MAMMALIAN DIVERSITY IN NEOTROPICAL LOWLAND RAINFORESTS: A PRELIMINARY ASSESSMENT ROBERT S. VOSS AND LOUISE H. EMMONS BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY NUMBER 230 NEW YORK: 1996 Recent issues ofthe Bulletin may be purchased from the Museum. Lists ofbackissues ofthe Bulletin, Novitates, andAnthropologicalPaperspublished duringthe lastfive years are available freeofcharge. Addressordersto: American MuseumofNatural HistoryLibrary, DepartmentD, Central ParkWest at 79th St., New York, New York 10024. TEL: (212) 769-5545. FAX: (212) 769-5009. E-MAIL: scipubseamnh.org MAMMALIAN DIVERSITY IN NEOTROPICAL LOWLAND RAINFORESTS: A PRELIMINARY ASSESSMENT ROBERT S. VOSS Associate Curator, Department ofMammalogy American Museum ofNatural History LOUISE H. EMMONS Research Associate, Division ofMammals NationalMuseum ofNatural History Smithsonia-n Institution BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Number 230, 115 pages, 26 figures, 14 tables Issued September 26, 1996 Reprinted February 1998 Price: $11.60 a copy Copyright © American Museum ofNaturalHistory 1996 ISSN0003-0090 CONTENTS Abstract .................... ..................................... 3 Resumen .......................................................... 3 Resumo .................... ..................................... 4 Introduction .......................................................... 5 Geographic, Ecological, and Taxonomic Scope ......... ........................ 6 Questions ofSemantics and Scale ................ ............................. 9 Taxonomic Problems ....................................................... 9 Inventory Methods ......................................................... 13 Marsupials .......................................................... 13 Xenarthrans.......................................................... 17 Bats ................. ........................................ 18 Primates 22 ................................................................... Carnivores .......................................................... 23 Ungulates .......................................................... 24 Rodents and Lagomorphs .................................................... 27 Summary ..... .................................................... 34 Inventory Results from Ten Localities .............. ............................ 35 Inferences from Geographic Range Data ............ ............................ 42 Marsupials .......................................................... 43 Xenarthrans................................................................. 45 Bats ................. ........................................ 45 Primates 49 ................................................................... Carnivores 51 .......................................................... Ungulates 52 .................................................................. Rodents and Lagomorphs .................................................... 52 Discussion .......................................................... 56 Sources ofDiversity Information ................ ............................. 56 Diversity in Amazonian Rainforests .............. 57 ............................ Diversity in Central American Rainforests .... .................................. 61 Diversity in Other Neotropical Rainforest Regions ........ 62 ..................... Comparisons with Other Biomes ................ ............................. 62 Generality and Causes ofObserved Diversity Patterns ....... ................... 66 Directions for Future Research ................... 68 ............................ Acknowledgments ......................................................... 70 References ........... 71 .............................................. Appendix 1: Introduction to the Appendices .......... 86 ........................... Appendix 2: Rainforest Mammals ofLa Selva .......... 87 ......................... Appendix 3: Rainforest Mammals ofBarro Colorado ........ 90 ..................... Appendix 4: Nonvolant Rainforest Mammals ofKartabo ....... 93 .................. Appendix 5: Rainforest Mammals ofthe Lower Arataye ....... 95 ................... Appendix 6: Rainforest Mammals ofthe Rio Cunucunuma ....... 97 ................. Appendix 7: Nonvolant Rainforest Mammals ofthe MCSE Reserves ..... 