%) was in creek- of O. yunganus differ conspicuously from side primary forest, and 1 was in primary western Amazonian samples in the frequency forest of unspeci®ed character. Microhabitat of occurrence of the hypothenar pad on the notes accompanying 15 specimens record 4 hindfoot (Musser et al., 1998: table 7). The captures under or beside logs, 3 captures at currently accepted provisional hypothesis, the bases of trees, 3 captures in dense un- that these and other differences among sam- dergrowth unsheltered by woody objects, 2 ples currently referred to O. yunganus rep- captures under tangled dead branches, 2 cap- resent intraspeci®c geographic variation, tures under fallen palm fronds, and 1 capture merits testing by additional collecting at in- among the stilt roots of a standing tree. termediate localities (op. cit.: p. 109), and by Our capture data from Paracou are broadly analyzing molecular sequence data from consistent with specimen counts from other Guianan and western Amazonian populations Amazonian localities (summarized by Mus- (as by Patton et al., 2000). ser et al., 1998) in suggesting that Oryzomys OTHER SPECIMENS EXAMINED: BoliviaÐ yunganus is less abundant than O. mega- Cochabamba, Charuplaya (BMNH 2.1.1.39 cephalus wherever these species occur sym- [holotype]). ColombiaÐCaquetaÂ, Tres Tron- patrically. Although we sometimes caught cos (FMNH 72036, 72051, 72066); Meta, La both species in the same trapline on the same Macarena (FMNH 58778, 58779, 87969, date, statistical comparisons of capture fre- 87970); Putumayo, RõÂo Mecaya (FMNH quencies by habitat (table 36) suggest that O. 72067). EcuadorÐPastaza, RõÂo Capahuari yunganus prefers moister primary forest hab- (FMNH 43268, 43271), RõÂo Yana Rumi itats (swamp or creekside formations) than (FMNH 43265). French GuianaÐArataye does O. megacephalus. The latter species (MNHN 1986.294, 1986.297, 1986.313, was also trapped above ground level on lia- 1986.881, 1986.883), Cacao (MNHN nas, and in secondary vegetation, situations 1986.490), Cayenne and Rorota (MNHN in which O. yunganus was not encountered. 1986.322, 1986.324, 1986.326, 1986.327, Future ecological studies of Amazonian ro- 1986.800±1986.803), Kaw (MNHN dent communities should test the hypothesis 1986.1105). PeruÐCuzco, Hacienda Cadena that O. yunganus is a habitat specialist by (FMNH 65704, 66399, 66401, 68630, comparison with O. megacephalus, but we 68631), Quincemil (FMNH 75242, 75253, caution that destructive sampling (or molec- 75254, 75257, 75259, 75261±75264, 75272; ular typing, as by Lavergne et al., 1997; HuaÂnuco, Chinchao (FMNH 23721, 23722), Steiner et al., 2000) will be necessary in or- Hacienda Buena Vista (FMNH 24544, der to obtain reliable taxonomic identi®ca- 24547, 24548); Loreto, RõÂo Pastaza (BMNH tions for this purpose. 54.421, 54.422, 54.425, 54.429, 54.430); San MartõÂn, Moyobamba (FMNH 19376, 19387, Rhipidomys nitela Thomas 19392), Puca Tambo (BMNH 26.5.3.31± Figure 53C, 61B, 62A 26.5.3.38, 26.5.3.40±26.5.3.42; FMNH 19787); SurinamÐNickerie, Kayserberg VOUCHER MATERIAL: AMNH 267021, Airstrip (FMNH 93284, 93286). 267580, 267582, 267583, 267594; MNHN FIELD OBSERVATIONS: All of our de®nite 1995.1011, 1995.1012. Total ϭ 7 specimens. records of Oryzomys yunganus at Paracou IDENTIFICATION: The ten nominal taxa of are based on collected specimens. Our 18 Rhipidomys based on type material collected vouchers represent only 16 sampling events, in the Guiana subregion of Amazonia appear however, because pairs of juvenile individu- to represent four valid species that can be als were taken in the same trap on two oc- readily distinguished by external and cran- casions. Of these 16 independent captures, iodental characters. Rhipidomys macconnelli 15 (94%) were in Sherman traps set on the de Winton (1900) (including subnubis Tate, ground, and one was in a pitfall. All of our 1939) and R. wetzeli Gardner (1989) have 126 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 37 Diagnostic Characters and Geographic Ranges of Rhipidomys mastacalis, R. nitela, and R. venezuelaea

long, soft fur and a primitive carotid arterial treated sclateri as a subjective junior syno- circulation (pattern 1 of Voss, 1988); they oc- nym of leucodactylus Tschudi (1844), the cur in montane and premontane vegetation type locality of which is in eastern Peru. associated with rocky outcrops of the Pan- Only a single specimen, the type of R. leu- tepui complex (Tate, 1939; Handley, 1976; codactylus aratayae, is currently known Gardner, 1989) and neither is known from from French Guiana. French Guiana (where such habitats are ab- The oldest Guianan name for the small sent). The other two species have shorter, lowland species is Rhipidomys nitela, but ni- coarser fur, a derived carotid arterial mor- tela has often been treated (e.g., by Cabrera phology (pattern 3 of Voss, 1988), and occur [1961] and Husson [1978]) as a junior syn- in lowland rainforest. One of the lowland onym of R. mastacalis (Lund, 1840), and rainforest species (represented by the Guian- some nominal taxa referable to nitela were an holotypes of sclateri Thomas [1887], bov- originally described as subspecies of R. ve- allii Thomas [1911b], and aratayae Guillotin nezuelae Thomas (1896). Based on our ex- and Petter [1984]) is large (HF ϭ 32±36 mm; LM ϭ 5.9±6.8 mm) with gray-based ventral amination of types and other material, we fur, whereas the other (represented by the agree with Musser and Carleton (1993) and Guianan holotypes of nitela Thomas [1901], with Tribe (1996) that nitela, mastacalis, and fervidus Thomas [1904], milleri Allen venezuelae represent three valid species with [1913b], and yuruanus Allen [1913b]) is diagnostic morphological and karyotypic at- small (HF ϭ 24±28 mm, LM ϭ 4.1±4.8 mm) tributes, and with discrete geographic ranges with (usually) pure white ventral fur. (table 37). Our six Paracou vouchers together Although the oldest Guianan name for the with 13 additional specimens subsequently large lowland species is Rhipidomys sclateri, collected at Les Nouragues (by F. Catze¯is Musser and Carleton (1993) and Tribe (1996) and his colleagues from Montpellier) are ap- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 127

parently the only examples known from TABLE 38 French Guiana. Measurements (mm) and Weights (g) of Adult For the most part, external and cranio- Rhipidomys nitela from French Guiana and dental measurements of French Guianan the Type Localitya specimens of Rhipidomys nitela agree closely with those of the Guyanese type series (table 38). Although relative tail length appears to be divergent in the two samples (averaging about 130% of head-and-body length in French Guianan material versus 115% in the type series), methodological artifacts might explain this proportional difference between small series of specimens measured in the ®eld by different collectors. By contrast, measurements of the hindfoot and of the molars (both measured by us) suggest that the populations in question do not differ much, if at all, in size. Because the French Guianan material additionally resembles the type series in qualitative characters, we interpret these samples as representing populations of a single species. Some of Husson's (1978: table 71) measurements of the hindfeet of Surinamese specimens that he identi®ed as Rhipidomys mastacalis nitela are smaller than any that we or Tribe (1996) observed for this species. If Husson's material was correctly identi®ed, which we do not doubt, it is likely that the feet were simply mismeasured. The single weight datum that Husson tabulated for this species (150 g) was obtained from a female with three near-term embryos and is obviously not comparable with our weights of nonpregnant animals. lated on the northern Caribbean coast of Co- All of the specimens that we refer to Rhip- lombia. We examined the voucher in ques- idomys nitela (see below) appear to represent tion (MHNG 1706.75, from Bonda, Depar- a morphologically cohesive taxon that is geo- tamento Magdalena), which differs from R. graphically limited to the Guianan and nitela by its pale bicolored tail and short (6 Southeastern subregions of Amazonia. Tribe mm) caudal tuft; in our opinion, this speci- (1996), however, recognized two isolated re- men represents another taxon, perhaps allied cords of R. nitela from outside Amazonia. to R. venezuelae despite its short (4.6 mm) One of these records, consisting of the type toothrow. series of R. nitela tobagi Goodwin (1961) Recently, Anderson (1997) reported Rhip- from Little Tobago Island, is equivocal in our idomys nitela from eastern Bolivia, a consid- judgment because the diagnostic presence of erable range extension that we attempted to a conspicuous caudal tuft (®g. 61B) cannot con®rm by examining his material. Of the be con®rmed from the two partially decayed two specimens that Anderson cited by num- ¯uid specimens of this taxon; possibly, these ber (AMNH 119406, UMMZ 156298), how- represent an insular form of the adjacent ever, we were only able to locate one. That mainland species R. venezuelae. The other specimen, UMMZ 156298, is an example of nonAmazonian record of R. nitela mapped Thomasomys resembling T. oreas Anthony by Tribe (1996: ®g. 7.6) is improbably iso- (1926). Therefore, as far as we have been 128 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

able to determine by direct examination of museum specimens, R. nitela appears to be an Amazonian endemic that does not occur west of the north-south zoogeographic axis represented by the Rio Negro and the Rio Madeira. Species of Oecomys externally resemble Rhipidomys by their large eyes, long vibrissae (extending well behind the pinnae when laid back alongside the head), short-broad hindfeet with semi-opposable ®fth digits, and long tufted tails. At Paracou, O. auyantepui and R. nitela are of similar size and might be confused in the ®eld. However, R. nitela and O. auyantepui differ in many external characters that are potentially useful for identi®- cation of specimens in hand. (1) Whereas the pinnae of R. nitela are blackish (contrasting in color with the fur of the head) and appear quite naked (a very sparse auricular pelage is visible only under magni®cation), the pinnae of O. auyantepui are not contrastingly colored and have a visible pelage of reddish- brown hairs. (2) The dorsal body pelage of R. nitela is very short (5±6 mm), somewhat coarse to the touch, and dull grayish-brown, but the dorsal fur of adult specimens of O. auyantepui is much longer (Ն10 mm), very soft to the touch, and lustrous reddish-brown in appearance. (3) The ventral pelage of R. nitela is almost entirely self-colored (pure white or cream to the roots of the hairs), except immediately along the ¯anks (where some hairs have gray bases and white tips); by contrast, the ventral fur of O. auyantepui is mostly gray-based, except on the throat and along the midline (where some hairs are pure white). (4) The hindfeet of R. nitela are always prominently marked by a dark band of brownish hairs that extends from the ankle over all or part of the metatarsus to the base of the toes, which are usually white (pigmented hairs extend onto the proximal phalanges of the middle toes in a few specimens, but the outermost digits are always complete- Fig. 61. Caudal morphology of Rhipidomys venezuelae (A, AMNH 131081) and R. nitela (B, ly white); by contrast, pedal markings are not AMNH 267021). The tail of R. venezuelae is uni- conspicuous in most specimens of O. auyan- formly covered by short hairs that are clearly vis- tepui, some of which have uniformly pale ible without magni®cation but do not conceal the feet. (5) Adult female specimens of Rhipi- underlying epidermal scales; there is no terminal domys have six mammae in postaxial, ab- tuft of conspicuously longer hairs. By contrast, dominal, and inguinal pairs (muroid mam- the caudal hairs of R. nitela increase in length mary loci are illustrated in Voss and Carle- from the base of that organ to the tip, where the ton, 1993: ®g. 8), but female Oecomys have longest hairs are concentrated to form a conspicuous terminal tuft. Scale bar ϭ 20 mm. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 129

eight mammae (the additional teat-pair is pectoral). Extended craniodental comparisons between Rhipidomys and Oecomys are unnec- essary for ®eldworkers, but it is relevant to note that these genera super®cially resemble one another by their shallow zygomatic notches, convergent and beaded supraorbital margins, and pentalophodont molars. Nev- ertheless, skulls found in stomachs or scat can be easily identi®ed by palatal architec- ture. The bony palate of Rhipidomys (®g. 62A) is ``short'' because it does not extend behind the molar rows, and the posterior palatal margin is biconcave because a small median palatal process is present; the posterior palatal pits are small, simple perforations. By contrast, the bony palate of Oecomys (®g. 62B) is ``long'' (extending behind the molar rows) with a typically arch-shaped posterior margin, and the posterior palatal pits are larger and often more complex. The only French Guianan species of Rhipidomys and Oecomys that are at all likely to be confused, R. nitela and O. auyantepui, can also be distinguished by carotid arterial morphology (pattern 3 versus pattern 1 [of Voss, 1988], respectively), but carotid morphology does not consistently differ between other representatives of these genera. OTHER SPECIMENS EXAMINED: BrazilÐ ParaÂ, Aramanay on Rio Tapajos (AMNH 94810±94813). French GuianaÐLes No- uragues (AMNH 269821; V-824, Ϫ825, Ϫ826, Ϫ831, Ϫ876, Ϫ886, Ϫ890, Ϫ891, Ϫ893, Ϫ905, Ϫ914). GuyanaÐPotaro-Si- paruni, Minnehaha Creek (AMNH 36331± 36336 [type series of milleri]); Upper Tak- utu-Upper Essequibo, Quatatat (BMNH 1.6.4.81±1.6.4.86 [type series of nitela]). VenezuelaÐBolõÂvar, La UnioÂn (BMNH 4.5.7.34, 4.5.7.35 [type series of fervidus]), La Vuelta (BMNH 4.5.7.36), RõÂo YuruaÂn (AMNH 30727±30735, 30737 [type series of Fig. 62. Posterior palatal morphology of yuruanus]), San Ignacio YuruanõÂ (AMNH Rhipidomys nitela (A, AMNH 267582) and Oec- omys auyantepui (B, ROM 103244). In Rhipido- mys (and other ``thomasomyines''; see Voss, ← 1993), the palate is said to be ``short'' because the mesopterygoid fossa (mpf) extends anteriorly to because the mesopterygoid fossa does not extend or between the last molars; additionally, postero- anteriorly to the last molars; also, the posterolat- lateral palatal pits (ppp) are small or absent in eral palatal pits of oryzomyines are large, some- ``thomasomyines''. By contrast, the palate of Oec- times complex, and frequently recessed in shallow omys and other oryzomyines is said to be ``long'' fossae. 130 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

257273±257275; MHNLS 7845, 7846, Amazonas, Brazil.19 Both are skins, original- 7848±7850, 7891, 7892, 7895, 7896, 7898, ly stuffed and mounted for exhibition, with 7899, 8072, 8074; USNM 448613±448616, skulls and mandibles subsequently extracted. 448618±448623, 448625±448628). In order to clarify the application of Cer- FIELD OBSERVATIONS: All of our de®nite colabes melanurus, we select as lectotype records of Rhipidomys nitela at Paracou are NMW 42010, an adult female collected in based on collected specimens. In 1992, one February 1834. The skull (®g. 64) is that of adult female was shot in the daytime as it a fully mature animal, with slightly swollen perched several meters above ground level in frontal sinuses and most cranial sutures the dark interior of a hollow tree (®g. 63). In fused; the cheekteeth are worn ¯at, but all 1993, the same tree cavity contained four in- essential details of the occlusal morphology dividuals, of which one adult female and two remain. The squamosal root of the right zygomatic arch is broken, as is the right pter- juveniles were shot and one juvenile escaped. ygoid process; both occipital condyles and One specimen was captured at ground level part of the basioccipital are missing. The skin in a pitfall, and two others were taken in plat- (®g. 65) is essentially intact, but the tail is form traps 14.5±15.2 m above the ground. partially broken away at the base and secured All of our vouchers were shot or trapped in to the body with thread. The paralectotype well-drained primary forest. (NMW B-1017) is a subadult female with deciduous premolars and unfused cranial su- ERETHIZONTIDAE tures. Like the lectotype, the tail of this specimen is partially broken away and tied to the Coendou melanurus (Wagner) body with thread. Figures 64, 65, 67, 68, 69, 70A, 71 DISTRIBUTION: Specimens that we examined and literature records that we judge to Only two specimens documented the oc- be reliable suggest that Coendou melanurus currence of this distinctive porcupine in ranges throughout the Guiana subregion of French Guiana prior to ®eldwork at Paracou: Amazonia (®g. 66). Although Emmons the mounted skin of an immature animal (1990, 1997), Alberico et al. (1999), and Ti- from ``Guyane'' (MNHN 1909.241), and an rira (1999) described the range of C. melan- adult skin-and-skull from St. Laurent du Ma- urus as extending into western Amazonia, all roni (MNHN 1909.242). Just a single indi- extralimital records that we investigated were vidual (AMNH 266565) was encountered in found to be based on misidenti®ed material our inventory, but simultaneous faunal res- of other species (see Comparisons, below). cue operations at the Petit Saut hydroelectric DESCRIPTION: Small long-tailed porcupines dam site resulted in 59 captures (Lemercier, (see measurements in table 39) with dorsal 1998; VieÂ, 1999), from which three speci- pelage composed of short quills more-or-less mens were salvaged as vouchers (MNHN concealed beneath a thick coat of long black- 1997.640, 1997.641, 1999.1080; F. Catze¯is, ish fur coarsely streaked with yellow. Quills personal commun.). Evidently, the species is bicolored (yellowish basally with dark- brown tips), about 3 cm or less in length, not rare, but cryptic. Because the morpho- densely covering dorsal surface of head, logical and geographic limits of Coendou neck, trunk, and upper limbs, but only ex- melanurus are not adequately documented in posed on cheeks and crown of head (except the literature, we redescribe the species be- where fur has come away in patches due to low, discuss its morphological variation, compare it with other congeners, and provide 19 A third specimen in the same museum (NMW B- new information about its geographic range. 1006), consisting of a stuffed skin with skull and man- TYPE MATERIAL: Wagner's (1842) original dibles inside, lacks locality information and is not part material of Cercolabes melanurus consists of of Natterer's collection from Barra do Rio Negro (Pel- two specimens in the Naturhistorisches Mu- zeln, 1883). According to the museum's acquisition cat- alogs and card ®les, NMW B-1006 was purchased in seum Wien collected by Johann Natterer at London by L. v. Fichtel in 1809 and may have origi- Barra do Rio Negro (ϭ Manaus), Estado nated from the Guianas (K. Bauer, personal commun.). 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 131

Fig. 63. Diurnal refuge of Rhipidomys nitela at Paracou. Specimens were collected in 1992 and 1993 from the dark internal cavity (arrow) of this large kimboto (Sapotaceae: Pradosia cochlearia), a canopy emergent in well-drained primary rainforest. 132 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 64. Dorsal, ventral, and lateral cranial views of NMW 42010, lectotype of Coendou melanurus (Wagner). All views approximately ϫ1.3. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 133

Fig. 65. Dorsal and ventral views of the skin of NMW 42010, lectotype of Coendou melanurus (Wagner). Both views approximately ϫ0.5. faulty preservation). Dorsal fur long and duced by coarse guard hairs with very long abundant, concealing quills from crown to yellow tips scattered abundantly among ®ner rump; individual hairs pale basally (among hairs that are entirely black or have only in- the quills), but emergent fur black heavily conspicuous pale tips). Ventral surface of streaked with pale yellow (a mass effect pro- body without offensive quills, covered 134 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 66. Geographic range of Coendou melanurus based on specimens examined and literature records. 1, BRAZIL, AmapaÂ, Serra do Navio; 2, BRAZIL, Amazonas, Manaus; 3, BRAZIL, ParaÂ, Lago Claimy; 4, FRENCH GUIANA, Paracou; 5, FRENCH GUIANA, St.-Laurent du Maroni; 6, GUYANA, Demerara-Mahaica, Georgetown; 7, GUYANA, Cuyuni-Mazaruni, Kartabo; 8, GUYANA, Upper Tak- utu-Upper Essequibo, 25 mi E Dadanawa; 9, GUYANA, Upper Takutu-Upper Essequibo, Nappi Creek; 10, SURINAM, Brokopondo, Afobaka (Husson, 1978); 11, SURINAM, Suriname, Paramaribo (Husson, 1978); 12, VENEZUELA, Amazonas, Sierra Parima (Linares, 1998); 13, VENEZUELA, BolõÂvar, Ima- taca (Ochoa, 1995); 14, VENEZUELA, Delta Amacuro, Delta del Orinoco (Linares, 1998). sparsely with short, coarse, brownish hairs but remainder of tail (except for naked pre- (usually frosted with whitish bands or tips) hensile surface) densely covered above and from chin to anus. Face with very short quills below with stiff black bristles. on forehead and cheeks but otherwise almost Frontal and nasal sinuses unin¯ated (re- naked, especially around eyes; facial vibris- sulting in a ¯attened dorsal pro®le from mid- sae consisting of long, black mystacial, ge- parietal region to nasal tips) or weakly in- nal, and superciliary hairs; submental vibris- ¯ated (resulting in a noticeable bulge over sae short and black. Pinnae rudimentary and the orbits). Rostrum usually short and very almost naked. Hands and feet covered dor- broad, not conspicuously excavated laterally sally with very coarse blackish hairs. Tail for origin of infraorbital muscle; nasals par- probably about as long as head-and-body on allel-sided or weakly convergent posteriorly, average; dorsum of proximal third with body with rounded posterior margins that extend pelage (quills and yellow-streaked black fur), well behind the premaxillae. Zygomatic 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 135

TABLE 39 Measurements (mm) of Coendou melanurus

arches (viewed from above) usually with wedge penetrating between third molar rounded lateral de¯ection at orbits, but some- crowns; bony roof of mesopterygoid fossa times convergent anteriorly with no orbital usually completely ossi®ed (sometimes with de¯ection (e.g., AMNH 266565); jugals tiny perforations but never large vacuities). moderately expanded dorsoventrally behind Auditory bullae very large and antero-pos- maxillary suture. Dorsolateral contours of teriorly elongated; roof of external auditory braincase weakly sculpted by bony scars of meatus with conspicuous bony ridge extend- M. temporalis in most specimens (but tem- ing from dorsal lip of bulla to malleus. poralis scars well marked in AMNH 70120), VARIATION: This species is extraordinarily the right and left scars always widely sepa- uniform in qualitative external characters de- rated (never joined to form a sagittal ridge spite the apparent plasticity of some morpho- or crest). metric and osteological traits. High variabil- Incisive foramina short and completely ity in external measurements (table 39) is contained by premaxillae (e.g., AMNH common in erethizontids because of the dif- 94174), or longer and bordered by maxillae ®culty of handling animals protected by posteriorly; left and right foramina recessed sharp quills; additionally, porcupine feet do in a common fossa and incompletely sepa- not resemble those of other rodents, and rated (e.g., AMNH 94174) or completely some ®eld collectors may have measured the separated by a stout bony septum and not hindfoot erroneously, from the rounded pos- recessed (e.g., AMNH 142955). Posterior di- terior margin of the plantar callosity instead astema usually distinctly trisulcate; palatal of the heel. As in other caviomorphs, quali- bridge usually without a distinct median keel, tative cranial and dental characters of ereth- or keel weakly developed (never a high crest izontids are annoyingly variable. Some dif- ¯anked by deep lateral gutters). Anterior ferences among the specimens of Coendou margin of mesopterygoid fossa a broad, blunt melanurus that we examined may be corre- 136 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 lated with age and sex, but with miniscule sons between Surinamese samples of melan- samples from widely scattered localities, in- urus and prehensilis were discussed by Hus- terpretation is dif®cult. son (1978), who misidenti®ed his material of COMPARISONS: Many faunal accounts and the former species as Sphiggurus insidiosus checklists (e.g., Cabrera, 1961; Husson, (see above and Remarks, below) 1978; Woods, 1993) have treated Coendou REMARKS: Coendou melanurus and other melanurus as a subjective junior synonym of long-furred Neotropical porcupines have of- C. insidiosus (Olfers, 1818), a taxon endemic ten been referred to the genus Sphiggurus F. to the Atlantic rainforests of southeastern Cuvier (most recently by Husson, 1978; Brazil. The history of this erroneous usage ConcepcioÂn and Molinari, 1991; Woods, was reviewed by Voss and Angermann 1993; Eisenberg and Redford, 1999). How- (1997), who rediscovered and redescribed ever, we agree with Handley and Pine (1992) the holotype of C. insidiosus, compiled geo- that Coendou and Sphiggurus are not mean- graphic data from other known specimens, ingfully diagnosable based on current knowl- and provided diagnostic comparisons with C. edge of morphological character variation, melanurus. Morphological differences be- and that the latter name should be treated as tween these highly distinctive species are a subjective junior synonym of the former. here illustrated for the ®rst time (®gs. 67± Sphiggurus melanurus Gray (1842), pub- 70). lished in the same year as Cercolabes me- Other species of South American porcu- lanurus Wagner, is based on a skin in the pines with long fur are easily distinguished Natural History Museum (London) bearing from Coendou melanurus by conspicuous the number 86a on a cardboard tag attached external characters. For example, Coendou to the hindfoot. This specimen agrees in ev- vestitus Thomas (1899c) and C. pruinosus ery essential respect with Wagner's type se- Thomas (1905) from Venezuela and Colom- ries and with the other material herein re- bia are smaller, shorter-tailed porcupines that ferred to Coendou melanurus. Waterhouse have long wire-like bristles mixed among the (1848: 425), the ®rst reviser in the sense of quills and woolly fur of the dorsal pelage. the International Code of Zoological Nomen- We have also examined several Ecuadorean clature (ICZN, 1999), can be considered to specimens previously misidenti®ed as C. me- have chosen Wagner's name to have prece- lanurus and determined that they represent dence over Gray's (a junior subjective syn- an undescribed species of the vestitus group; onym). this taxon super®cially resembles melanurus SPECIMENS EXAMINED: BrazilÐ``Brazil'' because it has long yellow-tipped bristles (BMNH specimen numbered 86a in J. E. that contrast with the black-tipped quills to Gray's manuscript catalog [type of melanu- produce a similarly streaked pelage pattern. rus Gray]); AmapaÂ, Serra do Navio (USNM However, the new species is substantially 394732); Amazonas, Manaus (NMW 42010, smaller (HF, 58±59 mm; CIL, 58.8±64.4 mm; B-1017); ParaÂ, Lago Claimy on Rio Jha- MTR, 14.1±15.2 mm), lacks a dense coat of munda (AMNH 94174). French GuianaÐ long adult fur, and has more acutely angled ``Guyane'' (MNHN 1909.241), Paracou mesopterygoid margins that penetrate more (AMNH 266565), St.-Laurent du Maroni deeply between the toothrows (Voss and da (MNHN 1909.242). GuyanaÐCuyuni-Ma- Silva, submitted). zaruni, Kartabo (AMNH 70120, 70131, Throughout the Guiana subregion of Ama- 142955); Demerara-Mahaica, Georgetown zonia, Coendou melanurus is sympatric with (FMNH 17762); Upper Takutu-Upper Esse- another congener, C. prehensilis (see below). quibo, 25 mi E Dadanawa (ROM 31984), Although melanurus is smaller than prehen- Nappi Creek near Letham (ROM 31683, silis, there is some overlapping variation in 31783, 31801). No locality dataÐ(NMW B- weights and linear dimensions from large 1006). sympatric samples (Richard-Hansen et al., FIELD OBSERVATIONS: Our single voucher 1999). Fortunately, the two species are un- is the only de®nite record we have of Coen- mistakeable in qualitative external characters dou melanurus from Paracou; surprisingly, (®g. 71). Diagnostic craniodental compari- none of the forestry personnel whom we in- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 137

Fig. 67. External morphology of Coendou melanurus (left, AMNH 266565) and C. insidiosus (right, AMNH 90119). Although these species are similar in possessing short bicolored quills concealed by long fur, C. melanurus has a much longer tail relative to head-and-body length, and its coarse blackish dorsal fur is heavily streaked with yellow- or white-tipped guard hairs; the caudal bristles (covering the tail between its furry base and naked prehensile tip) are jet black. By contrast, C. insidiosus is relatively shorter-tailed and has softer smoky-gray (or whitish) dorsal fur without contrastingly colored guard hairs; the caudal bristles are brownish, not black. Both views approximately ϫ0.3. 138 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 68. Dorsal and ventral cranial views of Coendou melanurus (left, AMNH 94174) and C. insidiosus (right, AMNH 90119). All views approximately life-size. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 139

Fig. 69. Left lateral cranial views of Coendou melanurus (top, AMNH 94174) and C. insidiosus (bottom, AMNH 90119). All views approximately life-size.

terviewed were aware that this species oc- the same color and odor when crushed as the curred in the area. This specimen was shot freshly dissected stomach contents. at night as it perched 15±20 m above the ground in the subcanopy of well-drained pri- Coendou prehensilis (Linnaeus) mary forest; its stomach was completely Figure 71 ®lled with a homogeneous, ®nely masticated, Although we did not directly observe this bright-green pulp. We surmise that the ani- species at Paracou, we found one of its dis- mal had been eating the new leaves of the tinctively large, tricolored quills on a dirt tree in which it was shot because these had road near our camp. P. Petronelli (personal 140 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 70. Morphology of the roof of the external auditory meatus in Coendou melanurus (A, AMNH 266565) and C. insidiosus (B, AMNH 90119). In C. melanurus, the roof of the meatus has a prominent bony ridge (br) that is lacking in C. insidiosus. Scale bars ϭ 1 mm. commun., 1993) told us that he had also scriptions are at the root of several nomen- found quills of this species lying on the clatural problems. ground in our study area, and that an ocelot (L. pardalis) killed at Paracou by a local Dasyprocta leporina (Linnaeus) hunter several years before our inventory work began had quills of C. prehensilis em- VOUCHER MATERIAL: AMNH 265955; bedded in its neck and shoulders. The species MNHN 1998.677. Total ϭ 2 specimens. has occasionally been observed in roadside IDENTIFICATION: The genus Dasyprocta has secondary growth by visiting researchers (G. never been revised and the current species- Dubost, personal commun.), but accurate level taxonomy (e.g., as summarized by Ca- counts of such observations are unavailable. brera, 1961; Emmons, 1990, 1997; Woods, 1993) is sorely in need of critical attention. Traditionally, the red- or yellow-rumped DASYPROCTIDAE agoutis of Amazonia have been referred to The Paracou fauna contains two dasyproc- D. aguti (Linnaeus, 1766), but the oldest tids, one species each of Dasyprocta and My- available name for these animals is unequiv- oprocta, the usual number known to occur ocally D. leporina (Linnaeus, 1758) (see sympatrically throughout most of Amazonia. Husson [1978] and Remarks, below). Although Guianan dasyproctids are easily Our voucher material closely resembles identi®ed in the ®eld and in the museum, Husson's (1978) description of Dasyprocta their technical names are still controversial leporina, which was based on the neotype due to unresolved issues of usage and syn- and other specimens from Surinam. The only onymy. Large, edible, and diurnal, dasyproc- noteworthy point of difference is the color of tids were among the ®rst mammals of the the long nape hairs, which are blackish in our Guianan fauna to be reported by European vouchers, whereas Husson (p. 459) stated travellers, whose inadequate published de- that ``[t]he anterior part of the dorsal surface 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 141

Fig. 71. Unvouchered photographs of living Coendou melanurus (top) and C. prehensilis (bottom, not to same scale) from Petit Saut, French Guiana (ca. 28 km SSW Paracou). Both animals are alarmed, so the pale bases of their erected quills are exposed. When C. melanurus is not alarmed and defensive, its short bicolored (black-tipped) quills are hidden beneath a dense coat of long fur (see ®g. 67). By contrast, the large tricolored (white-tipped) quills of C. prehensilis are always visible because the short sparse fur of this species is concealed among the quill bases. Also note the diagnostically bulging forehead and swollen muzzle of C. prehensilis, and the equally diagnostic blackish caudal bristles of C. melanurus. From color transparencies by J.-C. VieÂ. 142 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 of the body including the neck is olivaceous TABLE 40 gray speckled with yellowish'', the nuchal Measurements (mm) and Weights (kg) of hairs apparently not being distinctively col- Dasyprocta leporina from French Guiana and ored in the neotype. Husson's other speci- Surinam mens, however, were described (op. cit.) as having darker necks, and most of the Suri- namese agoutis we examined had blackish napes like our vouchers. As noted by Hus- son, the rump color in this species is also variable, ranging from clear yellow-orange to grizzled reddish-brown in the specimens we examined. Although our single adult voucher (AMNH 265955) is larger than any of the 12 Surinamese specimens measured by Husson (1978: table 82), the differences between homologous dimensions of AMNH 265955 and his largest example (RMNH 18235) are proportionately small (e.g., about 4% in maxillary toothrow length), and we do not regard them as taxonomically signi®cant. Morpho- metric variation among nine FMNH specimens from Surinam (table 40) likewise suggests that our voucher is large but probably not outside the normal size range of typical D. leporina. For future revisionary work, a large series of specimens collected by H. A. Beatty in the Wilhelmina Mountains of Su- rinam (see Specimens Examined, below) provides a useful sample of individual variation from a single local population of this nomenclaturally important species. Both Cabrera (1961) and Ojasti (1972) 1978). A single 19th-century RMNH speci- recognized several subspecies of Dasyprocta men of D. fuliginosa is also said to have aguti (ϭ D. leporina; see above and Re- come from Surinam (op. cit.). None of this marks, below) as valid, but the necessity for material, however, is accompanied by names a trinomial classi®cation remains to be con- of collectors, dates of collection, or any ad- vincingly established by a comprehensive ditional geographic information. Because we study of geographic variation. From the close have not seen any material referable to D. similarity between our voucher material and cristata or D. fuliginosa accompanied by def- the Surinamese specimens with which we inite evidence of origin from any of the compared them, the Paracou population Guianas, we assume that these old and poor- would be unambiguously referable to the ly documented Surinamese records are erro- nominate form if other subspecies were to be neous. As discussed by Goeldi and Hagmann recognized in a future revisionary study. Al- (1904), Thomas (1917), and Cabrera (1961), ternatively, it is possible that red-rumped the type of D. prymnolopha, said to be from agoutis include two or more separate species ``Guiana'' (Wagler, 1831: 619), was probably (as suggested by Krumbiegel, 1941), another collected somewhere along the Atlantic coast hypothesis that remains to be effectively test- of Brazil between the mouth of the Rio To- ed by critical analyses of specimen data. cantins and Bahia. Apparently, D. leporina is The type of Dasyprocta cristata (Geof- the only agouti species validly known from froy, 1803) and another specimen so identi- Guyana, Surinam, French Guiana, and ®ed in the RMNH are both zoo animals said Guianan Brazil. The black agouti (Dasyproc- to have been collected in Surinam (Husson, ta fuliginosa), however, occurs allopatrically 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 143

in the headwaters of the Orinoco (geograph- and Toccheton, 1969) and not a junior syn- ically part of the Guiana subregion of Ama- onym of D. leporina (contra Husson, 1978), zonia) in southernmost Amazonian Venezue- our examination of Linnaeus's (1766) biblio- la (Tate, 1939; Ojasti, 1972; Handley, 1976). graphic sources for Mus aguti does not sup- REMARKS: Although Dasyprocta aguti has port Thomas's (1898) statement that all were long been used as the technical name for one based on Marcgraf's species. Whereas it is or more geographic forms of red- or yellow- true that Piso's (1658: 102) and Ray's (1693: rumped agoutis (e.g., by Waterhouse, 1848; 226) accounts were obviously extracted from Thomas, 1917; Cabrera, 1961), the basis for Marcgraf (1648), Brisson's (1756: 143) this traditional usage is problematic. Because agouti description was based on an examined Linnaeus's (1766) original description of specimen (as previously noted by Tate, Mus aguti did not mention rump color, the 1935). Unfortunately, Brisson's description is identi®cation of aguti has been justi®ed pri- taxonomically uninformative (like Marc- marily by reference to the bibliographic graf's), and the geographic origin of his sources of his account (Marcgraf, 1648; Piso, specimen is not stated. However, Brisson was 1658; Ray, 1693; Brisson, 1756). Thomas employed in the natural history cabinet of the (1898) argued that all of Linnaeus's sources French naturalist ReÂaumur (Taton, 1970), so for Mus aguti could be traced back to the it is probable that all of the Neotropical spec- Brazilian animal described by the Dutch ex- imens seen by him were from French Gui- plorer-naturalist Georg Marcgraf. According ana. Additionally, Brisson (1756: 144) gave to Thomas, Marcgraf's agouti was a yellow- the geographic range of his agouti as ``Gui- rumped animal, but this inference is not con- ania & Brasilia'' and he cited BarreÁre's sistent with what is known about the South (1741) description of the `Àgouty'' of Cay- American travels of that author. enne. After ReÂaumur's death, his collection Marcgraf, an employee of the Dutch West was transferred to the Cabinet du Roi (Taton, India Company, principally resided at Recife 1970), which subsequently became part of in the northeastern state of Pernambuco, the MuseÂum National d'Histoire Naturelle. from which base he explored the region ef- Geoffroy (1803), however, recorded no fectively controlled by the 17th-century MNHN agouti material from the `àncien Dutch millitary occupation of northeastern cabinet'', nor does that museum now contain Brazil (Whitehead, 1979). However, red- (or any 18th-century specimen of Dasyprocta yellow-) rumped agoutis are not known to that might have been the one described by occur in Pernambuco (see range map in Em- Brisson (L. Granjon, personal commun.). mons, 1990, 1997), nor anywhere else within Finally, Linnaeus's (1766: 80) account of the geographic limits of Dutch Brazil (as Mus aguti gave the geographic range as mapped by Boxer, 1973). Instead, Marcgraf's ``Brasilia, Surinamo, Guiania'' despite the agouti was almost certainly the black-rumped fact that none of his cited references for this species of the northeastern Brazilian coast, a species mentioned Surinam. It is therefore taxon currently known as D. prymnolopha probable that Linnaeus himself either had (Wagler, 1831). Marcgraf's (1648) brief color seen specimens of Surinamese agoutis or had description and an accompanying woodcut of reliable reports of them from his many Dutch his `Àguti vel Acuti Brasiliensibus'' are too colleagues and aquaintances. Regrettably, no ambiguous to support or refute this inference, material of Dasyprocta that is certainly but a painting by one of Marcgraf's contem- known to have been examined by Linnaeus poraries in Dutch Brazil (reproduced by has apparently been identi®ed in the litera- Teixeira, 1995: vol. 5, p. 28) shows the local ture, nor have we succeeded in locating any species as a more-or-less reddish animal with nomenclaturally useful 18th-century speci- a blackish middorsal stripe over the rump, mens in museums known to contain Linnae- closely resembling D. prymnolopha of cur- an mammals. However, any Surinamese rent usage. agoutis seen by or reported to Linnaeus were Although this line of reasoning suggests necessarily red-rumped animals. that Dasyprocta aguti is a senior synonym of Linnaeus's Mus aguti is therefore compos- D. prymnolopha (as suggested by Carvalho ite, having been based directly or indirectly 144 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 on lost specimens of both the red- and the MyoproctaÐcommonly known as acou- black-rumped species. Because Thomas's chiesÐare restricted to Amazonia, where (1898) proposal to restrict aguti to Marc- most authors have recognized two species, graf's Brazilian animal was not equivalent to one ``red'' (or ``reddish'') and the other a lectotype designation, the application of the ``green'' (or ``greenish''), based primarily on name aguti is still an open question. In our coat-color differences. Despite this consen- judgment, it would not be desirable to re- sus, the diagnostic morphological characters place prymnolopha, the name by which the and geographic distribution of red and green black-rumped agouti of the Brazilian Atlantic acouchies have yet to be convincingly doc- coast has been long and consistently known, umented by any published revisionary study with aguti, the traditional name for the red- based on museum specimens, and the no- rumped species. We therefore select as the menclature of Myoprocta species is currently neotype for Mus aguti Linnaeus (1766) the confused. same specimen (RMNH 20752) that Husson Red and green acouchies can be distin- (1978) designated as the neotype of Mus le- guished unambiguously by a combination of porinus Linnaeus (1758). By this action external and cranial characters whose diag- Dasyprocta aguti becomes an objective ju- nostic value we tested by examining all of nior synonym of D. leporina and current us- the Myoprocta specimens (including types) age (Woods, 1993) is preserved. in ®ve American and European museums OTHER SPECIMENS EXAMINED: SurinamÐ (AMNH, BMNH, FMNH, MNHN, and Brokopondo, Locksie Hattie on the Saramac- USNM). In external appearance, most red ca River (FMNH 95757, 95758, 95760); Ma- acouchies are rich reddish-brown dorsally, rowijne, Paloemeu Camp (FMNH 95763, with uniformly orange or reddish underparts, 95765, 95767±95771); Nickerie, Wilhelmina whereas most green acouchies are drab yel- Mountains on West River (FMNH 95772± lowish- or grayish-brown dorsally with yel- 95778, 95790±95792); Para, Zanderij lowish underparts that are usually marked by (BMNH 1952.1152±1952.1154); Parama- white midventral streaks. Occasional skins of ribo, Paramaribo (BMNH 1952.1155, both the red and green species, however, 1952.1156); Saramacca, Dirkshoop (FMNH have somewhat intermediate pigmentation. 95761); Suriname, Carolina Kreek (FMNH Thus, a few red acouchies (especially zoo 95756). specimens) have rather drab fur (e.g., AMNH FIELD OBSERVATIONS: We saw or heard 130148, BMNH 5.11.1.19, FMNH 21786), Dasyprocta leporina almost every day that and a few green acouchies (e.g., AMNH we were in the forest at Paracou, often in the 68243, BMNH 54.608) have warmer pelage early morning or late afternoon, less fre- tones than usual. Some coat-color variation quently in the middle of the day, and only is geographic in origin (e.g., red acouchies rarely after dark (most nocturnal sightings from the Guianas have generally less satu- may have been of individuals frightened rated pigments than Brazilian specimens from their resting places in dense vegeta- from the north bank of the Amazon), but in- tion). Most individuals were encountered sin- dividual tonal differences also exist within gly, but a few groups of two or three indi- most large series (e.g., 16 AMNH skins of viduals were also seen. We recorded sight- green acouchies from San JoseÂ Abajo, Ec- ings in well-drained primary forest, swampy uador). Nevertheless, the difference in col- primary forest, creekside primary forest, and oration between red and green acouchies is roadside secondary growth. Both of our conspicuous when typical exemplars of both voucher specimens were shot. kinds are viewed side-by-side. Another obvious external difference is the Myoprocta acouchy (Erxleben) presence in all red acouchies of a distinct Figures 72A, 74, 75 rump patch of very long (60±80 mm) highly polished hairs that are typically much more VOUCHER MATERIAL: AMNH 266566; heavily pigmented than the fur of the sides, MNHN 1998.678. Total ϭ 2 specimens. middle back, and forequarters; in life, these IDENTIFICATION: Members of the genus hairs form a dark, glossy fringe that actually 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 145 extends beyond the rump to overlap the base of the tail (Emmons, 1990, 1997). Some specimens (e.g., AMNH 93043) have blackish rump hairs, but the usual color is a deep mahogany brown with or without inconspicuous basal bands of red. In a few Brazilian specimens (e.g., AMNH 94068, 94071; BMNH 20.7.1.23), the rump hairs are not substantially darker than the hairs of the sides and middle back, but they are still distinguishable from the fur of the latter parts by their length and high polish. By contrast, green acouchies never have a distinct rump patch, the fur over the hindquarters being essentially similar in length, color, and texture to the rest of the dorsal pelage. Red acouchies are, on average, larger than green acouchies, a comparison that is best appreciated by craniometric comparisons of representative series (table 41). Visually, most red acouchy skulls have noticeably larger toothrows and bullae, broader interorbits, and longer nasals than most green acouchy skulls. Although no univariate measurement is diagnostic, red and green acouchies have nonoverlapping morphometric distributions in multivariate ordinations (e.g., by principal components analysis; not shown). Of the several qualitative cranial traits that Tate (1939: table IV) used to distinguish red and green acouchies, the most consistently useful is the size of the ``palatal foramina'' (ϭ sphenopalatine vacuities), which perforate the bony roof of the mesopterygoid fossa; these are very narrow slits (Ͻ1 mm wide) in most specimens of red acouchies (®g. 72A), but they are wider (Ͼ1 mm) teardrop-shaped openings in most specimens of green acouchies (®g. 72B). Based on specimens that we examined, red acouchies occur throughout Guyana, Suri- nam, French Guiana, and Brazil north of the Amazon and east of the Rio Branco (®g. 73). We are aware of only two vouchered records of red acouchies outside the Guiana subregion of Amazonia: (1) one specimen (AMNH 37123) collected by Leo E. Miller Fig. 72. Fenestration of the bony roof of the on 21 April 1914 on the ``Lower SolimoÄes'', mesopterygoid fossa in typical examples of My- a locality that is now believed (see footnote oprocta acouchy (A, AMNH 266566) and M. 17) to correspond to ManacaparuÂ, a settle- pratti (B, FMNH 125085). The sphenopalatine ment on the north bank of the upper Amazon vacuities (spv) are usually present only as narrow just west of its con¯uence with the Rio Ne- (Ͻ1 mm) slits in the bony roof of the mesopter- gro (locality 5 in ®g. 73); and (2) two spec- ygoid fossa of M. acouchy, but in M. pratti these fenestrae are wider, teardrop-shaped apertures. 146 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 41 Comparisons of Craniodental Measurements (mm) from Representative Samples of Adult Red Acouchies (Myoprocta acouchy) and Green Acouchies (M. pratti)

imens (FMNH 50895, 50896) collected by ranges of red and green acouchies is due to A. M. Olalla on 9 August 1936 at Lago do the confused technical nomenclature for Baptista on the south bank of the Amazon these animals. between the Rio Madeira and the Rio Tapa- Erxleben's (1777) original description of jos (locality 3 in ®g. 73). Based on the latter Cavia acouchy mentioned only small size, record, we assume that unvouchered reports presence of a tail, and olivaceous coloration of unidenti®ed Myoprocta from other sites as characters distinguishing this species from along the south bank of the Amazon between other terrestrial hystricognaths then referred the Madeira and the Tocantins (e.g., George to the genus Cavia. Apparently, Erxleben did et al., 1988; Voss and Emmons, 1996: ap- not examine any specimens himself, but in- pendix 8) were probably of red acouchies. stead based his description of acouchy on the Acouchies are apparently unknown east of earlier accounts of des Marchais (1730), Bar- the Rio Tocantins (Carvalho and Toccheton, reÁre (1741), Buffon (1767), and Pennant 1969; Pine, 1973). (1771) that he cited as references. We ex- By contrast, green acouchies are exten- amined all of these early works and deter- sively distributed in western Amazonia (west mined that Buffon and Pennant contain noth- of the zoogeographic axis de®ned by the Rio ing more than rephrased versions of des Mar- Negro and the Rio Madeira), and they also chais' and BarreÁre's very brief reports about occur in the headwaters of the Orinoco in the ``Agouchi'' or ``Akouchy'' of Cayenne. southernmost Venezuela (Tate, 1939; Han- By way of description, des Marchais stated dley, 1976; Linares, 1998). Although a few only that the Agouchi is smaller and tastes published sources imply that red and green better than the Agouti (Dasyprocta leporina), acouchies occur sympatrically in eastern Ec- but BarreÁre (quoted verbatim by Tate [1935: uador (LoÈnnberg, 1925) or eastern Colombia 331] and Husson [1978: 471]) mentioned a (Emmons, 1990, 1997), our specimen data tail and olive coloration. suggest that red and green acouchies are al- There has never been any question that lopatrically distributed. At least some of the Erxleben's sources were describing a My- historical uncertainty about the geographic oprocta from French Guiana, but the refer- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 147

Fig. 73. Geographic range of Myoprocta acouchy based on specimens examined. 1, BRAZIL, Ama- paÂ, Serra do Navio; 2, BRAZIL, Amazonas, Boca Rio Piratucu; 3, BRAZIL, Amazonas, Lago do Bap- tista; 4, BRAZIL, Amazonas, Lago do Serpa; 5, BRAZIL, Amazonas, Lower SolimoÄes; 6, BRAZIL, Amazonas, Santo Antonio do Amatari; 7, BRAZIL, ParaÂ, Colonia do Veado; 8, BRAZIL, ParaÂ, Faro; 9, BRAZIL, ParaÂ, Monte Alegre; 10, ParaÂ, Cachoeira Porteira; 11, Roraima, ConceicËaÄo; 12, BRAZIL, Roraima, Serra Grande; 13, FRENCH GUIANA, Paracou; 14, FRENCH GUIANA, Saut Macaque; 15, FRENCH GUIANA, St.-Laurent du Maroni; 16, FRENCH GUIANA, Tamanoir; 17, GUYANA, Cuyuni- Mazaruni, Kartabo; 18, GUYANA, Pomeroon-Supenaam, Supinaam River; 19, GUYANA, Potaro-Si- paruni, Potaro; 20, GUYANA, Upper Demerara-Berbice, Moraballi; 21, GUYANA, Upper Takutu- Upper Essequibo, Dadanawa; 22, SURINAM, Brokopondo, Locksie Hattie; 23, SURINAM, Marowijne, Paloemeu Camp; 24, SURINAM, Nickerie, Kaiserberg Airstrip; 25, SURINAM, Nickerie, Wilhelmina Mountains. ence to olive coloration has led to con¯icting justify the word being applied to them.'' Ca- taxonomic applications of the epithet acou- brera (1961) agreed, noting that Erxleben's chy. Thomas (1926: 639) argued that this original sources were not professional natu- name applies to the reddish species based on ralists and might have been describing the the geographic origin of des Marchais' and species inaccurately from memory. By con- BarreÁre's observations, noting that ``many trast, Tate (1939), Carvalho (1962), and Hus- specimens of this animal are of a somewhat son (1978) applied the name acouchy to the olivaceous tone, which, in the absence of green species based on BarreÁre's color de- [M.] pratti [the green acouchy], might easily scription; according to these authors, the red 148 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 acouchy should be called M. exilis (Wagler, can be provisionally referred to M. pratti 1831). pending a comprehensive study of this The crux of this disagreement is whether group. geography or color is to be given greater im- OTHER SPECIMENS EXAMINED: BrazilÐ portance in applying the name acouchy. In AmapaÂ, Serra do Navio (USNM 546313); the absence of an extant type,20 the issue can- Amazonas, Boca Rio Paratucu on Rio Ja- not be de®nitely resolved, and both of the mundaÂ (AMNH 94073±94075), Lago do con¯icting usages mentioned above are cur- Baptista on Rio Amazonas (FMNH 50895, rent in the literature (e.g., Woods, 1993; Em- 50896), Lago do Serpa on Rio Amazonas mons, 1990, 1997). Since the geographic da- (FMNH 50897), ``Lago do Taraci on Rio Ne- tum is de®nite whereas the color description gro'' (BMNH 27.8.11.52), Lower SolimoÄes is subject to interpretation, we favor Thom- (AMNH 37123), Santo Antonio do Amatari as's (1926) and Cabrera's (1961) usage, (AMNH AMNH 92886±92888, 93043); which also avoids the absurdity of making ParaÂ, ``Castanhal on Rio JamundaÂ'' (AMNH French GuianaÐwhere only red acouchies 94076), Colonia do Veado (BMNH are known to occurÐthe type locality for the 12.5.11.9), Faro (AMNH 94068±94072), green species. In order to ®x this application Monte Alegre (BMNH 20.7.1.23), Cachoeira of Myoprocta acouchy and thereby stabilize Porteira (USNM 546296, 546297), San JoseÂ the species-level nomenclature of Myoproc- on Rio JamundaÂ (AMNH 94077); Roraima, ta, we select as the neotype of Cavia acouchy ConceicËaÄo (FMNH 20007), Serra Grande Erxleben (1777) our adult Paracou voucher, (FMNH 20019). French GuianaÐSaut Ma- AMNH 266566, consisting of a well-pre- caque (MNHN 1962.1329), St.-Laurent du served skin (®g. 74), skull (®g. 75), and post- Maroni (MNHN 1909.243), Tamanoir on cranial skeleton; measurements of this spec- Mana River (FMNH 21783, 21785±21787), imen are provided in table 42 along with no other locality data (MNHN 1962.1330, those of other conspeci®c adults from French 1974.268). GuyanaÐ``Bonasica on Essequi- Guiana. bo River'' (AMNH 36493 [type of demer- Our assignment of nominal taxa in the ge- arae], BMNH 12.6.5.28, 13.5.23.6± nus Myoprocta to either the red or green spe- 13.5.23.9), ``Kuitaro River'' (USNM cies groups (table 43) is based on ®rst-hand 338969±338971), ``Manarica Creek on Es- examination of types and/or original descrip- sequibo River'' (BMNH 13.6.8.11), ``Moon tions. Although some Brazilian specimens of Mountains'' (BMNH 11.6.7.45), ``Supinaam red acouchies are larger and redder than most River'' (BMNH 20.7.1.20±20.7.1.22); Cu- specimens from Guyana, Surinam, or French yuni-Mazaruni, Kartabo (AMNH 8178); Po- Guiana, all red acouchies closely resemble taro-Siparuni, Potaro (BMNH 3.4.6.6, one another and we see no compelling evi- 3.4.6.7); Upper Demerara-Berbice, ``Com- dence that more than a single species is rep- ackka on Demerara River'' (BMNH resented. Green acouchies, however, appear 5.11.1.19), Moraballi (BMNH 34.6.30.57± to exhibit signi®cant geographic variation 34.6.30.61); Upper Takutu-Upper Essequibo, and may eventually prove to be a complex Dadanawa (USNM 339670). SurinamÐ of closely related species; nevertheless, all Brokopondo, Locksie Hattie (FMNH 95753 [two specimens with this number]); Marow- 20 According to Husson (1978: 472), ``we may con- ijne, Paloemeu Camp on Tapahoni River sider BarreÁre's specimen to be the lectotype of this spe- (FMNH 95593, 95754, 95787); Nickerie, cies'', but it is unclear whether BarreÁre actually saw an Kaiserberg Airstrip on Zuid River (FMNH acouchy himself or was merely repeating what he had 93270±93277), Wilhelmina Mountains on been told about it by natives. Geoffroy (1803: 167) mentioned an old Paris museum specimen (``Individu prov- West River (FMNH 95755). enant de l'ancien cabinet'') of Cavia acuschi [sic] that FIELD OBSERVATIONS: We frequently heard might have been seen by BarreÁre, but Rode (1945) did the alarm calls of Myoprocta acouchy and we not list it in his catalog of MNHN rodent types, and the caught glimpses of ¯eeing individuals on museum does not now contain any 18th-century specimens of Myoprocta (L. Granjon, personal commun.). In many occasions, but we seldom obtained an the absence of any indication to the contrary, we assume unobstructed view of the animal in repose. that the lectotype no longer exists. Our single adult voucher was taken on the 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 149

Fig. 74. Dorsal and ventral views of the skin of AMNH 266566, neotype of Cavia acouchy Erxleben, 1777 (ϭ Myoprocta acouchy). Both views about ϫ0.3. 150 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 75. Dorsal, ventral, and left lateral views of the skull of AMNH 266566, neotype of Cavia acouchy Erxleben, 1777 (ϭ Myoprocta acouchy). All views approximately ϫ1.3. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 151

TABLE 42 TABLE 43 Measurements (mm) and Weight (kg) of the Species Groups, Nominal Taxa, and Type Neotype and Other Specimens of Myoprocta Localities of Acouchies (Genus Myoprocta) acouchy from French Guiana

ground in a leghold trap in swampy primary forest, and a juvenile specimen was subsequently taken in a Victor trap tied to the base of a broad liana 50 cm above the ground in well-drained primary forest. Four unvouchered observations recorded in our ®eldnotes document the presence of this species in well-drained primary forest, swampy primary forest, creekside primary forest, and secondary vegetation. Most of our sightings were diurnal, usually in the very early morning or late afternoon, but a few animals were ¯ushed from their hiding places at night.

CUNICULIDAE Only a single cuniculid species occurs at Paracou. The lowland paca is unmistakable in external and craniodental characters and cannot be confused with any other species of Guianan mammal.

Cuniculus paca (Linnaeus) IDENTIFICATION: Our specimens are topo- VOUCHER MATERIAL: AMNH 265954, types of this species (the type locality of 266567, 266569; MNHN 1998.679. Total ϭ which was restricted to French Guiana by 4 specimens. Tate, 1935) and agree closely with Husson's 152 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

(1978) detailed description and illustrations chimys. Based on collections from other lo- of Surinamese material. Selected measure- calities in French Guiana and Surinam (ap- ments (mm) and weights (kg) of two adult pendix 1), it seems probable that two addi- male vouchers (AMNH 265954, MNHN tional echimyids could occur locally. Al- 1998.679) are: HBL 650, 739; LT 0, 11; HF though species of Proechimys can only be 115, 121; Ear 52, 53; CIL 138.1, 134.4; LD distinguished from one another with adult 54.3, 52.7; MTR 27.3, 30.5; LN 52.4, 51.0; specimens in the hand, other French Guianan LIB 41.2, 42.4; ZL 72.3, 77.0; ZB 98.4, echimyids can be identi®ed at a distance by 105.2; Wt 9.5, 9.2. well-marked external characters (Emmons, The lowland paca ranges from southern 1990, 1997) if a suf®ciently clear view is ob- Mexico to Paraguay, and several subspecies tained. Unfortunately, although the taxa in have been recognized as valid by authors (e.g., question are readily distinguished, their no- Krumbiegel, 1940b; Cabrera, 1961). Although menclature is still problematic in some cases. Krumbiegel (1940b) cited pelage and cranial characters to justify his trinomial distinctions, Makalata didelphoides (Desmarest) no comprehensive review of geographic vari- Our only record of this species at Paracou ation in this species based on the very large is an unambiguous sighting by DPL of a sol- museum collections now available for study itary adult perched on a tree trunk 3 m above has yet been attempted. In the absence of such a small stream in primary forest at 10:30 a critical undertaking, a subspeci®c nomencla- hours on 27 October 1992. ture does not seem warranted at present; how- In most of the older literature (e.g., Tate, ever, our material would obviously represent 1935, 1939; Cabrera, 1961) this species is the nominate race if trinomials were to be rec- called Echimys armatus (I. Geoffroy), but ognized in any future revision. Husson (1978) proposed the new genus Mak- In a recently published opinion, the Inter- alata with armatus as type species in rec- national Commission on Zoological Nomen- ognition of the well-marked craniodental dif- clature (ICZN, 1998) ruled that Cuniculus ferences between this taxon and Echimys was available from Brisson (1762), thus re- chrysurus (the type species of Echimys G. placing Agouti LaceÂpeÁde (1799) as the oldest Cuvier). For the use of didelphoides to re- valid generic name for the lowland paca. place armatus as the oldest available name FIELD OBSERVATIONS: We collected four for the red-nosed tree rats of the Guiana sub- specimens of Cuniculus paca at Paracou and region of Amazonia, see Emmons (1993b). recorded 24 unvouchered observations in our ®eldnotes. Habitat information accompany- Mesomys sp. ing 26 records include 7 encounters (27%) in well-drained primary forest, 4 (15%) in VOUCHER MATERIAL: AMNH 266596. swampy primary forest, 6 (23%) in creekside IDENTIFICATION: The genus Mesomys has primary forest, 1 (4%) in primary forest of never been revised and the identi®cation of unspeci®ed character, and 9 (35%) in sec- its constituent species has long been prob- ondary vegetation; all encounters were noc- lematic. Woods (1993) recognized M. didel- turnal. Pacas were encountered singly, except phoides (Desmarest), M. hispidus (Desma- in rare cases when we witnessed agonistic rest), M. leniceps Thomas, M. obscurus encounters between two adults or saw an (Wagner), and M. stimulax Thomas as valid adult female accompanied by a juvenile. species, but Emmons (1993b) showed that Most individuals were sighted on the ground, didelphoides and obscurus both belong in the but one individual was submerged in a genus Makalata, and Patton et al. (2000) stream under a fallen tree, apparently at- subsequently described a new species, Me- tempting to hide underwater. somys occultus. Therefore, current usage would now recognize four valid species of ECHIMYIDAE Mesomys: M. hispidus (including ecaudatus We captured or observed four echimyid Wagner, ferrugineus GuÈnther, and spicatus species at Paracou, including representatives Thomas as synonyms; after Woods, 1993), of the genera Makalata, Mesomys, and Proe- M. leniceps, M. occultus, and M. stimulax. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 153

In order to identify our single Paracou samples have mean toothrow lengths greater voucher, we consulted the original descrip- than 7 mm (e.g., the series measured by Pat- tions of all nominal taxa currently referred to ton et al., 2000: tables 56, 58). Small-toothed Mesomys, and we examined the holotypes of and large-toothed Mesomys are sympatric at every named form except ecaudatus. In ad- IgarapeÂ Amorim on the left (west) bank of dition, we measured relevant series of spec- the Rio Tapajos (specimens in AMNH), but imens in the AMNH, USNM, and BMNH to elsewhere these morphometric classes appear assess geographic variation in morphometric to be allopatric. Although it is geographically characters. Below we explain why, despite part of the Guiana subregion of Amazonia, this effort, we are still unable to con®dently the Venezuelan state of Amazonas is inhab- assign a speci®c epithet to our material. ited by large-toothed Mesomys (e.g., nine Of all the material we examined, the most USNM specimens with a mean maxillary distinctive is the holotype of Mesomys leni- toothrow length of 7.2 mm), an observation ceps from the Andean highlands of eastern consistent with the presence of other western Peru. This specimen (BMNH 26.8.6.61) is Amazonian taxa (e.g., Dasyprocta fuliginosa, distinguished from all other congeners by its Myoprocta pratti; see above) in that area. much longer tail (relative to head-and-body Although western Amazonian Mesomys length), ®ner and denser tail hairs, ®ner and specimens have often been identi®ed as M. softer spines, more convergent toothrows, hispidus (e.g., by Cabrera, 1961; da Silva and posteriorly constricted incisive foramina, and Patton, 1993; Patton et al., 1994, 2000; Voss much smaller bullae. By contrast, Amazo- and Emmons, 1996), this usage appears to be nian specimens of Mesomys (including all of incorrect. The maxillary toothrow of the type the remaining nominal taxa in this genus) of hispidus21 measures 6.4 mm, well outside have relatively shorter tails, coarser spines the observed range of variation in this di- and tail hairs, less convergent toothrows, mension for any known western Amazonian posteriorly unconstricted incisive foramina, sample. Unfortunately, it is not known ex- and larger bullae. Although we searched dil- actly where the type was collected. Desma- igently for external and craniodental charac- rest (1817) gave the type locality as ter variation among Amazonian samples of ``l'AmeÂrique meÂridionale'', but the specimen Mesomys, only maxillary toothrow length was almost certainly obtained by Alexandre appears to offer any broadly useful morpho- Rodriguez Ferreira,22 whose known collect- logical basis for taxonomic inference. The ®rst example of Mesomys to be re- 21 Rode's (1945) catalog of MNHN rodent types lists ported from any of the Guianas was a Suri- that of Echimys hispidus as a mounted specimen (num- namese specimen that Husson (1978) iden- ber 1806 in the old ``montage'' series of the MNHN) ti®ed as M. stimulax. We have examined ®ve from which the skull had been extracted and lost. The additional specimens (table 44) from the skull, however, identi®ed as that of the type by a BMNH label in Thomas's handwriting, is still preserved as A- Guiana subregion, including our Paracou 7668 in the nearby Laboratoire d'Anatomie CompareÂe. voucher, all of which essentially resemble Emmons (1993b) and RSV independently examined Husson's material and appear to represent the both skin and skull, which we judge to be correctly as- same taxon despite substantial variation in sociated in the absence of any indication to the contrary. 22 The 19th-century wooden base on which the type some cranial dimensions (plausibly attribut- skin of Mesomys hispidus is still mounted bears the in- able to age differences) and in external pro- scription ``du Cabinet de Lisbonne 1808'', indicating portions (perhaps attributable to measure- that this specimen was part of the plunder that Napo- ment artifacts). Consistent with Husson's leon's troops shipped to Paris following the sack of Lis- identi®cation, these Guianan specimens have bon. That shipment apparently included the entire collection of natural history specimens assembled in Brazil small toothrows (MTR ϭ 5.9±6.6 mm) that between 1783 and 1792 by Ferreira, an employee of the fall within the range of variation exhibited Museu Royal d'Ajuda (Hershkovitz, 1987a). Signi®- by specimens referable to stimulax from cantly, the primates illustrated in Ferreira's ``Viagem Fi- southeastern Amazonia (right-hand column, losoÂ®ca'' include two taxa (Chiropotes satanas chiropotes and Saguinus midas midas; op. cit.: table 2) that table 44). By contrast with these small- must have been collected or observed in the Guiana sub- toothed Guiana and Southeastern subregion region of Amazonia, where the type of Mesomys hispi- samples, all western Amazonian Mesomys dus might also have been taken. 154 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 44 Measurements (mm) and Weights (g) of Adult Specimens of Small-Toothed Mesomys from the Guiana and Southeastern Subregions of Amazonia

ing itinerary (reproduced by Hershkovitz, small-toothed animal (like hispidus) or not. 1987a: ®g. 3) was largely con®ned to Ama- Until the type of ecaudatus is examined and zonian Brazil. Tate (1935) proposed restrict- measured, the oldest available name based on ing the type locality of hispidus to Borba, a a large-toothed Mesomys specimen is ferru- settlement on the right bank of the lower Rio gineus GuÈnther (1876) from northeastern Madeira where Natterer collected the type of Peru. ecaudatus in 1830 (Pelzeln, 1883). Tate's ac- Based simply on toothrow measurements, tion was biologically arbitrary but served to our Paracou voucher and other similar spec- justify his synonymization of hispidus and imens from the Guiana subregion of Ama- ecaudatus, the types of which he had not zonia could justi®ably be referred to Meso- seen. Unfortunately, two species of Mesomys mys hispidus, of which stimulax is perhaps with divergent toothrow measurements are only a junior synonym. However, Patton et known to occur in the inter¯uvial region be- al. (2000) reported high cytochrome-b se- tween the Madeira and the Tapajos (e.g., at quence divergence between small-toothed IgarapeÂ Amorim, see above) and it is not Mesomys specimens from opposite sides of known whether the type of ecaudatus is a the Amazon, a result that clearly indicates the 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 155

insuf®ciency of toothrow length as a basis French Guiana, respectively. Although both for taxonomic inference and brings into authors recognized a common large species sharper focus the problem of exactly where and a rarer small species, they gave different the type of hispidus was collected (see foot- accounts of diagnostic nonmetrical charac- note 22). For the moment, these are insoluble ters and they used different nomenclature. problems for which the only appropriate in- Husson (1978) identi®ed his large Suri- terim solution is to leave our voucher mate- namese specimens as Proechimys guyannen- rial unnamed. sis (E. Geoffroy, 1803), and his small spec- OTHER SMALL-TOOTHED SPECIMENS EXAM- imens as P. warreni Thomas (1905). In Hus- INED FROM THE GUIANA SUBREGION: BrazilÐ son's opinion, measurements of the cheek- AmapaÂ, Serra do Navio (USNM 543283). teeth provided the best diagnostic criterion: French GuianaÐLes Nouragues (V-924). 18 specimens of the large Surinamese spe- GuyanaÐ Potaro-Siparuni, 5 km SE Sura- cies had maxillary toothrow lengths of 8.3± ma (ROM 103346); Upper Takutu-Upper 9.2 mm, whereas 6 specimens of the small Essequibo, Maipaima Creek in Kanuku species had cheektooth measurements of Mountains (LHE 968, an uncataloged speci- 7.2±7.7 mm. Husson also reported that his men to be deposited in UG). large species was more abundant, and had FIELD OBSERVATIONS: Our single record of more rufous pelage, a relatively shorter tail, Mesomys from Paracou is based on a speci- less appressed caudal hairs, and a wider me- men taken in a Victor trap tied to a broad sopterygoid notch in the back of the hard liana 1.9 m above the ground in swampy pri- palate than his small species (op. cit.: 429± mary forest. 438). Husson based his identi®cation of the large species primarily on Geoffroy's (1803) Proechimys J. A. Allen description of pelage color in guyannensis because the skull of the type (not measured The last comprehensive revision of the ge- by Geoffroy) was thought to have been lost nus Proechimys was Moojen's (1948) mono- (Rode, 1945). Husson's identi®cation of the graph, but subsequent research has resulted small species, however, was based on direct in substantial modi®cations of his pioneering external and craniodental comparisons with taxonomic synthesis. In particular, studies of the type of warreni. well-sampled local faunas have proven cru- Petter (1978) reported the rediscovery of cial for distinguishing patterns of morpho- the type skull of Proechimys guyannensis logical, karyotypic, and molecular variation and assigned this name to the smaller (and within and among sympatric species of Proe- rarer) of the two forms found in French Gui- chimys, especially in western Amazonia ana; the larger (and more abundant) form, where four or more occur at some localities apparently undescribed, was then named as (Patton and Gardner, 1972; da Silva, 1998; a new species, P. cuvieri. However, whereas Patton et al., 2000). To date, however, no Husson found that the large and small Suri- equivalent studies of character variation namese species were nonoverlapping in max- within and among sympatric Proechimys illary toothrow length, Petter reported over- species from other Amazonian subregions lapping variation in this measurement, with have been published. observed ranges of 7.0±8.5 mm for guyan- Until quite recently, only a single wide- nensis and 8.1±9.3 mm for cuvieri. Petter did spread species of Proechimys, variously not mention the other distinguishing charac- identi®ed as P. guyannensis (E. Geoffroy, ters discussed by Husson, but he noted kar- 1803) or as P. cayennensis (Desmarest, yotypic differences (2N ϭ 40 in guyannensis, 1817), was recognized in the Guianas (e.g., 2N ϭ 28 in cuvieri) and differences in the by Tate, 1935; Cabrera, 1961). In 1978, how- number of enamel islands of the lower ever, A. M. Husson (at the Rijksmuseum van cheekteeth (two per tooth in guyannensis, Natuurlijke Historie, Leiden) and F. Petter (at three in cuvieri). The karyotypes of guyan- the MuseÂum National d'Histoire Naturelle, nensis and cuvieri were subsequently de- Paris) independently reported the occurrence scribed in greater detail by Reig et al. (1979). of two sympatric forms in Surinam and in In an important review of morphological 156 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 76. Frequency histogram of maxillary toothrow length (MTR) for all measured French Guianan specimens of Proechimys (N ϭ 72; see text). variation in Proechimys, Patton (1987) iden- both species are based on types collected in ti®ed several qualitative characters distin- French Guiana (Petter, 1978). Additionally, guishing clusters of populations assigned to the application of these names to the sym- his cuvieri and guyannensis species groups. patric Surinamese forms reported by Husson However, Patton's study included no French remains to be evaluated. In order to identify Guianan material, so the diagnostic value of our Paracou vouchers, and to provide a basis the characters he investigated is unknown for for future revisionary research with these the local populations sampled by our study. species, we examined most of the Proechi- Indeed, Guillotin and Ponge (1984: 287) mys specimens from French Guiana currently doubted that cuvieri and guyannensis could held in American and European museums. be distinguished by ``meÂthodes morpholo- Except as noted below, our results are based giques classiques'', a conclusion based on bi- on measurements and qualitative character variate and multivariate analyses that showed data obtained from fully adult but nonsenes- no obvious morphometric discontinuity cent animals, herein de®ned as members of among French Guianan specimens represent- age categories 8 and 9 of Patton and Rogers ing both of the karyomorphs reported by Pet- (1983). The following results of our analyses ter (1978) and Reig et al. (1979). The ap- broadly overlap those independently ob- parent lack of diagnostic external characters tained by Catze¯is and Steiner (2000), who between guyannensis and cuvieri has also examined many of the same specimens. been an impediment for ecologists unable to Maxillary toothrow measurements of non- identify these species by nondestructive sam- senescent adult Proechimys from French pling in ®eld studies (e.g., Forget, 1991). Guiana form two distributions (®g. 76): a The morphological characters of French sparse cluster of small specimens with cheek- Guianan populations of Proechimys cuvieri teeth measuring 6.9±8.0 mm, and a denser and P. guyannensis are important because cluster of larger specimens with longer tooth- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 157