99 .......... Appendix 8: Rainforest Mammals ofthe Lower Rio Xingu ....... 101 ................. Appendix 9: Rainforest Mammals ofBalta ........... 103 ........................... Appendix 10: Rainforest Mammals ofCocha Cashu/Pakitza ...... 106 ................ Appendix 11: Rainforest Mammals ofCuzco Amazonico ....... 109 .................. Appendix 12: Amazonian Primate Inventories .........111.......................I.I.I Appendix 13: Rainforest Mammals ofthe Maracaibo Basin ...... 114 ................. 1996 VOSS AND EMMONS: RAINFOREST MAMMAL DIVERSITY 3 ABSTRACT Informationaboutthemagnitudeandgeograph- related with inventory duration, and that special ic distribution ofmammalian diversity in Neo- methods are required to add elusive species to tropicallowlandrainforestsisimportantforeval- faunal lists. The range data at hand also suggest uatingresearchandconservationprioritiesinCen- several geographic patterns that should be tested tralandSouthAmerica. Althoughrelevantinven- withcarefullyfocussedfieldwork. (1)Mammalian torydataarerapidlyaccumulatingintheliterature, diversity in Amazonia is probably greatest in the real site-to-site diversity differences are hard to westernsubregion(betweentheRioNegroandthe identifybecausemanyconfoundingfactorscanaf- Rio Madeira, where over 200 species might be fectthesizeandcompositionoffaunallists.Herein sympatric at some localities), least in the Guiana we assess the available information about Neo- subregion(eastoftheNegroandnorthoftheAm- tropical rainforest mammal diversity and suggest azon),andintermediateinsoutheasternAmazonia guidelines forfutureworkbyreviewinginventory (east ofthe Madeira and south ofthe Amazon). methods, documentinganddiscussingfaunallists (2) Geographic variation in Amazonian diversity from ten localities, and summarizing geographic chiefly involves marsupials, bats, primates, and range data to predict diversity patterns that can rodents; by contrast, xenarthran, carnivore, and be tested by field and museum research. ungulatefaunasareremarkablyuniformacrossthe All inventory methods are biased because each entireregion. (3) InCentralAmericanrainforests, is suitable forcollecting or observing only a frac- aconspicuousandapparentlymonotonicdiversity tion ofthe morphologically and behaviorally di- gradient extends from eastern Panama (where versemammalianfaunathatinhabitsNeotropical mammalian diversity is within the range ofAm- rainforests. Hence, many methods must be used azonianvalues)tosouthern Mexico (wheremam- incombinationtocensuswholecommunities. Al- malian diversity may be less than anywhere else though no combination ofmethods can be guar- ontherainforestedNeotropicalmainland).Mam- anteedtoproducecompleteinventories,theomis- maliandiversityincoastalVenezuelanandsouth- sion ornonintensive application ofanyofseveral eastern Brazilian rainforests is difficult to assess essentialmethodsprobablyguaranteesincomplete with existing literature and collection resources, results. We recommend nine methods that, used butneitherregionislikelytobeasdiverseasAma- intensivelyandincombination, shouldmaximize zonia. the efficiency offuture inventory fieldwork. Despiteafewdissentingvoices,theliteratureof Ten rainforest mammal inventories selected as NewWorldmammalogyprovidescompellingev- exemplars illustrate several common problems: idencethatmammaliandiversity,asmeasuredby sampling effort is highly variable from study to sympatric species richness, is greatest in lowland study,speciesaccumulationcurvesarenotasymp- tropical rainforests and decreases along gradients totic for any fauna, essential field methods were ofincreasinglatitude,elevation,andaridity.Thus, omitted in every case, and some localities were the mammalian faunas ofwestern Amazonia are partiallydefaunatedbyhunterspriortoinventory. the mostdiverse ofany in the Americas and per- Meaningful diversity comparisons are therefore hapsintheworld.Webrieflydiscussthegenerality impossible without a major investment in addi- andcausesofobserveddiversitypatternsinterms tional fieldwork at each site. ofcontemporaryecologyandhistorical scenarios. Geographic range data provide an essential al- Significant advances in understanding mam- ternative source ofdiversity estimates. Compari- malian diversity patterns in Neotropical rainfo- sons ofinventory results with geographic