TABLE 45 Measurements (mm) and Weights (g) of Adult Proechimys cuvieri and P. guyannensis from French Guianaa

rows (8.2±9.3 mm). Following Petter (1978), sence of contact between the bony capsules we associate the name guyannensis with the containing Jacobson's organ and a median smaller-toothed form based on the toothrow process of the maxillary bone. Initially, we dimensions of the holotype (MNHN distinguished complete septa formed by slen- 1995.1395, an old adult), which has an al- der maxillary processes from complete septa veolar measurement of 7.8 mm and an esti- formed by robust maxillary processes, but in- mated crown measurement of 7.3 mm (table termediate conditions made this additional 45). The larger-toothed specimens that we re®nement too arbitrary for con®dent scor- measured include the type of cuvieri, MNHN ing. 1977.774, which has a crown-length tooth- We scored the development of the canal row measurement of 8.5 mm. To supplement transmitting the infraorbital nerve in the ¯oor this essentially univariate distinction, we ex- of the infraorbital foramen (®g. 78) using the amined qualitative variation in craniodental, external, and genitalic characters to deter- numerical coding suggested by Patton mine whether or not Proechimys with small (1987): no groove present (1), or groove versus large cheekteeth differ in other re- moderately developed (2), or groove well de- spects. ®ned (3). Scoring these conditions as well as We scored the septum that separates the intermediate states (1.5, 2.5) was accom- right and left incisive foramina as ``com- plished by reference to exemplar specimens plete'' or ``incomplete'' (®g. 77). Variation of Brazilian Proechimys guyannensis previ- in this character among the specimens at ously examined by J. L. Patton (personal hand is determined by the presence or ab- commun.). Therefore, we presume that our 158 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 77. Morphology of the incisive foramina and the posterior palate in Proechimys guyannensis (A, AMNH 266595) and P. cuvieri (B, AMNH 266592). Whereas the bony septum that separates the right and left incisive foramina is usually incomplete in P. guyannensis, the septum is always complete in P. cuvieri. Additionally, the incisive foramina of P. cuvieri are usually constricted posteriorly by well-developed lateral ¯anges of the maxillary bone, the posterior palate often has a prominent median keel, and the mesopterygoid fossa is typically broad and shallow. By contrast, the bony relief of the posterior palate is less prominent in most specimens of P. guyannensis, in which the mesopterygoid fossa is often narrower, more acutely angled, and penetrates farther between the toothrows. Scale bar ϭ 5 mm. trait-frequency data and his (Patton, 1987: ta- Patton, 1987), we did not score the shape of ble 3) are comparable. the mesopterygoid notch as a separate char- We also followed Patton's (1987) numer- acter. ical coding convention for the depth of the Finally, we recorded the number of inter- mesopterygoid fossa (®g. 77): not extending nal folds on the second mandibular molar to the posterior margin of M3 (1), or extend- (corresponding to the ``õÃlots d'eÂmail'' of Pet- ing to the posterior half of M3 (2), or exter, 1978); compound (Y-shaped) folds were tending to the anterior half of M3 (3), or ex- each counted as one-and-a-half folds. As not- tending to the posterior half of M2 (4), or ed by Patton (1987), internal fold number extending to the anterior half of M2 (5). Be- and morphology change with toothwear, so cause the angle formed by the posterior pal- it is particularly important that this character atal margins of the mesopterygoid is corre- be scored among individuals of approximate- lated with the depth of penetration of the fos- ly equivalent age. sa between the toothrows (deeper fossae Trait frequencies of these four qualitative have more acutely angled palatal margins; characters differ signi®cantly between our 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 159

operationally de®ned samples of cuvieri (with MTR Ն 8.2 mm) and guyannensis (with MTR Յ 8.0 mm) despite the small number of guyannensis available for scoring (table 46). The presence or absence of a complete incisive septum and the development of an infraorbital groove are the two qualitative characters most consistently correlated with maxillary toothrow length: all large-toothed specimens (cuvieri) have a complete incisive septum and either lack an infraorbital groove entirely or have a very weakly de®ned groove; by contrast, most small-toothed specimens have an incomplete incisive septum and well-de®ned infraorbital grooves. Although most specimens of both tooth-size classes have a moderately deep mesopterygoid fossa and 2½ folds on m2, no specimens of small-toothed rats in our sample have either a very shallow mesopterygoid fossa or m2s with 3 folds. It is noteworthy that the qualitative traits of the specimen with a maxillary toothrow length of 8.0 mm (MNHN 1981.48, which could be interpreted as an outlier of either tooth-size class) link it unambiguously with guyannensis, and that specimens with ``atypical'' states for one qualitative character (e.g., AMNH 267047, a small-toothed animal that lacks any trace of an infraorbital groove) are not atypical in other qualitative respects. Altogether, these results strongly support the hypothesis that the discrete tooth-size classes associated with the names cuvieri and guyannensis represent valid species, a conclusion that is further bol- stered by external, genitalic, and karyotypic comparisons. As noted by Malcolm (1992), the tail of Proechimys cuvieri is visibly hairier than that of P. guyannensis. This difference is caused by the individual caudal hairs, which curve outward from beneath each epithelial scale in P. cuvieri, where they can easily be seen standing away from the caudal surface (®g. 79, left). By contrast, the individual hairs are appressed to the caudal surface in P. guy- Fig. 78. Morphology of the ¯oor of the infraorbital foramen in French Guianan Proechimys, ← illustrating alternative conditions of the canal for the infraorbital nerve scored as character states in by moderately developed lateral ¯ange (P. cuvieri table 46. A, Floor of infraorbital foramen smooth, [AMNH 266574] scored as ``2''); C, nerve canal without nerve canal (P. cuvieri [AMNH 266572] present, de®ned by highly developed lateral ¯ange scored as ``1''); B, nerve canal present, de®ned (P. guyannensis [AMNH 266595] scored as ``3''). 160 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 46 Comparisons of Qualitative Character-State Frequencies Between Proechimys cuvieri and P. guyannensisa

annensis, where they are dif®cult to see without magni®cation; in consequence, the tail appears to be smooth and naked (®g. 79, right). The dorsal body pelage of Proechimys cuvieri is, on average, redder (more saturated) than the generally drab (grayish or yellowish brown) fur of P. guyannensis, but there is suf®cient overlap in color among the specimens at hand that this contrast is useful for Fig. 79. Caudal pelage of Proechimys cuvieri ®eld identi®cation only in combination with (left, AMNH 267029) and P. guyannensis (right, size and other characters. This seems to be AMNH 267037). Tails of cuvieri are conspicu- the only signi®cant species color difference ously hairier than those of guyannensis, an exter- in the material we examined. For example, nal difference that is useful for ®eld identi®cation we did not see any diagnostically useful dif- of these sympatric congeners. ferences in ventral fur color (pure white in most specimens of both species), tail pig- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 161

mentation (distinctly bicolored in most spec- chimys cuvieri can be characterized as large imens of both species), or hindfoot markings. rats that contrast with adults of the smaller The male genitalia differ strikingly in sympatric species P. guyannensis by their shape between Proechimys cuvieri and P. longer (Ն8.2 mm) maxillary toothrows (ver- guyannensis. Whereas the penis of cuvieri is sus Յ8.0 mm in guyannensis), complete in- short and very broad, that of guyannensis is cisive septum (vs. septum usually incomplete long and slender, an obvious contrast that is in guyannensis), incisive foramina with larg- re¯ected in the highly divergent bacular mor- er posterolateral ¯anges (vs. foramina usu- phologies illustrated by Patton (1987: ®gs. 5, ally with smaller ¯anges in guyannensis), 10) for members of his cuvieri and guyan- palatal bridge with better developed median nensis species groups. Although this char- keel (vs. palate unkeeled or weakly keeled in acter cannot be used to identify most muse- guyannensis), infraorbital groove absent or um study skins (few of which have the penis usually weakly developed (vs. groove usu- attached), it is potentially useful for ®eld ally well developed in guyannensis), mesop- identi®cations because the penis can be ex- terygoid fossa often shallow and broad (vs. truded by retracting the prepuce of live ani- deeper and narrower in many guyannensis), mals. lower molars usually with more than two in- Karyotypic data recorded on skin tags of ternal folds (vs. lower molars often with two MNHN specimens indicate that different dip- folds in guyannensis), tail visibly hairier (vs. loid counts are associated with the divergent apparently smooth and naked in guyannen- morphological phenotypes described above sis), pelage often reddish (vs. drab in most and corroborate Petter's (1978) and Reig et guyannensis), penis short and very broad (vs. al.'s (1979) taxonomic assignments. Thus, longer and narrower in guyannensis), and ®ve French Guianan specimens that we ex- diploid karyotype with 28 chromosomes (vs. amined with recorded karyotypes of 2N ϭ 2N ϭ 40 in guyannensis). 28 (MNHN 1972.639, 1974.263, 1974.266, In our experience, fully adult Proechimys 1981.36, 1998.315) represent the cuvieri in French Guiana can be identi®ed to species morphotype, whereas two specimens that we in the ®eld with considerable con®dence us- examined with 2N ϭ 40 (MNHN 1983.376, ing external measurements (e.g., hindfoot 1998.312) represent the guyannensis mor- length, table 45) and the qualitative external photype. In the following species accounts we sum- traits described above. However, taxonomic marize diagnostic characters, comment on assignments of juvenile and subadult Proe- Husson's (1978) identi®cations of Surina- chimys (with unmolted or incompletely molt- mese material, and evaluate the probable ed soft, gray, immature pelage) are always geographic distribution of Proechimys cuvi- problematic. Tooth impressions (Malcolm, eri and P. guyannensis based on our assess- 1992) and molecular markers (Steiner et al., ment of character variation in the material at 2000) are potentially useful tools for identi- hand. fying young animals that should be incor- porated in future ®eld studies. Otherwise, many individuals will inevitably remain un- Proechimys cuvieri Petter identi®ed in ecological research based on Figures 77±80 nondestructive sampling, especially during VOUCHER MATERIAL: AMNH 266570± the rainy season, when a considerable frac- 266575, 266578, 266580±266582, 266588, tion of the population consists of young an- 266589, 266591, 266592, 266594, 267025± imals (Guillotin, 1982). 267030, 267032, 267034, 267039, 267041, Specimens that we examined document 267045, 267599, 267601±267603, 269122; the sympatry of Proechimys cuvieri and P. MNHN 1998.685±1998.699. Total ϭ 46 guyannensis at several localities in French specimens. Guiana, including Arataye, Cayenne, Florida, IDENTIFICATION: Based on the preceding MontsineÂry, and St.-EugeÁne, as well as Par- synthesis of character data from French acou. We presume that these species are also Guianan material, adult specimens of Proe- co-distributed elsewhere, certainly in Suri- 162 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 80. Dorsal, ventral, and lateral views of skulls of Proechimys cuvieri (left, AMNH 266592) and P. guyannensis (right, AMNH 266595). All views approximately ϫ1.5. nam and perhaps throughout the Guiana sub- Patton et al. (2000) reported that the vomer region of Amazonia. (an element of the incisive septum) is ex- According to Patton et al. (2000), Pro- posed in most (25 out of 34) specimens from echimys cuvieri is widely distributed in Ama- the Rio JuruaÂ, this bone is not exposed in zonia, including large parts of the Guianan, any of our 22 adult Paracou vouchers. If geo- southeastern, and western subregions. Al- graphic patterns of variation in this and other though cuvieri is said to be relatively uni- morphological characters were found to be form in morphological characters throughout consistently correlated with mtDNA haplo- this enormous range, signi®cant mtDNA se- type divergence, it would be reasonable to quence divergence (7±9%) exists among sev- infer that two or more species could be rep- eral geographic clusters of populations sam- resented among the samples currently re- pled by those authors (op. cit.). We have not ferred to this species. attempted to evaluate geographic variation REMARKS: Husson (1978) identi®ed Suri- among Amazonian populations of cuvieri- namese material of this species as Proechi- like spiny rats for this faunal report, but it is mys guyannensis on the basis of Geoffroy's relevant to note that typical (French Guian- (1803) and Desmarest's (1817) color descrip- an) cuvieri appears to differ signi®cantly tions of the type, and on the assumption that from western Amazonian material in some the type of guyannensis probably represented morphological traits. For example, whereas the commoner of the two Guianan forms. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 163

Color alone, however, is not a reliable basis Proechimys guyannensis (E. Geoffroy) for species identi®cation, and the other mor- Figures 77±80 phological details mentioned in Geoffroy's VOUCHER MATERIAL: AMNH 266576, and Desmarest's descriptions are likewise in- 266577, 266586, 266595, 267037, 267038, suf®cient to determine which local species 267047; MNHN 1998.682±1998.684. Total they had in hand. Petter's (1978) rediscovery ϭ 10 specimens. of the long-lost type skull of P. guyannensis IDENTIFICATION: External and craniodental ®nally resolved the identity of that taxon and characters that distinguish this species from indicated the necessity of naming the larger Proechimys cuvieri in French Guiana are dis- form as a new species. cussed above and need not be repeated here. OTHER SPECIMENS EXAMINED: French Gui- Patton (1987) mapped the distribution of anaÐArataye (MNHN 1983.378; USNM the guyannensis species group of Proechimys 548450±548452), Cacao (MNHN 1981.107), as extending throughout eastern Amazonia Cayenne (MNHN 1970.223, 1974.263), together with adjacent parts of the northern Florida (MNHN 1981.46, 1981.50), Piste Venezuelan coast and the Brazilian Cerrado. St.-EÂ lie (MNHN 1982.523), St.-EugeÁne Besides the type species, Patton listed the (MNHN 1995.3220±1995.3222, 1995.3224, following taxa as group members: cherriei 1998.314, 1998.315, 1998.1821), SauÈl Thomas, roberti Thomas, vacillator Thomas, (MNHN 1977.774 [holotype], 1981.23, oris Thomas, warreni Thomas, boimensis Al- 1981.24, 1981.26, 1981.29±1981.33, len, arescens Osgood, riparum Moojen, and 1981.36), Station FRG near MontsineÂry arabupu Moojen (see Cabrera, 1961, for bib- (MNHN 1986.1129), Trois-Sauts (MNHN liographic references). Patton believed that 1981.54, 1981.56), no other locality data more than one valid species was represented (MNHN 1972.639, 1974.266). by these names, emphasizing morphological FIELD OBSERVATIONS: All of our unambig- differences between samples from the Gui- uous records of Proechimys cuvieri from ana subregion (for which the oldest available Paracou are based on collected specimens. name is guyannensis), and those from south Of our 46 vouchers, 18 (39%) were taken in of the Amazon (for which the oldest name is Sherman traps, 14 (30%) were shot, 7 (15%) roberti). were taken in Victor traps, 4 (11%) were tak- The morphological trait frequencies re- en in Tomahawk traps, 2 (4%) were taken in ported herein for typical Proechimys guyan- Conibear traps, and 1 was taken in a pitfall. nensis (from French Guiana) resemble those Forty-four specimens (96%) were shot or tabulated by Patton (1987) for guyannensis- trapped at ground level, but 2 specimens group samples from north of the Amazon, (4%) were taken in traps tied to lianas 0.7± which we provisionally regard as conspecif- 1.0 m above the ground. Habitat data record- ic. By contrast, specimens from southeastern ed for 45 specimens include 10 captures Amazonia have divergent trait frequencies (22%) in well-drained primary forest, 3 cap- (op. cit.) and also differ from north-bank tures (7%) in swampy primary forest, 10 samples in karyotypes and cytochrome-b se- captures (22%) in creekside primary forest, quences (Weksler et al., 2001). Apparently, 2 captures (4%) in primary forest of unspec- all recently collected Proechimys from Ve- i®ed character, and 20 captures (44%) in sec- nezuelan coastal rainforests (north of the Ori- ondary vegetation. Microhabitat notes ac- noco) are referable to other species groups companying 28 specimens trapped at ground (Aguilera and Corti, 1994; Aguilera et al., level record captures made under masses of 1995; Corti and Aguilera, 1995). Therefore, fallen branches and other debris (8 speci- P. guyannensis appears to be an Amazonian mens), under logs (5), in unsheltered sites in endemic largely, but perhaps not exclusively, dense understory vegetation (5), beside logs distributed in the Guiana subregion.23 (3), at the bases of trees (2), among stilt roots (1), on top of a log (1), inside a hollow log 23 The type locality of Echimys cherriei Thomas, a (1), on a smooth branch fallen over a stream taxon listed by Patton (1987) as a member of the guy- (1), and under an overhanging stream bank annensis species group, was incorrectly mapped (op. (1). cit.: ®g. 1) in the easternmost coastal rainforest region 164 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

REMARKS: Surinamese material of this spe- TABLE 47 cies was identi®ed as Proechimys warreni by Comparison of Capture-Habitat Frequencies Husson (1978), who used the name P. guy- between Proechimys cuvieri and P. annensis for the larger species identi®ed as guyannensis Trapped at Paracoua P. cuvieri in this report (see the preceding account). Although Proechimys guyannensis (Geof- froy, 1803) has long been recognized as a valid name (e.g., by Moojen, 1948; Hersh- kovitz, 1948b; Cabrera, 1961; Patton and Gardner, 1972; Patton, 1987), this epithet was rejected as unavailable by Woods (1993), who used the replacement name cayennensis Desmarest (1817) instead. In our opinion, Mus guyannensis and other names ®rst published by Geoffroy (1803) are unambiguously available from that work for the reasons clearly explained by Hershkovitz FIELD OBSERVATIONS: All of our de®nite (1955) and Holthuis (1963). records of Proechimys guyannensis at Para- OTHER SPECIMENS EXAMINED: French Gui- anaÐArataye (MNHN 1983.381; USNM cou are based on collected specimens. Of our 548454±548456), Cayenne (MNHN 10 vouchers, 3 (30%) were captured in Vic- 1983.374, 1983.376, 1995.1395 [holotype]), tor traps, 5 (50%) in Sherman traps, and 2 ``Fleuve Oyapock'' (MNHN 1983.375), (20%) were shot. Eight of our vouchers Florida (MNHN 1981.48, 1982.601), Mon- (80%) were shot or trapped on the ground, tsineÂry (MNHN 1986.1124), Nancibo but 2 (20%) were taken in traps tied to lianas (MNHN 1986.1130), Petit Saut (MNHN 0.5±1.6 m above the ground. Recorded hab- 1998.312), St.-EugeÁne (MNHN 1994.128), itat data include 9 captures (90%) in well- no other locality data (MNHN 1981.88, drained primary forest and 1 (10%) in creek- 1981.103). side primary forest. By comparison with Proechimys cuvieri, which we caught with almost equal frequency in primary forest and 33 ← secondary growth, P. guyannensis was cap- of northern Venezuela, where only P. trinitatis is cur- tured signi®cantly more often in primary for- rently known to occur (see range maps in Aguilera and est (table 47). Corti [1994], Aguilera et al. [1995], and Corti and Aguilera [1995]). In fact, the type locality of cherriei is Munduapo (ϭ Monduapo according to Paynter, 1982), ANALYSES OF SAMPLING on the right bank of the upper Orinoco at 4Њ54ЈN, 67Њ48ЈW in the Guiana subregion of Amazonia. We encountered nonvolant mammals on a Other concepts of Proechimys guyannensis are not daily basis at Paracou even when we made consistent with the taxonomic and geographic limits rec- no deliberate effort to collect or observe ognized in this report. In particular, the synonymy of P. them. Tamarins twittered and plunged in the cayennensis (ϭ P. guyannensis in our usage) provided by Woods (1993) includes columbianus Thomas (a canopy around our camp in the mornings, member of Patton's [1987] semispinosus species group); agoutis ¯ed barking through the undergrowth hylaea Moojen, leioprimna Moojen, nesiotes Moojen, at midday, kinkajous whistled above our bat and rattinus Thomas (members of Patton's goeldii spe- nets at night, and armadillos bolted across cies group); ochraceus Osgood (a member of Patton's the path as we returned to camp later in the trinitatus species group); and villacauda Moojen (a member of Patton's longicaudatus species group). Ap- evening. Therefore, every day we spent in parently, Woods' association of these nominal taxa with the ®eld was potentially informative as an guyannensis followed Moojen's (1948) generic revision, interval of nonvolant mammal sampling, a historically important but now outdated reference. Un- whether or not we cataloged specimens or less shown to be incorrect by subsequent research, Pat- ton's carefully documented allocations of nominal taxa recorded noteworthy sightings on a given to species groups should be used as the basis for all date. Similarly, although no more than two future classi®cations of Proechimys. persons were ever simultaneously committed 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 165

Fig. 81. Results of nonvolant mammal sampling at Paracou from 1991 to 1994, where we recorded a total of 50 species in 202 days using all direct methods of nonvolant faunal inventory (trapping, hunting/census, and miscellaneous; table 48). ``Minimum known diversity'' (64 species) includes the additional records obtained by previous researchers in our study area, together with second-hand observations that we documented by interviewing local residents. to the nonvolant mammal survey, all inven- day throughout the remainder of that ®eld tory personnel made signi®cant observations season and the next; in 1993 and 1994, the of nonvolant species from time to time. Be- average rate of species accumulation was less cause a substantial amount of diversity data than 0.1 species/day. Although these data thus accumulated involuntarily, it is dif®cult suggest an approaching asymptote, ®eld days to quantify meaningfully the time or effort were not commensurate units of sampling ef- expended on the nonvolant mammal inven- fort throughout the course of our inventory tory at Paracou. In the absence of any better because different methods were used in dif- sampling units, however, ®eld days can be ferent years, and because some methods were used to obtain a general overview of our re- more effective, or were used more intensive- sults. ly, early in the inventory than later (table 49). Counting only those recordsÐcollected Furthermore, the minimum known diversity specimens and unvouchered observationsÐ of nonvolant mammals at Paracou (64 spe- that resulted from our own efforts (excluding cies, based on additional records from inter- interview results; table 48), we documented views and other sources) is so far above the the occurrence of 50 nonvolant species over levelling terminal portion of our graph as to the 202 dates that we worked in the ®eld suggest that the latter is misleading about from 1991 to 1994. After an initially rapid completeness. Analyzing the results of each rate of about 1.6 species/day in the ®rst two inventory method separately therefore pro- weeks of 1991 (®g. 81), species accumula- vides a better picture of sampling effective- tion abruptly levelled to around 0.2 species/ ness. 166 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 48 Summary of Nonvolant Mammal Records at Paracou by Species and Methoda 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 167

TABLE 48Ð(Continued)

SAMPLING RESULTS FROM DIFFERENT 8 nights, the longest interval was 14 nights, METHODS and the average interval was 12 nights. In None of the methods that we used to in- total, we trapped on 71 dates in 1991 and ventory the nonvolant mammal fauna at Par- 1992 with an average of 84 Victor/Sherman acou recorded all of the species known to traps deployed per night. Supplementary occur in our study area. Instead, each method ground-level trapping with other equipment appeared to be maximally effective for some (Tomahawks, Conibears, and legholds) was taxa and relatively ineffective for others. Be- much more intensive in 1991 than in any low we summarize our principal sampling re- subsequent year, but the numbers of supple- sults method by method and explain some of mentary traps deployed each night were not the factors that affect our subsequent assess- consistently recorded. ments of complementarity and completeness. Combining results from conventional trap- CONVENTIONAL TRAPPING: Most conven- ping with all commercially available equip- tional trapping at or near ground level (0±3 ment, we captured a total of 162 individuals m by our convention) was accomplished in representing 18 species of marsupials and ro- 1991 and 1992, when ten Victor/Sherman dents, including our only examples of Mon- traplines were established at widely separat- odelphis brevicaudata and Mesomys sp. (ta- ed sites within our 3-km sampling radius. ble 48). Species accumulated quickly at ®rst: Whereas some Victor/Sherman traplines con- we took 15 species in our ®rst 74 captures, tained over 100 traps and extended for 2 km for an initial average rate of almost one new or more, most were shorter and contained species per ®ve individuals trapped (®g. 82). fewer traps. The shortest interval that any However, only three additional species were Victor/Sherman trapline was operational was represented among the last 88 captures by 168 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 49 Methods and Effort Used to Sample Nonvolant Mammal Diversity at Paracou, 1991±1994

this method, or about one new species per 30 ibears were also occasionally taken in Vic- individuals trapped. tors. However, Victors and Shermans never In general, Victors and Shermans took captured squirrels or any adult individuals of smaller species than did Tomahawks, Coni- Didelphis, Metachirus, Philander, or Myo- bears, and legholds (®g. 83), but there was procta. Similarly, Tomahawks, Conibears, some taxonomic overlap (table 50). For ex- and legholds never took Marmosops, Mono- ample, both classes of traps commonly took delphis, Neacomys, Oryzomys, or Mesomys adult Proechimys, and juveniles of some at Paracou. Therefore, both kinds of equip- large marsupials (e.g., Didelphis marsupialis, ment provided useful information about local Metachirus nudicaudatus, Philander opos- diversity. sum) usually taken in Tomahawks and Con- The unimpressive numbers of individuals

Fig. 82. Results of conventional trapping for nonvolant mammals at Paracou, using all commercially available equipment (Victor rat traps, Sherman live traps, Tomahawks, Conibears, and legholds) at or near ground level (0±3 m above the ground). A total of 162 captures representing 18 species of marsupials and rodents were recorded from 1991 to 1994. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 169

Fig. 83. Frequency histograms of body weights of nonvolant mammals (marsupials and rodents) taken at or near ground level by different trapping methods. Sample sizes (N) are recorded weights, which were not obtainable from all trapped individuals. ``Other traps'' includes captures made using Tomahawks, Conibears, and legholds (combined with Victor/Sherman captures in the ®rst column of table 48). captured by conventional trapping at Paracou In effect, trap success (®g. 84, indicated by is explained by a dramatic decline in trap the slope of the solid line in the upper graph) success that we experienced over the course changed abruptly when, after about 2600 of our inventory. This phenomenon is best trap-nights in 1991, traplines previously de- illustrated by the results obtained with Victor ployed for general collecting in productive and Sherman traps, for which accurate counts habitats (e.g., well-drained primary forest) of the numbers set each night are available. were shifted to less productive sites (e.g., 170 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 50 ping combined (®g. 82). This paradox refects Frequency Data for Species Taken at Ground the late appearance in Victor traps of juvenile Level in Different Conventional Trap Types individuals of two species (Didelphis mar- at Paracoua supialis, Myoprocta acouchy) previously taken only in Tomahawks, Conibears, or legholds, together with new taxa not taken by any other method (Monodelphis and Meso- mys). Clearly, the apparent completeness of Victor/Sherman species sampling as assessed by trap-nights (®g. 84, dashed line in upper graph) is just an artifact of declining trap success. ARBOREAL TRAPPING: We used arboreal platform traps only in 1993, when 25 trapping stations were established at approximately 20-m intervals along a transect through both well-drained and swampy primary forest. Trapping platforms were installed between 7.2 m and 19 m above the ground, with an average platform height of 13.7 m. From the date when the ®rst arboreal trap station was operational (29 July) to the date when all platform trapping was discon- tinued (15 September), we trapped continuously for 47 nights with an average of 21 trapping stations operational per night. In total, we logged 1002 station-nights, which is equivalent to 2004 trap-nights if both Sher- mans and Tomahawks are counted. Seven- teen individual marsupials and rodents were captured, for an overall success rate (captures per trap-night ϫ 100) of only 0.8%. Our 17 arboreal trap captures represent six species, none of which were taken exclusive- along streams) and baited for certain target ly by this method (table 48). The temporal species expected to occur there. Furthermore, pattern of captures (®g. 85, solid line in up- trap success in comparable situations was per graph) suggests that traps were avoided never as high in subsequent ®eld seasons as for the ®rst two weeks after installation (our it was in 1991. For example, we made 3.6 ®rst arboreal capture was made after 454 captures per 100 trap-nights using Victor and trap-nights) and that more prolonged trap- Sherman traps for general collecting in pri- ping could have resulted in many more cap- mary forest in 1991, but only 0.6 captures tures (trap success was higher in the last per 100 trap-nights using the same equip- week of operation than in any previous in- ment in the same habitat in 1992 (an 83% terval). Although species accumulation plot- decline). In total, we recorded only 128 cap- ted against captures (®g. 85, bottom graph) tures in 5960 Victor/Sherman trap-nights at shows some indication of levelling, the num- Paracou, for an overall success rate (captures ber of individuals taken is too small to plau- per trap-night ϫ 100) of 2.1%. sibly suggest any asymptotic value for tax- Species accumulation plotted against cap- onomic diversity obtainable by this method. tures for Victor/Sherman trapping (®g. 84, PITFALL TRAPPING: This method was used lower graph) shows no sign of an asymptote only in 1993, when ®ve pitfall traplines were despite the ¯attening terminal portion of the installed at widely separated locations in our analogous graph for all conventional trap- study area: two in well-drained primary for- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 171

Fig. 84. Results of ground-level (conventional) trapping with Victor rat traps and Sherman live traps at Paracou, where 124 identi®able captures of 16 species of marsupials and rodents were obtained with 5960 trap-nights of effort using this equipment in 1991 and 1992. Note the abrupt change of slope for cumulative captures after about 2600 trap-nights in the upper graph. est, one in swampy primary forest, and two taken up (13 September), we trapped for 54 in creekside primary forest. From the date consecutive nights. Except for the ®rst week when the ®rst pitfall trapline was operational of trapping (when the lines were still being (21 July) to the date when all pitfalls were installed), 55 buckets per night (11 in each 172 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 line) were deployed as pitfalls throughout all personnel alert to ¯eeting encounters with this interval. In total, we logged 2783 buck- wary taxa locally persecuted as game (e.g., et-nights and captured 45 individual mam- deer and monkeys). Given these and other mals, for an overall success rate (captures per problems with quantifying observational ef- bucket-night ϫ 100) of 1.6%. fort, ®eld days are probably no worse than Our 45 mammalian pitfall captures24 rep- any alternative unit. resent 12 species of marsupials and rodents, Diurnal mammal sightings accumulated at of which two (Neacomys dubosti and Neus- an initially rapid rate during our ®rst several ticomys oyapocki) were not taken by any oth- weeks at Paracou, but few new species were er method (table 48). Most individual mam- subsequently recorded by this method (®g. mals captured in pitfalls were small, the great 87. Over the entire course of our inventory, majority (84%) weighing less than 30 g (®g. we recorded 16 species from diurnal sight- 83). Pitfall captures accumulated rapidly in ings (table 48). Although many of these were the ®rst few weeks of trapping (®g. 86, solid not found by us using other inventory meth- line in upper graph), when we recorded about ods, most were previously known to the local three captures per 100 bucket-nights on av- residents whom we interviewed. Oddly erage; thereafter, our capture rate declined to enough, the only species uniquely recorded a more-or-less steady rate of about 0.8 cap- by a diurnal observation was the usually noc- tures per 100 bucket-nights. Although spe- turnal rodent Makalata didelphoides. cies accumulation plotted against bucket- NOCTURNAL HUNTING: Most of our obser- nights (®g. 86, dashed line in upper graph) vations and collections of nonvolant noctur- suggests that faunal sampling with pitfalls nal mammals resulted from deliberate hunt- was nearly complete, species accumulation ing (including sight censuses when few or no plotted against captures (®g. 86, lower graph) specimens were taken). Unfortunately, the does not show a convincing asymptote. duration of nightly hunts was not quanti®ed DIURNAL HUNTING: Because most of our in 1991 when this method was used most sightings and collections of diurnal mammals intensively. However, an estimate can be resulted from chance encounters, sampling based on the fact that RSV and DPL both effort is hard to quantify for this method. On hunted every night on 58 consecutive dates average, all inventory personnel spent at in that year; assuming a typical hunt duration least several daylight hours in the forest per of 3.0 hours (the known average for hunts by calendar date, so the total accumulation of the same personnel in 1992) yields a proba- time available for diurnal mammal observa- ble total effort of about 350 hours devoted to tions was considerable: about 2500 hours this method in 1991. Hunt durations were (634 person-days ϫ 4 hours/person-day) is a consistently recorded in subsequent ®eld sea- plausible estimate. However, much of the sons: we devoted 96.7 hours to this method time that we spent in the forest by day was in 1992, 79.2 hours in 1993, and 64.4 hours not conducive to observing cryptic species in 1994. In total, we systematically hunted at (e.g., sloths and pygmy squirrels), nor were night on 123 dates from 1991 to 1994 at Par- acou, for a total cumulative effort of about 24 Among the many amphibians and reptiles captured 590 hours. in pitfall traps at Paracou were examples of Leptodac- tylus sp., Bufo cf. typhonius, Otophryne pyburni, Cae- Nevertheless, chance observations made cilia tentaculata, Rhinatrema bivittatum, Coelodactylus while bat netting contributed so many im- amazonicus, Gonatodes annularis, Anolis chrysolepis, portant records of nonvolant mammals (in- Plica plica, Plica umbra, Mabuya sp., Ameiva ameiva, cluding our ®rst of Caluromys philander, Kentropyx calcarata, Arthrosaura kockii, Bachia ¯aves- Hyladelphys kalinowskii, Oligoryzomys ful- cens, Iphisa elegans, Leposoma guianense, Neusticurus bicarinatus, Neusticurus rudis, Tretioscincus agilis, Am- vescens, and Priodontes maximus) that it is phisbaena fuliginosa, Typhlops reticulatus, Atractus sp., misleading to ignore this source of nocturnal Erythrolamprus aesculapii, Helicops angulatus, Liophis observations. Indeed, whereas our deliberate sp., Oxyrhopus sp., Pseustes poecilonotus, Umbrivaga hunting effort per ®eld season declined from sp., and Leptomicrurus collaris. Voucher material of these taxa (as provisionally identi®ed by J. A. Campbell, 1991 to 1994, our bat netting effort increased C. J. Cole, and D. R. Frost) are preserved in the AMNH in almost direct proportion (all inventory Department of Herpetology. personnel were usually engaged in one activ- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 173

Fig. 85. Results of arboreal platform trapping at Paracou, where 17 captures representing six species of marsupials and rodents were obtained with 2004 trap-nights of effort in 1993. ity or the other after dusk, seldom returning sense to plot species accumulation by noc- to camp before midnight), so the total time turnal observation as a function of calendar actually or potentially available to observe dates rather than dedicated search time (®g. nonvolant nocturnal mammals per ®eld date 87, solid line). probably remained more-or-less constant Over the 202 dates that we worked at Par- throughout our inventory. It therefore makes acou from 1991 to 1994, we recorded 31 spe- 174 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 86. Results of pitfall trapping at Paracou, where 45 captures representing 12 species of marsupials and rodents were recorded with 2783 bucket-nights of effort in 1993. cies of nonvolant mammals by nocturnal ob- mulation by nocturnal observation was steep- servation, ®ve of which were not recorded est in the ®rst several weeks and decreased by any other method (Gracilinanus emiliae, subsequently, no convincing asymptotic val- Dasypus kappleri, Leopardus wiedii, Oligo- ue is indicated. In total, we recorded 271 ryzomys fulvescens, Coendou melanurus; ta- nocturnal observations that were unambigu- ble 48). Although our rate of species accu- ously identi®able to species, but this sum 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 175

Fig. 87. Results of diurnal and nocturnal hunting/census at Paracou, where 16 species of nonvolant mammals were recorded by the former method and 31 species by the latter. In lieu of more appropriate measures of effort (see text), we simply plot species accumulation against ®eld dates (202 total) for both methods. does not include numerous unvouchered ob- Potos ¯avus, Cuniculus paca, Philander servations (mostly of common taxa) from opossum, and Chironectes minimus account 1991, nor did we record chance encounters for over half of the data in table 51. By con- with common species while bat netting. trast, ten species represented by only one re- Therefore, only the smaller frequency classes cord each include some that may be genu- for this method (table 48) represent accurate inely uncommon, but others that are simply counts. inconspicuous (silent, small, and/or with More interpretable frequency data were weak eyeshine), dif®cult to identify without obtained from 1992 to 1994, when every specimens in hand, or usually diurnal (e.g., identi®able observation (including sightings Gracilinanus emiliae, Marmosa murina, Na- and distinctive sounds, except howler mon- sua nasua, Neacomys paracou, Oecomys ru- key vocalizations) were recorded in the tilus, Oryzomys megacephalus, Proechimys course of 240.3 hours of systematic nocturnal guyannensis, Myoprocta acouchy). hunting (table 51). Unidenti®ed sightings and INTERVIEWS: Interviews resulted in 33 pos- sounds (e.g., distant eyeshine, crashing nois- itive identi®cations of local species, of which es in the undergrowth) were not consistently 12 were not recorded by any other method recorded. On average, we made 1.1 identi®- (Bradypus tridactylus, Cabassous unicinctus, able observations per hour of nighttime hunt- Cyclopes didactylus, Myrmecophaga tridac- ing from 1992 to 1994, a rate that did not tyla, Cebus apella, Saimiri sciureus, Speo- differ substantially among the four persons thos venaticus, Herpailurus yagouarondi, principally involved in this activity (RSV, Leopardus pardalis, Puma concolor, Galictis DPL, R. W. Kays, and L. H. Emmons). vittata, Tapirus terrestris; table 48). Most of Most identi®able nocturnal observations at the species uniquely recorded from inter- Paracou represent a few common species views are either uncommon throughout the with bright eyeshine, distinctive vocaliza- Neotropics (e.g., Cabassous unicinctus, Gal- tions, and/or noisy habits. Thus, records of ictis vittata) or are locally uncommon from 176 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 51 Leopardus pardalis). Only two relatively Nocturnal Hunting/Census Results, 1992±1994 large and easily identi®ed local species (Chi- ronectes minimus, Coendou melanurus) were not familiar to our interviewees. Although interviewees often estimated the relative frequency with which mammals were observed (e.g., see accounts for Bra- dypus tridactylus and Mazama gouazoubira, above), accurate counts of second-hand observations were only available for the rarest species. Interview records known to have been based on single observations include those for Cabassous unicinctus, Priodontes maximus, Pithecia pithecia, Saimiri sciureus, and Galictis vittata. Only two second-hand observations are known for Cyclopes didactylus and Nasua nasua; three for Speothos venaticus and Leopardus pardalis; and four for Myrmecophaga tridactyla, Panthera onca, and Puma concolor. SUMMARY: Although each inventory method that we used to sample nonvolant mammal diversity at Paracou produced a different list of species, some methods were clearly more productive and/or produced more distinctive lists than others. Relevant quantitative comparisons (table 52) include the total number of species recorded by each method, the number of unique species, and pairwise complementarity values. The latter re¯ect the extent to which two methods provide non- redundant diversity information: low complementarity values imply high redundancy, whereas high complementarity implies low redundancy (see table footnote for compu- tational details). Clearly, interviews produced more species records (33) than any other single method used in our nonvolant inventory, as well as the largest number (12) and proportion (36%) of unique species. Nocturnal hunting ranks next by these criteria, followed by conventional trapping. Pitfall trapping and diurnal hunting each recorded some unique species, but arboreal trapping did not and was also least productive in terms of total records. The lowest complementarity value calculated from our sampling data (60%) corre- overhunting (e.g., Cebus apella, Tapirus ter- sponds to the comparison of diurnal hunting restris); others, however, are perhaps com- with interviews, methods that were substan- mon but seldom seen because they are cryp- tially redundant for the obvious reason that tic (e.g., Bradypus tridactylus) or wary (e.g., almost all diurnally active mammals at Par- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 177

TABLE 52 Quantitative Comparisons of Six Nonvolant Mammal Inventory Methods Used at Paracou

acou were known to local inhabitants. Simi- ESTIMATING COMPLETENESS larly, the second-lowest complementarity value we computed (64%) corresponds to the Because none of our species accumulation comparison of pitfall trapping with conven- graphs shows a convincing asymptote, it is tional trapping, methods that produced reasonable to expect that more species could broadly overlapping lists of small marsupials have been recorded with additional sampling and rodents. By contrast, other pairs of meth- effort by each method. Visual comparisons ods (such as pitfall trapping and interviews of these graphs, however, suggest that asymptotic values for species accumulation on the one hand, or arboreal trapping and may have been more closely approached by diurnal hunting on the other) are 100% com- some methods (e.g., conventional trapping, plementary because they produced no re- ®g. 82) than by others (e.g., pitfall trapping, corded species in common. ®g. 86). In order to assess sampling com- Overall, 22 species were each recorded by pleteness by less subjective criteria, we com- only one of the six methods we used to sam- pared observed species counts with predicted ple the nonvolant fauna at Paracou. If the two values obtained by nonparametric extrapola- species uniquely taken in conventional traps tions. by previous researchers (Nectomys melanius, The logic of extrapolating unobserved spe- Oryzomys macconnelli) are added to this to- cies richness from incomplete samples was tal, then about 38% of the known nonvolant recently reviewed by Colwell and Codding- fauna (64 species) were recorded by a single ton (1994), and the application of certain method each. If miscellaneous records (e.g., nonparametric methods to our Paracou bat scavenged material, occasional observations data was explained by Simmons and Voss of distinctive spoor) are discounted for an ad- (1998: 182±184). As in that study, we ®rst ditional four species (Marmosa murina, consider the results of applying Chao's Panthera onca, Tayassu pecari, Coendou (1984) estimator, which is based on the total prehensilis; table 48), then a total of 28 spe- number of observed species, Sobs, the number cies (44% of the known fauna) was not re- of singletons (species recorded only once), a, dundantly recorded. Of the 36 redundantly and the number of doubletons (species re- recorded species, 20 were recorded by two corded only twice), b. The expected total methods each, 14 by three methods, and 2 by number of species, S*, is then given by the four methods. expression 178 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 53 Analysis of Sampling Completeness Using Chao's (1984) Species Richness Estimator for Methodologically De®ned Partitions of the Paracou Nonvolant Mammal Inventory (See text for explanation of symbols and computation.)

*2 S ϭ Sobs ϩ (a /2b). (arboreal trapping, diurnal hunting, nocturnal hunting, interviews) are not consistently We constructed approximate 95% con®dence ranked inter se by completeness. Perhaps the intervals (Ϯ2 SD) using a formula for the best synthesis of these results is obtained by variance given by Colwell and Coddington taking average completeness values across (1994) on the assumption that S* is normally distributed (after Chao, 1987), and we esti- all four sets of extrapolation ®gures, which mated completeness as the percentage (S / yields the sequence: conventional trapping Ͼ obs arboreal trapping and interviews nocturnal S*) ϫ 100. Ͼ The results of such calculations (table 53) hunting Ͼ diurnal hunting Ͼ pitfall trapping. suggest that sampling by some inventory Pooling sampling results from all nonvo- methods was indeed more complete than lant inventory methods, eight species are sampling by others, although con®dence in- each represented in our data by single re- tervals are wide enough to include the ob- cords (Gracilinanus emiliae, Cabassous un- served number of species in every case. Of icinctus, Saimiri sciureus, Galictis vittata, particular interest is the apparent near-com- Neacomys dubosti, Coendou melanurus, pleteness of conventional trapping on the one Makalata didelphoides, Mesomys sp.) and hand, and the relative incompleteness of pit- four species are represented by just two re- fall trapping on the other, estimates that cords each (Priodontes maximus, Cyclopes more-or-less coincide with our subjective in- didactylus, Nectomys melanius, Neusticomys terpretation of species accumulation graphs. oyapocki). Using Chao's (1984) extrapola- On the other hand, the rather high complete- tion procedure (table 53), the predicted total ness estimate for arboreal trapping is dif®cult nonvolant species richness at Paracou is 72 to reconcile with our scant capture success species, an estimate that suggests our non- using this method. volant inventory is 89% complete. Other Given the inherent uncertainty of all ex- nonparametric estimators yield S* values in trapolation procedures, other estimates of S* the range of 69±74 species and correspond- are of interest in order to bracket the range ing completeness estimates of 86±93% (table of plausible inferences that can be based on 54). our sampling data. We therefore used three Obviously, neither the presence of addi- additional nonparametric species-richness es- tional nonvolant species in the Paracou fauna timators to assess inventory completeness nor their absence can be proven without ad- (table 54). By each estimator, our conven- ditional ®eldwork. However, it is noteworthy tional-trapping data appear to be the most that the numbers of missing species implied complete (although not always by a large by these statistical extrapolations (5±10, cal- margin), and our pitfall-trapping data the culated as S* minus Sobs) correspond closely least complete. The remaining data partitions with the number of additional species that 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 179

TABLE 54 Analyses of Sampling Completeness Using Alternative Species Richness Estimatorsa

could be expected to occur locally on the ba- parable intensity throughout our study. Fi- sis of known geographic and ecological dis- nally, the same labor constraints dictated that tributions (appendix 1). Presumably, that most of our nonvolant mammal sampling short list (Didelphis albiventris, Marmosa was restricted to primary forest with only lepida, Cebus olivaceus, Leopardus tigrinus, minimal coverage of secondary vegetation. Oecomys bicolor, Oecomys rex, Rhipidomys As a consequence of these limitations, we leucodactylus, Echimys chrysurus, Isothrix cannot provide the same detail of analysis for sinnamariensis) includes the likeliest candi- nonvolant mammal sampling that we were dates for discovery by future inventory ef- previously able to accomplish with our bat forts at Paracou. data. For example, available nonvolant mammal records from Paracou are simply too DISCUSSION OF THE NONVOLANT sparse, too spatiotemporally biased, and/or FAUNA too methodologically heterogeneous to val- Although a few nonvolant species were al- idly compare the faunas of different local ways common and easily observed at Para- habitats, or to compare rates of species ac- cou, most were uncommon or cryptic. Our cumulation in different ®eld seasons. Addi- sampling data are therefore much sparser for tionally, the obvious effects of hunting on lo- nonvolant mammals (583 records total, cal game populations (see Primates, above), summed over all methods excluding inter- together with the dramatic decline in abun- views) than for bats (3126 records; Simmons dance of small rodents and marsupials after and Voss, 1998). Furthermore, because most 1991 (see results for Conventional Trapping, protocols for sampling nonvolant mammal above), suggest that we were not effectively diversity were very labor-intensive, available sampling the same nonvolant fauna through- effort had to be allocated among just a few out the course of our project. The data at methods in each ®eld season; as a result, hand are therefore less than ideal for testing none were applied continuously or with com- ecological or methodological hypotheses. 180 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 88. Twelve Neotropical rainforest localities from which nonvolant mammal diversity data were compiled for this report. See footnotes to table 55 for geographic coordinates and references.

Nevertheless, nonparametric species-rich- 55, and a complete species-by-locality matrix ness estimators applied to the pooled results is provided in appendix 2. of all methods (including interviews) suggest that our nonvolant inventory is about 90% TAXONOMIC COMPOSITION AND complete, and it therefore seems unlikely that BIOGEOGRAPHY faunistic inferencesÐabout taxonomic composition, zoogeographic relationships, spe- In terms of higher-level taxonomic com- cies richness, etc.Ðwill be substantially al- position, the nonvolant mammals of Paracou tered by future ®eldwork at Paracou. In order represent a typical Neotropical lowland rain- to provide a comparative context for dis- forest fauna. Seven of the eight orders com- cussing such topics, we surveyed the litera- monly found in Central and South American ture and identi®ed 11 other Neotropical rain- rainforests (Marsupialia, Xenarthra, Pri- forest sites, 2 in Central America and 9 in mates, Carnivora, Perissodactyla, Artiodac- Amazonia, from which large species lists of tyla, Rodentia) are represented in our inven- nonvolant mammals have been published. tory; only Lagomorpha, which is never rep- These localities are mapped in ®gure 88, the resented by more than a single species at any taxonomic distribution of nonvolant species rainforest site, is absent. All of the families richness at each site is summarized in table and genera of nonvolant mammals in the Par- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 181

TABLE 55 Taxonomic Distribution of Nonvolant Mammalian Species Richness in 12 Neotropical Rainforest Faunas (Table entries are numbers of species present.) 182 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 89. Analysis of faunal similarity among nonvolant mammal inventories from 12 Neotropical rainforest localities. Faunal similarity was quanti®ed between each pair of localities (i, j) by Jaccard's coef®cient, Jij ϭ Cij/Tij, where Cij is the number of species common to both faunas and Tij is the total number of species in both faunas combined (Tij ϭ Ni ϩ Nj Ϫ Cij). Localities were clustered by the unweighted pair-group method using arithmetic averages (Sneath and Sokal, 1973) for comparability with other recent biogeographic analyses of Neotropical vertebrates (e.g., da Silva and Sites, 1995). The bottom scale shows the clustering level in units of percent faunal similarity (J ϫ 100). No historical or other causal interpretation is implied by these results, nor do we assume that a hierarchical model is necessarily appropriate except as a convenient summary graphic. acou fauna are likewise widespread rainfo- French Guianan locality only 136 km to the rest taxa. SSE, and least similar to that from La Selva, In terms of species composition, Paracou a Costa Rican locality almost 3400 km to the clusters (®g. 89) with ®ve other faunas from WNW. Overall, these results clearly indicate the Guiana subregion of Amazonia, next with a pattern of increasing faunal resemblance a single southeastern Amazonian site, then with increasing geographic proximity, and with a discrete grouping of three western that the Paracou fauna in particular is most Amazonian localities, and lastly with a pair similar to others from northeastern Amazon- of Central American faunas. Pairwise mea- ia. sures of faunal resemblance (table 56) indi- Paracou nonvolant mammals can be sorted cate that the Paracou nonvolant inventory is (table 57) into seven groups based on their most similar to that from Arataye, another pattern of distribution among the four Neo- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 183

TABLE 56 Percent Similarity Among Nonvolant Mammal Faunas from 12 Neotropical Rainforest Inventory Sitesa (Table entries are Jaccard's coef®cientb ϫ 100.)

tropical lowland rainforest regions recog- videns (A), Marmosops pinheiroi (A), Mon- nized by Voss and Emmons (1996). The larg- odelphis brevicaudata (G), Dasypus kappleri est distributional group (pattern 1) includes (A), Saguinus midas (G), Ateles paniscus 24 species that are endemic to Amazonia (G), Pithecia pithecia (G), Saimiri sciureus (A), or are even more narrowly restricted to (A), Sciurillus pusillus (A), Sciurus aestuans the Guianan (G) subregion of Amazonia: Hy- (G), Neacomys dubosti (G), Neacomys par- ladelphys kalinowskii (A), Marmosops paracou (G), Neusticomys oyapocki (G), Oeco- mys auyantepui (G), Oecomys rutilus (G), TABLE 57 Oryzomys macconnelli (A), Oryzomys yun- Rainforest Distribution Patterns of Nonvolant ganus (A), Rhipidomys nitela (A), Coendou Paracou Mammals melanurus (G), Myoprocta acouchy (A), Me- somys sp. (G), Proechimys cuvieri (A), and Proechimys guyannensis (A). Therefore, a substantial fraction (38%) of the Paracou fauna is distinctively Amazonian or Guianan in character. The next-largest distributional group (pattern 2) includes 18 species that (as currently recognized by taxonomists) occur in all four Neotropical rainforest regions. These essentially pan-Neotropical taxa include Chironec- tes minimus, Metachirus nudicaudatus, Das- ypus novemcinctus, Myrmecophaga tridactyla, Speothos venaticus, Herpailurus yagouarondi, Leopardus pardalis, Leopardus wiedii, Panthera onca, Puma concolor, Eira 184 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 barbara, Galictis vittata, Potos ¯avus, Ma- ed to rainforest habitats: many are eurytopic zama americana, Mazama gouazoubira, Pe- and occur in a wide range of other Neotrop- cari tajacu, Tayassu pecari, and Cuniculus ical biomes. For example, at least 16 of the paca. Pending future revisionary study (see 18 species with distributional pattern 2 are below), this group of species, comprising also known from the Llanos (Eisenberg et al., about 28% of the fauna, is zoogeographically 1979; IbaÂnÄez, 1981), the Caatinga (Willig uninformative. and Mares, 1989), the Cerrado (Redford, By contrast, other Paracou nonvolant 1983; Fonseca and Redford, 1984; Mares et mammals are neither Amazonian endemics al., 1989), or the Pantanal (Schaller, 1983). (as currently recognized) nor pan-Neotropi- On the other hand, none of the taxa that we cal, but have distributions that suggest a va- recorded at Paracou are consistently associ- riety of inter-regional connections. Thus, ated elsewhere with nonforest habitats. A nine species occur in both Amazonian and few Paracou species (Marmosa murina, Sai- Coastal Venezuelan rainforests, but not in the miri sciureus, Oligoryzomys fulvescens) are other two regions (pattern 3: Gracilinanus perhaps never found in unbroken tracts of emiliae, Bradypus tridactylus, Choloepus di- Guianan rainforest, but these can be charac- dactylus, Cabassous unicinctus, Priodontes terized as inhabitants of the rainforest edge maximus, Alouatta seniculus, Nectomys me- that probably occur along riverbanks, blow- lanius, Oryzomys megacephalus, Makalata downs, and other natural openings at undis- didelphoides), seven species occur in all rain- turbed sites. Several unambiguously nonfo- forest regions except the trans-Andean (pat- rest mammals might inhabit the coastal sa- tern 4: Caluromys philander, Marmosa mur- vannas and salt marshes just north of our ina, Micoureus demerarae, Tamandua tetra- study area (e.g., Odocoileus cariacou, Holo- dactyla, Tapirus terrestris, Coendou prehen- chilus sciureus, Zygodontomys brevicauda), silis, Dasyprocta leporina), and four species but none were encountered within our sam- occur in all rainforests except the Atlantic pling radius. Therefore, although some non- region of southeastern Brazil (pattern 5: Di- volant Paracou mammals are not rainforest delphis marsupialis, Philander opossum, Cy- specialists, the fauna as a whole includes clopes didactylus, Oligoryzomys fulvescens). only species that are known to occur in rain- One species (Nasua nasua) occurs in all rainforested landscapes and appears to constitute forest regions except the Coastal Venezuelan an ecologically homogeneous assemblage. (pattern 6), and another (Cebus apella) occurs only in Amazonia and the Atlantic rain- SPECIES RICHNESS forest region of southeastern Brazil (pattern 7).25 Comparisons of nonvolant mammal spe- Of course, not all of the nonvolant species cies richness among Neotropical rainforest that belong to the Paracou fauna are restrict- inventories are complicated by many factors. As documented in the preceding analyses of 25 Because so few genera of nonvolant mammals in sampling results, different inventory methods the Paracou fauna have been revised, it is prudent to effectively sample different sets of species, consider how the memberships of these distributional and increased effort with any method gen- groups are likely to change with future taxonomic re- erally produces more species. Therefore, search. In fact, morphological and/or molecular data are already available to suggest that many widely distributed even in the absence of real intersite diversity nonvolant ``species'' are composite (i.e., consist of two differences, long-term inventories (or those or more diagnosable geographic forms that should be using more methods) would be expected to recognized as valid taxa; see Remarks above for Calu- obtain larger species lists than short-term in- romys philander, Marmosa murina, Metachirus nudicaudatus, Micoureus demerarae, Philander opossum, ventories (or those using fewer methods). and Mazama gouazoubira). Although it is also possible Differences in the ecological scope of inven- that a few species currently thought to be Amazonian tory ®eldwork can also affect species rich- endemics will eventually be found to occur in other rain- ness comparisons because sites with greater forest regions, the likely net outcome of future research will be to decrease signi®cantly the number of valid spe- habitat diversity will tend to be richer than cies with widespread distributions and to increase the sites with fewer habitats. Finally, large mam- number of species with more restricted ranges. mals are likely to be more abundant, and 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 185

TABLE 58 Methodological Comparisons among Nonvolant Mammal Inventories from Three Neotropical Rainforest Localities

therefore more easily recorded, at undis- have more species). To better exemplify re- turbed localities than at sites which have gional and subregional differences in species been partially defaunated by hunting. richness, we focus our comparisons on two Published descriptions of inventory ®eld- sites where prolonged ®eldwork may have work (cited in the footnotes to table 55) doc- compensated in some degree for methodo- ument intersite differences in all of these po- logical incompleteness (table 58). tentially confounding factors: (1) Although La Selva (with over 30 years of faunal ob- ®eldworkers at all sites used conventional servations) probably provides the best-sam- trapping, diurnal hunting, and nocturnal pled nonvolant mammal fauna from any hunting to sample nonvolant mammal diver- Central American lowland rainforest, and sity, other methods (pitfall trapping, arboreal Manu (with over 20 years) provides an ap- trapping, and interviews) were less consis- propriate western Amazonian counterpart. tently applied, and no previous inventory has Although it is not possible to estimate sam- used all six methods in combination. (2) pling completeness at either of these sites by Overall sampling effort is hard to quantify extrapolation (requisite frequency data are for nonvolant mammal surveys, but the du- not available), surrogate estimates can be ration of inventory work at each site (a based on the number of rainforest species ex- monotonic correlate of effort) ranged from pected at each locality from geographic less than a year at some sites to many years range data. Such calculations (see Voss and at others. (3) Whereas most sites were locat- Emmons, 1996: table 10) suggest that the La ed on rivers or lakes that provide habitat for Selva nonvolant inventory is about 95% semiaquatic, riparian, and forest-edge spe- complete and the Manu nonvolant inventory cies, others were in well-drained uplands re- about 85% complete. By any of the extrap- mote from large bodies of water. (4) Subsis- olations computed earlier in this report, the tence hunting probably eliminated some pri- Paracou nonvolant fauna falls between La mates and other large species prior to inven- Selva and Manu in estimated completeness tory ®eldwork at several sites, whereas other (86±93%). Therefore, known species rich- sites were pristine. ness and estimated completeness are inverse- As a consequence of such disparities, tab- ly correlated among these three inventories. ulated species counts are undoubtedly affect- Graphical comparisons (®g. 90) summa- ed by sampling artifacts to a greater or lesser rize the principal higher-taxonomic patterns extent. The most that can be said from these of species richness among La Selva, Paracou, data, without introducing additional assump- and Manu. Intersite differences in known tions or corrections, is that nonvolant species species richness are most pronounced for richness at Paracou more nearly resembles Marsupialia, Primates, and Glires (Rodentia that from other well-sampled sites in the Gui- ϩ Lagomorpha). For each of these clades, La ana subregion of Amazonia (e.g., Kartabo, Selva is the least diverse site; Paracou and Arataye) than they do Central American sites Manu are equivalent in known marsupial di- (which have fewer species for most orders) versity, but Manu is more diverse than Par- or western Amazonian sites (which tend to acou for both primates and rodents. Paracou 186 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 90. Species richness comparisons for higher-level clades at three Neotropical rainforest sites. For each clade, bar height indicates the number of species at La Selva (L), Paracou (P), and Manu (M); percentages above each bar indicate the relative contribution of those species to the entire nonvolant fauna at each site. Artiodactyla and Perissodactyla are combined as Ungulata, Rodentia and Lagomorpha as Glires. has the most known xenarthran species of Although some of these faunal differences any site, but the least number of known car- are probably artifactual,26 the short lists of nivore species; ungulate species richness is expected species for each locality (Voss and invariant. Overall, known nonvolant species Emmons, 1996: appendices 2, 10; this report: richness increases by 12 from La Selva to appendix 1) suggest that future inventory Paracou, and by 15 from Paracou to Manu; work is unlikely to change the overall picture from La Selva to Manu, the net increase in very much. For example, primates are likely species richness (27 species) amounts to 52% to remain in last place at La Selva, as are of the former fauna. Given the previously ungulates at Paracou and Manu. A few noted inverse correlation between known clades may switch ranks at some sites (e.g., species richness and estimated inventory marsupials may prove to be more diverse completeness, the true diversity gradient than primates at Manu), but the larger diver- from La Selva to Paracou to Manu is likely sity differences (among clades at each local- to be even steeper than these quantities imity, and among localities for the same clade) ply. are probably robust. Whereas Glires is the most diverse clade Clearly, exemplar comparisons can be at each locality (accounting for an almost over-interpreted, and it would be unwise to constant proportion, 33±35%, of the local make sweeping generalizations about geo- nonvolant species), the remaining clades are graphic diversity patterns based on just these not consistently ranked by species richness. three sites. Nevertheless, the nonvolant mam- At La Selva, Carnivora is the most speciose mal fauna at Paracou is probably represen- after Glires, distantly followed by Xenarthra; Marsupialia and Ungulata are next, and Pri- 26 Three uncontrolled factors seem most likely to have mates is last. At Paracou, however, Marsu- affected our faunal comparisons, First, bodies of water suitable for otters and other large semiaquatic or riparian pialia is more speciose than Carnivora, which mammals are present at La Selva and Manu, but not at is closely followed by Xenarthra; Primates is Paracou. Second, the long duration of inventory work at next, followed by Ungulata. At Manu, Car- La Selva and Manu has probably been insuf®cient to nivora is again the most speciose clade after fully compensate for the lack of methods suitable for detecting all of the small and elusive nonvolant species Glires, but Primates is almost as diverse, fol- at those sites. Third, interviews at Paracou may not have lowed closely by Marsupialia; Xenarthra is compensated fully for the partial defaunation of our next and Ungulata is last. study area prior to this inventory. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 187 tative of that throughout most of the Guiana semiaquatic (Chironectes, Lontra), others subregion of Amazonia, which would appear forage at or near the water's edge (Procyon, to be richer in species (especially of marsu- Nectomys, Hydrochoerus) and are perhaps pials and rodents) than Central American better described as riparian. Unfortunately, it rainforest faunas, but less species-rich (es- is dif®cult to classify species by such subtle pecially in primates) than western Amazo- (but perhaps important) behavioral distinc- nian faunas. Although such regional and sub- tions in the frequent absence of relevant ®eld regional diversity contrasts could have his- studies. torical explanations, it is nevertheless appro- Despite these and other problems, some priate to explore their consequences for the guilds of nonvolant rainforest mammals have ecological structure of contemporary com- been widely and more-or-less consistently munities, which may suggest other causal recognized (e.g., arboreal folivores; Mont- factors. gomery, 1978), and others can readily be de- ®ned to label obvious clusters of ecologically TROPHIC GUILDS AND OTHER TOPICS similar species (e.g., terrestrial granivore/frugivores; Peres, 1999). By adopting or modi- Several previous studies have tabulated the fying some previously recognized ecobehav- ecobehavioral attributes of nonvolant rain- ioral categories and recognizing a few new forest mammals at various Neotropical in- ones, we were able to sort most of the non- ventory sites (e.g., Eisenberg and Thoring- volant taxa from La Selva, Paracou, and ton, 1973; Janson and Emmons, 1990; Peres, Manu into a reasonably small number of 1999), but none have provided a comprehen- guilds (table 59). Obviously, some of these sive trophic classi®cation. Indeed, there are are very broadly de®nedÐsuch as terrestrial many problems in attempting to do so. animalivoresÐand could be further subdi- Whereas most rainforest bats can be sorted vided for more detailed analyses. Alterna- into guilds based on reasonably straightfor- tively, many could be combined for the pur- ward dietary and behavioral attributes (Bon- poses of broader generalizations (e.g., about accorso, 1979; Kalko et al., 1996; Simmons primary versus secondary consumers). In ef- and Voss, 1998), many nonvolant mammals fect, we tried to achieve a balance between are harder to characterize ecologically. Nu- excessively ®ne distinctions on the one hand merous species feed on fruit when it is sea- and insuf®cient resolution on the other while sonally abundant, but switch to different re- minimizing the number of arbitrary decisions sources when fruit is scarce; for example, necessary to assign taxa to different catego- woody browse (Mazama americana; Bod- ries. mer, 1990), seeds (Dasyprocta leporina; We primarily used natural history data Henry, 1999), or animals (many didelphids; summarized by Emmons (1997) to assign Atramentowicz, 1988). The poorly docu- taxa to guilds, but we consulted other refer- mented or completely unknown food habits ences for supplementary information. When of some nonvolant taxa (e.g., Glironia, Sci- two or more references suggested different urillus, many murids) is another obstacle to guild assignments for the same taxon, we trophic classi®cation. based our ®nal assignment on the most de- Foraging substrate is an important factor tailed available ®eld study or the most thor- in de®ning nonvolant guilds that introduces ough literature review. Unavoidably, many additional ambiguities because many species guild assignments are arguable (how much do not restrict their activities to one substrate more fruit in the diet should distinguish a type. Whereas some taxa are unambiguously frugivore from an omnivore?), and some are terrestrial or arboreal, numerous trapping and frankly speculative (e.g., dietary categories observational studies (e.g., Janson and Em- for most murids). However, insofar as pos- mons, 1990; Malcolm, 1991; Woodman et sible we used consistent criteria from locality al., 1995; this study) suggest that others are to locality, so the results are hopefully un- scansorialÐactive both in trees and on the biased for comparative purposes. ground. Similarly, whereas some species that ARBOREAL FOLIVORES: This guild is repre- forage primarily in water are unambiguously sented by just three species at Paracou, two 188 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 59 Guild Composition of Nonvolant Mammal Faunas from Three Neotropical Rainforest Localitiesa 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 189

of which are diurnal (Bradypus tridactylus also occurs at La Selva. Furthermore, where- and Alouatta seniculus) and one nocturnal as western Amazonian tamarins (members of (Choloepus didactylus). The same number of the Saguinus fuscicollis and S. mystax arboreal folivores, belonging to the same groups, both of which occur at Manu) appear genera, are present at most Neotropical rain- to be primarily frugivorous (Peres, 1993a), forest sites except where a fourth taxon the single tamarin that occurs at Paracou, S. (Dactylomys) feeds in bamboo and other midas, is not (Pack et al., 1999). Because dense vegetation growing on rich alluvial only one arboreal frugivore (Cebus oliva- soils (as at Manu; Emmons, 1997). ceus) is among the species that might have Although it is probable that howler popu- been locally extirpated by hunting prior to lations had been depleted by hunting in our our inventory (or that might yet be discov- study area, it is unlikely that local sloth pop- ered by future ®eldwork; appendix 1), it ulations were signi®cantly affected. For seems unlikely that this guild is grossly un- Choloepus, the only taxon for which intersite derrepresented at Paracou as a consequence density comparisons are possible, our sight- of sampling inadequacy. ing rate by walked nocturnal census (0.08 in- All of the diurnal arboreal frugivores in dividuals per 10 hours; table 51) is about our study areaÐAteles paniscus, Cebus apel- one-eighth the rate reported by Emmons la, Saimiri sciureusÐwere very uncommon, (1984)27 at Limoncocha, a western Amazo- the ®rst two probably as a consequence of nian locality with rich soils, and one-fourth persistent hunting. The only commonly en- the average sighting rate on Barro Colorado countered species belonging to this guild Island (Glanz, 1982). Although our infre- (Caluromys philander, Potos ¯avus) were quent views of Choloepus and our inability both nocturnal. In fact, our sighting rate for to obtain even a single sighting of Bradypus kinkajous at Paracou (3.08 per 10 hours of suggest that sloths are locally rare, these taxa walked nocturnal census) is comparable to were among the commonest mammals res- the highest rate reported by Glanz (1982: ta- cued from primary forest ¯ooded by a hy- ble 4) from Barro Colorado Island, and is droelectric dam at Petit Saut, about 28 km higher than any ®gures that Emmons (1984: SSW of our study area (VieÂ, 1999), where table 4) reported from four Amazonian lo- their joint density was estimated to be about calities. This result seems anomalous in the 2.6 individuals/km2 (Taube et al., 1999). context of other indications that the Paracou ARBOREAL FRUGIVORES: The most conspic- environment is not rich in fruit resources, but uous intersite differences in trophic compo- it is possible that kinkajous are locally abun- sition concern this guild, which contains only dant as a consequence of ecological release ®ve species at Paracou, a mere 8% of the following the virtual extirpation of compet- total nonvolant fauna. By contrast, six arbo- ing diurnal species. Our sighting rate for Cal- real frugivores constitute 12% of the less uromys philander (0.17 per 10 hours), how- speciose nonvolant fauna of La Selva, and ever, is low by comparison with most re- 13 species constitute 16% of the more spe- ported rates for congeneric species on Barro ciose fauna of Manu. The striking de®cit of Colorado Island (C. derbianus; Glanz, 1982: arboreal frugivores at Paracou is partly, but table 4) and in western Amazonia (C. lana- not entirely, a consequence of the previously tus; Emmons, 1984: table 4). It is noteworthy noted faunal differences in primate diversity. that trapping and sighting data for C. philan- Thus, although the Manu fauna includes sev- der at Paracou are consistent with the results eral frugivorous primate genera not found at of other recent ®eld studies (Charles-Domi- Paracou, it also includes additional nonpri- nique et al., 1981; Malcolm, 1991) in sug- mate arboreal frugivores (Caluromysiops, gesting that this is one of the most exclu- Bassaricyon), one of which (Bassaricyon) sively arboreal of Amazonian marsupials. ARBOREAL GRANIVORE/FRUGIVORES: This guild consists of nine species at Paracou, rep- 27 Emmons (1984) reported her census results as sightings per 10 km, and her walking speed as 0.6±1.5 km/ resenting about 14% of the known nonvolant hr. In comparing our results with hers, we assume that fauna. Roughly similar proportions of the La she walked about 1 km per hour on average. Selva and Manu faunas are arboreal grani- 190 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 vore/frugivores, despite geographic differ- cies from Barro Colorado Island (Glanz, ences in taxonomic composition. Among 1982: table 4) and is higher than either of the other contrasts, Central American faunas two previously reported rates from Amazo- contain no granivorous monkeys, whereas nian localities (Emmons, 1984: table 4). squirrels are more speciose in both Central ARBOREAL INSECTIVORES: This guild in- America and western Amazonia than they cludes only two species at Paracou and at all are in the Guiana subregion. other adequately sampled Neotropical rain- Members of this guild at Paracou include forest localities. Our ®eld data contribute no two diurnal species, one a primate (Pithecia new natural history information about either pithecia) and the other a squirrel (Sciurus Cyclopes or Tamandua. aestuans); the remainder, three murids (Oec- TERRESTRIAL HERBIVORES: The Paracou omys auyantepui, O. rutilus, Rhipidomys ni- fauna includes no species assignable to this tela) and four caviomorphs (Coendou melan- guild, nor (apparently) does any other known urus, C. prehensilis, Makalata didelphoides, rainforest fauna from the Guiana subregion Mesomys sp.), are all nocturnal. All of the of Amazonia. members of this guild except Sciurus aes- TERRESTRIAL GRANIVORE/FRUGIVORES: tuans and the murids were very uncommon Thirteen species belong to this guild at Par- at Paracou. Most of our scant ®eld observa- acou, where they constitute 20% of the non- tions for these taxa are more-or-less consis- volant fauna. Similar proportions of terres- tent with previously published natural history trial granivore/frugivores are present in the data (summarized by Emmons, 1997), but La Selva and Manu faunas despite differenc- the diurnal sighting of Makalata didelphoides in the taxonomic composition of this es is an inexplicable oddity. group from site to site. ARBOREAL GUMMIVORES: No mammal in At Paracou, terrestrial granivore/frugi- the Guiana subregion of Amazonia is known vores include four large diurnal species (Pe- to subsist primarily on saps and gums, al- cari tajacu, Tayassu pecari, Dasyprocta le- though it is possible that the unknown diet porina, Myoprocta acouchy), one large noc- of Sciurillus pusillus includes some type of turnal species (Cuniculus paca), and eight plant exudate (Emmons, 1997). small nocturnal species (Neacomys dubosti, ARBOREAL OMNIVORES: Five Paracou spe- N. paracou, Oligoryzomys fulvescens, Ory- cies are assigned to this guild, constituting zomys macconnelli, O. megacephalus, O. 8% of the known fauna. Proportions of ar- yunganus, Proechimys cuvieri, P. guyannen- boreal omnivores in the La Selva and Manu sis). Both of the peccaries, locally hunted for faunas are similar. meat, were uncommon or wary; only Dasy- At Paracou, arboreal omnivores consist of procta and Myoprocta were regularly seen a single diurnal primate (Saguinus midas) (or heard) by us, but neither was really com- and at least four nocturnal marsupials (Di- mon. By contrast, our sighting rate for pacas delphis marsupialis, Marmosa murina, Mi- (1.1 individuals per 10 hours of walked noc- coureus demerarae, Philander opossum). Of turnal census) is about equal to the average the marsupials, our data suggest that Micou- sighting rate (1.3 per 10 hours) for ®ve Am- reus is the most consistently arboreal, fol- azonian inventories tabulated by Emmons lowed by Didelphis and Philander; Paracou (1984: table 4), and is similar to sighting records are too scanty to place Marmosa rates from two of the three nocturnal census murina in this ranking, but Charles-Domi- episodes recorded for Barro Colorado Island nique et al. (1981) suggested that this species (Glanz, 1982: table 4). It is possible that pa- is primarily active in the forest understory, cas are relatively common at Paracou for the like P. opossum. Most members of this guild same reason that kinkajous are, because local were common in our study area, although populations of most competing species in the numbers of D. marsupialis declined abruptly diurnal fauna have been depleted by hunting. after 1991; only M. murina was consistently Small nocturnal members of this guild ac- uncommon. Our nocturnal sighting rate for counted for 68% of our captures using all P. opossum (0.96 per 10 hours) is compara- conventional traps, and 80% of our captures ble to the highest rate reported for this spe- using just Victor and Sherman traps. Inter- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 191 estingly, this is the only guild at Paracou and admittedly heterogeneous guild compris- where sympatry between congeneric forms is es a higher fraction (25%) of the fauna at the rule rather than the exception: Neacomys, Paracou than at either La Selva or Manu. Oryzomys, and Proechimys are each repre- Three terrestrial animalivores at Paracou are sented by two or more species in the local diurnal (Monodelphis, Speothos, Galictis), fauna. Although most of the small nocturnal six are active both by day and at night (Myr- granivore/frugivores we recorded at Paracou mecophaga, Herpailurus, Leopardus pardal- occur syntopically in well-drained primary is, Panthera, Puma), and the remaining eight forest (the only exception is Oligoryzomys are nocturnal (Marmosops parvidens, M. pin- fulvescens, found exclusively in roadside heiroi, Metachirus nudicaudatus, Dasypus secondary growth), our capture results sug- kappleri, D. novemcinctus, Cabassous, Prio- gest that some closely related species may dontes, Leopardus weidii). All of the terres- differ in their use of other habitats. Thus, trial-animalivorous carnivores at Paracou (fe- Oryzomys yunganus occurs more frequently lids, Speothos, and Galictis) were uncommon than O. megacephalus in swampy and creek- and most were only recorded by us through side forest, whereas Proechimys cuvieri oc- interviews. Monodelphis, Cabassous, Dasy- curs more frequently than P. guyannensis in pus kappleri, Priodontes, and Myrmecopha- secondary growth. To the best of our knowl- ga were likewise rarely collected, observed, edge, ecological differences between O. me- or recorded from interviews. Only four local gacephalus and O. yunganus have not been members of this guild were encountered of- observed previously, but Malcolm's (1992) ten enough to provide noteworthy behavioral report of differential habitat use by P. cuvieri data or density estimates. and P. guyannensis near Manaus agrees with Based on their highly carnassialized den- our sampling results for these species. Nea- titions, small size, and the absence of mac- comys paracou and N. dubosti occur syntop- roscopic seeds in feces, we assume that spe- ically at Paracou and are known from sym- cies of Marmosops are predominantly, if not patric collections elsewhere in the Guiana exclusively, animalivorous. Although species subregion, but available information is too of this genus have opposable halluces and sparse to suggest how (or if) these congeners prehensile tails, our nocturnal observations differ ecologically. and trapping results suggest that M. parvi- TERRESTRIAL FOLIVORE/FRUGIVORES: Three dens and M. pinheiroi are predominantly ter- ungulate species occupy this guild at all com- restrial, seldom ascending slender lianas or pared inventory sites, where they account for narrow stems more than a meter or two only 4±6% of each fauna. All of the Paracou above the forest ¯oor; none were taken in species were persecuted by local hunters and arboreal platform traps. Malcolm (1991) only two were actually observed by us. Our likewise trapped M. parvidens only on the ®eldnotes contain no noteworthy natural his- ground, despite his intensive program of ar- tory information about either Mazama or boreal platform trapping. Woodman et al. Tapirus, but a recent analysis of tapir diets (1995) captured M. noctivagus equally often in French Guiana (Henry et al., 2000) was on the ground and above ground, but none based in part on material collected near Par- of their above-ground traps were set higher acou. than 2 m. Patton et al. (2000: appendix A) TERRESTRIAL OMNIVORES: Two members of listed several species of Marmosops as hav- this guild are present at Paracou, La Selva, ing been taken in ``canopy'' traps at some and Manu, where they constitute an almost collecting localities along the Rio JuruaÂin constant minor fraction of the known non- western Amazonia, but their text summaries volant fauna. Although both Paracou species of habitat data (op. cit.: 53±61) indicate that (Eira barbara and Nasua nasua) are large, none were actually trapped at heights ex- diurnal, and not locally hunted for food, nei- ceeding 2 m; instead, the majority (82%) of ther was commonly observed in our study all captures of this genus on the Rio JuruaÂ and no noteworthy natural history observa- were on the ground. Because many unam- tions were recorded about them. biguously terrestrial mammals (including TERRESTRIAL ANIMALIVORES: This large Oryzomys, Myoprocta, and Proechimys in 192 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 our study) will occasionally ascend lianas to obtained by this method, which might use- reach baited traps, infrequent captures of fully be applied to detect the presence of Marmosops in low vegetation is not incon- congeneric species at other Amazonian lo- sistent with our provisional guild assignment calities. for this genus. Metachirus has been variously character- GENERAL DISCUSSION ized in standard references as an arboreal Although convenient for the purposes of omnivore (Walker, 1964) or a semiarboreal data analysis and publication, our separate frugivore (Hunsaker, 1977; Streilein, 1982), treatments of the bats and nonvolant mam- but these ecobehavioral descriptors appear to mals from Paracou provide an incomplete be unsupported by ®eld data. Instead, our picture of the fauna as a whole. For example, capture results agree with observations from taxonomically comprehensive estimates of many other trapping programs (e.g., Miles et local species richness have yet to be dis- al., 1981; Malcolm, 1991; Patton et al., 2000) cussed and compared with those from other in suggesting that this is one of the most ex- inventory sites, nor has it been possible until clusively terrestrial of all New World mar- now to analyze similarities and differences supials. Similarly, our examination of stom- between bats and nonvolant mammals in ach contents from 10 Paracou specimens is terms of their biogeographic relationships consistent with evidence from other dietary and trophic structure. Such topics bridge the studies (Mathews, 1977; Santori et al., 1996; separate research traditions exempli®ed by Freitas et al., 1997) that Metachirus is almost many previous studies of rainforest mammal entirely insectivorous. Our nocturnal sighting faunas that have treated only bats or nonvo- rate for Metachirus (0.42 per 10 hours) is lant species in substantive detail (e.g., Eisen- within the range of values reported for this berg and Thorington, 1973; Glanz, 1982; taxon by Emmons (1984: table 4) at other Brosset and Charles-Dominique, 1990; Jan- Amazonian sites. son and Emmons, 1990; Ascorra et al., 1993, SEMIAQUATIC/RIPARIAN FOLIVORES: This 1996; MedellõÂn, 1993; Kalko et al., 1996; guild is not represented at Paracou, where Peres, 1999; Patton et al., 2000). The follow- suitable habitat for capybaras (Hydrocho- ing discussion therefore contributes to a syn- erus) is absent. Semiaquatic-folivorous mam- thesis of mammalian diversity studies in mals never represent more than a tiny frac- Neotropical rainforests, an objective for tion of any Neotropical rainforest fauna. which future research priorities are suggested SEMIAQUATIC/RIPARIAN OMNIVORES: Only a in our ®nal chapter. single species assignable to this guild is present at Paracou and at Manu; no ecological SPECIES RICHNESS AND INVENTORY equivalent is known from Central American COMPLETENESS rainforests. The Paracou species (Nectomys melanius) is apparently very uncommon and We recorded a total of 142 species of was not trapped by us. mammals at Paracou, of which 128 were SEMIAQUATIC/RIPARIAN ANIMALIVORES: captured or observed as result of our own This guild comprises a smaller fraction of the efforts in the ®eld from 1991 to 1994. The fauna at Paracou than at either La Selva or remaining 14 species were either recorded Manu, where large bodies of water support from interviews with local residents or were resident otter populations. At Paracou, this identi®ed from museum voucher material guild is only represented by two species, collected in our study area by previous re- both of which are nocturnal. Water opossums searchers. Because our total species accu- (Chironectes minimus) were sighted fre- mulation results (®g. 91) do not indicate a quently in our study area (0.7 individuals per convincing asymptote, it is reasonable to as- 10 hours of nocturnal census) and are obvi- sume that we could have recorded more spe- ously common in local streams, but we re- cies at Paracou with additional ®eldwork, corded no other signi®cant natural history in- perhaps including some species not previ- formation about this species. Our pitfall cap- ously observed or collected by others. tures of Neusticomys oyapocki are the ®rst To explore the range of plausible estimates 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 193

Fig. 91. Species accumulation for all mammals at Paracou, where we recorded 128 species in the course of 202 ®eld days from 1991 to 1994. ``Minimum known diversity'' (142 species) includes additional records obtained from interviews, together with specimens previously collected in our study area by other investigators.

of true species richness for the entire Paracou tensive that most species were each observed mammal fauna, we summed the results of three or more times (i.e., information in the four nonparametric extrapolation methods larger frequency classes is ``nonnegligible'' applied separately28 to our sampling data for sensu Chao [1984, 1987]). By contrast, the bats and nonvolant mammals (table 60). Giv- jackknife methods yield statistically consis- en the inherent uncertainty of any extrapo- tent estimators (Burnham and Overton, 1979) lation procedure, it is encouraging that the that should converge on true species richness range of total richness values predicted by as sampling effort increases. In fact, the these methods is really quite narrow (155± numbers of species for bats, nonvolant mam- 168 spp.). The results from Chao's methods mals, and total mammals predicted by the should probably be interpreted as lower second-order jackknife (in the ®fth column bounds for plausible inference from our data of table 60) are very close to ecogeographic because sampling effort was suf®ciently in- expectations based on the known northern French Guianan source fauna. 28 It makes sense to perform separate extrapolations Estimating the overall completeness of our and sum the results because bats and nonvolant mam- mammal inventory clearly depends on which mals were sampled with different methods on different of these extrapolations of true species rich- schedules. In effect, however, extrapolations based on ness is used. Whereas Chao's methods sug- the pooled data for bats and nonbats are very close to the tabulated values using Chao's methods (157 spp. for gest completeness values (observed/extrapo- CHAO1, 156 spp. for CHAO2) and are identical with lated species richness ϫ 100) in the range of the tabulated values using the jackknife estimators. 90±92%, the jackknife methods suggest low- 194 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 60 Observed and Extrapolated Mammalian Species Richness at Paracou Compared with Expectations from Ecogeographic Range Data

er values of about 85±87%. These differenc- ni®cantly with additional ®eldwork in our es are not large, however, and even the low- study area. est estimate (based on the second-order boot- The taxonomic dominance of bats in Neo- strap) suggests that the mammal fauna at Par- tropical rainforest mammal faunas is evi- acou is at least as well sampled as those at denced by all large species lists obtained Central American ®eld stations with decades- with modern collecting methods (table 61), long histories of biological research. For ex- but bats may be proportionately more diverse ample, faunal sampling at La Selva is per- or less diverse at some localities than they haps only 85% complete after more than 30 are at Paracou. Wilson (1990) suggested that years of ®eld research at that site, whereas bats probably represent about 60% of the faunal sampling near Barro Colorado Island fauna at La Selva, a somewhat higher ®gure is perhaps only 78% complete after more than is supported by observed species counts than 70 years (Voss and Emmons, 1996: ta- from that site, but approximately what could ble 10). be expected there based on ecogeographic Bats constitute about 55% of the known range data (Voss and Emmons, 1996: appen- Paracou mammal fauna, and our extrapola- dix 2). Similar proportions will perhaps be tions suggest that similar proportions (54± found to characterize other Central American 56%) would have been obtained with addi- faunas, where most nonvolant clades (nota- tional sampling effort. In order of decreasing bly marsupials, primates, and rodents) are richness, rodents are clearly a distant second conspicuously less diverse than in Amazonia. to bats, representing only 15% of the ob- By contrast, bats may constitute a smaller served total species, followed by marsupials proportion of western Amazonian mammal (8%), carnivores (7%), xenarthrans (6%), faunas, where many nonvolant clades are primates (4%), and ungulates (4%). The maximally diverse. Bat communities have small list of expected nonvolant species (ap- not been intensively sampled over multiple pendix 1) suggests that this rank ordering by years at any western Amazonian site, how- relative diversity is unlikely to change sig- ever, with the result that existing species 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 195