expec- rests will require systematic revisions of many tations(diversitypredictionsbasedonrangedata) problematicgeneraandan aggressive program to suggestthatallexistinginventoriesareincomplete, inventory poorly sampled areas while opportu- thatthedegreeofincompletenessisinverselycor- nities to do so yet remain. RESUMEN La informaci6n sobre la magnitud y distribu- diversidad, de sitio a sitio, son dificiles de iden- ciongeogrfaficade ladiversidad de los mamiferos tificardebidoaquemuchosfactoresconfusospue- en la selva neotropical es importante para la ev- denafectareltamaiioylacomposiciondelaslistas aluaci6ndelasprioridadesdeinvestigacionycon- faunisticas. Aqui, evaluamos la informacion dis- servacionenCentroySurAmerica.Aunquedatos ponible acerca de la diversidad de mamiferos de de inventarios relevantes se estan acumulando la selva hiumeda neotropical y sugerimos lineam- rapidamente en laliteratura, reales diferencias de ientosparafuturostrabajos,alrevisarmetodosde 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 230 inventario,documentarydiscutirlistasfaunisticas La diversidad de mamiferos en la Amazonia es de diez localidades, y al resumir datos de distri- probablementesuperiorenlasubregionoccidental bucion geografica para predecir patrones de div- (entreel Rio Negro yel Rio Madeira, donde mas ersidad que pueden ser probados por investiga- de200especiespodriansersimpatricasenalgunas ciones de museo y de campo. localidades),menorenlasubregionguyana(aleste Todos los metodos de inventario son sesgados del Rio Negro y norte del Rio Amazonas), e in- debido a que cada uno es apropiado para la co- termedia en la Amazonia suroriental (al este del leccion u observacion de solo una fraccion de la RioMadeiraysurdelRioAmazonas). (2)Lavar- diversidad morfologica y conductual de la fauna iaciongeograficaenladiversidadamazonicaprin- mastozoologica que habita la selva hiumeda neo- cipalmente involucra marsupiales, murcielagos, tropical. Deaquique, muchos metodosdebenser primates,yroedores;encontraste, lafaunadeed- usadosencombinacion acensos decomunidades entados, carnivoros, yungulados sonremarcable- integras. Aunque ninguna combinacion dv meto- menteuniformesatravesdetodalaregion. (3)En dospuedegarantizarelproducirinventarioscom- las selvas hCumedas centroamericanas, una gra- pletos, la omision o aplicacion no intensiva de diente de diversidad conspicua y aparentemente algunodelosvariosmetodosesencialesprobable- monotonica se extiende desde Panama oriental mentegarantizaresultadosincompletos.Nosotros (donde la diversidad de mamiferos es dentro del recomendamosnuevemetodosque,usadosinten- rangodevaloresamazonicos)hastaelsurdeMex- sivamenteyencombinacion, debenmaximizarla ico(dondeladiversidaddemamiferosdelaselva eficienciadefuturasexploracionesdecampopara h'umeda puede ser menor que en cualquier otro inventarios. lugardel territorio neotropical). La diversidad de Diez inventarios de mamiferos en selvas hui- mamiferos en las selvas venezolanas costeiias y medas seleccionados como ejemplares ilustran brasilerassurorientalesesdificildeevaluarconlos varios problemas comunes: el esfuerzo de mues- recursosdeliteraturaycoleccionesexistentes,pero treoesaltamentevariabledeestudioaestudio,las ninguna de estas regiones es probablemente tan curvasdeacumulaciondeespeciesno sonasimp- diversa como la Amazonia. totas para ninguna fauna, metodos esenciales de A pesar de unas pocas voces discrepantes, la campofueronomitidosencadacaso, yenalgunas literaturamastozoologica delNuevo Mundo pro- localidades su fauna fue parcialmente extirpada vee evidencias concluyentes que la diversidad de porcaceriaspreviasalinventario. Comparaciones mamiferos,medidaporlariquezadeespeciessim- dediversidadsignificativassonporelloimposibles patricas, es superior en la selva baja tropical y sinunainversion mayordetrabajodecampoadi- decrece alo largo de gradientesde incremento de cional en cada sitio. latitud, elevacion, y aridez. Asi, las faunas de Datosderangosgeograficos proveenunafuente mamiferosdelaAmazoniaoccidentalsonlasmas alternativa esencial de estimados de diversidad. diversascomparadaacualquierotraenlasAmer- Comparaciones de resultados de inventarios con icas y quizas del mundo. Brevemente discutimos esperadosgeograficos (predicciones de diversidad la generalidad y causas de los patrones de