TABLE 61 zonia) and at Balta (in western Amazonia) Bat and Nonvolant Species Richness at Ten tend to support this interpretation (®g. 92). Neotropical Rainforest Inventory Sitesa For any speci®ed level of sampling effort after the ®rst several person-weeks at each site, the number of species recorded at Balta is consistently higher than that recorded on the Rio Xingu, which in turn is consistently higher than that recorded at Paracou. There- fore, the large species list we eventually obtained at Paracou is plausibly explained by our prolonged total sampling effort (634 person-days), and by our subsequent use (after 1991) of more methods to capture elusive species. It is noteworthy that the rank-ordering of inventory sites by species richness implied by ®gure 92 is the same as that expected from geographic range data, which predict maximal mammalian diversity in the western part of Amazonia, intermediate diversity in the southeast, and minimal diversity in the Guiana subregion (Voss and Emmons, 1996). Although three data points are not suf®cient to establish a signi®cant correlation between counts (e.g., from Balta, Manu, and Cuzco expected and observed diversity, this result AmazoÂnico; table 61) probably underesti- is consistent with other lines of evidence sug- mate relative bat diversity. Based on expect- gesting that the Paracou mammal community ed species lists (Voss and Emmons, 1996: ap- is a typically Guianan assemblage whose es- pendices 9±11) it seems likely that bats will sential characteristics are largely determined eventually be found to comprise about 50% by its biogeographic context. That context of the total mammal fauna at most western now requires more precise de®nition before Amazonian localities. its implications for community phenomena Although a few more mammalian species other than species richness can be assessed. are currently known from Paracou than from any other Neotropical rainforest site for BIOGEOGRAPHY which published faunal lists are currently available (table 61, last column), this result Our previous biogeographic analyses of is hard to interpret due to inventory differ- Paracou bats and nonvolant mammals sug- ences in many of the confounding factors gested that these groups show essentially previously discussed in our comparative similar spatial patterns of faunal relationships analyses of bat and nonvolant species rich- with other rainforest inventories. In both ness. Among such factors, sampling methods analyses, Paracou ®rst clustered with other and effort seem likely to be the most impor- inventory sites from the Guiana subregion of tant: because we used many inventory meth- Amazonia, next with sites from other Ama- ods over multiple ®eld seasons at Paracou, zonian subregions, and lastly with Central our species list could be artifactually larger American sites (Simmons and Voss, 1998: than those from more diverse sites where ®g. 76; this report: ®g. 89). Obviously, faunal faunal sampling was substantially less com- similarity is correlated with geographic prox- plete. imity in both datasets, but bats and nonvolant Species accumulation graphs from our ®rst mammals differ in other quantitative biogeo- (1991) ®eld season at Paracou and from graphic phenomena that merit further analy- methodologically comparable ®eldwork on sis. the lower Rio Xingu (in southeastern Ama- To obtain strictly comparable biogeo- 196 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 92. Species accumulation curves for all mammals at three Neotropical rainforest inventory sites (see ®g. 88). Person-days were calculated by multiplying the number of inventory workers by the number of productive ®eld days at each site (excluding dates when no sampling occurred). The Paracou curve is from our ®rst (1991) ®eld season, when we used the same methods as those employed at the other localities (primarily ground-level mistnetting, searching for bat roosts, conventional trapping, and hunting/census). The Xingu curve represents collections and sight records made by USNM personnel near their base camp from 13 August to 23 October 1986 (collections at other sites up- and down-river are excluded; see Voss and Emmons, 1996: appendix 8). The Balta curve represents collections made by A. L. Gardner and J. L. Patton from 1966 to 1971, but omits miscellaneous specimens obtained at the same locality by other researchers not continuously engaged in mammal inventory work (see Voss and Emmons, 1996: appendix 9). graphic data for bats and nonvolant mam- Among this common set of geographic sam- mals, we computed matrices of pairwise fau- ples, bat faunal similarity ranges from a min- nal similarity (scaled as percentages in table imum value of 31% (for the comparison Bar- 62) and airline distances (table 63) among all ro Colorado-Xingu) to a maximum of 68% ten inventory sites at which presence/absence (Balta-Manu), with a mean of 46% and a data for both groups were obtained within the standard deviation of 9%. By contrast, non- same study area (La Selva, Barro Colorado, volant faunal similarity ranges from 10% (for Imataca, Paracou, Arataye, Cunucunuma, La Selva-Xingu) to 78% (Paracou-Arataye), Xingu, Balta, Manu, Cuzco AmazoÂnico).29 with a mean of 34% and a standard deviation of 16%. At this spatial scale (airline distances 29 Manaus was omitted from these computations because the bat list was compiled from collections made at widely scattered sites around the city (Reis and Per- Other inventory sites omitted from these matrix calcu- acchi, 1987), whereas the nonvolant mammal list was lations are those from which only bat or nonvolant mam- based on geographically restricted sampling within the mal data are available (Jenaro Herrera, Kartabo, Piste MCSE Reserves (Voss and Emmons, 1996: appendix 7). St.-EÂ lie). 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 197

TABLE 62 Percent Similarity among Bat and Nonvolant Mammal Faunas at Ten Neotropical Rainforest Inventory Sitesa (Table entries are Jaccard's coef®cient ϫ 100.)

among these sites range from 136 to 3788 declines with distance at different average km with a mean of 2038 km and a standard rates for the two groups (®gs. 93, 94). Linear deviation of 954 km), bat faunas are there- regression suggests that faunal divergence fore more similar to one another (on average) averages about 14% per 1000 km for non- and are less variable in composition than volant mammals, almost twice the estimated nonvolant mammal faunas. rate of about 8% per 1000 km for bats. In Matrix correlations (assessed for statistical addition, the residual variance in faunal sim- signi®cance by random permutations; Smou- ilarity valuesÐthe variation ``unexplained'' se et al., 1986) indicate that the inverse re- by airline distanceÐis substantially greater lationship between faunal similarity and air- for nonvolant mammals than it is for bats, line distance is equally strong for bats (r ϭ with an F-ratio (computed from the mean- Ϫ0.81, p ϭ 0.001) and nonvolant mammals squared-error estimates of the respective re- (r ϭϪ0.82, p ϭ 0.001), but faunal similarity gressions) of about 2.8.

TABLE 63 Matrix of Airline Distances (km) among Ten Neotropical Rainforest Inventory Sitesa 198 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 93. Percent faunal similarity (Jaccard's coef®cient of faunal similarity ϫ 100) plotted against airline distance for all 45 pairwise comparisons among ten rainforest bat inventories (tables 62, 63).

The magnitude of the residual variation in gu lists are taxa with eastern Amazonian nonvolant faunal similarity is aptly illustrated ranges that are wholly or largely restricted to by comparisons between Paracou and two al- either the north side of the Amazon (e.g., most equidistant inventory sites. Although Bradypus tridactylus, Saguinus midas, Al- Imataca is slightly further from Paracou ouatta seniculus, Ateles paniscus, Pithecia (1023 km) than is Xingu (994 km), percent pithecia, Oecomys auyantepui, Oecomys ru- nonvolant faunal similarity between Paracou tilus, Coendou melanurus, Proechimys guy- and Imataca is 61%, whereas nonvolant fau- annensis) or to the south side (e.g., Bradypus nal similarity between Paracou and Xingu is variegatus, Saguinus niger, Alouatta belze- only 36%. Of course, faunal similarity values bul, Aotus infulatus, Callicebus moloch, Oec- estimated from incomplete inventory data are omys paricola, Oryzomys emmonsae, Oxy- subject to sampling error, which plausibly ac- mycterus amazonicus, Proechimys goeldii, counts for some differences between the Par- Proechimys oris). By contrast, bat faunal acou and Xingu species lists (many ubiqui- similarity does not differ dramatically be- tous carnivore taxa, for example, were not tween Paracou-Imataca (56%) on the one recorded by USNM ®eldworkers during their hand and Paracou-Xingu (49%) on the other, short visit to the Xingu site; Voss and Em- and only two bat species appear to have mons, 1996: appendix 8). An additional fac- range limits de®ned by the lower Amazon tor of obvious importance, however, is the (Tonatia schulzi and Lasiurus atratus, both role of the lower Amazon as a barrier to non- known only from the north side). The same volant faunal exchange: at least 23 of the set of inventory comparisons therefore sug- mismatches (species present in one fauna but gests that major rivers and other large-scale not the other) between the Paracou and Xin- habitat discontinuities are less effective as 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 199

Fig. 94. Percent faunal similarity (Jaccard's coef®cient of faunal similarity ϫ 100) plotted against airline distance for all 45 pairwise comparisons among ten nonvolant rainforest mammal inventories (tables 62, 63). dispersal barriers for bats than for nonvolant discontinuities are less effective as faunal mammals. barriers for bats than they are for nonvolant Different average dispersal abilities of bats mammals. Thus, most Paracou bats (42 spe- on the one hand and nonvolant mammals on cies, representing 54% of the local chirop- the other could explain the disparate results teran fauna; table 64) occur in all four Neo- of regressing faunal similarity on airline dis- tropical rainforest regions, another 19 bat tance for these groups. Because widely sep- species (24%) occur in three out of four rain- arated localities will usually have more in- forest regions, and an additional 11 species tervening barriers between them than adja- (14%) occur in two regions. Altogether, 92% cent localities, faunas composed of taxa with of Paracou bats are apparently undifferenti- lower average dispersal abilities should di- ated across one or more major landscape fea- verge more rapidly with distance (on aver- tures delimiting the Neotropical rainforest bi- age) than faunas composed of more vagile ota. Only six species (about 8% of the local taxa. However, because barriers sometimes chiropteran fauna) are Amazonian endemics. do occur between nearby sites (as those on Nonvolant Paracou mammals show con- opposite river banks), the residual variance trasting patterns of species membership in should also be larger for faunas composed of these distributional categories, especially the taxa with lower average dispersal abilities. ®rst and last. Only 18 species (comprising Patterns of species distribution among just 28% of the nonvolant fauna) occur in all Neotropical rainforest regions separated by four Neotropical rainforest regions, whereas high mountains, major rivers, and xero- 24 species (fully 38% of the nonvolant fau- morphic vegetation (®g. 95) are consistent na) are Amazonian endemics. To facilitate with the hypothesis that large-scale habitat statistical testing of these frequency differ- 200 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 95. Distribution of lowland rainforest on the Central and South American mainland (see Voss and Emmons, 1996, for references). Major habitat discontinuities delimit four distinct regions: (1) the Andes and the SerranõÂa de PerijaÂ separate trans-Andean rainforests to the west from Venezuelan coastal rainforests and Amazonian rainforests to the east; (2) the Llanos and the lower Orinoco separate Ve- nezuelan coastal rainforests to the north and west from Amazonia to the south and east; (3) an arid diagonal of xeromorphic biomes (Caatinga, Cerrado, Chaco) separates Amazonia to the northwest from the Atlantic rainforests (Mata Atlantica) to the southeast. The location of our inventory site at Paracou is indicated by the arrow. ences, we dichotomously classi®ed all Para- demism in rainforest bats and nonvolant cou mammals as either ``widespread'' (pre- mammals, we applied biogeographic parsi- sent in two or more rainforest regions) or mony methods (as originally described by ``endemic'' (restricted to Amazonia). The hy- Rosen, 1988, 1992) to identify repeated pat- pothesis that Paracou bats and nonvolant terns of species-sharing among inventory mammals do not differ in relative endemism sites. Parsimony analyses of endemism (and implied dispersal ability) can be con®- (PAE) were implemented with heuristic dently rejected based on the resulting 2 ϫ 2 search algorithms in PAUP* (version 4.0b3a; contingency analysis (table 65). Swofford, 2000), which were applied sepa- To explore the geographic structure of en- rately to the presence/absence data for bats 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 201

TABLE 64 TABLE 65 Comparative Rainforest Distribution Patterns Comparison of Endemicity in Paracou Bats of Paracou Bats and Nonvolant Mammals and Nonvolant Mammalsa

A spatially compact grouping of four nonvolant inventories from the coastal watershed of the Guianan subregion of Amazonia (Im- ataca, Kartabo, Arataye, Paracou) is recovered by PAE with moderately strong (71%) bootstrap support, but this endemicity set is immediately joined with similar (74%) support by the nonvolant inventory from Ma- naus, a site deep in the continental interior near the con¯uence of the Rio Negro and the Amazon. Although the next nonvolant inventory site to join is Cunucunuma (another deep-continental site), this grouping has only and nonvolant mammals; an all-zero opera- marginal (57%) bootstrap support. Finally, tional geographic unit was added to both da- PAE recovers an even more inclusive group- tasets to root the resulting networks. Relative ing composed of these six sites plus the non- support for different locality groupings was volant inventory from the Rio Xingu, which estimated by bootstrap resampling with 1000 joins the others with moderate (68%) boot- replicates using 100 random-addition se- strap support despite its isolated position quences per replicate. south of the Amazon. The geographic patterns recovered with Of course, some of this recovered structure Ն50% bootstrap support by PAE are a subset is due to the absence of inventory data from of those previously obtained from UPGMA many geographically intermediate localities. clustering by Jaccard's coef®cient for both Inevitably, the addition of new faunal lists datasets (®gs. 96, 97). It is also noteworthy with novel combinations of species will re- that no PAE grouping that received more duce the distinctiveness of some currently than 50% bootstrap support in one dataset well-supported groupings in future analyses. con¯icts with any PAE grouping with equiv- A more general criticism of these results alent support in the other. Instead, a Central could justi®ably invoke the inappropriateness American/Amazonian faunal dichotomy is of hierarchical models for analyzing data that indicated for both bats and nonvolant mam- perhaps lack a natural hierarchical struc- mals, as is a well-supported grouping of ture.30 Certain spatial patterns recovered in three adjacent inventory sites in southwestern Amazonia. However, PAE based on the 30 This criticism, which de Queiroz and Good (1997) nonvolant data supports additional intra-Am- and others have levelled at phenetic clustering methods azonian faunal relationships that are not con- as conventionally used to explore the structure of simi- sistently recovered by PAE from the bat data. larity data, applies equally to parsimony algorithms as 202 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 96. Results of biogeographic analyses of 14 rainforest bat faunas (see Simmons and Voss [1998: table 75] for geographic coordinates and other locality information). Solid contours enclose groups obtained from UPGMA clustering by Jaccard's coef®cient of faunal similarity that were also recovered with Ն50% bootstrap support from a parsimony analysis of endemicity (PAE; see text for explanation). Broken contours show UPGMA similarity clusters that receive Ͻ50% bootstrap support from PAE. Paracou and Piste St.-EÂ lie (an adjacent inventory site in northern French Guiana) were treated as separate terminals (OTUs) in both analyses, but are here combined for visual clarity. The raw (presence/absence) data on which both clustering and PAE were based are those tabulated by Simmons and Voss (1998: appendix 2) with the corrections noted in table 62 of this report. our analyses, however, are strongly support- sustain meaningful biogeographic interpre- ed by external evidence and seem likely to tation. One such pattern is the cluster of inventories that includes Paracou and most, but ← not all, of the other analyzed sites from the used to reconstruct the relationships of continental faunas Guiana subregion of Amazonia. that may have routinely exchanged taxa by dispersal (or Although the area enclosed by the inter- ``dispersion'' sensu Cracraft, 1994) over evolutionary time. connected waters of the Orinoco, the Rio Ne- The most that can be said for either analytic method in our applications is that the results can be mapped on geography gro, and the lower Amazon has traditionally in such a way that spatial patterns (or the absence of such been recognized as a convenient unit for bio- patterns) in the data are readily apparent. geographic analysis (e.g., by Tate, 1939; 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 203

Fig. 97. Results of biogeographic analyses of 12 nonvolant rainforest mammal faunas (see table 55 for geographic coordinates and other locality information). Solid and broken contours enclosing various groups were drawn following the same conventions explained in the caption to ®gure 96. The raw (presence/absence) data on which both clustering and PAE were based are tabulated in appendix 2.

Hoogmoed, 1979; Mori, 1991; Voss and Em- Guianas (e.g., Dadanawa) and Guianan Bra- mons, 1996), the Guiana subregion of Ama- zil (e.g., Faro, Serra do Navio). zonia is actually inhabited by two distinctly The fauna that occurs east of the Casi- different rainforest mammal faunas (®g. 98). quiare in southern Venezuela includes at least One of these consists of an apparently allo- 19 species, all of which have more-or-less chthonous (non-Guianan) assemblage that is extensive western Amazonian distributions represented by collections from Cunucunu- (table 66). Most of these do not range much ma, Esmeralda, Boca Mavaca, Neblina Base farther into the Guiana subregion, but a few Camp, and several other localities east of the have been collected or observed as far east- Casiquiare in southern Venezuela. The other ward as the right (west) bank of the Rio fauna, distinctively Guianan in taxonomic Branco, which apparently constitutes a sig- composition, is represented by collections ni®cant geographic limit to this fauna in Bra- from Imataca, Kartabo, Paracou, Arataye, zil: Ateles belzebuth (see Nunes et al., 1988), Manaus, and numerous other sites in the Callicebus torquatus (Hershkovitz, 1990), 204 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Fig. 98. The Guiana subregion of Amazonia as traditionally delimited by the Orinoco, the Rio Negro, and the lower Amazon; an arrow indicates the Casiquiare Canal, which connects the headwaters of the Orinoco and the Rio Negro in southern Venezuela. Closed circles (●) show inventory sites and other collecting localities with typically Guianan rainforest mammal faunas; triangles (᭡) show localities with faunas that include many allochthonous (non-Guianan) species of rainforest mammals. Geographic coordinates and references (which list museum collections where voucher specimens are preserved) for Cunucunuma, Imataca, Kartabo, and Manaus (ϭ MCSE Reserves) are provided in table 55. Equivalent information about three other localities is available in the literature: Esmeralda (Tate, 1939; Handley, 1976), Boca Mavaca (Handley, 1976), and Neblina Base Camp (Gardner, 1988). Collections from Dad- anawa (2Њ50ЈN, 59Њ30ЈW) are in the ROM and USNM. Collections from Locksie Hattie (5Њ10ЈN, 55Њ28ЈW) are in the FMNH. Collections from the vicinity of Faro (2Њ11ЈS, 56Њ44ЈW) are in the AMNH. Collections that we examined from the Serra do Navio (0Њ59ЈN, 52Њ03ЈW) are in the USNM.

Sciurus igniventris (Emmons, 1997: map the left bank of the Essequibo or its upper 147), and Oecomys concolor (G. G. Musser, tributaries in Guyana: Bassaricyon gabbii personal commun.). In Venezuela, some al- (based on the single historical record dis- lochthonous taxa appear to reach their east- cussed by Tate, 1939) and Caluromys lanatus ern range limits at or near the RõÂo Caura (see Emmons, 1993a). (e.g., Philander andersoni, Ateles belzebuth) A few other non-Guianan mammals are or the RõÂo CaronõÂ (e.g., Aotus trivirgatus, known from scattered localities on the north Callicebus torquatus; Linares, 1998). Appar- bank of the lower Amazon. These could be ently, only two members of this fauna reach members of a distinctive whitewater-¯ood- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 205

TABLE 66 plain biota (e.g., as mapped by Salo and RaÈs- Allochthonous (non-Guianan) Mammals that aÈnen, 1989: ®g. 1), the taxonomic composi- Occur East of the Casiquiare Canal in tion of which might include both western Southern Venezuelaa Amazonian elements (e.g., Glironia venusta; da Silva and Langguth, 1989) and central Amazonian endemics that occur on both sides of the river (e.g., Makalata grandis; Emmons, 1997, personal commun.). The apparently narrow distributions of Dactylomys dactylinus and Bradypus variegatus along the north bank of the lower Amazon may also be restricted to ¯oodplain habitats. By contrast, the remainder of the Guiana subregion (east of the RõÂo CaronõÂ-Rio Branco and north of the Amazonian ¯oodplain) is inhabited by a strikingly homogeneous and unmistakably autochthonous fauna, at least 17 members of which have suf®ciently con- gruent range limits to usefully de®ne a Guianan center of mammalian endemism (table 67). Although a few taxa that we consider Guianan endemics are known to occur west of the RõÂo CaronõÂ (e.g., Monodelphis brevicaudata, Chiropotes satanas chiropotes, Proechimys guyannensis) or along the south bank of the lower Amazon (Marmosops parvidens, Marmosops pinheiroi, Bradypus tridactylus, Myoprocta acouchy), range overlap is most extensive in the core area whose pe- riphery is suggested by the closed circles in ®gure 98. Other typically Guianan but nonendemic species that have similar distributional limits within the subregion include Philander opossum (replaced by P. andersoni in southern Venezuela), Cebus olivaceus (replaced by C. albifrons in southern Vene- zuela), Sciurus aestuans (replaced by S. gilvigularis in southern Venezuela), Dasyprocta leporina (replaced by D. fuliginosa in southern Venezuela), Myoprocta acouchy (replaced by M. pratti in southern Venezuela), and a small-toothed form of Mesomys with distinctive mtDNA sequences (replaced by a large-toothed congener with divergent mitochondrial genes in southern Venezuela). This geographic segregation of distinct rainforest mammal faunas in the Guiana subregion strikingly resembles the avifaunal patterns previously mapped by Cracraft (1985: ®g. 3). In particular, his concept of an ``Im- eri'' avifauna that extends from northwestern Amazonia into southern Venezuela is consistent with our interpretation of the southern 206 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 67 Taxa with Overlapping Geographic Distributions that De®ne a Guianan Center of Mammalian Endemism

Venezuelan mammal fauna as essentially invites yet wider organismal comparisons to non-Guianan.31 Likewise, his ``Guyanan'' test the generality of these results. center of avian endemism appears to be geo- Only a few relevant compilations of dis- graphically identical with our similarly tributional data are available for other higher named mammalian center. Such congruence taxa, but several show similar geographic patterns. In particular, mapped range limits 31 Few mammals can be con®dently identi®ed as Imeri endemics (sensu Cracraft, 1985) based on current taxo- for rainforest lizards (Dixon, 1979: ®g. 9:8), nomic revisions. Although possible examples include snakes (Dixon, 1979: ®g. 9:9), and Lecythi- Saguinus inustus (mapped by Hershkovitz, 1977: ®g. daceae (Mori, 1991: ®g. 1) suggest that most X.15), Aotus trivirgatus (see Hershkovitz, 1983: ®g. 2), Guianan endemics in these groups likewise and Cacajao melanocephalus (see Hershkovitz, 1987b), most mammals that occur in the Imeri region seem to do not occur west of the RõÂo CaronõÂ and the be widespread in other parts of western Amazonia. Rio Branco. Possibly, the broad lower reach- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 207

es of both riversÐtogether with the mostly (Urosciurus), Scolomys, and Dactylomys is savanna-covered highlands from which they likewise uniquely Guianan. ariseÐhave been historically effective bar- Of course, mammalian faunal composition riers to faunal and ¯oral exchange between is not constant within the Guianan center. For the rainforested Imeri and Guyana lowlands. example, some species that have been re- Alternatively (or additionally), these rivers corded near Kartabo (e.g., Neacomys gui- might approximate the geographic limits of anae; table 19, footnote b) or Manaus (Sa- some distinctive combination of soils and cli- guinus bicolor; Hershkovitz, 1977: ®g. X15) mate to which the endemic Guianan rainfo- are not known to occur in French Guiana. rest biota is uniquely adapted. Whatever his- Similarly, a few species that we collected at torical and/or contemporary-ecological fac- Paracou are currently known only from ad- tors might explain this phenomenon, how- ditional sites in French Guiana, AmapaÂ, or ever, it seems clear that the Guianan center eastern Surinam (Molossus barnesi, Neaco- of mammalian endemism is part of a general mys dubosti, Neusticomys oyapocki). Such pattern of biotic differentiation shared with narrowly distributed taxa suggest the exis- other sympatric groups of rainforest organ- tence of subcenters of mammalian ende- isms. mism, perhaps corresponding in location to The Paracou mammal fauna is a typically several of the ``refugia'' that Prance (1982: Guianan assemblage that includes at least 14 ®g. 11.9) postulated to explain distributional of the 17 endemics listed in table 67 together phenomena in the Guianan rainforest ¯ora. with many nonendemic but typically Guian- In particular, Prance's ``East Guianan'' en- an species (e.g., Philander opossum, Ametri- demicsÐrainforest plants with small ranges da centurio, Sciurillus pusillus, Dasyprocta centered on French Guiana and AmapaÂ (e.g., leporina). Although we recorded a few taxa Corythophora amapaensis; see Mori and at Paracou that were previously unknown as Prance, 1987: ®g. VI-1)Ðmay represent el- elements of the Guianan fauna (Hyladelphys ements of the same narrowly endemic biota kalinowskii, Saccopteryx gymnura, Micro- that we sampled at Paracou and share the nycteris homezi, M. schmidtorum), these are same history of geographic isolation and/or more plausibly interpreted as widespread but local adaptation. Much future collecting, elusive species than as biogeographic anom- however, will be required to convincingly alies. By contrast, most of the other mam- document geographic range limits for East mals we recorded at Paracou are known from Guianan endemic mammals, which do not numerous additional Guianan localities from appear to account for more than a small frac- Imataca or Kartabo to Manaus and the Serra tion (about 2%) of the Paracou fauna. do Navio. The biogeographic character of the Para- COMMUNITY STRUCTURE cou fauna is equally apparent, however, in our failure to record many taxa that are wide- Using the information previously summa- spread in western and/or southeastern Ama- rized in our separate guild analyses of bats zonia. In fact, the shared absence of such and nonvolant species, we classi®ed all Par- species as Glossophaga comissarisi, Carollia acou mammals by diel activity (nocturnal, di- castanea, Rhinophylla ®sherae, Enchisthenes urnal, both), substrate use (aerial, arboreal, hartii, Platyrrhinus infuscus, Sphaeronycter- terrestrial, semiaquatic), and trophic role is toxophyllum, Sturnira magna, and Urod- (primary consumers, secondary consumers, erma magnirostrum more readily character- omnivores) to examine basic aspects of com- izes Guianan bat inventories (and accounts munity-wide patterns of resource use (table for their cohesion in faunal cluster analyses) 68). Obviously, a more complete ecological than does the shared presence of such rare representation of the entire fauna could be endemics as Tonatia schulzi and Lasiurus obtained by cross-classifying species using atratus. Among nonvolant Amazonian in- the same criteria (e.g., to count diurnal-ar- ventories, the shared absence of Callithrix, boreal primary consumers, nocturnal-terres- Cebuella, Cacajao, Callicebus, Lagothrix, trial secondary consumers, etc.) and by tak- Bassaricyon, Microsciurus, giant squirrels ing other physiologically signi®cant traits 208 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

TABLE 68 TABLE 69 Summary Ecobehavioral Traits Comparison of Trophic Structure between of Paracou Mammals Rainforest Mammal Communities at Two Amazonian Localitiesa

1973; BourlieÁre, 1989). The Paracou fauna itself may be undersampled for arboreal species, however, which comprise a somewhat smaller proportion of the fauna at this site than at other Neotropical rainforest localities previously analyzed for substrate use (e.g., Kay and Madden, 1997: table 30.4). Based on the list of additional nonvolant species that might yet be recorded in our study area (appendix 1), future inventory work could plausibly bring the numbers of arboreal and terrestrial species in the local community into closer conformance with ratios observed in other Neotropical faunas. Apparently, the only distinctive community-wide aspect of resource use at Paracou concerns trophic structure. Whereas most (such as body weight) into account. How- species of mammals in both New World and ever, faunal analyses based on such compre- Old World tropical rainforests are said to be hensive treatments are dauntingly complex primary consumers (feeding primarily on (Skalli and Dubost, 1986; Dubost, 1987) and fruits, seeds, nectar, leaves, or other plant tis- beyond the scope of this report. For many sues; BourlieÁre, 1973, 1989), species of sec- ecological researchers, the appropriate units ondary consumers (faunivores) outnumber of community analysis represent biomass primary consumers by a substantial margin (e.g., kg/km2; Eisenberg and Thorington, at Paracou. Contingency tests suggest that 1973; Janson and Emmons, 1990; Peres, this difference is highly signi®cant for some 1999) rather than species, but taxonomic di- pairwise comparisons with other well-known versity (not energy ¯ow) is the subject of in- Amazonian faunas. At Manu, for example, terest in the following discussion. the numbers of primary and secondary con- The Paracou fauna does not appear to be sumers are almost inversely proportional to unusual with respect to patterns of either diel those observed at Paracou (table 69), and it activity or substrate use if allowance is made therefore seems appropriate to consider al- for the undersampling of bats in most Neo- ternative hypotheses that might account for tropical rainforest inventories on which pre- the preponderance of secondary consumers vious conjectures about whole-community in our study. structure have been based (e.g., by Fleming, A partial explanation clearly involves sam- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 209 pling artifacts. Primary consumers constitute effort at Paracou (®g. 99), trophic compari- most of the mammalian biomass in all Neo- sons with less intensively worked sites are tropical rainforest communities studied to obviously problematic. date (Eisenberg and Thorington, 1973; Jan- Indeed, future ®eldwork at Manu will son and Emmons, 1990; Peres, 1999), where probably add more secondary than primary frugivores and granivores in particular tend consumers to the species list from that site to be larger-bodied and/or to maintain higher because 33 out of the 48 additional species population densities than most faunivores. In that could still be expected there (Voss and addition, some standard inventory methods Emmons, 1996: appendix 10) are faunivores. are known to produce trophically biased However, at least part of the difference in samples of rainforest mammal communities; trophic structure between the mammal com- mistnets, for example, are primarily effective munities at Manu and Paracou is not artifac- for capturing frugivorous bats (Fleming et tual. Many taxa of primary consumers that al., 1972; LaVal and Fitch, 1977; Kalko et occur at Manu are widespread in western al., 1996; Simmons and Voss, 1998), and Amazonia but have no known ecological conventional traps are most effective for cap- equivalents at Paracou or in other core- turing frugivorous-granivorous rodents (Voss Guianan faunas (e.g., Caluromysiops, Cal- and Emmons, 1996; this report). For these lithrix, Aotus, Callicebus, Lagothrix, Bassar- reasons, short-term and/or methodologically icyon, Microsciurus, giant squirrels [Uros- limited inventories probably tend to overes- ciurus], Dinomys, Dactylomys, Sylvilagus), timate the ratio of primary to secondary con- and several taxa of primary consumers com- sumers in local faunas. Given the inverse mon to both inventories are consistently correlation between known relative primary more speciose at western Amazonian than at consumer diversity and cumulative sampling Guianan localities (e.g., Carollia, Platyrrhin-

Fig. 99. Percent primary consumers in the known mammal fauna at Paracou (excluding species recorded only from interviews or collections made by previous researchers) on each day of our inventory from 1991 to 1994. From initially high values early in our ®eldwork, the proportion of the known fauna represented by primary consumers declined almost monotonically throughout the last half of our ®eldwork to a ®nal value of 39%. 210 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 us, Uroderma, Saguinus, Proechimys). By cal lowlands can hardly be overemphasized. contrast, no taxon of primary consumers in Estimates of both local richness and faunal the Guianan fauna seems to lack an ecolog- complementarity (endemism) rely crucially ical equivalent in western Amazonia, and no on the ability of investigators to distinguish taxon of primary consumers common to both taxa, an ability that is compromised by our faunas is known to be consistently more spe- currently inadequate knowledge of species ciose at Guianan than at western Amazonian limits in many Neotropical clades. Just how localities. Because secondary consumers ap- much remains to be learned is suggested by pear to be about equally diverse in both Am- the taxonomic results of this study, wherein azonian subregions, at least part of the ob- we applied the most basic of all species cri- served trophic divergence between the Par- teria (morphological diagnosability) to assign acou and Manu communities seems to re¯ect names to voucher material collected at Par- a biogeographic gradient that principally af- acou. fects the species richness of primary consum- Based on character differences described ers.32 or referenced in Simmons and Voss (1998) and this report, the following 23 species SUGGESTIONS FOR FUTURE WORK (newly described or resurrected from syn- Our previous report on the Paracou bat onymy) should be recognized as valid: Mar- fauna explained the importance for future mosa quichua, Marmosops bishopi, Marmo- rainforest mammal diversity research of (1) sops juninensis, Marmosops pinheiroi, Mon- improving inventory ef®ciency, (2) establish- odelphis glirina, Monodelphis palliolata, ing standards for reporting inventory data, Centronycteris centralis, Peropteryx trinita- and (3) using quantitative methods for com- tis, Micronycteris brosseti, Micronycteris paring inventory results from different sites. homezi, Micronycteris microtis, Mimon coz- Although illustrated with chiropteran exam- umelae, Eptesicus andinus, Eptesicus chiri- ples, all of the recommendations made in that quinus, Molossops paranus, Molossus bar- report are likewise broadly applicable to non- nesi, Saguinus niger, Neacomys dubosti, volant mammalian surveys. Rather than re- Neacomys paracou, Nectomys melanius, iterate such essentially methodological points Oecomys auyantepui, Coendou melanurus, below, the following topics concern oppor- and Mesomys ferrugineus. Of course, we tunities for advancing the biogeographic syn- were not the ®rst to recognize many of these thesis outlined in our general discussion of species as valid. Most of the resurrected taxa the Paracou fauna. in this list were originally named as full species and were long recognized as such until synonymized (often with little or no explicit REVISIONARY TAXONOMY justi®cation) by uncritical advocates of the The importance of continued revisionary ``polytypic'' species concept. More recently, taxonomic research for mammalian biogeo- Handley (1976), Brosset and Charles-Domi- graphic studies in the rainforested Neotropi- nique (1990), and other cited authors antici- pated some of our taxonomic conclusions. 32 Most of the data on which this conclusion is based However, none of these species were recog- were summarized by Voss and Emmons (1996), who did not, however, consider their implications for community nized as valid by Wilson and Reeder (1993), trophic structure. Our statement about the absence of whose checklist serves as an adequate sum- Guianan primary consumers lacking ecological equiva- mary of the prevailing taxonomic consensus lents in western Amazonia acknowledges the dietary at the time we began our inventory work. By similarity between Chiropotes and Cacajao (following contrast, our results suggest that only two Ayres, 1989), and our assessment of secondary consumer diversity as approximately equal in Guianan and west- species that Wilson and Reeder recognized as ern Amazonian habitats is based on the fact that several valid should be synonymized: Choeroniscus faunivorous bat taxa (e.g., emballonurids, phyllostomi- intermedius (with C. minor), and Nectomys nes, molossids) have yet to be adequately sampled at parvipes (with N. melanius). any western Amazonian site. Voss and Emmons (1996: table 13) listed Caluromys (a marsupial primary con- The net change in taxonomic diversity re- sumer) as maximally speciose in the Guianan subregion, sulting from research with Paracou voucher but most core-Guianan faunas have only C. philander. material, 21 species, represents an increase 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 211 of almost 2% over the total Neotropical ditional sites in AmapaÂ. Special attention mammal fauna recognized prior to this study should be devoted to the potential existence (1145 species according to Patterson, 1994). of a distinctive mammalian fauna in the Am- Because Guianan mammals are relatively azonian white-water ¯oodplain vis-aÂ-vis the well known as a consequence of long acces- Guianan terra ®rme in each of these inter¯u- sibility to European naturalists, similarly de- vial zones. tailed analyses of voucher material collected Faunal surveys along the south bank of the in more remote areas (e.g., in western Ama- Amazon are equally important to test the hy- zonia; Patton et al., 2000) seem certain to pothesis that putative Guianan endemics are result in substantially larger diversity incre- not widely distributed there. Indeed, the ef- ments. ®cacy of the lower Amazon as a barrier to Predictably, the improved taxonomic res- faunal dispersion should be tested by repli- olution resulting from such research will cated transects through vaÂrzea and terra ®rme contribute to a much-needed assessment of habitats on opposite banks (after Patton et al., biogeographic congruence between mam- 2000). Such transects should devote compa- malian and other biotic distributional data. rable effort to sampling bat and nonvolant The close correspondence between spatial mammalian faunas in order to test the con- patterns of mammalian and avian endemism jecture that riverine barriers are not equally in the Guiana subregion of Amazonia, for ex- effective for these groups. ample, is documented herein by numerous Species distributions in the transitional distributional data previously obscured by area between the western Amazonian (``Im- synonymy. Based on our preliminary assess- eri'') fauna of southern Venezuela and the ment of many extralimital alpha-taxonomic Guianan fauna of eastern Venezuela and problems, we anticipate that future research western Guyana likewise merit targeted ®eld- will provide strong mammalian support for work and careful mapping. Of special inter- many of the endemic patterns ®rst remarked est are the geographic boundaries between and clearly delimited for Amazonian birds by such replacing-species pairs as Philander an- Cracraft (1985). Many of the molecular re- dersoni/P. opossum, Bradypus variegatus/B. sults summarized by Patton et al. (1997, tridactylus, Ateles belzebuth/A. paniscus, Ce- 2000) tend to support the same conclusion, bus albifrons/C. olivaceus, Dasyprocta fulig- in addition to suggesting other patterns of en- inosa/D. leporina, and Myoprocta pratti/M. demism that may exist within recognized acouchy. Because most faunal transitions in avian centers. Plausibly, future revisionary the Amazonian biota are currently thought to research incorporating both morphological coincide abruptly with very large riverine and molecular data will afford the strongest barriers, the apparently gradual (or stepwise) basis for assessing the generality of biogeo- loss of western Amazonian taxa and their re- graphic congruence across major clades of placement with Guianan forms across hun- Amazonian organisms. dreds of kilometers of more-or-less continuous Venezuelan lowland forest (Huber and MAPPING GUIANAN ENDEMICITY Alarcon, 1982) is of exceptional biogeographic interest. Much additional collecting is needed before the geographic range limits of mammals HISTORICAL CONNECTIONS WITH OTHER identi®ed herein as Guianan endemics can be AREAS OF ENDEMISM mapped with greater con®dence. Notably useful would be thoroughly vouchered faunal Phylogenetic analyses of mammalian surveys from northern Brazil, especially (1) clades that include endemic Guianan taxa in Amazonas and Roraima states between the will contribute to an historical assessment of Rio Negro and the Rio Branco (where most the isolation and assembly of this well- Guianan endemics are apparently absent), (2) marked center of endemism. Maximally in- in ParaÂ state between the Trombetas and the formative analyses with this objective should Jari (an area from which no large faunal lists include morphological and/or molecular ter- are apparently available), and (3) from ad- minals from as many areas of Amazonian en- 212 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263 demism as possible in order to clearly distin- such will prove to be the case. If so, then guish among alternative biogeographic sce- appropriate explanations for these commu- narios. Minimally, taxa inhabiting each of the nity characteristics should be based on his- Napo/Imeri, Inambari, and RondoÃnia/ParaÂ/ torical and/or ecological circumstances com- BeleÂm centers (after Cracraft, 1985) would mon to the Guianan fauna as a whole rather appear necessary for any analysis to be in- than on the environmental peculiarities of formative about the historical biogeography any particular study site. of Guianan endemics. Certain paleoenvironmental reconstruc- Unfortunately, the few available phyloge- tions, for example, have suggested that netic datasets that approximate these condi- Guianan rainforests were simultaneously iso- tions give contradictory biogeographic indi- lated and reduced in area during the Pleis- cations. The most impressive, Canavez et tocene: by an enormous freshwater lake (e.g., al.`s (1999) analysis of ␤2-microglobulin se- Frailey et al., 1988: ®g. 8), or by savannas quences from Saguinus species, suggests that and other arid vegetation (e.g., Clapperton, the sister taxon to the endemic Guianan lin- 1993: ®g. 8). Both scenarios lack convincing eage S. bicolor ϩ S. midas is S. niger from independent support (Colinvaux, 1996), but southeastern Amazonia (ParaÂ/BeleÂm), a re- either might explain the lower mammalian sult that implies relatively recent vicariance species richness of modern Guianan rainfo- or dispersal across the lower Amazon. A rests (as residual insular effects). Neither ar- similar scenario is implied by several of the eal reduction nor isolation, however, plausi- cytochrome-b sequence analyses reported by bly accounts for the conspicuous trophic bias Patton et al. (2000: e.g., ®gs. 171A, 171C), observed in our comparisons of Guianan but sister-group relationships between Guian- with western Amazonian mammal commu- an and western Amazonian haplotypes in nities. other analyses by the same researchers (op. The ancient soils produced by in-situ cit.: e.g., ®gs. 172B, 172D) imply relatively chemical weathering of the Guiana Shield more recent faunal connections across the are routinely characterized as exceptionally Rio Negro. Although it is crucial to draw the nutrient-poor (Jordan, 1985; Sanchez, 1989), distinction between haplotypes and taxa in especially by comparison with the younger these mitochondrial-gene studies, the prima alluvium (mechanically weathered from the facie implication of such incongruence is that Andes) that forms the mineral substrate not all Guianan taxa share the same biogeo- throughout much of western Amazonia. Poor graphic history. Whether any one historical soils have been hypothesized to negatively pattern better characterizes the biogeography affect the species richness of rainforest mam- of Guianan endemics than another will re- mals by constraining plant productivity (Em- quire much more phylogenetic research to mons, 1984; Gentry and Emmons, 1987), convincingly establish. and it seems intuitively obvious that plant productivity should more directly impact pri- CAUSAL EXPLANATIONS FOR COMMUNITY mary consumer than faunivore diversity. Ge- COMPOSITION ology therefore offers a super®cially plausible explanation for two of the most distinc- The hypothesis that the mammalian com- tive contrasts between Guianan and western munity at Paracou is broadly representative Amazonian mammal faunas. of the Guianan fauna with respect to species The Guianan center of endemism is not, richness and trophic structure should be test- however, coextensive with the Guiana ed with methodologically similar inventory Shield, which extends hundreds of kilome- ®eldwork at other Guianan sites. Given the ters west of the Rio Negro to the base of the essential taxonomic homogeneity of this cen- Andes (Gibbs and Barron, 1993), so geology ter of endemism, however, together with the cannot be the whole story. Nor do the cli- marked tendency for inventory results to matic maps that we have examined suggest converge on geographic expectations with any sharp change in the amount or season- suf®cient sampling effort (Voss and Em- ality of annual rainfall that coincides with the mons, 1996: ®g. 24), it seems probable that hypothesized limits of the Guianan center of 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 213 endemism. Although multifactorial models tance in locating ancient types and other his- of plant productivity might reveal some eco- torically important specimens in Paris, also logically signi®cant combination of edaphic merit special mention here. Paula Jenkins and climatic variables that broadly character- generously hosted several visits to the Nat- izes Guianan versus other Amazonian land- ural History Museum, patiently examined scapes, it does not seem plausible that eco- and measured type material at our request, logical gradients alone could result in sharply and provided other valuable information de®ned areas of endemism in the absence of about specimens and archival material in historical barriers to dispersal. Given that London. FrancËois Catze¯is of the Institut des gradients exist, however, and that taxa differ Sciences de l'EÂ volution (Montpellier) sent us in their ecological requirements, it would be much useful material collected in the course interesting to explore the possibility (with of his own research in French Guiana, with- computer simulations) that semipermeable out which our samples of several taxa would dispersal barriers might attract species have been impoverished. In addition, we boundaries over many generations in much thank our colleagues at the following muse- the same way that they appear to attract step ums (spelled out on p. 17) for loans and other clines in gene frequencies (Endler, 1977). Al- technical assistance: CM (D. A. Schlitter, S. ternative scenarios for the historical assem- B. McLaren, and J. R. Wible), EBRG (F. Bis- bly of Amazonian biotas have tended to view bal), FMNH (B. D. Patterson), MHNG (F. J. major rivers as effective dispersal barriers Baud), MHNLS (D. Lew), MUSM (V. Pa- (e.g., Wallace, 1852; Cracraft, 1985; Ayres checo), MVZ (J. L. Patton), NMW (K. Bauer and Clutton-Brock, 1992), or as mere hydro- and F. Spitzenberger), RMNH (C. Smeenk), logical boundaries between ecologically dis- ROM (M. D. Engstrom and B. K. Lim), tinctive terrains (Tuomisto and Ruokolainen, USNM (M. D. Carleton), and ZMB (R. An- 1997). The nature of faunal transitions be- germann). tween mammalian centers of endemism, We are especially mindful of the gener- however, suggests that a more interactive osity of colleagues who shared their unpub- conception of biotic assembly might better ®t lished data, notes, manuscripts, or advice the facts at hand. about numerous taxonomic and distributional problems. In particular, we consulted the late ACKNOWLEDGMENTS C. O. Handley, Jr., concerning species limits in Monodelphis; M. D. Carleton and G. G. In addition to the persons and institutions Musser concerning character variation and previously acknowledged (in part 1) for their taxonomy in Oecomys and Oryzomys;J.L. support of our ®eld and museum research on Patton concerning numerous problems in Paracou mammals, we extend our apprecia- Marmosops, Philander, Neacomys, and tion here to those who made special contri- Proechimys; M. D. Engstrom, V. A. Funk, butions to the nonvolant mammal inventory. and B. K. Lim for assistance with Guyanese In particular, we thank Roland W. Kays, who map coordinates; and F. Catze¯is concerning helped construct pitfall traplines and arboreal many details of taxonomic records from platform traps, monitored and recorded pit- French Guiana. For advice on nomenclatural fall captures, hunted, prepared specimens, issues, we thank A. L. Gardner and A. Gen- and contributed in many other ways to the try. Of course, none of these individuals are success of our 1993 ®eld season. We are also responsible for any errors of fact or interpre- grateful to Louise H. Emmons for her pro- tation that might remain in our systematic ac- ductive visit to our ®eld site in 1994, and to counts. Andrea L. Pef¯ey for at least one spectacular Critically reading a manuscript the size of nonvolant mammal sighting that would oth- ours is no small favor, and we are corre- erwise have escaped our attention. spondingly indebted to our reviewers (M. D. The heroic efforts of Laurent Granjon and Carleton, F. Catze¯is, L. H. Emmons, A. L. Michel Tranier to arrange many crucial loans Gardner, and J. L. Patton) for their many of marsupials and rodents from the MuseÂum helpful comments. F. Catze¯is generously d'Histoire Naturelle, and their expert assis- translated our abstract. 214 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

As always, we are grateful for the logis- 1915b. Notes on American deer of the genus tical support of the AMNH Department of Mazama. Bull. Am. Mus. Nat. Hist. 34: Mammalogy staffÐespecially Pat Brunauer, 521±553. Neil Duncan, Brian Kraatz, and Bob Rand- 1916a. New mammals collected on the Roo- allÐall of whom contributed directly or in- sevelt Brazilian expedition. Bull. Am. directly to the production of this report. Pat Mus. Nat. Hist. 35: 523±530. 1916b. Mammals collected on the Roosevelt Wynne responded to the usual pressure with Brazilian Expedition, with ®eld notes her usual grace, producing all of the line art by Leo E. Miller. Bull. Am. Mus. Nat. for this report with characteristic skill and Hist. 35: 559±610. enthusiasm. Peter Goldberg expertly printed 1919. Notes on the synonymy and nomencla- our ®eld photographs and produced most of ture of the smaller spotted cats of trop- the color and black-and-white images of ical America. Bull. Am. Mus. Nat. Hist. specimens in our systematic accounts. An- 41: 341±419. gela Klaus patiently helped us obtain the Allen, J. A., and F. M. Chapman SEM images in ®gures 39 and 43. 1893. On a collection of mammals from the island of Trinidad, with descriptions of REFERENCES new species. Bull. Am. Mus. Nat. Hist. 5: 203±234. Aguilera, M., and M. Corti Anderson, S. 1994. Craniometric differentiation and chro- 1997. Mammals of Bolivia, taxonomy and mosomal speciation of the genus Proe- distribution. Bull. Am. Mus. Nat. Hist. chimys. Z. SaÈugetierk. 59: 366±377. 231: 652 pp. Aguilera, M., O. A. Reig, and A. PeÂrez-Zapata Anthony, H. E. 1995. G- and C-banding karyotypes of spiny 1921a. Mammals collected by William Beebe rats (Proechimys) of Venezuela. Rev. at the British Guiana tropical research Chil. Hist. Nat. 68: 185±196. station. Zoologica 3: 265±286. Alberico, M., V. Rojas-DõÂaz, and J. Gregorio M. 1921b. New mammals from British Guiana 1999. Aporte sobre la taxonomõÂa y distribu- and Colombia. Am. Mus. Novitates 19: cioÂn de los puercoespines (Rodentia: 7 pp. Erethizontidae) en Colombia. Rev. 1926. Two new rodents from Bolivia. Am. Acad. Colomb. Cienc. 23(supl.): 595± Mus. Novitates 239: 1±3. 612. Anthony, H. E., and G.H.H. Tate Allen, J. A. 1935. Notes on South American Mammalia. 1900. Descriptions of new American marsu- No. 1. Sciurillus. Am. Mus. Novitates pials. Bull. Am. Mus. Nat. Hist. 13: 780: 13 pp. 191±199. Archer, M. 1901. A preliminary study of the North 1976a. The basicranial region of marsupicar- American opossums of the genus Di- delphis. Bull. Am. Mus. Nat. Hist. 14: nivores (Marsupialia), interrelation- 149±195. ships of carnivorous marsupials, and 1902. A preliminary study of the South af®nities of the insectivorous marsupial American opossums of the genus peramelids. Zool. J. Linn. Soc. 59: Didelphis. Bull. Am. Mus. Nat. Hist. 217±322. 16: 249±279. 1976b. The dasyurid dentition and its relation- 1904. New mammals from Venezuela and ships to that of didelphids, thylacinids, Colombia. Bull. Am. Mus. Nat. Hist. borhyaenids (Marsupicarnivora) and 20: 327±335. peramelids. Aust. J. Zool. Suppl. Ser. 1912. Mammals from western Colombia. 39: 34 pp. Bull. Am. Mus. Nat. Hist. 31: 71±95. Ascorra, C. F., D. L. Gorchov, and F. Cornejo 1913a. New mammals from Colombia and Ec- 1993. The bats from Jenaro Herrera, Loreto, uador. Bull. Am. Mus. Nat. Hist. 32: Peru. Mammalia 57: 533±552. 469±484. Ascorra, C. F., S. Solari T., and D. E. Wilson 1913b. New South American Muridae. Bull. 1996. Diversidad y ecologõÂa de los quiroÂpter- Am. Mus. Nat. Hist. 32: 597±604. os en Pakitza. In D. E. Wilson and A. 1915a. Review of the South American Sciuri- Sandoval (eds.), Manu: the biodiversity dae. Bull. Am. Mus. Nat. Hist. 34: of southeastern Peru: 593±612. Lima: 147±309. Editorial Horizonte. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 215

Atramentowicz, M. Linn. Soc. London (Zool.) 9: 83±217 ϩ 1988. La frugivorie opportuniste de trois mar- pls. 5±7. supiaux didelphideÂs de Guyane. Rev. Bisbal, F. J. Ecol. (Terre Vie) 43: 47±57. 1991. DistribucioÂn y taxonomõÂa del venado Augustiny, G. matacaÂn (Mazama sp. ) en Venezuela. 1942. Die Schwimmanpassung von Chironec- Acta Biol. Venez. 13: 89±104. tes. Z. f. Morph. Okol. Tiere Berlin 39: Bodini, R., and R. PeÂrez-HernaÂndez 276±319. 1987. Distribution of the species and subspe- AÂ vila-Pires, F. D. de cies of cebids in Venezuela. Fieldiana 1958. ContribucËaÄo ao conhecimento dos cer- Zool. (new ser.) 39: 231±244. võÂdos Sul-Americanos (RuminantiaÐ Bodmer, R. E. Cervidae). An. Acad. Bras. Cienc. Rio 1990. Responses of ungulates to seasonal inde Janeiro 30: 585±598. undations in the Amazonian ¯oodplain. 1959. As formas Sul-Americanas do ``Veado- J. Trop. Ecol. 6: 191±201. viraÂ''. An. Acad. Bras. Cienc. 31: 547± Boeseman, M. 556. 1970. The vicissitudes and dispersal of Al- 1964. MamõÂferos colecionados na regiaÄo do bertus Seba's zoological specimens. Rio Negro (Amazonas, Brasil). Bol. Zool. Meded. (Leiden) 44: 177±206 ϩ Mus. Paraense EmõÂlio Goeldi (nova 4 pls. ser.), Zool. 42: 1±23. Bonaccorso, F. J. Ayres, J. M. 1979. Foraging and reproductive ecology in a 1989. Comparative feeding ecology of the Panamanian bat community. Bull. Flor- Uakari and Bearded Saki, Cacajao and ida State Mus., Biol. Sci. 24: 359±408. Chiropotes. J. Human Evol. 18: 697± Bonvincino, C. R., P. S. D'Andrea, R. Cerqueira, 716. and H. N. SeuaÂnez Ayres, J. M., and T. H. Clutton-Brock 1996. The chromosomes of Nectomys (Ro- 1992. River boundaries and species range size dentia, Cricetidae) with 2n ϭ 52, 2n ϭ in Amazonian primates. Am. Nat. 140: 56, and interspeci®c hybrids (2n ϭ 54). 531±537. Cytogenet. Cell Genet. 73: 190±193. Azara, F. de BourlieÁre, F. 1801. Essais sur l'histoire naturelle des quad- 1973. The comparative ecology of rain forest rupeÁdes de la Province du Paraguay, 2 mammals in Africa and tropical Amer- vols. Paris: Charles Pougens. ica: some introductory remarks. In B. Baker, R. J., B. F. Koop, and M. W. Haiduk J. Meggers, E. S. Ayensu, and W. D. 1983. Resolving systematic relationships with Duckworth (eds.), Tropical forest eco- G-bands: a study of ®ve genera of systems in Africa and South America: South American cricetine rodents. Syst. a comparative review: 279±292. Wash- Zool. 32: 403±416. ington, DC: Smithson. Inst. Press. BarreÁre, P. 1989. Mammalian species richness in tropical 1741. Essai sur l'histoire naturelle de la rainforests. In M. L. Harmelin-Vivien France EÂ quinoxiale [etc. ]. Paris: Piget. and F. BourlieÁre (eds.), Vertebrates in Barros, M. A., O. A. Reig, and A. Perez-Zapata complex tropical systems: 153±168. 1992. Cytogenetics and karyosystematics of New York: Springer-Verlag. South American oryzomyine rodents Boxer, C. R. (Cricetidae: Sigmodontinae). IV. Kar- 1973. The Dutch in Brazil, 1624±1654. Ham- yotypes of Venezuelan, Trinidadian, den, CN: Archon Books. and Argentinian water rats of the genus Brants, A. Nectomys. Cytogenet. Cell Genet. 59: 1827. Het geslacht der muizen door Linnaeus 34±38. opgesteld, volgens de tegenswoordige Beebe, W. toestand der wetenschap in familien, 1925. Studies of a tropical jungle; one-quarter geslachten en soorten verdeeld. Berlyn of a square mile of jungle at Kartabo, [Berlin]: Gedrukt ter Akademische British Guiana. Zoologica 6: 4±193. Boekdrukkery. Bensley, B. A. Bresslau, E. 1903. On the evolution of the Australian Mar- 1920. The mammary apparatus of the Mam- supialia; with remarks on the relation- malia in the light of ontogenesis and ships of marsupials in general. Trans. phylogenesis. London: Methuen. 216 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Brisson, M. Cienc. Nat. ``Bernardino Rivadavia,'' 1756. Le reÁgne animal, diviseÂ en IX classes, Cienc. Zool. 4(1): 1±307. ou meÂthode contenant la division geÂ- 1961. CataÂlogo de los mamõÂferos de AmeÂrica neÂrale des animaux en IX classes, & la del Sur [part 2]. Rev. Mus. Argentino division particulieÁre des deux pre- Cienc. Nat. ``Bernardino Rivadavia,'' mieÁres classes, scËavoir de celle des Cienc. Zool. 4(2): 309±732. QuadrupeÁdes & de celle des CetaceÂes, Canavez, F. C., M. A. M. Moreira, F. Simon, P. en ordres, sections, genres & espeÂces Parham, and H. N. SeuaÂnez [etc. ]. Paris: Cl. Jean-Baptiste Bauche. 1999. Phylogenetic relationships of the Cal- 1762. Regnum animale in classes IX distri- litrichinae (Platyrrhini, Primates) based butum, sive synopsis methodica sistens on ␤2-microglobulin DNA sequences. generalem animalium distributionem in Am. J. Primatol. 48: 225±236. classes IX, & duarum primarum clas- Carleton, M. D., and G. G. Musser 1989. Systematic studies of oryzomyine ro- sium, Quadrupedum scilicet & cetaceo- dents (Muridae, Sigmodontinae): a syn- rum, particularem divisionem in ordi- opsis of Microryzomys. Bull. Am. Mus. nes, sectiones, genera & species. Lug- Nat. Hist. 191: 83 pp. duni Batavorum [Leiden]: Theodorum 1995. Systematic studies of oryzomyine ro- Haak. dents (Muridae: Sigmodontinae): de®- Brosset, A., and P. Charles-Dominique nition and distribution of Oligoryzomys 1990. The bats from French Guiana: a taxo- vegetus (Bangs, 1902). Proc. Biol. Soc. nomic, faunistic, and ecological ap- Washington 108: 338±369. proach. Mammalia 54: 509±560. Carter, S. K., and F. C. W. Rosas Brown, J. C. 1997. Biology and conservation of the giant 1971. The description of mammals 1. The ex- otter Pteronura brasiliensis. Mammal ternal characters of the head. Mammal. Rev. 27: 1±26. Rev. 1: 151±168. Carvalho, C. T. de Buffon, G. L. de 1962. Lista preliminar dos mamõÂferos do 1767. Histoire naturelle, geÂneÂrale et particu- AmapaÂ. Pap. Avul. Dept. Zool. Secr. lieÁre, avec la description du cabinet du Agric. SaÄo Paulo 15: 283±297. Roi [1st ed. ], vol. 15. Paris: Imprimer- Carvalho, C. T. de, and A. J. Toccheton ie Royale. 1969. MamõÂferos do nordeste do ParaÂ, Brazil. 1776. Histoire naturelle, geÂneÂrale et particu- Rev. Biol. Trop. 15: 215±226. lieÁre, servant de suite aÁ l'histoire des Catze¯is, F., and C. Steiner animaux quadrupeÁdes. SuppleÂment, 2000. Nouvelles donneÂes sur la morphologie vol. 3. Paris: Imprimerie Royale. compareÂe et la distribution des rats eÂpi- 1789. Histoire naturelle, geÂneÂrale et particuli- neux Proechimys cuvieri et P. cayen- eÁre, servant de suite aÁ l'histoire des ani- nensis (Echimyidae: Mammalia) en maux quadrupeÁdes. SuppleÂment, vol. 7. Guyane francËaise. Mammalia 64: 209± Paris: Imprimerie Royale. 220. Burnham, K. P., and W. S. Overton Catze¯is, F., C. Richard-Hansen, C. Fournier- Chambrillon, A. Lavergne, and J. VieÂ 1979. Robust estimation of population size 1997. BiomeÂtrie, reproduction et sympatrie when capture probabilities vary among chez Didelphis marsupialis et D. albi- animals. Ecology 60: 927±936. ventris en Guyane francËaise (Didelphi- Cabanis, J. dae: Marsupialia). Mammalia 61: 231± 1848. Saeugethiere. In R. Schomburgk (ed.), 243. Reisen in Britisch-Guiana in den Jahren Chao, A. 1840±1844. Dritter Theil [vol. 3], Ver- 1984. Nonparametric estimation of the num- such einer Fauna und Flora von Bri- ber of classes in a population. Scand. J. tisch-Guiana: 766±786. Leipzig: J. J. Stat. 11: 265±270. Weber. 1987. Estimating the population size for cap- Cabrera, A. ture-recapture data with unequal catch- 1919. Genera mammalium: Monotremata, ability. Biometrics 43: 783±791. Marsupialia. Madrid: Mus. Nac. Cienc. Charles-Dominique, P. Nat. 1993. EÂ cologie des mammifeÁres forestiers de 1958. CataÂlogo de los mamõÂferos de AmeÂrica Guyane francËaise: origines biogeÂogra- del Sur [part 1]. Rev. Mus. Argentino phiques et organisation du peuplement. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 217

In H. L. Raymond (ed.), Gestion de Cuvier, G. l'eÂcosysteme forestier et ameÂnagement 1798. Tableau eÂleÂmentaire de l'histoire natu- de l'espace reÂgional: 145±152. Cay- relle des animaux. Paris: Baudouin. enne: Sepanguy. Czernay, S. Charles-Dominique, P., M. Atramentowicz, M. 1987. Die Spiessehirsche und Pudus. Witten- Charles-Dominique, et al. berg Lutherstadt: A. Ziemsen. 1981. Les mammifeÁres frugivores arboricoles Decker, D. M. nocturnes d'une foreÃt Guyanaise: inter- 1991. Systematics of the coatis, genus Nasua relations plantes-animaux. Rev. Ecol. (Mammalia: Procyonidae). Proc. Biol. (Terre Vie) 35: 341±435. Soc. Washington 104: 370±386. Clapperton, C. M. de Queiroz, K., and D. A. Good 1993. Nature of environmental change in 1997. Phenetic clustering in biology: a cri- South America during the Last Glacial tique. Q. Rev. Biol. 72: 3±30. Maximum. Paleogeogr. Paleoclimatol. Desmarest, A. G. 1817. Echimys, Echimys [entry for]. In Nou- Paleoecol. 101: 189±208. veau dictionnaire d'histoire naturelle Colinvaux, P. A. appliqueÂe aux arts, aÁ l'agriculture, aÁ 1996. Quaternary environmental history and l'eÂconomie rurale et domestique, aÁla forest diversity in the Neotropics. In meÂdecine, etc. Tome X: 54±59. Paris: J.B.C. Jackson, A. F. Budd, and A. G. Deterville. Coates (eds.), Evolution and environ- Ditch®eld, A. D. ment in tropical America: 359±405. 1996. The phylogeography of Neotropical Chicago: Univ. Chicago Press. bats [Ph. D. diss., Univ. California at Colwell, R. K., and J. A. Coddington Berkeley]. Ann Arbor, MI: Univ. Mi- 1994. Estimating terrestrial biodiversity cro®lms. through extrapolation. Philos. Trans. R. Dixon, J. R. Soc. London B 345: 101±118. 1979. Origin and distribution of reptiles in ConcepcioÂn, J. L., and J. Molinari lowland tropical rainforests of South 1991. Sphiggurus vestitus pruinosus (Mam- America. In W. E. Duellman (ed.), The malia, Rodentia, Erethizontidae): the South American herpetofauna: its ori- karyotype and its phylogenetic impli- gin, evolution, and dispersal. Univ. cations, descriptive notes. Stud. Neo- Kansas Mus. Nat. Hist. Monogr. 7: trop. Fauna Environ. 4: 237±241. 217±240. Corti, M., and M. Aguilera Dubost, G. 1995. Allometry and chromosomal speciation 1987. Une analyse eÂcologique de deux faunes of the casiraguas Proechimys (Mam- de mammifeÁres forestiers tropicaux. malia, Rodentia). J. Zool. Syst. Evol. Mammalia 51: 415±436. Res. 33: 109±115. Dubost, G., and F. Petter Cracraft, J. 1978. Une espeÁce nouvelle de ``rat-peÃcheur'' 1985. Historical biogeography and patterns of de Guyane francËaise: Daptomys oya- differentiation within the South Amer- pocki sp. nov. (Rongeurs, Cricetidae). ican avifauna: areas of endemism. In P. Mammalia 42: 435±439. A. Buckley et al. (eds.), Neotropical or- Ducke, A., and G. A. Black 1953. Phytogeographic notes on the Brazilian nithology (Ornithol. Monogr. 36): 49± Amazon. An. Acad. Bras. Cienc. 25: 1± 84. Washington, DC: Am. Ornithol. 46. Union. Duplaix, N. 1994. Species diversity, biogeography, and 1980. Observations on the ecology and be- the evolution of biotas. Am. Zool. 34: havior of the giant river otter Pteronura 33±47. brasiliensis in Suriname. Rev. Ecol. Creighton, G. K. (Terre Vie) 34: 495±620. 1984. Systematic studies on opossums (Di- Eisenberg, J. F. delphidae) and rodents (Cricetidae) 1989. Mammals of the Neotropics, vol. 1. [Ph. D. diss., Univ. Michigan]. Ann Ar- The northern Neotropics. Chicago and bor, MI: Univ. Micro®lms. London: Univ. Chicago Press. Cruz Lima, E. da Eisenberg, J. F., and K. H. Redford 1945. Mammals of Amazonia. Vol. 1, Gen- 1999. Mammals of the Neotropics, vol. 3. eral introduction and primates. BeleÂm The central Neotropics. Chicago and do ParaÂ: Mus. Para. Emilio Goeldi. London: Univ. Chicago Press. 218 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Eisenberg, J. F., and R. W. Thorington, Jr. Fleming, T. H., E. T. Hooper, and D. E. Wilson 1973. A preliminary analysis of a Neotropical 1972. Three Central American bat communi- mammal fauna. Biotropica 5: 150±161. ties: structure, reproductive cycles, and Eisenberg, J. F., M. A. O'Connell, and P. V. Au- movement patterns. Ecology 53: 555± gust 569. 1979. Density, productivity, and distribution Flower, W. H. of mammals in two Venezuelan habi- 1867. On the development and succession of tats. In J. F. Eisenberg (ed.), Vertebrate the teeth in the Marsupialia. Philos. ecology in the northern Neotropics, pp. Trans. R. Soc. London 157: 631±641 ϩ 187±207. Washington, DC: Smithson. pls. 29, 30. Inst. Press. Fonseca, G.A.B. da, and K. H. Redford Ellerman, J. R. 1984. The mammals of IBGE's ecological re- 1941. The families and genera of living ro- serve, BrasõÂlia, and an analysis of the dents, vol. 2. Family Muridae. London: role of gallery forests in increasing di- British Mus. (Nat. Hist). versity. Rev. Brasil. Biol. 44: 517±523. Elliot, D. G. Forget, P.-M. 1912. A review of the Primates, vol. 1. Mon- 1991. Scatterhoarding of Astrocaryum para- ogr. Am. Mus. Nat. Hist 1: i±cxxvi, 1± maca by Proechimys in French Guiana: 317, i±xxxviii. comparison with Myoprocta exilis. Emmons, L. H. Trop. Ecol. 32: 155±167. 1984. Geographic variation in densities and 1996. Removal of seeds of Carapa procera diversities of non-¯ying mammals in (Meliaceae) by rodents and their fate in Amazonia. Biotropica 16: 210±222. rainforest in French Guiana. J. Trop. 1990. Neotropical rainforest mammals, 1st Ecol. 12: 751±761. ed. Chicago and London: Univ. Chi- 1997. Effect of microhabitat on seed fate and cago Press. seedling performance in two rodent- 1993a. Mammals. In T. A. Parker and A. B. dispersed tree species in rain forest in Forsyth (eds.), A biological assessment French Guiana. J. Ecol. 85: 693±703. of the Kanuku Mountain region of Forget, P.-M., F. Mercier, and F. Collinet southwestern Guyana (RAP Working 1999. Spatial patterns of two rodent-dispersed Papers 5): 28±31. Washington, DC: rain forest trees Carapa procera (Me- Conservation International. liaceae) and Vouacapoua americana 1993b. On the identity of Echimys didelphoi- (Caesalpiniaceae) at Paracou, French des Desmarest, 1817 (Mammalia: Ro- Guiana. J. Trop. Ecol. 15: 301±313. dentia: Echimyidae). Proc. Biol. Soc. Frailey, C. D., E. L. Lavina, A. Rancy, and J. P. Washington 106: 1±4. de Souza Filho 1997. Neotropical rainforest mammals, 2nd 1988. A proposed Pleistocene/Holocene lake ed. Chicago and London: Univ. Chi- in the Amazon Basin and its signi®- cago Press. cance to Amazonian geology and bio- Endler, J. A. geography. Acta Amazonica 18: 119± 1977. Geographic variation, speciation, and clines. Princeton, NJ: Princeton Univ. 143. Press. Freitas, S. R., D. A. de Moraes, R. T. Santori, and Engel, H. R. Cerqueira 1937. The life of Albert Seba. Sven. LinneÂ- 1997. Habitat preference and food use by Me- SaÈllsk. AÊ rsskr. 20: 75±100. tachirus nudicaudatus and Didelphis Erxleben, I. C. P. aurita (Didelphimorphia, Didelphidae) 1777. Systema regni animalis per classes, or- in a restinga forest at Rio de Janeiro. dines, genera, species, varietates, cum Rev. Brasil. Biol. 57: 93±98. synonymia et historia animalium. Clas- Gardner, A. L. sis I, Mammalia. Lipsiae [Leipzig]: Im- 1973. The systematics of the genus Didelphis pensis Weygandianis. (Marsupialia: Didelphidae) in North Fleming, T. H. and Middle America. Spec. Publ. Mus. 1971. Population ecology of three species of Texas Tech Univ. 4: 81 pp. Neotropical rodents. Univ. Michigan 1988. The mammals of Parque Nacional Ser- Mus. Zool. Misc. Publ. 143: 77 pp. ranõÂa de la Neblina, Territorio Federal 1973. Numbers of mammal species in North Amazonas, Venezuela. In C. Brewer- and Central American forest commu- Carias (ed.), Cerro de la Neblina, re- nities. Ecology 54: 555±563. sultados de la expedicioÂn 1983±1987: 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 219

695±765. Caracas: FundacioÂn para el George, T. K., S. A. Marques, M. de Vivo, L. C. Desarrolo de las Ciencias FõÂsicas, Ma- Branch, N. Gomes, and R. Rodrigues temaÂticas, y Naturales. 1988. Levantamento de mamõÂferos do Par- 1989. Two new mammals from southern Ven- naÐTapajoÂs. Brasil Florestal 63: 33± ezuela and comments on the af®nities 41. of the highland fauna of Cerro de la Gibbs, A. K., and C. N. Barron Neblina. In K. H. Redford and J. F. Ei- 1993. The geology of the Guiana Shield. New senberg (eds.) Advances in Neotropical York: Oxford Univ. Press. mammalogy, pp. 411±424. Gainesville: Glanz, W. E. Sandhill Crane Press. 1982. The terrestrial mammal fauna of Barro Colorado Island: censuses and long- 1993. Order Didelphimorphia. In D. E. Wil- term changes. In E. G. Leigh, Jr. (ed.), son and D. M. Reeder (eds.), Mammal The ecology of a tropical forest, sea- species of the World, 2nd ed., pp. 15± sonal rhythms and long-term changes: 23. Washington, DC: Smithsonian Inst. 455±468. Washington, DC: Smithson. Press. Inst. Press. 1999. Cervus gouazoubira Fischer, 1814 (cur- Gmelin, J. F. rently Mazama gouazoubira; Mamma- 1788. Systema naturae per regna tria naturae, lia, Artiodactyla): proposed conserva- secundum classes, ordines, genera, spe- tion as the correct original spelling. cies; cum characteribus, differentiis, Bull. Zool. Nomencl. 56: 262±265. synonymis, locis. Vol. 1: i±xii, 1±500. Gardner, A. L., and G. K. Creighton Lipsiae [Leipzig]: Georg. Emanuel 1989. A new generic name for Tate's (1933) Beer. [Thirteenth edition of Linnaeus's microtarsus group of South American Systema Naturae, edited by Gmelin.] mouse opossums (Marsupialia: Didel- Goeldi, E. A., and G. Hagmann phidae). Proc. Biol. Soc. Washington 1904. Prodromo de um catalogo critico, com- 102: 3±7. entado da colleccËaÄo de mammiferos no Genoways, H. H., and S. L. Williams Museo do ParaÂ. Bol. Mus. Goeldi 4: 1980. Results of the Alcoa Foundation±Suri- 38±122. name expeditions. I. A new species of Goldman, E. A. bat of the genus Tonatia (Mammalia: 1912. New mammals from eastern Panama. Phyllostomidae). Ann. Carnegie Mus. Smithson. Misc. Coll. 60(2): 18 pp. 49: 203±211. Gompper, M. E., and D. M. Decker Genoways, H. H., S. L. Williams, and J. A. Groen 1998. Nasua nasua. Mamm. Species 580: 9 1981. Results of the Alcoa Foundation±Suri- pp. name expeditions. V. Noteworthy re- Goodwin, G. G. cords of Surinamese mammals. Ann. 1953. Catalogue of type specimens of recent Carnegie Mus. 50: 319±332. mammals in the American Museum of Natural History. Bull. Am. Mus. Nat. Gentry, A. H., and L. H. Emmons Hist. 102: 207±412. 1987. Geographic variation in fertility, phe- 1961. The murine opossums (genus Marmo- nology, and composition of the under- sa) of the West Indies, and the descrip- story of Neotropical forests. Biotropica tion of a new subspecies of Rhipidomys 19: 216±227. from Little Tobago. Am. Mus. Novita- Geoffroy Saint-Hilaire, E. tes 2070: 20 pp. 1803. Catalogue des mammifeÁres du MuseÂum Gray, J. E. National d'Histoire Naturelle. [Paris: 1842. Descriptions of some new genera and lacking publisher's imprint and never ®fty unrecorded species of Mammalia. offered for sale, but printed copies Ann. Mag. Nat. Hist. 10: 255±267. widely distributed (Holthuis, 1963). Gribel, R., and V. A. Taddei Both the copy examined by Hershkov- 1989. Notes on the distribution of Tonatia itz (1955) in Washington, D.C., and schulzi and Tonatia carrikeri in the that examined by us in London are Brazilian Amazon. J. Mammal. 70: clearly not proof sheets or other arti- 871±873. facts disallowed by Article 9 of the In- Groves, C. P. ternational Code of Zoological Nomen- 1993. Order Primates. In D. E. Wilson and D. clature (ICZN, 1999).] M. Reeder (eds.), Mammal species of 220 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

the world, 2nd ed.: 243±277. Washing- Heltshe, J. F., and N. E. Forrester ton, DC: Smithson. Inst. Press. 1983. Estimating species richness using the Grubb, P. jackknife procedure. Biometrics 39: 1± 1993. Order Artiodactyla. In D. E. Wilson 11. and D. M. Reeder (eds.), Mammal spe- Henry, O. cies of the world, 2nd ed.: 377±414. 1994. Saisons de reproduction chez trois ron- Washington, DC: Smithson. Inst. Press. geurs et un artiodactyle en Guyane Guillotin, M. francËaise, en fonction des facteurs du 1982. Place de Proechimys cuvieri (Rodentia, milieu et de l'alimentation. Mammalia Echimyidae) dans les peuplements mi- 58: 183±200. cromammaliens terrestres de la foreÃt 1996. The in¯uence of sex and reproductive guyannaise. Mammalia 46: 299±318. state on diet preference in four terres- Guillotin, M., and F. Petter trial mammals of the French Guianan 1984. Un Rhipidomys nouveau de Guyane rainforest. Can. J. Zool. 75: 929±935. francËaise, R. leucodactylus aratayae 1999. Frugivory and the importance of seeds ssp. nov. (Rongeurs, CriceÂtideÂs). Mam- in the diet of the orange-rumped agouti malia 48: 541±544. (Dasyprocta leporina) in French Gui- Guillotin, M., and J. F. Ponge ana. J. Trop. Ecol. 15: 291±300. 1984. Identi®cation de deux espeÁce de ron- Henry, O., F. Feer, and D. Sabatier geurs de Guyane francËaise, Proechimys 2000. Diet of the lowland tapir (Tapirus ter- cuvieri et Proechimys guyannensis restris L.) in French Guiana. Biotropica (Echimyidae) par l'analyse des corres- 32: 364±368. pondances. Mammalia 48: 287±291. HernaÂndez-Camacho, J. Guillotin, M., G. Dubost, and D. Sabatier 1977. Notas para una monografõÂa de Potos 1994. Food choice and food competition ¯avus (Mammalia: Carnivora) en Co- among the three major primate species lombia. Caldasia 11: 147±181. of French Guiana. J. Zool. London 233: Hershkovitz, P. 551±579. 1944. A systematic review of the Neotropical GuÈnther, A. water rats of the genus Nectomys (Cri- 1876. On some new mammals from tropical cetinae). Univ. Michigan Mus. Zool. America. Proc. Zool. Soc. London Misc. Publ. 58: 88 pp. ϩ 4 pls., folding 1876: 743±751. map. Gyldenstolpe, N. 1948a. Mammals of northern Colombia, pre- 1932. A manual of Neotropical sigmodont ro- liminary report no. 3: water rats (genus dents. K. Sven. Vetenskapsakad. Nectomys), with supplemental notes on Handl., Ser. 3, Band 11(3): 164 pp. ϩ related forms. Proc. U.S. Natl. Mus. 98: 18 pl. 49±56. Hall, E. R. 1948b. The technical name of the Virginia deer 1981. Suggestions for collecting and prepar- with a list of the South American ing study specimens of mammals. In E. forms. Proc. Biol. Soc. Washington 61: R. Hall (ed.), The mammals of North 41±45. America 2: 1120±1132. New York: Wi- 1949. Mammals of northern Colombia, pre- ley. liminary report no. 4: monkeys (Pri- Handley, C. O., Jr. mates), with taxonomic revisions of 1976. Mammals of the Smithsonian Venezue- some forms. Proc. U.S. Natl. Mus. 98: lan Project. Brigham Young Univ. Sci. 323±427. Bull., Biol. Ser. 20(5): 89 pp. ϩ map. 1955. Status of the generic name Zorilla 1996. New species of mammals from north- (Mammalia): nomenclature by rule or ern South America: bats of the genera by caprice. Proc. Biol. Soc. Washington Histiotus Gervais and Lasiurus Gray 68: 185±192. (Chiroptera: Vespertilionidae). Proc. 1959. Nomenclature and taxonomy of the Biol. Soc. Washington 109: 1±9. Neotropical mammals described by Ol- Handley, C. O., Jr., and R. H. Pine fers, 1818. J. Mammal. 40: 337±353. 1992. A new species of prehensile-tailed por- 1960. Mammals of northern Colombia, pre- cupine, genus Coendou LaceÂpeÁde, from liminary report no. 8: Arboreal rice Brazil. Mammalia 56: 237±244. rats, a systematic revision of the sub- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 221