diver- basadosendatosderango) sugierenquetodoslos sidadobservadosenterminosdeecologiacontem- inventariosexistentessonincompletos,queelgra- poranea y escenarios historicos. dodeloincompleto esinversamentecorrelacion- Avances significativos en el entendimiento de adoconladuraciondelinventario,yquemetodos patronesdediversidaddemamiferos,enlasselvas especiales sonrequeridos paraincluiraquellases- neotropicales, requeriran de revisiones sistemati- peciesevasivas en las listasfaunisticas. Losdatos cas de muchos generos problematicos y un pro- de rango disponibles tambien sugieren variospa- gramaagresivoparainventariarareaspobremente trones geograficos que deberian ser probados con muestreadas, mientras que las oportunidades de trabajosdecampocuidadosamenteenfocados. (1) hacerlo aun existan. RESUMO Informac6es referentes 'amagnitude e distribui- ciar o tamanho e a composirao de uma lista de ,cao geograCfica da diversidade de mamfferos nas fauna. Neste trabalho nos avaliamos as informa- florestas neotropicais sao importantes na avalia- c6es existentes sobre a diversidade da fauna de ,cao de prioridadespara a pesquisa e conservacao mamiferos em florestas neotropicais e sugerimos nas Americas do Sul e Central. Apesar do rapido diretrizes para trabalhos futuros, atrav6s de revi- acumulo de dados provenientes de inventArios sao de metodos de inventariamento, documenta- faunisticos na literatura especializada, diferen,as ,cao e discussao das listas de fauna de dez locali- reais de diversidade entre localidades sao de dificil dades nas Americas do Sul e Central, alem da constatacao, ja que muitos fatores podem influen- compila,co de dados de distribui,cao geografica 1996 VOSS AND EMMONS: RAINFOREST MAMMAL DIVERSITY 5 para a predic,o de padroes de diversidade passiv- tradanaAmaz6niaeprovavelmente maiornasub- eis de serem testados por pesquisa de campo e regiao ocidental (entre os rios Negro e Madeira, consultas 'a coleq6es de museus. onde mais de duzentas especies podem ser sim- Todos os metodos de inventariamento sao par- patricas em algumas localidades), mais pobre na cias na medida que cada ume6 adequado paracol- sub-regiao das Guianas (lestedorioNegroenorte eta e/ou observac,o de apenas uma frac,ao darica do rio Amazonas) e intermediaria no sudeste da fauna de mamfferos neotropicais, que exibe uma Amazonia (lestedo rio Madeirae suldorioAma- grande diversidade morfologica e comportamen- zonas). (2) A varia,co geografica da diversidade tal. Consequentemente diferentes metodos devem amazonica deve-se principalmente aos marsu- ser combinados para o censo adequado da co- piais, quiropteros, primatas, e roedores, ao passo munidade como um todo. Ainda que nenhuma que as faunas de edentados, carnivoros, e ungu- combina,cao de m6todos possa garantir um inven- lados sao praticamente as mesmas ao longo de tario completo, a omissao e/ou aplicac,o nao in- toda a regiao. (3) Nas florestas da America Cen- tensiva de metodos essenciais provavelmente re- tral o evidente e aparentemente monotono gra- sultani em um inventario incompleto. Recomen- diente de diversidade estende-se desdeaparteori- damos nove m6todos que, se combinados e inten- ental do Panama (onde a diversidade de mamifer- samente aplicados, deverio maximizar a eficiencia os apresenta-se em nivel equivalente aos encon- de inventariamentos da fauna de mamiferos. trados na regiao amaz6nica) ate o sudeste do A partir da analise de dez inventarios de mam- Mexico (onde a diversidade de mamiferos de iferos, considerados exemplares, uma s6rie de mata tropical pluvial ali existente deve ser menor problemas recorrentes foram identificados: o es- do que em qualquer outra parte dos neotr6picos forgo amostralealtamente varRiveldeestudopara continental). A diversidade de mamifferos na re- estudo, as curvas de acumulac,o de especies nao giao costeira da Venezuela e na regiao sudeste do sao assintoticas para nehuma fauna, m6todos es- Brasil e de dificil avaliarcao a partir dos dados senciais para os estudos de campo sao constante- atualmente disponiveis na literatura e do acervo mente omitidos e, finalmente, alguns dos estudos existente em cole,coes de museus. No entanto, foram efetivados em localidades parcialmente al- pode-se afirmar que nenhuma dessas regioes po- teradas por atividades de ca,a anteriores a reali- deni apresentar-se como portadora de indices de zasao dos inventarios. Cabe-se dizerque, em fun- diversidade para mamfferos em niveis equivalen- c,o do exposto, andlises confiaveis em termos de tes aos existentes na Amaz6nia. compara,96es entre padroes de diversidade fazem- Apesar de algumas vozes discordantes, a liter- sepraticamente impossiveis semqueconsideraveis atura cientifica relativa aos mamiferos do novo investimentos em trabalhos adicionais de campo mundo apresenta evidencias suficientes de que a sejam efetivados para cada um dos sitios acima diversidade de mamiferos, medidapelariquezade considerados. especies simpatricas, e maior nas florestas tropi- Dados de distribuisao geograficaconstituemuma cais de terras baixas e decresce com os gradientes alternativa essencial para a elaboraqdo de esti- de aumento em latitude, altitude, e aridez. Pode- mativas de diversidade. Comparaq6es entreresul- se dizerentao que afaunade mamiferos daAma- tados de inventarios e expectativas geogrificas zonia ocidental 6 mais diversaque qualquer outra (previs6es de diversidade com base em dados de nas Amr6ricas e talvez no mundo. Aproveitamos distribuisao) sugeremquetodososinventAriosex- para discutir a abrangencia e causas dos padroes istentes sao incompletos, que os graus de incom- de diversidade observados em termos da ecologia pletitude estao inversamente correlacionados com contemporanea e de cenarios historicos. adurarco dos invent6rios, e que metodos especiais Cabe salientarqueavancos significativosnoen- para inventarios sao requeridos de maneira a que tendimento dos padroes de diversidade das flores- se possa adicionarespecies raras e de dificil amos- tas tropicais irao requererrevisoes sistemadticas dos tragemas listas defauna. Osdadosdedistribui,cao generos mais problematicos, assim como o esta- disponiveis no momento tambe6m sugerem varios belecimento de umprograma mais agressivo para padr6es geograficos quedeverao sertestados atra- o inventariamento de areas pobremente investi- ves de trabalhos de campo cuidadosamente dire- gadas, enquanto as oportunidades para tal ainda cionados. (1) A diversidade de mamiferos encon- existem. INTRODUCTION The accelerating pace of deforestation in systems whose biological diversity is still humidtropicallowlandsworldwidethreatens largely unexplored (Whitmore and Sayer, the continued existence of magnificent eco- 1992). Tragically, lowlandrainforestsarenow 6 BULLETIN AMERICAN MUSEUM OFNATURAL HISTORY NO. 230 only a memory in some regions where they and taxonomic scope ofour review and ex- were once extensive (Fonseca, 1985; Por, plain relevant issues ofsemantics, scale, and 1992). Evenwhere large tracts remain uncut, taxonomy. We summarize methods for col- hunting has extirpated populations of key lecting and observing mammals in rainfo- predators and large frugivores along roads rests, discusstheirsamplingbiases(ifknown), andnavigablerivers,compromisingthelong- and suggest an essential core of techniques termsurvivalofnaturalcommunitiesinmost that should be used in combination to max- accessible areas (Redford, 1992; Terborgh, imize inventory efficiency. We compare spe- 1992). Thus, opportunities to inventory the cies lists from ten Neotropical rainforest lo- biotasofundisturbedrainforests, andtostudy calities and discuss the limited conclusions the ecology ofrainforest species under pris- that can be drawn from incomplete data ob- tine conditions, are rapidly dwindling. tained by unstandardized and methodologi- Assessments of current knowledge about cally biased collecting. We evaluate distri- themagnitudeandgeographicdistributionof butional data from published and unpub- biologicaldiversityinNeotropicalrainforests lished sources and estimate minimal and are urgently needed to evaluate priorities for maximal species counts that could be ex- research and conservation in Central and pected in Central American and Amazonian South America. Much of what can yet be rainforests. We discuss discrepancies be- learned from faunal surveys in zoologically tweenexpecteddiversity(fromdistributional unexplored areas and much ofwhat can still data) and observed diversity (from invento- be saved by effectively locating protected ar- ries) in terms ofsampling artifacts and other eas may depend crucially on using informa- limitations of the information at hand. Fi- tionalreadyathandtomaximumadvantage. nally, we summarize