genus Oecomys, genus Oryzomys. Proc. demann, 1808. Fieldiana Zool. (new U.S. Natl. Mus. 110: 513±568. ser.) 86: 103 pp. 1976. Comments on generic names of four- Hill, W. C. O. eyed opossums (family Didelphidae). 1957. Primates, comparative anatomy and Proc. Biol. Soc. Washington 89: 295± taxonomy, vol. 3. Pithecoidea, Platyr- 304. rhini (families Hapalidae and Callimi- 1977. Living New World monkeys (Platyrrhi- conidae). New York: Interscience Publ. ni) with an introduction to Primates, Hoffmann, R. S., C. G. Anderson, R. W. Thoring- vol. 1. Chicago: Univ. Chicago Press. ton, Jr., and L. R. Heaney 1981. Philander and four-eyed opossums 1993. Family Sciuridae. In D. E. Wilson and once again. Proc. Biol. Soc. Washing- D. M. Reeder (eds.), Mammal species ton 93: 943±946. of the world, 2nd. ed.: 419±465. Wash- 1983. Two species of night monkeys, genus ington, DC: Smithson. Inst. Press. Aotus (Cebidae, Platyrrhini): a prelim- Holthuis, L. B. inary report on Aotus taxonomy. Am. 1963. Comments on the proposed suppression J. Primatol. 4: 209±243. of Zorilla I. Geoffroy, 1826. Bull. Zool. 1984. Taxonomy of squirrel monkeys genus Nomencl. 20: 242. Saimiri (Cebidae, Platyrrhini): a pre- Honacki, J. H., K. E. Kinman, and J. W. Koeppl liminary report with description of a 1982. Mammal species of the world, a taxo- hitherto unnamed form. Am. J. Prima- nomic and geographic reference [1st tol. 7: 155±210. ed.]. Lawrence, KS: Allen Press. 1985. A preliminary taxonomic review of the Hoogmoed, M. S. South American bearded saki monkeys 1979. The herpetofauna of the Guianan re- genus Chiropotes (Cebidae, Platyrrhi- gion. In W. E. Duellman (ed.), The ni), with the description of a new sub- South American herpetofauna: its ori- species. Fieldiana Zool. (new ser.) 27: gin, evolution, and dispersal, pp. 241± 46 pp. 279. Univ. Kansas Mus. Nat. Hist. 1987a. A history of the Recent mammalogy of Monogr. 7. the Neotropical Region from 1492 to 1983. De verspreiding van Sylvilagus brasi- 1850. Fieldiana Zool. (new ser.) 39: liensis (L., 1785) (Leporidae) in Suri- 11±98. name. Lutra 26: 35±45. 1987b. Uacaries, New World monkeys of the Howell, A. B. genus Cacajao (Cebidae, Platyrrhini): a 1924. Individual and age variation in Micro- preliminary taxonomic review with the tus montanus yosemite. J. Agric. Res. description of a new subspecies. Am. J. 28: 977±1015 ϩ 1 pl. Primatol. 12: 1±53. Huber, O., and C. Alarcon 1987c. The taxonomy of South American sa- 1988. Mapa de vegetacioÂn de Venezuela. Ca- kis, genus Pithecia (Cebidae, Platyrrhi- racas: MARNR. ni): a preliminary report and critical re- Hunsaker, D. view with the description of a new spe- 1977. Ecology of New World marsupials. In cies and a new subspecies. Am. J. Pri- D. Hunsakere (ed.), The biology of matol. 12: 387±468. marsupials, pp. 95±156. New York: Ac- 1990. Titis, New World monkeys of the genus ademic Press. Callicebus (Cebidae, Platyrrhini): a Hurault, J. preliminary taxonomic review. Fieldi- 1961. Les noirs reÂfugieÂs Boni de la Guyane ana Zool. (new ser.) 55: 109 pp. francËaise. Mem. Inst. Fr. d'Afrique 1992. The South American gracile mouse Noire (Dakar) 63: i±xii, 1±363, folding opossums, genus Gracilinanus Gardner charts. and Creighton, 1989 (Marmosidae, Husson, A. M. Marsupialia): a taxonomic review with 1957. Notes on the primates of Surinam. notes on general morphology and rela- Stud. Fauna Surinam Other Guyanas 2: tionships. Fieldiana Zool. (new ser.) 70: 13±40 ϩ 8 pls. 56 pp. 1978. The mammals of Suriname. Leiden: E. 1997. Composition of the family Didelphidae J. Brill. Gray, 1821 (Didelphoidea: Marsupia- IbaÂnÄez, C. J. lia), with a review of the morphology 1981. BiologõÂa y ecologõÂa de los murcieÂlagos and behavior of the included four-eyed del Hato ``El FrõÂo,'' Apure, Venezuela. opossums of the genus Philander Tie- DonÄana Acta Vertebr. 8(4): 271 pp. 222 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

ICZN tebrate communities, pp. 503±553. San 1998. Opinion 1894. Regnum Animale ..., Diego: Academic Press. Ed. 2 (M. J. Brisson, 1762): rejected for Kay, R. F., and R. H. Madden nomenclatural purposes, with the con- 1997. Paleogeography and paleoecology. In servation of the mammalian generic R. F. Kay et al. (eds.), Vertebrate pale- names Philander (Marsupialia), Ptero- ontology in the Neotropics: the Mio- pus (Chiroptera), Glis, Cuniculus and cene fauna of La Venta, Colombia, pp. Hydrochoerus (Rodentia), Meles, Lutra 520±550. Washington DC: Smithson. and Hyaena (Carnivora), Tapirus (Pe- Inst. Press. rissodactyla), Tragulus and Giraffa Kays, R. W. (Artiodactyla). Bull. Zool. Nomencl. 1999. Food preferences of kinkajous (Potos 55: 64±71. ¯avus): a frugivorous carnivore. J. 1999. International code of zoological no- Mammal. 80: 589±599. menclature, 4th ed. London: Interna- Kellogg, R., and E. A. Goldman tional Trust for Zoological Nomencla- 1944. Review of the spider monkeys. Proc. ture. U.S. Natl. Mus. 96: 1±45. Janos, D. P., C. T. Sahley, and L. H. Emmons Kortlucke, S. M. 1995. Rodent dispersal of vesicular-arbuscu- 1973. Morphological variation in the kinka- lar mycorrhizal fungi in Amazonian jou, Potos ¯avus (Mammalia: Procyon- Peru. Ecology 76: 1852±1858. idae), in Middle America. Occas. Pap. Jansa, S., and R. S. Voss Mus. Nat. Hist. Univ. Kansas 17: 36 2000. Phylogenetic studies on didelphid mar- pp. supials I. Introduction and preliminary Krumbiegel, V. I. results from nuclear IRBP gene se- 1940a. Die SaÈugetiere der SuÈdamerika-Expe- quences. J. Mamm. Evol. 7: 43±77. ditionen Prof. Dr. Kriegs. 5. Schwi- Janson, C. H., and L. H. Emmons mmbeutler. Zool. Anz. 132: 63±72. 1990. Ecological structure of the non¯ying 1940b. mammal community at Cocha Cashu Die SaÈugetiere der SuÈdamerika-Expe- Biological Station, Manu National ditionen Prof. Dr. Kriegs. 7. Pakas. Park, Peru. In A. H. Gentry (ed.), Four Zool. Anz. 132: 223±238. Neotropical rainforests: 314±338. New 1941. Die SaÈugetiere der SuÈdamerika-Expe- Haven: Yale Univ. Press. ditionen Prof. Dr. Kriegs. 8. Agutis. Jordan, C. F. Zool. Anz. 133: 97±113. 1985. Soils of the Amazon rainforest. In G. LaceÂpeÁde, [B. G.E. de] T. Prance and T. E. Lovejoy (eds.), Key 1799. Tableau des divisions, sous-divisions, environments: Amazonia, pp. 83±94. ordres et genres des mammifeÁres. Paris: Oxford: Pergamon. Plassan. Julien-LaferrieÁre, D. Langguth, A. 1991. Organisation du peuplement de marsu- 1974. Ecology and evolution in the South piaux en Guyane francËaise. Rev. Ecol. American canids. In M. W. Fox (ed.), (Terre Vie) 46: 125±144. The wild canids, their systematics, be- 1994. Catalogue des types de marsupiaux havioral ecology, and evolution: 192± (Marsupialia) du MuseÂum National 206. New York: Van Nostrand Rein- d'Histoire Naturelle, Paris. Mammalia hold. 58: 1±39. LaVal, R. K., and H. S. Fitch Julliot, C., and D. Sabatier 1977. Structure, movements, and reproduc- 1993. Diet of the red howler monkey (Al- tion in three Costa Rican bat commu- ouatta seniculus) in French Guiana. Int. nities. Occas. Pap. Mus. Nat. Hist. J. Primatol. 14: 527±550. Univ. Kansas 69: 28 pp. Kahn, M. C. Lavergne, A., O. Verneau, J. L. Patton, and F. M. 1931. Djuka, the bush negroes of Dutch Gui- Catze¯is ana. New York: Viking Press. 1997. Molecular discrimination of two sym- Kalko, E. K. V., C. O. Handley, Jr., and D. Han- patric species of opossum (genus Di- dley delphis: Didelphidae) in French Gui- 1996. Organization, diversity, and long-term ana. Mol. Ecol. 6: 889±891. dynamics of a Neotropical bat com- Lemercier, I. M. J. munity. In M. L. Cody and J. A. Small- 1998. Coendou prehensilis et Coendou me- wood (eds.), Long-term studies of ver- lanurus, deux porcs-eÂpics d'Amerique 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 223

du Sud. Thesis, Ecole Nationale VeÂteÂr- Lunde, D. P., and W. A. Schutt, Jr. inaire de Toulouse. 1999. The peculiar carpal tubercles of male Lim, B. K., and M. D. Engstrom Marmosops parvidens and Marmosa 2001. Species diversity of bats (Mammalia: robinsoni (Didelphidae: Didelphinae). Chiroptera) in Iwokrama Forest, Guy- Mammalia 63: 495±504. ana, and the Guianan subregion: impli- Malcolm, J. R. cations for conservation. Biodiv. Con- 1990. Estimation of mammalian densities in serv. 10: 613±657. continuous forest north of Manaus. In Lim, B. K., M. D. Engstrom, R. M. Timm, R. P. A. H. Gentry (ed.), Four Neotropical Anderson, and L. C. Watson rainforests, pp. 339±357. New Haven: 1999. First records of 10 bat species in Guy- Yale Univ. Press. ana and comments on diversity of bats 1991. Comparative abundances of Neotropi- in Iwokrama Forest. Acta Chiropterol- cal small mammals by trap height. J. ogica 1: 179±190. Mammal. 72: 188±192. Linares, O. 1992. Use of tooth impressions to identify 1998. MamõÂferos de Venezuela. Caracas: So- and age live Proechimys guyannensis ciedad Conservacionista Audubon de and P. cuvieri (Rodentia: Echimyidae). Venezuela. J. Zool. London 227: 537±546. Linnaeus, C. Marcgraf, G. 1758. Systema naturae per regna tria naturae, 1648. Historiae rerum naturalium Brasiliae secundum classes, ordines, genera, spe- [etc.]. In G. Piso and G. Marcgraf, His- cies, cum characteribus, differentiis, toria naturalis Brasiliae [etc.]. Leiden: synonymis, locis. 10th ed., tomus 1. F. Hackium, and Amsterdam: L. Elzev- Holmiae [Stockholm]: Laurentii Salvii. irium. [For the complete citation of this 1766. Systema naturae per regna tria naturae, work, see Husson (1978).] secundum classes, ordines, genera, spe- Marchais, Chevalier des cies, cum characteribus, differentiis, 1730. Voyage du Chevalier des Marchais en synonymis, locis. 12th ed., tomus 1. GuineÂe, isles voisines, et aÁ Cayenne, Holmiae [Stockholm]: Laurentii Salvii. fait en 1725, 1726 & 1727 (4 vols.). LoÈnnberg, E. Paris: Pierre Prault. [This work, invari- 1921. A second contribution to the mammal- ably cited as by des Marchais, was ac- ogy of Ecuador with some remarks on tually authored by J. B. Labat.] Caenolestes. Ark. Zool. 14: 104 pp. ϩ Mares, M. A., J. K. Braun, and D. Gettinger 1 pl. 1989. Observations on the distribution and 1925. Notes on some mammals from Ecua- ecology of the mammals of the cerrado dor. J. Mammal. 6: 271±275. grasslands of central Brazil. Ann. Car- Luckett, W. P. negie Mus. 58: 1±60. 1993. An ontogenetic assessment of dental Marques, S. A., and D. C. Oren homologies in therian mammals. In F. 1987. First Brazilian record for Tonatia S. Szalay, M. J. Novacek, and M. C. schulzi and Sturnira bidens (Chirop- McKenna (eds.), Mammal phylogeny tera: Phyllostomidae). Bol. Mus. Par. 1: 182±204. New York: Springer-Ver- EmõÂlio Goeldi (ser. zool.) 3: 159±160. lag. Marshall, L. G. 1994. Suprafamilial relationships within Mar- 1978a. Glironia venusta. Mamm. Species 107: supialia: resolution and discordance 3 pp. from multidisciplinary data. J. Mamm. 1978b. Chironectes minimus. Mamm. Species Evol. 2: 255±283. 109: 6 pp. Lund, P. W. Mathews, A. G. A. 1840. [Preprint of] Blik paa Brasiliens Dyrev- 1977. Studies on termites from the Mato erden foÈr sidste Jordomvaeltning. Tre- Grosso state, Brazil. Rio de Janeiro: die Afhandling: Fortsaettelse af Patte- Acad. Bras. Cienc. dyrene. K. Dansk. Vidensk. Selskab. Matschie, P. Afhandl.: 219±272 ϩ pls. xiv±xxiv. 1916. Bemerkungen uÈber die Gattung Didel- [Preprint 56 pp., repaginated; not seen. phis L. Sitzungsber. Ges. Naturfor- Journal issue paginated as above pub- schender Freunde, Berlin 1916(1): lished 1841.] 259±272. 224 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

McCarthy, T. J., and C. O. Handley, Jr. lia±Marsupialia). Acta Cient. Venez. 1987. Records of Tonatia carrikeri (Chirop- 35: 407±413. tera: Phyllostomidae) from the Brazili- Montgomery, G. G. (ed.) an Amazon, and Tonatia schulzi in 1978. The ecology of arboreal folivores. Guyana. Bat Res. News 28: 20±23. Washington, DC: Smithson. Inst. Press. McCarthy, T. J., P. Robertson, and J. Mitchell Moojen, J. 1988. The occurrence of Tonatia schulzi 1942. Sobre os ``ciurõÂdeos'' das colecËoÄes do (Chiroptera: Phyllostomidae) in French Museu Nacional, do Departamento de Guiana with comments on the female Zoologia de S. Paulo e do Museu Par- genitalia. Mammalia 52: 583±584. aense Emilio Goeldi. Bol. Mus. Nac. MedellõÂn,R.A. (Rio de Janeiro), nov. ser. (Zool.) 1: 52 1993. Estructura y diversidad de una comu- pp. nidad de murcieÂlagos en el troÂpico huÂ- 1943. Algunos mamõÂferos colecionados no medo mexicano. In R. A. MedellõÂn and nordeste do Brasil. Bol. Mus. Nac. (Rio G. Ceballos (eds.), Avances en el es- de Janeiro), nov. ser. (Zool.) 5: 14 pp. tudio de los mamõÂferos de MeÂxico ϩ 3 pls. (Publ. Espec. Asoc. Mex. Mastozool. 1948. Speciation in the Brazilian spiny rats 1): 333±354. MeÂxico, DF: Asoc. Mex. (genus Proechimys, family Echimyi- Mastozool. dae). Univ. Kansas Publ. Mus. Nat. MedellõÂn, R. A., A. L. Gardner, and J. M. Aranda Hist. 1: 301±401. 1998. The taxonomic status of the YucataÂn Moore, J. C. brown brocket, Mazama pandora 1959. Relationships among living squirrels of (Mammalia: Cervidae). Proc. Biol. Soc. the Sciurinae. Bull. Am. Mus. Nat. Washington 111: 1±14. Hist. 118: 153±206. Menegaux, M. A. Morales-SaÂnchez, J. E. 1902. Catalogue des mammifeÁres rapporteÂs 1983. Notas sobre Marmosa lepida Thomas, par M. Geay de la Guyane francËaise en 1888 (Polyprotodontia: Didelphidae). 1898 et 1900. Bull. Mus. d'Hist. Nat. Lozania (Acta Zool. Colombiana) 45: 7 8: 490±496. pp. Miles, M. A., A. A. de Souza, and M. M. PoÂvoa Mori, S. A. 1981. Mammal tracking and nest location in 1987. Introduction. In S. A. Mori (ed.), The Brazilian forest with an improved Lecythidaceae of a lowland Neotropi- spool-and-line device. J. Zool. London cal forest: La FumeÂe Mountain, French 195: 331±347. Guiana. Mem. New York Bot. Gard. Miranda-Ribeiro, A. de 44: 1±8. 1919. Os veados do Brasil segundo as collec- 1991. The Guayana lowland ¯oristic prov- cËoÄes Rodon e de varios museus nacion- ince. C. R. Soc. Biogeogr. 67: 67±75. aes e estrangeiros. Rev. Mus. Paulista Mori, S. A., and G. H. Prance 11: 209±307. 1987. Phytogeography. In S. A. Mori (ed.), Molina, C., C. GarcõÂa, and N. Abad The Lecythidaceae of a lowland Neo- 1995. Notas sobre la distribucioÂn de Dacty- tropical forest: La FumeÂe Mountain, lomys dactylinus en Venezuela. Mem. French Guiana. Mem. New York Bot. Soc. Cienc. Nat. La Salle 55(143): 41± Gard. 44: 55±71. 45. Musser, G. G., and M. D. Carleton Molina, M., and J. Molinari 1993. Family Muridae. In D. E. Wilson and 1999. Taxonomy of Venezuelan white-tailed D. M. Reeder (eds.), Mammal species deer (Odocoileus, Cervidae, Mamma- of the world: 501±755. Washington, lia) based on cranial and mandibular DC: Smithson. Inst. Press. traits. Can. J. Zool. 77: 632±645. Musser, G. G., and A. L. Gardner Mondol®, E., and G. Medina P. 1974. A new species of the ichthyomyine 1957. ContribucioÂn al conocimiento del ``per- Daptomys from Peru. Am. Mus. Novi- rito de agua'' (Chironectes minimus tates 2537: 23 pp. Zimmermann). Mem. Soc. Cienc. Nat. Musser, G. G., M. D. Carleton, E. M. Brothers, La Salle 17: 141±155. and A. L. Gardner Mondol®, E., and R. PeÂrez-HernaÂndez 1998. Systematic studies of oryzomyine ro- 1984. Una nueva subespecie de zariguÈeya del dents (Muridae: Sigmodontinae): diag- grupo Didelphis albiventris (Mamma- noses and distributions of species for- 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 225

merly assigned to Oryzomys ''capito''. Ojasti, J., R. Guerrero, and O. E. HernaÂndez P. Bull. Am. Mus. Nat. Hist. 236: 376 pp. 1992. MamõÂferos de la ExpedicioÂn de Tapi- Mustrangi, M., and J. L. Patton rapeco, Estado Amazonas, Venezuela. 1997. Phylogeography and systematics of the Acta Biol. Venez. 14: 27±40. slender mouse opossum Marmosops Olds, N., and S. Anderson (Marsupialia, Didelphidae). Univ. Ca- 1987. Notes on Bolivian mammals 2. Taxon- lif. Publ. Zool. 130: 86 pp. omy and distribution of rice rats of the Myers, P., and M. D. Carleton subgenus Oligoryzomys. Fieldiana 1981. The species of Oryzomys (Oligoryzo- Zool. (new ser.) 39: 261±281. mys) in Paraguay and the identity of Olfers, I. [F. J. M.] von Azara's ``Rat sixieÁme ou Rat aÁ Tarse 1818. Bemerkungen zu Illiger's Ueberblick Noir''. Misc. Publ. Mus. Zool. Univ. der SaÈugthiere nach ihrer Vertheilung Michigan 161: 41 pp. uÈber die Welttheile, ruÈcksichtlich der Nagorsen, D. W., and R. L. Peterson SuÈdamericanischen Arten (Species). In 1980. Mammal collectors' manual. Misc. W. L. von Eschwege (ed.), Journal von Publ. R. Ontario Mus. Life Sci. (un- Brasilien, oder vermischte Nachrichten numbered): 79 pp. aus Brasilien, auf wissenschaftlichen Napier, P. H. Reisen gesammelt: 192±237. Weimar: 1976. Catalogue of primates in the British Im Verlage des Gr. H. S. priv. Landes- Museum (Natural History) part 1: fam- Industries-Comptoirs. ilies Callitrichidae and Cebidae. Lon- Oliveira, T. G. de don: Brit. Mus. (Nat. Hist.). 1998. Leopardus wiedii. Mamm. Species 579: Natori, M., and T. Hanihara 6 pp. 1992. Variations in dental measurements be- Osgood, W. H. tween Saguinus species and their sys- 1915. New mammals from Brazil and Peru. tematic relationships. Folia Primatol. Field Mus. Nat. Hist. Publ. Zool. Ser. 58: 84±92. 10: 187±198. Norconk, M. A., R. W. Sussman, and J. Phillips- 1921. A monographic study of the American Conroy marsupial Caenolestes, with a descrip- 1996. Primates of Guyana shield forests: Ven- tion of the brain of Caenolestes by C. ezuela and the Guianas. In M. A. Nor- Judson Herrick. Field Mus. Nat. Hist. conk, A. L. Rosenberger, and P. A. Gar- Publ. Zool. Ser. 14(1): 162 pp. ϩ 22 ber (eds.), Adaptive radiations of Neo- pls. tropical primates: 69±83. New York: 1943. The mammals of Chile. Field Mus. Nat. Plenum Press. Hist. Publ. Zool. Ser. 30: 268 pp. Nunes, A. P., J. M. Ayres, E. S. Martins, and J. Pacheco, V., and E. Vivar de Sousa e Silva 1996. Annotated checklist of the non¯ying 1988. Primates of Roraima (Brazil). I. North- mammals at Pakitza, Manu Reserve eastern part of the territory. Bol. Mus. Zone, Manu National Park, Peru. In D. Para. EmõÂlio Goeldi (ser. zool.) 4: 87± E. Wilson and A. Sandoval (eds.), 100 ϩ 9 pls. Manu, the biodiversity of southeastern Ochoa, J. Peru: 577±592. Washington, DC: 1995. Los mamõÂferos de la regioÂn de Imataca, Smithson. Inst. Press. Venezuela. Acta Cient. Venez. 46: Pack, K. S., O. Henry, and D. Sabatier 274±287. 1999. The insectivorous-frugivorous diet of Ochoa, J., and P. Soriano the golden-handed tamarin (Saguinus 1991. A new species of water rat, genus midas midas) in French Guiana. Folia Neusticomys Anthony, from the Andes Primatol. 70: 1±7. of Venezuela. J. Mammal. 72: 97±103. Palma, R. Eduardo, and A. E. Spotorno Ochoa, J., P. J. Soriano, D. Lew, and M. Ojeda C. 1999. Molecular systematics of marsupials 1993. Taxonomic and distributional notes on based on the rRNA 12S mitochondrial some bats and rodents from Venezuela. gene: the phylogeny of Didelphimor- Mammalia 57: 393±400. phia and of the living fossil microbioth- Ojasti, J. eriid Dromiciops gliroides Thomas. 1972. RevisioÂn preliminar de los picures o Mol. Phylogenet. Evol. 13: 525±535. agutõÂes de Venezuela (Rodentia, Dasy- Patterson, B. D. proctidae). Mem. Soc. Cien. Nat. La 1992. Mammals in the Natural History Mu- Salle 32: 159±204. seum, Stockholm, collected in Brazil 226 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

and Bolivia by A. M. Olalla during Paynter, R. A., Jr., and M. A. Traylor, Jr. 1934±1938. Fieldiana Zool. (new ser.) 1991. Ornithological gazetteer of Brazil (2 66: 42 pp. vols. ). Cambridge, MA: Mus. Comp. 1994. Accumulating knowledge on the di- Zool., Harvard Univ. mensions of biodiversity: systematic Pelzeln, A. von perspectives on Neotropical mammals. 1883. Brasilische SaÈugethiere. Resultate von Biodivers. Lett. 2: 79±86. Johann Natterer's Reisen in den Jahren Patton, J. L. 1817 bis 1835. Verhl. kaiserl.-koÈnigl. 1987. Species groups of spiny rats, genus zool.-bot. Ges. Wien 33 (Suppl.): 140 Proechimys (Rodentia: Echimyidae). pp. Fieldiana Zool. (new ser.) 39: 305±345. Pennant, T. Patton, J. L., and L. H. Emmons 1771. Synopsis of quadrupeds. Chester: J. 1985. A review of the genus Isothrix (Roden- Monk. tia, Echimyidae). Am. Mus. Novitates Peres, C. A. 2817: 14 pp. 1993a. Diet and feeding ecology of saddle- Patton, J. L., and A. L. Gardner back (Saguinus fuscicollis) and mous- 1972. Notes on the systematics of Proechimys tached (S. mystax) tamarins in an Am- (Rodentia: Echimyidae), with emphasis azonian terra ®rme forest. J. Zool. Lon- on Peruvian forms. Occas. Pap. Mus. don 230: 567±592. Zool. Louisiana State Univ. 44: 30 pp. 1993b. Notes on the ecology of buffy saki Patton, J. L., and M. A. Rogers monkeys (Pithecia albicans Gray, 1983. Systematic implication of nongeo- 1860): a canopy seed-predator. Am. J. graphic variation in the spiny rat genus Primatol. 31: 129±140. Proechimys (Echimyidae). Z. SaÈuge- 1999. The structure of nonvolant mammal tierk. 48: 363±370. communities in different Amazonian Patton, J. L., and M.N.F. da Silva forest types. In J. F. Eisenberg and K. 1997. De®nitions of species of pouched four- H. Redford (eds.), Mammals of the eyed opossums (Didelphidae: Philan- Neotropics 3: 564±581. Chicago and der). J. Mammal. 78: 90±102. London: Chicago Univ. Press. Patton, J. L., S. F. do Reis, and M.N.F. da Silva PeÂrez-HernaÂndez, R., P. Soriano, and D. Lew 1996. Relationships among didelphid marsu- 1994. Marsupiales de Venezuela. Caracas: pials based on sequence variation in the Cuadernos Lagoven. mitochondrial cytochrome b gene. J. Peters, W. Mamm. Evol. 3: 3±29. 1861. UÈ ber einige merkwuÈrdige Nagethiere Patton, J. L., M.N.F. da Silva, M. C. Lara, and M. (Spalacomys indicus, Mus tomentosus A. Mustrangi und Mus squamipes) des KoÈnigl. zool- 1997. Diversity, differentiation, and the his- ogischen Museums. Abh. KoÈnigl. torical biogeography of nonvolant Akad. Wiss. Berlin 1860: 139±156 ϩ 2 small mammals of the Neotropical for- pls. ests. In W. F. Laurance and R. O. Bier- Petter, F. regaard, Jr. (eds.), Forest remnants: 1978. EÂ pideÂmiologie de la leishmaniose en ecology, management, and conserva- Guyane francËaise, en relation avec tion of fragmented communities: 455± l'existence d'une espeÁce nouvelle de 465. Chicago: Univ. Chicago Press. Rongeurs EÂ chimyideÂs, Proechimys cu- Patton, J. L., M.N.F. da Silva, and J. R. Malcolm vieri sp. n. C. R. Acad. Sci. Paris, ser. 1994. Gene genealogy and differentiation D, 287: 261±264. among arboreal spiny rats (Rodentia: 1979. Une nouvelle espeÁce de rat d'eau de Echimyidae) of the Amazon basin: a Guyane francËaise, Nectomys parvipes test of the riverine barrier hypothesis. sp. nov. (Rongeurs, Cricetidae). Mam- Evolution 48: 1314±1323. malia 43: 507±510. 2000. Mammals of the Rio JuruaÂ and the evo- Pine, R. H. lutionary and ecological diversi®cation 1973. Mammals (exclusive of bats) of BeleÂm, of Amazonia. Bull. Am. Mus. Nat. ParaÂ, Brazil. Acta AmazoÃnica 3: 47±79. Hist. 244: 306 pp. 1981. Reviews of the mouse opossums Mar- Paynter, R. A., Jr. mosa parvidens Tate and Marmosa in- 1982. Ornithological gazetteer of Venezuela. victa Goldman (Mammalia: Marsupia- Cambridge, MA: Mus. Comp. Zool., lia: Didelphidae) with description of a Harvard Univ. new species. Mammalia 45: 55±70. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 227

Pine, R. H., and C. O. Handley, Jr. 1978 (Rodentia, Echimyidae). Mam- 1984. A review of the Amazonian short-tailed malia 43: 501±505. opossum Monodelphis emiliae (Thom- Reis, N. R. dos, and A. L. Peracchi as). Mammalia 48: 239±245. 1987. QuiroÂpteros da regiaÄo de Manaus, Piso, G. Amazonas, Brazil (Mammalia, Chrop- 1658. De Indiae utriusque re naturali et med- tera). Bol. Mus. Par. EmõÂlio Goeldi, ser. ica libri quatuordecim, quorum conten- zool. 3: 161±182. ta pagina sequens exhibit. Amsteldae- Richard-Hansen, C., J.-C. VieÂ, N. Vidal, and J. dami [Amsterdam]: Apud L. et D. El- KeÂravec zevirios. 1999. Body measurements on 40 species of Pocock, R. I. mammals from French Guiana. J. Zool. 1913. Description of a new species of agouti (London) 247: 419±428. (Myoprocta). Ann. Mag. Nat. Hist. Rode, P. 8(12): 110±111. 1938. Catalogue des types de mammifeÁres du 1941. The races of the ocelot and the margay. MuseÂum National d'Histoire Naturelle, Field Mus. Nat. Hist. Publ. Zool. Ser. I. Ordre des Primates, A.ÐSous-ordre 27: 319±369. des simiens. Paris: Mus. Nat. d'Hist. Pons, J.-M., and L. Granjon Nat. 1998. Liste des mammifeÁres de Guyane fran- 1945. Catalogue des types de mammifeÁres du cËaise. Arvicola 10: 12±15. MuseÂum National d'Histoire Naturelle, Prance, G. T. IV. Ordre des rongeurs. Paris: Mus. 1982. Forest refuges: evidence from woody Nat. d'Hist. Nat. angiosperms. In G. T. Prance (ed.), Bi- Rosen, B. R. ological diversi®cation in the tropics: 1988. From fossils to earth history: applied 137±158. New York: Columbia Univ. historical biogeography. In A. A. My- Press. ers and P. S. Giller (eds.), Analytical Price, R. biogeography: 437±481. London: 1976. The Guiana maroons, a historical and Chapman and Hall. bibliographic introduction. Baltimore: 1992. Empiricism and the biogeographical Johns Hopkins Univ. Press. black box: concepts and methods in Ray, J. marine palaeobiogeography. Palaeo- 1693. Synopsis methodica animalium quad- geogr. Palaeoclimatol. Palaeoecol. 92: rupedum et sepentini generis. London: 171±205. S. Smith & B. Walford. Salo, J., and M. RaÈsaÈnen Hierarchy of landscape patterns in Redford, K. H. 1989. western Amazonia. In L. B. Holm- 1983. Lista preliminar de mamõÂferos do Nielsen, I. C. Nielsen, and H. Balslev Parque Nacional das Emas. Brasil Flo- (eds.), Tropical forests: botanical dy- restal 55: 29±33. namics, speciation, and diversity: 35± Reig, O. A. 45. London: Academic Press. 1977. A proposed uni®ed nomenclature for Sanchez, P. A. the enamelled components of the molar 1989. Soils. In H. Lieth and M. J. A. Werger teeth of the Cricetidae (Rodentia). J. (eds.), Tropical rainforest ecosystems Zool. (London) 181: 227±241. (Ecosystems of the world, vol. 14B): Reig, O. A., J. A. W. Kirsch, and L. G. Marshall 73±88. Amsterdam: Elsevier. 1987. Systematic relationships of the living Sanderson, I. T. and Neocenozoic American ``opossum- 1949. A brief review of the mammals of Su- like'' marsupials (suborder Didelphi- riname (Dutch Guiana), based upon a morphia), with comments on the clas- collection made in 1938. Proc. Zool. si®cation of these and of the Creta- Soc. London 119: 755±789. ceous and Paleogene New World and Santori, R. T., D. A. de Moraes, and R. Cerqeira European metatherians. In M. Archer 1996. Diet composition of Metachirus nudi- (ed.), Possums and opossums: studies caudatus and Didelphis aurita (Mar- in evolution, pp. 1±89. Sydney: Surrey supialia, Didelphoidea) in southeastern Beatty. Brazil. Mammalia 59: 511±516. Reig, O. A., M. Tranier, and M. A. Barros Schaller, G. B. 1979. Sur l'identi®cation chromosomique de 1983. Mammals and their biomass on a Bra- Proechimys guyannensis (E. Geoffroy, zilian ranch. Arq. Zool. (SaÄo Paulo) 31: 1803) et de Proechimys cuvieri Petter, 1±36. 228 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Schlegel, H. Skalli, A., and G. Dubost 1876. Monographie des Singes. Leiden: E. J. 1986. Comparaison de deux faunes de mam- Brill mifeÁres tropicaux par l'analyse des Schreber, J. C. D. characteÁres de leur mode de vie. Cah. 1778. Die SaÈugthiere in Abbildungen nach Anal. DonneÂes 11: 403±440. der Natur mit Beschreibungen, dritter Smouse, P. E., J. C. Long, and R. R. Sokal Theil. Erlangen: Wolfgang Walther. 1986. Multiple regression and correlation ex- [Plates published separately in supple- tensions of the Mantel test of matrix ments; see Sherborn, 1891.] correspondence. Syst. Zool. 35: 627± Seba, A. 632. 1734. Locupletissimi rerum naturalium the- Sneath, P. H. A., and R. R. Sokal sauri accurata descriptio, et iconibus ar- 1973. Numerical taxonomy. San Francisco: ti®ciosissimis expressio, per universam W. H. Freeman. physices historiam, [etc.], vol. 1. Am- Steiner, C., P. Sourrouille, and F. Catze¯is stelaedami [Amsterdam]: Janssonio- 2000. Molecular characterization and mito- Waesbergios. chondrial sequence variation in two Shaw, G. sympatric species of Proechimys (Ro- dentia: Echimyidae) in French Guiana. 1800. General zoology or systematic natural Biochem. Syst. Ecol. 28: 963±973. history, vol. 1, part 2. Mammalia. Lon- Stephens, L., and M. A. Traylor, Jr. don: G. Kearsley. 1985. Ornithological gazetteer of the Guian- Sherborn, C. D. as. Cambridge, MA: Mus. Comp. 1891. On the dates of the parts, plates, and Zool., Harvard Univ. text of Schreber's `Saugthiere.' Proc. Straney, D. O. Zool. Soc. London 1891: 587±592. 1978. Variance partitioning and nongeograph- 1922. Index animalium [section 2, part 1: ic variation. J. Mammal. 59: 1±11. 1801±1850]. London: British Museum. Streilein, K. E. Silva, M. M. S., N.M.S. Harmani, E.F.B. GoncËal- 1982. Behavior, ecology, and distribution of ves, and W. Uieda South American marsupials. In M. A. 1996. Bats from the metropolitan region of Mares and H. H. Genoways (eds.), SaÄo Paulo, southeastern Brazil. Chirop- Mammalian biology in South America: tera Neotropical 2: 39±41. 231±250. Pittsburgh: Pymatuning Lab. Silva, M.N.F. da Ecol. 1998. Four new species of spiny rats of the Swofford, D. L. genus Proechimys (Rodentia: Echimyi- 2000. PAUP*. Phylogenetic analysis using dae) from the western Amazon of Bra- parsimony (* and other methods), vers. zil. Proc. Biol. Soc. Washington 111: 4. Sunderland, MA: Sinauer Assoc. 436±471. Tate, G.H.H. Silva, M.N.F. da, and A. Langguth 1931. Brief diagnoses of twenty-six apparent- 1989. A new record of Glironia venusta from ly new forms of Marmosa (Marsupia- the lower Amazon, Brazil. J. Mammal. lia) from South America. Am. Mus. 70: 873±875. Novitates 493: 14 pp. Silva, M.N.F. da, and J. L. Patton 1933. A systematic revision of the marsupial genus Marmosa, with a discussion of 1993. Amazonian phylogeography: mtDNA the adaptive radiation of the murine sequence variation in arboreal echi- opossums (Marmosa). Bull. Am. Mus. myid rodents (Caviomorpha). Mol. Nat. Hist. 66: 250 pp. ϩ 26 pls., folding Phylogenet. Evol. 2: 243±255. tables. Silva, N. J. da, Jr., and J., W. Sites, Jr. 1935. The taxonomy of the genera of Neo- 1995. Patterns of diversity of Neotropical tropical hystricoid rodents. Bull. Am. squamate reptile species with emphasis Mus. Nat. Hist. 68: 295±447. on the Brazilian Amazon and the con- 1939. Mammals of the Guiana Region. Bull. servation potential of indigenous re- Am. Mus. Nat. Hist. 76: 151±229. serves. Conserv. Biol. 9: 873±901. 1947. Results of the Archibold Expeditions, Simmons, N. B., and R. S. Voss no. 56. On the anatomy and classi®ca- 1998. The mammals of Paracou, French Gui- tion of the Dasyuridae (Marsupialia). ana: a Neotropical lowland rainforest Bull. Am. Mus. Nat. Hist. 88: 97±156. fauna, part 1. Bats. Bull. Am. Mus. 1948. Results of the Archibold Expeditions, Nat. Hist. 237: 219 pp. no. 60. Studies in the Peramelidae 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 229