diversitypatterns with- To date, however, progress in reviewing di- inandamongNeotropicalrainforestregions, versitydataforNeotropical rainforestorgan- contrast mammalian faunas in rainforests isms has been limited chiefly to trees, but- with those ofotherNew World biomes, dis- terflies, frogs, squamate reptiles, and birds cussecologicalandhistoricalhypothesespro- (e.g., Duellman, 1988; Heyer, 1988; Haffer, posed to explain relevant diversity phenom- 1990; Gentry, 1992; Prance, 1994; Beccaloni ena,andconcludewithrecommendationsfor and Gaston, 1995; Silva and Sites, 1995). future research. This report is the first attempt to compre- hensively review published and unpublished GEOGRAPHIC, ECOLOGICAL, AND information about mammalian diversity in TAXONOMIC SCOPE Neotropical lowlandrainforests. Theneedfor synthesis is now acute because relevant in- LowlandtropicalrainforestsontheCentral ventory data are rapidly accumulating in the and South American mainland are distrib- absence ofany context for meaningful com- utedinfourregionsmore-or-lessisolatedfrom parisons (e.g., Fonseca and Kierulff, 1988; one another by intervening mountain ranges Georgeetal., 1988;MascarenhasandPuorto, and open vegetation (fig. 1). These regions 1988; Ochoa et al., 1988; Malcolm, 1990; definethegeographiclimitsofourreviewand Woodman et al., 1991; March and Aranda, provide a convenient framework for faunal 1992; Ascorra et al., 1993; Pacheco et al., analyses and comparisons. 1993;Medellin, 1994; Timm, 1994; Hutterer Trans-Andean rainforests extend from the et al., 1995). The essential problem with in- Mexican state of Veracruz southeastward ventorycomparisons, distinguishing real site- (chiefly along the Caribbean versant of the to-site diversity differences from sampling Central American isthmus) to northwestern artifacts that can affect the size and compo- SouthAmerica. InSouthAmerica, trans-An- sition of species lists, has not hitherto been dean rainforests extend southward along the addressed in the mammalogical literature. Pacific littoral to northwestern Ecuador and Alternative sources of diversity data have eastward across the lower Cauca and Mag- likewise not been evaluated in studies ofthe dalena valleys to the western flanks of the Neotropical rainforest fauna. Serrania de Perija in northern Colombia; an Belowwedefinethegeographic, ecological, isolated enclave once cloaked the northern 1996 VOSS AND EMMONS: RAINFOREST MAMMAL DIVERSITY 7 Fig. 1. DistributionoflowlandrainforestontheCentralandSouthAmericanmainland(afterPrance, 1989,andothers). Thepredominantnaturalvegetationinshadedregionsisrainforestexceptwherelocal soils and steep climaticgradients (e.g., rainshadows andcoastal deserts too small to showon a map of this scale) produce otherplant formations. Unshaded areas are mostly covered with montane orxero- morphic vegetation. The nomenclature forrainforestedregionsand subregionsisdefinedintheaccom- panying text. andeastern foothills ofthe SierraNevada de Coastal Venezuelan rainforests occur dis- Santa Marta. The wettest ofthese forests, in continuouslyfromtheMaracaibo Basineast- western Colombia and northwestern Ecua- ward to the Orinoco delta. The largest en- dor,areknownastheChoco(Gentry, 1982b). claveoncecoveredmuchofthesouthwestern Deforestation and settlement have been ex- Maracaibo lowlands, but human interven- tensive throughout the humid trans-Andean tionthroughoutnorthernVenezuelahasbeen lowlands, but pristine habitat may persist in extensive (Huber and Alarcon, 1988) and it some remote areas. Trans-Andean rainfo- seems unlikely that any pristine habitat now rests are separated from Amazonian rainfo- remains. The lower Orinoco and the llanos rests by the Andes and from coastal Vene- separate coastal Venezuelan rainforests from zuelan rainforests by the Serrania de Perija Amazonian habitats to the east and south. and by arid vegetation on the Peninsula de Amazonian rainforests (the hylaea of Guajira. Humboldt) cover most of the Guianas, the 8 BULLETIN AMERICAN MUSEUM OFNATURAL HISTORY NO. 230 upper Orinoco catchment, and the Amazon specimens ofNeotropical mammals are ac- valleyitself(DuckeandBlack, 1953). Wefol- companied by detailed habitat descriptions low Wallace (1854) in recognizing zoogeo- andwethereforeuse"rainforest" tosubsume graphicsubdivisionsofAmazoniadefinedby the entire