(Marsupialia). Bull. Am. Mus. Nat. America. Ann. Mag. Nat. Hist. 7(3): Hist. 92: 313±346. 152±155. Taton, R. 1899c. Descriptions of new Neotropical mam- 1970. Brisson, Mathurin-Jacques [entry for]. mals. Ann. Mag. Nat. Hist. 7(4): 278± In H. Berger and C. B. Ballot (eds.), 288. Dictionary of scienti®c biography 2: 1900. New South American mammals. Ann. 473±475. New York: Charles Scrib- Mag. Nat. Hist. 7(5): 148±153. ner's Sons. 1901. On a collection of mammals from the Taube, E., J.-C. VieÂ, P. Fournier, C. Genty, and J.- Kanuku Mountains, British Guiana. M. Duplantier Ann. Mag. Nat. Hist. 7(8): 139±154. 1999. Distribution of two sympatric species 1902. On the geographic races of the kinka- of sloths (Choloepus didactylus and jou. Ann. Mag. Nat. Hist. 7(9): 266± Bradypus didactylus) along the Sinna- 270. mary River, French Guiana. Biotropica 1903. New forms of Sciurus, Oxymycterus, 31: 686±691. Kannabateomys, Proechimys, Dasy- [Teixeira, D. M. (ed.)] procta, and Caluromys from South 1995. Brasil-HolandeÃs (Dutch-Brazil), 5 vols. America. Ann. Mag. Nat. Hist. 7(11): Rio de Janeiro: Editora Index. [Repro- 487±493. ductions of important color illustrations 1904. New Sciurus, Rhipidomys, Sylvilagus, of the people, fauna, and ¯ora of Dutch and Caluromys from Venezuela. Ann. Brazil from 17th-century sources, in- Mag. Nat. Hist. 7(14): 33±37. cluding the original illustrations of spe- 1905. New Neotropical Chrotopterus, Sciu- cies ®rst described by Marcgraf rus, Neacomys, Coendou, Proechimys, (1648).] and Marmosa. Ann. Mag. Nat. Hist. Thomas, O. 7(16): 308±314. 1882. On a collection of rodents from north 1907. On Neotropical mammals of the genera Peru. Proc. Zool. Soc. London 1882: Callicebus, Reithrodontomys, Cteno- 98±111 ϩ pl. IV. mys, Dasypus, and Marmosa. Ann. 1887. On the small Mammalia collected in Mag. Nat. Hist. 7(20): 161±168. Demerara by Mr. W. L. Sclater. Proc. 1909. New species of Oecomys and Marmosa Zool. Soc. London 1887: 150±153 ϩ from Amazonia. Ann. Mag. Nat. Hist. pl. xix. 8(3): 378±380. 1888. Catalogue of the Marsupialia and 1910. Mammals from the River Supinaam, Monotremata in the collection of the Demerara, presented by Mr. F. V. British Museum (Natural History). MacConnell to the British Museum. London: Taylor and Francis. Ann. Mag. Nat. Hist. 8(6): 184±189. 1892. On the probable identity of certain 1911a. The mammals of the tenth edition of specimens, formerly in the Lidth de Linnaeus; an attempt to ®x the types of Jeude collection, and now in the British the genera and the exact bases and lo- Museum, with those ®gured by Albert calities of the species. Proc. Zool. Soc. Seba in his `Thesaurus' of 1734. Proc. London 1910: 120±158. Zool. Soc. London 1892: 309±318. 1911b. Three new South American mammals. 1894. On two new Neotropical mammals. Ann. Mag. Nat. Hist. 8(7): 113±115. Ann. Mag. Nat. Hist. 6(13): 436±439. 1917. A new agouti from the Moon Moun- 1896. On new small mammals from the Neo- tains, southern British Guiana, with tropical Region. Ann. Mag. Nat. Hist. notes on other species. Ann. Mag. Nat. 6(18): 301±314. Hist. 8(20): 259±261. 1897. Notes on some S.-American Muridae. 1920. On mammals from the lower Amazon Ann. Mag. Nat. Hist. 6(19): 594±501. in the Goeldi Museum, Para. Ann. 1898. Descriptions of new mammals from Mag. Nat. Hist. 9(6): 266±283. South America. Ann. Mag. Nat. Hist. 1921. Three new species of Marmosa, with a 7(2): 265±275. note on Didelphys waterhousi Tomes. 1899a. On some small mammals from the Dis- Ann. Mag. Nat. Hist. 9(7): 519±523. trict of Cuzco, Peru. Ann. Mag. Nat. 1926. On some mammals from the middle Hist. 7(3): 40±44. Amazons. Ann. Mag. Nat. Hist. 9(17): 1899b. On new small mammals from South 635±639. 230 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

Tirira, D. Voss, R. S. 1999. MamõÂferos del Ecuador. Quito: Socie- 1988. Systematics and ecology of ichthy- dad para la InvestigacioÂn y Monitoreo omyine rodents (Muroidea): patterns of de la Biodiversidad Ecuatoriana. morphological evolution in a small Tomes, R. F. adaptive radiation. Bull. Am. Mus. Nat. 1860. Notes on a second collection of mam- Hist. 188: 258±493. mals made by Mr. Fraser in the Repub- 1991. An introduction to the Neotropical mu- lic of Ecuador. Proc. Zool. Soc. London roid rodent genus Zygodontomys. Bull. 1860: 211±221. Am. Mus. Nat. Hist. 210: 113 pp. Tranier, M. 1992. A revision of the South American spe- 1976. Nouvelles donneÂes sur l'eÂvolution non cies of Sigmodon (Mammalia: Muri- paralleÁle du caryotype et de la mor- dae) with notes on their natural history phologie chez les PhyllotineÂs (Ron- and biogeography. Am. Mus. Novitates geurs, CriceÂtideÂs). C. R. Acad. Sci. Par- 3050: 56 pp. is, ser. D, 283: 1201±1203. 1993. A revision of the Brazilian muroid ro- Tribe, C. J. dent genus Delomys with remarks on 1996. The Neotropical rodent genus Rhipi- ``thomasomyine'' characters. Am. Mus. domys (Cricetidae: Sigmodontinae)Ða Novitates 3073: 44 pp. taxonomic revision. Unpubl. Ph.D. Voss, R. S., and R. Angermann diss., Univ. College London. 1997. Revisionary notes on Neotropical por- Trouessart, E.-L. cupines (Rodentia: Erethizontidae). 1. 1897. Catalogus mammalium tam viventium Type material described by Olfers quam fossilium, nov. ed., fasc. III (Ro- (1818) and Kuhl (1820) in the Berlin dentia II: Myomorpha, Hystricomor- Zoological Museum. Am. Mus. Novi- pha, Lagomorpha). Berlin: R. FriedlaÈn- tates 3214: 44 pp. der. Voss, R. S., and M. D. Carleton Tschudi, J. J. von 1993. A new genus for Hesperomys molitor 1844±1846. Untersuchungen uÈber die Fauna Winge and Holochilus magnus Hersh- Peruana. St. Gallen: von Scheitlin und kovitz (Mammalia, Muridae) with an Zollikofer. [Mammalian accounts pub- analysis of its phylogenetic relation- lished in different years as follows: pp. ships. Am. Mus. Novitates 3085: 39 pp. 1±76 in 1844, pp. 77±244 in 1845; pp. Voss, R. S., and L. H. Emmons 245±262 in 1846; see Sherborn, 1922.] 1996. Mammalian diversity in Neotropical Tuomisto, H., and K. Ruokolainen lowland rainforests: a preliminary as- 1997. The role of ecological knowledge in sessment. Bull. Am. Mus. Nat. Hist. explaining biogeography and biodiver- 230: 115 pp. sity in Amazonia. Biodiv. Conserv. 6: Voss, R. S., J. L. Silva L., and J. A. Valdes L. 347±357. 1982. Feeding behavior and diets of Neotrop- Twigg, G. I. ical water rats, genus Ichthyomys 1962. Notes on Holochilus sciureus in British Thomas, 1893. Z. SaÈugetierk. 47: 364± Guiana. J. Mammal. 43: 369±374. 369. 1965. Studies on Holochilus sciureus berbi- Wagler, J. censis, a cricetine rodent from the 1831. BeytraÈge von Wagler zur Sippe Dasy- coastal region of British Guiana. Proc. procta Illig. Isis (von Oken) 24: 617± Zool. Soc. London 145: 263±283. 622. Tyndale-Biscoe, H., and M. Renfree Wagner, A. 1987. Reproductive physiology of marsupi- 1842. Diagnosen neuer Arten brasilischer als. Cambridge: Cambridge Univ. SaÈugthiere. Arch. Naturg. 8(1): 356± Press. 362. VieÂ, J.-C. 1844. Die SaÈugthiere in Abbildungen nach 1999. Wildlife rescuesÐthe case of the Petit der Natur, mit Beschreibungen von Dr. Saut hydroelectric dam in French Gui- Johann Christian Daniel von Schreber. ana. Oryx 33: 115±125. Supplementband, vierte Abtheilung. VieÂ, J.-C., V. Volobuev, J. L. Patton, and L. Gran- [Part four of Wagner's supplement to jon Schreber's multivolume work.] Erlang- 1996. A new species of Isothrix (Rodentia: en: in der Expedition des Schre- Echimyidae) from French Guiana. ber'schen SaÈugthier[e] und des Es- Mammalia 60: 393±406. per'schen Schmetterlingswerkes, etc. 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 231

Walker, E. P. The Johan Maurits van Nassau Sticht- 1964. Mammals of the world, 3 vols. Balti- ing. more: Johns Hopkins Press. Williams, S. L., H. H. Genoways, and J. A. Groen Wallace, A. R. 1983. Results of the Alcoa Foundation±Suri- 1852. On the monkeys of the Amazon. Proc. name expeditions. VII. Records of Zool. Soc. London 1852: 107±110. mammals from central and southern Waterhouse, G. R. Suriname. Ann. Carnegie Mus. 52: 1841. Marsupialia or pouched animals [vol. 329±336. 24 of The Naturalist's Library, edited Willig, M. R., and M. A. Mares by W. Jardine]. Edinburgh: W. H. Li- 1989. Mammals from the Caatinga: an updat- zars. ed list and summary of recent research. 1848. A natural history of the Mammalia, vol. Rev. Brasil. Biol. 49: 361±367. 2, containing the Rodentia or gnawing Wilson, D. E. Mammalia. London: Hippolyte Bailli- 1990. Mammals of La Selva, Costa Rica. In eÁre. A. H. Gentry (ed.), Four Neotropical Weksler, M., C. R. Bonvincino, I. B. Otazu, and rainforests: 273±286. New Haven, CT: J. S. Silva, Jr. Yale Univ. Press. 2001. Status of Proechimys roberti and P. Wilson, D. E., and D. M. Reeder oris (Rodentia: Echimyidae) from east- 1993. Mammal species of the world, 2nd ed. ern Amazonia and central Brazil. J. Washington, DC: Smithson. Inst. Press. Mammal. 82: 109±122. Winton, W. E. de Wetzel, R. M. 1900. Report on a collection made by Messrs 1975. The species of Tamandua Gray (Eden- F.V. McConnell and J. J. Quelch at tata, Myrmecophagidae). Proc. Biol. Mount Roraima in British GuianaÐ Soc. Washington 88: 95±112. Mammalia. Trans. Linn. Soc. London 1982. Systematics, distribution, ecology, and (Zool.) 8: 52. conservation of South American eden- Woodman, N., N. A. Slade, R. M. Timm, and C. tates. In M. A. Mares and H. H. Gen- A. Schmidt oways (eds.), Mammalian biology in 1995. Mammalian community structure in South America: 345±375. Univ. Pitts- lowland tropical Peru, as determined by burgh Pymatuning Lab. Ecol. Spec. removal trapping. Zool. J. Linn. Soc. Publ. Ser. 6: 539 pp. 113: 1±20. Woods, C. A. 1985. The identi®cation and distribution of 1993. Suborder Hystricognathi. In D. E. Wil- Recent Xenarthra (ϭ Edentata). In G. son and D. M. Reeder (eds.), Mammal G. Montgomery (ed.), The evolution species of the world, 2nd ed.: 771±806. and ecology of armadillos, sloths, and Washington, DC: Smithson. Inst. Press. vermilinguas: 5±21. Washington, DC: Wozencraft, W. C. Smithson. Inst. Press. Â 1993. Order Carnivora. In D. E. Wilson and Wetzel, R. M., and F. D. de Avila-Pires D. M. Reeder (eds.), Mammal species 1980. Identi®cation and distribution of the of the World, 2nd ed.: 279±348. Wash- Recent sloths of Brazil (Edentata). Rev. ington, DC: Smithson. Inst. Press. Brasil. Biol. 40: 831±836. Wroe, S. Wetzel, R. M., and E. Mondol® 1997. A reexamination of proposed morphol- 1979. The subgenera and species of long- ogy-based synapomorphies for the fam- nosed armadillos, genus Dasypus. In J. ilies of Dasyuromorphia (Marsupialia). F. Eisenberg (ed.), Vertebrate ecology I. Dasyuridae. J. Mamm. Evol. 4: 19± in the northern Neotropics: 43±63. 52. Washington, DC: Smithson. Inst. Press. Zanchin, N. I. T., A. Langguth, and M. S. Mattevi Whitehead, P.J.P. 1992. Karyotypes of Brazilian species of 1979. Georg Markgraf and Brazilian zoology. Rhipidomys (Rodentia, Cricetidae). J. In E. van den Boogaart, H. R. Hoetink, Mammal. 73: 120±122. and P. J. P. Whitehead (eds.), Johan Zimmermann, E. A. W. Maurits van Nassau-Siegen 1604± 1780. Geographische Geschichte des Mens- 1679, a humanist prince in Europe and chen, und der vierfuÈssigen Thiere, vol. Brazil. Essays on the occasion of the 2. Leipzig: In der Weygandschen Buch- tercentenary of his death. The Hague: handlung. 232 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

APPENDIX 1

Nonvolant Mammals Previously Reported from French Guiana or Surinam, but Not Recorded at Paracou

Among the mammals that we did not collect or (Didelphis albiventris, Marmosa lepida, Leopar- observe in the course of our ®eldwork at Paracou dus tigrinus, Oecomys bicolor, Oecomys rex, are 23 indigenous nonvolant and nonaquatic spe- Rhipidomys leucodactylus, Echimys chrysurus, Is- cies known from other localities in French Guiana othrix sinnamariensis). or Surinam. Below we cite the original literature 1. Didelphis albiventris: White-eared opossums for published records and note the existence of were ®rst reported from Surinam by Genoways et voucher material for unpublished range exten- al. (1981) and from French Guiana by Julien-Laf- sions. Habitat associations attributed to species in errieÁre (1991). Additional French Guianan records the following accounts are based on cited pub- were discussed by Catze¯is et al. (1997), who lished sources, or were inferred from specimen provided compelling documentation that this spe- data if no published information was available. cies inhabits primary lowland rainforest. French Guianan localities mentioned below are 2. Marmosa lepida: This elusive species has mapped in ®gure 1. long been known from Surinam (Thomas, 1888; Our list omits several problematic Surinamese Tate, 1933; Husson, 1978), but the ®rst specimen records whose doubtful validity has been dis- from French Guiana (MNHN 1998.306) was only cussed elsewhere: Blarina pyrrhonota (see Hus- recently collected by F. Catze¯is at Les No- son, 1978), and Dasyprocta cristata and D. fulig- uragues. The habitat associations of this species inosa (see account for D. leporina, above). Like- are not well documented, but many of the local- wise, we discount as obvious misidenti®cations ities from which it has been reported (especially Menegaux's (1902) French Guianan records of by Tate [1933] and Morales-SaÂnchez [1983]) are Sciurus variabilis (ϭ S. granatensis), Peromyscus in densely rainforested landscapes. mexicanus, Echimys spinosus (ϭ Euryzygomato- 3. Euphractus sexcinctus: Surinamese records mys spinosus), and Mazama dichotoma (ϭ Blas- of this rare armadillo were discussed by Husson tocerus dichotomus). Honacki et al.'s (1982) de- (1978), but the species has not been reported from scription of the range of Monodelphis americana French Guiana. Apparently, E. sexcinctus is re- as extending to French Guiana and Surinam is stricted to savannas, open woodlands, and decid- also clearly erroneous. Pons and Granjon (1998) uous forests; no vouchered records from rainforest listed Oecomys roberti and O. trinitatis from are known (Wetzel, 1982). French Guiana, but we have not personally ex- 4. Cebus olivaceus: Apparently, the only pub- amined any material unambiguously assignable to those taxa from either French Guiana or Surinam. lished French Guianan records of this rainforest Of the 23 species listed below, 12 are unlikely monkey are from Les Nouragues (Julliot and Sa- to occur in our study area because they are pri- batier, 1993) and Saut ParareÂ (Guillotin et al., marily associated with savannas and other open 1994), but the species is probably widely distrib- habitats (Euphractus sexcinctus, Cerdocyon thous, uted. Three specimens in Paris (MNHN Odocoileus cariacou, Holochilus sciureus, Sig- 1962.1344, 1962.1345, 1962.4143) are from ``Ri- modon alstoni, Zygodontomys brevicauda, Cavia vieÁre Marowini'', two others (MNHN 1978.321, aperea, Sylvilagus brasiliensis), or because they 1978.322) are from ``Crique Elepoussing'', and a are semiaquatic or littoral species for which suit- sixth (AC 1972.176) is from Koulimapopann. able riverine habitats are not locally available Husson (1978) summarized information about (Lontra longicaudis, Pteronura brasiliensis, Pro- Surinamese records. cyon cancrivorus, Hydrochoeris hydrochaeris). 5. Chiropotes satanas: This large and conspic- Another species (Neacomys guianae) probably uous rainforest monkey is apparently widespread occurs in rainforest, but its known range does not in Surinam (Husson, 1978; Hershkovitz, 1985), extend to French Guiana. One rainforest species but only two vouchered French Guianan records (Chiropotes satanas) is de®nitely known from are accompanied by de®nite locality data: (1) two French Guiana only in the extreme south of the specimens (MNHN 1978.326, 1978.327) collect- department, near the Brazilian frontier. The re- ed between Trois Sauts and Mont Saint-Marcel, maining nine species are perhaps widespread in and (2) another pair (AC 1972.173, 1972.174) northern French Guiana and might have been lo- collected at Koulimapopann; both localities are in cally extirpated in the historic past (Cebus oliva- or near the Tumuc-Humac mountains near the ceus) or could plausibly be expected to turn up in Brazilian border. The exact provenance of the our study area with additional inventory effort specimens that Schlegel (1876: 224) reported 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 233 from ``Oyapock aÁ Cayenne'' and from ``Oya- to mangrove swamps, tidal marshes, and river- pock'' is unknown. bank habitats (Beebe, 1925; Sanderson, 1949; 6. Cerdocyon thous: This savanna-dwelling ca- Husson, 1978). nid (Langguth, 1974) is known from several Sur- 11. Odocoileus cariacou: The type locality of inamese localities (Husson, 1978) but has not the Guianan white-tailed deer (a distinct species been reported from French Guiana. from O. virginianus; see Molina and Molinari, 7. Leopardus tigrinus: The original description 1999) was restricted to ``Guyane, coastal French of this species was based on an 18th-century spec- Guiana'' by Hershkovitz (1948b: 44) and further imen said to be from Cayenne (Allen, 1919; Hus- restricted to Cayenne by AÂ vila-Pires (1958). son, 1978), but we are not aware of any recently There appear to be no other published records of collected material from French Guiana. A single this deer from French Guiana accompanied by def- unvouchered sighting from Les Nouragues (P. inite locality information, but two mounted spec- Charles-Dominique, personal commun.) appears imens in the MNHN are from ``Guyane'' (L. to be the only known modern French Guianan re- Granjon, personal commun.). Surinamese records cord, but Husson (1978) described several speci- were discussed by Husson (1978), who summa- mens from Surinam. Oncillas occur in a wide rized published and unpublished accounts sug- range of habitats, including lowland rainforest gesting that Guianan white-tailed deer are primar- (Emmons, 1990, 1997). ily associated with savannas and other open hab- 8. Lontra longicaudis: This rainforest species itats. (Emmons, 1990, 1997) is probably widely distrib- 12. Holochilus sciureus: Marsh rats of the ge- uted in both French Guiana (MNHN specimens nus Holochilus were apparently ®rst reported are from ``Crique Saunier'', ``RivieÁre Kaw'', ``Ri- from French Guiana by Charles-Dominique vieÁre Oyapock'', and Trois Sauts) and Surinam (1993), who identi®ed them as H. brasiliensis. (Husson, 1978). Lontra longicaudis has been Guianan marsh rats, however, more closely re- sighted near Paracou (in a deep tributary stream semble H. sciureus in the craniodental characters of the lower Sinnamary River and in mangroves discussed and illustrated by Voss and Carleton along Crique Malmanoury; P. Petronelli, personal (1993). Apparently, the only known French commun.), but never within the 3-km sampling Guianan specimens (in the MNHN) are from the radius around our camp, where streams are prob- IÃle de Cayenne. Surinamese collection localities ably too shallow and/or intermittent to support ot- for Holochilus were reported by Husson (1978) ters. and Genoways et al. (1981). Like other conge- 9. Pteronura brasiliensis: The type locality of neric species, H. sciureus exclusively inhabits this semiaquatic rainforest species was restricted open habitats, notably wet grass- and sedge-dom- to French Guiana based on 18th-century obser- inated communities, including marshes, rice vations (see Husson, 1978), but few specimens ®elds, wet meadows, and sugarcane plantations are apparently known from the department. Two (e.g., Twigg, 1962, 1965; Husson, 1978; Geno- old live-mounts in the MNHN are from ``Cay- ways et al., 1981). enne'' (L. Granjon, personal commun.), and there 13. Neacomys guianae: Although frequently reis a single skin in the same museum from the ported from French Guiana (e.g., by Guillotin, vicinity of Kaw (MNHN 1990.647, previously 1982; Charles-Dominique, 1993; Voss and Em- misidenti®ed as Lutra longicaudis; F. Catze¯is, mons, 1996: appendix 5), all of the specimens of personal commun.). However, recent sightings French Guianan Neacomys that we examined rep- have been reported from Les Nouragues, Kaw, resent other species (see the accounts for N. du- Crique Malmanoury, and Crique AngeÂlique (F. bosti and N. paracou, above). Apparently, the Catze¯is and P. Charles-Dominique, personal range of true Neacomys guianae extends from commun.). Giant otters are still common and Venezuela eastward through Guyana to Surinam widespread in Surinam (Husson, 1978; Duplaix, (see table 19, footnote b) and does not include 1980; Carter and Rosas, 1997). French Guiana at all. Although nothing de®nite 10. Procyon cancrivorus: The crab-eating rac- has been recorded in the literature concerning the coon was originally described from French Gui- habitats where specimens that we can certainly ana by Cuvier (1798), and two old mounted spec- identify as N. guianae were collected, other con- imens from ``Cayenne'' are in the MNHN (L. generic species all occur in lowland or lower Granjon, personal commun.). Recent records of montane rainforests. French Guianan raccoons are from Petit Saut (VieÂ, 14. Oecomys bicolor: French Guianan speci- 1999) and Kourou (F. Catze¯is, personal com- mens in the MNHN document the occurrence of mun.). Husson (1978) summarized Surinamese re- this species at Arataye, Cacao, SauÈl, Trois-Sauts, cords of this species. Field observations from the and ``RivieÁre Approuague''. Husson (1978) re- Guianas suggest that P. cancrivorus is restricted corded numerous specimens that he identi®ed as 234 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

O. bicolor from Surinam, but Husson's material 20. Hydrochoeris hydrochaeris: Capybaras may have included unrecognized specimens of O. were reported from French Guiana by Charles- rutilus. Most specimens of O. bicolor that we ex- Dominique (1993), but only one museum speci- amined from French Guiana and elsewhere were men is apparently available from the department collected in rainforested landscapes. (MNHN 1977.540, collected at Trois-Sauts). The 15. Oecomys rex: French Guianan specimens in closest known sighting to our study area is from the MNHN document the occurrence of this spe- the Petit Saut hydroelectric dam site (VieÂ, 1999). cies on the ``RivieÁre Approuague'', and at Ara- Husson (1978) summarized information about taye, SauÈl, and Trois Sauts. The species has yet Surinamese specimens and ®eld observations. Al- to be reported from Surinam. All specimens of O. though widely distributed in rainforested land- rex that we have examined were collected in pre- scapes, capybaras are semiaquatic and never oc- dominantly rainforested landscapes. cur away from rivers, lakes, or marshes (Emmons, 16. Rhipidomys leucodactylus: The only record 1990, 1997). of this elusive arboreal species from French Gui- 21. Echimys chrysurus: This arboreal rat has ana is the holotype of Rhipidomys leucodactylus apparently been recorded from only four localities aratayae collected at Saut ParareÂ on Crique Ara- in French Guiana: (1) one specimen (MNHN taye (Guillotin and Petter, 1984). No records are 1982.1076) collected at Organobo by G. Dubost known from Surinam. Most specimens of R. leu- in 1966, (2) two specimens captured and released codactylus have been collected in rainforested in the course of an ecological study at Arataye landscapes (Tribe, 1996). (Guillotin, 1982), (3) three specimens captured 17. Sigmodon alstoni: This species has been and released in the course of another ecological collected in Surinam (Husson, 1978; Williams et study at Les Nouragues (F. Catze¯is, personal al., 1983; Voss, 1992) and AmapaÂ (Carvalho, commun.), and (4) 29 specimens captured and re- 1962), and there is one historical record of a leased during faunal rescue operations at Petit French Guianan specimen from Camopi (Mene- Saut (VieÂ, 1999). Husson (1978) summarized lo- gaux, 1902). Geographic and ecological data sum- cality information from Surinamese material. Ap- marized by Voss (1992) indicate that this is a grassland species that does not occur in rainforest. parently, E. chrysurus is only known to occur in 18. Zygodontomys brevicauda: Surinamese and rainforest (Emmons, 1990, 1997). French Guianan records of this species were sum- 22. Isothrix sinnamariensis: This species was marized by Voss (1991), who also documented recently described from two specimens collected the exclusively nonforest habitats occupied by in primary rainforest on the Sinnamary River, mainland forms of Zygodontomys. Zygodontomys about 22 km upstream from the Petit Saut hydro- reigi, named as a new species by Tranier (1976) electric dam site (VieÂ et al., 1996). Although from French Guianan material, is a junior syno- known from but one locality, this elusive arboreal nym of Z. b. microtinus (see Voss, 1991). rat could be expected to occur in lowland rain- 19. Cavia aperea: This widespread savanna forest anywhere in French Guiana. species (Voss, 1991: 92) is known from Surinam 23. Sylvilagus brasiliensis: Surinamese records by four specimens collected at a single locality of this rabbit were summarized by Husson (1978) near the Brazilian border (Husson, 1978; Williams and Hoogmoed (1983); apparently, all are from et al., 1983). No French Guianan records are savannas or open anthropogenic habitats. The spe- known. cies has not been reported from French Guiana.

APPENDIX 2

Species Matrix for 12 Nonvolant Rainforest Mammal Inventories Below we provide the data on which our quan- Species records are based on this report and titative comparisons of nonvolant rainforest mam- references cited in table 55, but in some cases we mal faunas are based. The matrix consists of bi- scored the presence/absence of higher taxa (e.g., nary presence/absence records (0 ϭ absent, 1 ϭ Neacomys, Nectomys, Mesomys, and the parvi- present) for 176 taxa at 12 Neotropical localities. dens group of Marmosa) due to unresolved issues From left to right, the matrix columns represent of voucher identi®cation at one or more inventory La Selva, Barro Colorado, Imataca, Kartabo, Par- sites. Therefore, not all columns in this matrix acou, Arataye, Cunucunuma, MCSE Reserves sum to the same species totals as the correspond- (Manaus), Xingu, Balta, Manu (Cocha Cashu/ ing rows of table 55. Note that Rhipidomys gard- Pakitza), and Cuzco AmazoÂnico. neri Patton et al. (2000) and Proechimys pattoni 2001 VOSS ET AL.: RAINFOREST MAMMAL FAUNA 235 da Silva (1998) are here used for the species re- 48. Alouatta palliata: 11000 00000 00 spectively identi®ed as ''Rhipidomys cf. couesi'' 49. Alouatta seniculus: 00111 11101 11 and ''Proechimys sp. nov.'' by Voss and Emmons 50. Aotus infulatus: 00000 00010 00 (1996: appendices 9±11). The squirrel monkeys 51. Aotus lemurinus: 11000 00000 00 from Balta reported as Saimiri sciureus by Voss 52. Aotus nigriceps: 00000 00001 11 and Emmons (1996: appendix 9) are identi®ed be- 53. Aotus trivirgatus: 00000 01000 00 low as S. boliviensis following Hershkovitz 54. Ateles chamek: 00000 00001 10 (1984). Other differences in taxonomic usage be- 55. Ateles geoffroyi: 10000 00000 00 tween Voss and Emmons' (1996) appendices and 56. Ateles paniscus: 00011 10100 00 the following list are explained in the systematic 57. Callicebus brunneus: 00000 00000 10 accounts above. 58. Callicebus cupreus: 00000 00001 00 1. Caluromys derbianus: 11000 00000 00 59. Callicebus moloch: 00000 00010 00 2. Caluromys lanatus: 00000 01101 11 60. Callicebus torquatus: 00000 01000 00 3. Caluromys philander: 00011 11110 00 61. Cebus albifrons: 00000 00001 11 4. Caluromysiops irrupta: 00000 00000 10 62. Cebus apella: 00001 10111 11 5. Chironectes minimus: 11011 11001 00 63. Cebus capucinus: 11000 00000 00 6. Didelphis albiventris: 00100 10000 00 64. Cebus olivaceus: 00110 11000 00 7. Didelphis marsupialis: 11111 11111 11 65. Chiropotes satanas: 00000 01110 00 8. Glironia venusta: 00000 00000 10 66. Lagothrix lagotricha: 00000 00001 11 9. Gracilinanus agilis: 00000 00001 10 67. Pithecia irrorata: 00000 00000 10 10. Gracilinanus emiliae: 00001 00000 00 68. Pithecia monachus: 00000 00001 00 11. Hyladelphys kalinowskii: 00001 00000 00 69. Pithecia pithecia: 00111 11100 00 12. Marmosa lepida: 00000 10000 00 70. Saimiri boliviensis: 00000 00001 11 13. Marmosa mexicana: 10000 00000 00 71. Saimiri sciureus: 00011 11010 00 14. Marmosa murina: 00111 11111 11 72. Atelocynus microtis: 00000 00001 11 15. Marmosa robinsoni: 01000 00000 00 73. Speothos venaticus: 00111 10101 00 16. Marmosops noctivagus: 00000 00001 11 74. Herpailurus yaguarondi: 11111 10001 11 17. Marmosops ''parvidens'': 00001 10111 11 75. Leopardus pardalis: 11111 11101 11 18. Metachirus nudicaudatus: 01111 10111 11 76. Leopardus tigrinus: 00100 10000 00 19. Micoureus demerarae: 00111 11110 00 77. Leopardus wiedii: 11111 11001 10 20. Micoureus regina: 00000 00001 11 78. Panthera onca: 11111 10101 11 21. Monodelphis adusta: 00000 00000 01 79. Puma concolor: 11111 10101 11 22. Monodelphis brevicaudata: 00111 11110 00 80. Conepatus semistriatus: 10000 00000 00 23. Monodelphis glirina: 00000 00000 10 81. Eira barbara: 11111 11101 11 24. Philander andersoni: 00000 01000 00 82. Galictis vittata: 11011 10001 11 25. Philander mcilhennyi: 00000 00001 00 83. Lontra longicaudis: 11110 11101 11 26. Philander opossum: 11101 10111 11 84. Mustela spp.: 10000 00000 01 27. Bradypus tridactylus: 00111 10100 00 85. Pteronura brasiliensis: 00110 10001 11 28. Bradypus variegatus: 11000 00011 11 86. Bassaricyon gabbii: 11000 00001 10 29. Choloepus didactylus: 00111 11110 00 87. Bassariscus sumichrasti: 01000 00000 00 30. Choloepus hoffmanni: 11000 00001 11 88. Nasua narica: 11000 00000 00 31. Cabassous centralis: 11000 00000 00 89. Nasua nasua: 00111 11111 10 32. Cabassous unicinctus: 00111 00001 00 90. Potos ¯avus: 11111 11111 11 33. Dasypus kappleri: 00111 11101 00 91. Procyon cancrivorus: 01010 00001 10 34. Dasypus novemcinctus: 11111 11111 11 92. Procyon lotor: 10000 00000 00 35. Priodontes maximus: 00111 11101 10 93. Tapirus bairdii: 11000 00000 00 36. Cyclopes didactylus: 11011 11101 10 94. Tapirus terrestris: 00111 11111 11 37. Myrmecophaga tridactyla: 10111 11101 11 95. Pecari tajacu: 11111 10111 11 38. Tamandua mexicana: 11000 00000 00 96. Tayassu pecari: 11111 11101 11 39. Tamandua tetradactyla: 00111 11111 11 97. Mazama americana: 11111 11111 11 40. Callimico goeldii: 00000 00000 10 98. Mazama gouazoubira: 00011 10100 10 41. Callithrix pygmaea: 00000 00000 10 99. Odocoileus cariacou: 11000 00000 00 42. Saguinus fuscicollis: 00000 00000 11 100. Microsciurus alfari: 11000 00000 00 43. Saguinus imperator: 00000 00001 10 101. Microsciurus ¯aviventer: 00000 00000 10 44. Saguinus midas: 00011 10100 00 102. Sciurillus pusillus: 00011 10000 00 45. Saguinus niger: 00000 00010 00 103. Sciurus aestuans: 00111 10000 00 46. Saguinus oedipus: 01000 00000 00 104. Sciurus gilvigularis: 00000 00110 00 47. Alouatta belzebul: 00000 00010 00 105. Sciurus granatensis: 11000 00000 00 236 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 263

106. Sciurus ignitus: 00000 00001 11 142. Tylomys panamensis: 01000 00000 00 107. Sciurus igniventris: 00000 01000 00 143. Tylomys watsoni: 10000 00000 00 108. Sciurus spadiceus: 00000 00001 11 144. Coendou bicolor: 00000 00001 11 109. Sciurus variegatoides: 10000 00000 00 145. Coendou melanurus: 00111 00000 00 110. Orthogeomys cherriei: 10000 00000 00 146. Coendou mexicanus: 10000 00000 00 111. Heteromys desmarestianus: 11000 00000 00 147. Coendou prehensilis: 00101 11110 00 112. Melanomys caliginosus: 10000 00000 00 148. Coendou rothschildi: 01000 00000 00 113. Neacomys spp.: 00111 11111 11 149. Hydrochoeris hydrochaeris: 01110 10011 114. Nectomys spp.: 00111 11011 11 10 115. Neusticomys oyapocki: 00001 00000 00 150. Dinomys branickii: 00000 00001 10 116. Neusticomys peruviensis: 00000 00001 10 151. Dasyprocta fuliginosa: 00000 01000 00 117. Neusticomys venezuelae: 00010 00000 00 152. Dasyprocta leporina: 00111 10110 00 118. Nyctomys sumichrasti: 10000 00000 00 153. Dasyprocta punctata: 11000 00000 00 119. Oecomys auyantepui: 00111 10100 00 154. Dasyprocta variegata: 00000 00001 11 120. Oecomys bicolor: 01110 11111 11 155. Myoprocta acouchy: 00011 10110 00 121. Oecomys concolor: 00000 01000 00 156. Myoprocta pratti: 00000 01001 11 122. Oecomys paricola: 00000 00010 00 157. Cuniculus paca: 11111 11111 11 123. Oecomys superans: 00000 00001 11 158. Dactylomys spp.: 00000 00011 10 124. Oecomys rex: 00110 10100 00 159. Diplomys labilis: 01000 00000 00 125. Oecomys roberti: 00010 00010 01 160. Echimys chrysurus: 00000 10110 00 126. Oecomys rutilus: 00011 10000 00 161. Hoplomys gymnurus: 10000 00000 00 127. Oecomys trinitatis: 01000 00010 00 162. Isothrix bistriata: 00000 00000 01 128. Oligoryzomys spp.: 11101 10001 11 163. Isothrix pagurus: 00000 00100 00 129. Oryzomys bolivaris: 10000 00000 00 164. Makalata didelphoides: 00111 10010 00 130. Oryzomys emmonsae: 00000 00010 00 165. Makalata occasius: 00000 00000 10 131. Oryzomys macconnelli: 00111 10101 10 166. Mesomys spp.: 00001 11111 11 132. Oryzomys megacephalus: 00111 10110 00 167. Proechimys brevicauda: 00000 00001 11 133. Oryzomys nitidus: 00000 00001 11 168. Proechimys cuvieri: 00011 10110 00 134. Oryzomys perenensis: 00000 00001 11 169. Proechimys goeldii: 00000 00010 00 135. Oryzomys talamancae: 01000 00000 00 170. Proechimys guyannensis: 00101 11100 00 136. Oryzomys yunganus: 00001 10001 01 171. Proechimys oris: 00000 00010 00 137. Oxymycterus spp.: 00000 00010 10 172. Proechimys pattoni: 00000 00001 10 138. Rhipidomys gardneri: 00000 00000 11 173. Proechimys semispinosus: 11000 00000 00 139. Rhipidomys leucodactylus: 00000 11000 00 174. Proechimys simonsi: 00000 00001 11 140. Rhipidomys nitela: 00111 10110 00 175. Proechimys steerei: 00000 00001 11 141. Sigmodontomys alfari: 10000 00000 00 176. Sylvilagus brasiliensis: 11100 00011 11