local matrix ofnatural vegetation theRioNegro,theRioMadeira,andthelow- characterized by an evergreen canopy oftall erAmazon: the Guianasubregion(eastofthe treesatinventorysiteswithinthefourregions Negro and north ofthe Amazon), southeast- defined above. Not included are savannas ernAmazonia(eastoftheMadeiraandsouth (sabanas, campos), shrublands (arbustales, oftheAmazon),andwesternAmazonia(west campinas, muri scrub), marshes, and other oftheNegroandMadeira).IAdiagonalswath kinds ofopen vegetation that may occur on of semiarid vegetation on the Brazilian patches ofrocky, sandy, or waterlogged soils Shield-caatinga, cerrado, and chaco-sep- within otherwise rainforested landscapes arates Amazonia from the humid Atlantic (Cooper, 1979; Anderson, 1981; Pires and lowlands ofsoutheastern Brazil. Prance, 1985). Somemammalsfoundinsuch Atlantic rainforests once extended (Por, nonforest enclaves never enter undisturbed 1992) along the eponymous coastline of rainforest and exist as paleoclimatically iso- southeastern Brazil from Cabo Sao Roque latedrepresentatives offaunaswidespread in (ca. 5°S) to the Rio Taquari (ca. 30°S). Most other biomes (Voss, 1991). ofthe Atlantic littoral region ofsoutheastern Although transitions to nonforest habitats Brazil is now completely deforested, how- (e.g., savannas and shrublands) are often ever,andthebiotasofmost(perhapsall)sur- sharply defined, lowland rainforest some- vivingfragmentshavebeenimpoverishedby times forms broad ecotones with montane huntingandselectivelogging(Fonseca, 1985). ("cloud") and deciduous ("dry") forests. We Within thesegeographic limits, the climax arbitrarily chose 1000 m above sea level to vegetationbelowabout 1000melevationand limitoursurveyofthelowlandrainforestfau- receiving about 2000 mm or more ofannual na, but montane forests can occur at lower rainfallbroadlyconformstoRichards'(1952) elevations on outlying ridges and small cor- classic description of "Tropical [nonmon- dilleras (the Massenerhebung effect; Grubb, tane]Rainforest,"forwhichweuse"lowland 1977). Forpracticalreasons (lackofrelevant rainforest" (after Grubb etal., 1963) or sim- datafrommostmammalinventorysites), we ply "rainforest" (without modifiers) as syn- ignore subtlephenological distinctionsamong onyms below. Although plant communities forestsongradients ofdecreasingrainfalland atmanylocalitiesinthewetNeotropicallow- include in our survey any with an evergreen lands exhibit conspicuous floristic and phys- canopywhetherornotasmallnumberoftree iognomicvariationcorrelatedwithseralstage speciesare seasonallyleafless (e.g., the "sem- and edaphic conditions (e.g., in riverine ideciduous"forestonBarroColoradoIsland; floodplains; PuhakkaandKalliola, 1995), few Foster and Brokaw, 1982). The mammalian fauna indigenous to low- I Wallace (1854) also used the upperAmazon (Soli- landrainforestsontheNeotropicalmainland m6es-Maraii6n)topartitionAmazoniaintofaunalunits, (within the geographic and ecological limits but this river is (arguably) oflesser zoogeographic im- definedabove) is presently known to include portance thantheMadeira, Negro, andlowerAmazon. 170 genera in 9 orders and 35 families (ex- Wallace's "Ecuadordistrict" and "Perudistrict," sepa- cludinghumans,domesticatedandcommen- ratedbytheupperAmazon, are therefore combined in our western Amazonian subregion. His "Guiana dis- sal species, sirenians, andcetaceans; table 1). trict"exactlycorrespondstoourGuianasubregionand A simple tabulation ofdistribution patterns his"Brazildistrict"tooursoutheasternAmazoniansub- (table 2) indicates that whereas half (85) of region.OfWallace'sfourdistricts,onlyGuianahasper- these genera occur from Central America or sisted with essentially the same name and geographic the Choco to southeastern Brazil, the re- definition in the zoological literature (e.g., Tate, 1939; Hoogmoed, 1979). Wallace'shypothesisthatlargeAm- mainderhave more restricted ranges thatre- azonianriversare importantzoogeographicboundaries veal significant faunal divergence among the was corroborated by a recent quantitative analysis of four rainforest regions previously identified. primate distributions (Ayres and Clutton-Brock, 1992) Although regional totals for genera (bottom but needs to be reevaluated when reliable range maps becomeavailable forothertaxa. of table 1) hint at geographic differences in