An Introduction to Marmosops (Marsupialia: Didelphidae)

Home , Didelphis, Gracilinanus, Marmosa, Marmosops, Pando Department, Thylamys

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3466, 40 pp., 12 ®gures, 13 tables December 30, 2004

An Introduction to Marmosops (Marsupialia: Didelphidae), with the Description of a New Species from Bolivia and Notes on the Taxonomy and Distribution of Other Bolivian Forms

ROBERT S. VOSS,1 TERESA TARIFA,2,3 AND ERIC YENSEN4

CONTENTS Abstract ...... 2 Introduction ...... 2 Taxonomic Accounts ...... 5 Marmosops Matschie, 1916 ...... 5 Marmosops creightoni, new species ...... 11 Notes on the Taxonomy and Distribution of Other Bolivian Forms ...... 24 Marmosops bishopi ...... 24 Marmosops impavidus ...... 25 Marmosops noctivagus ...... 27 Marmosops ocellatus ...... 29 Unidenti®ed Marmosops ...... 31 Discussion ...... 32 Acknowledgments ...... 33 References ...... 34 Appendix 1: Type localities of nominal species of Marmosops ...... 37 Appendix 2: Gazetteer of examined specimens of Bolivian Marmosops ...... 39

1 Division of Vertebrate Zoology (Mammalogy), American Museum of Natural History. 2 ColeccioÂn Boliviana de Fauna, La Paz. 3 Current address: Museum of Natural History, Albertson College, Caldwell, ID 83605. 4 Department of Biology, Albertson College, Caldwell, ID 83605.

ABSTRACT In order to facilitate much-needed revisionary research on Marmosops, we summarize the currently accepted species-level taxonomy, provide full bibliographic citations for original descriptions of all 36 included nominal taxa, map their type localities, and list their type material (if known). We rediagnose the genus Marmosops, compare it with three other didelphid genera to which misidenti®ed specimens of Marmosops have often been referred, and review the phylogenetic evidence that Marmosops is monophyletic. After describing a new species from the eastern-slope montane forests of Bolivia, we review the taxonomy of other Bolivian congeners based on morphological characters and published cytochrome-b gene sequences. Among our taxonomic results, we synonymize albiventris Tate (1931), dorothea Thomas (1911), and yungasensis Tate (1931) with M. noctivagus (Tschudi, 1845). By contrast, M. ocellatus (Tate, 1931), currently considered a synonym of dorothea, appears to be a valid species. Whereas published range maps of Bolivian species of Marmosops are demonstrably based on misidenti®ed material and show little correspondence with known environmental factors, locality records based on specimens examined for this report make much more ecogeographic sense.

INTRODUCTION ued to be treated as a junior synonym of Marmosa, even in classi®cations that used Didelphid marsupials represent the only subgeneric categories (e.g., Cabrera, 1958; substantially intact radiation of metatherian Reig et al., 1985). The current recognition of mammals in the New World, where they con- Marmosops as a distinct taxon dates from stitute a distinctive component of terrestrial Gardner and Creighton (1989), who raised all vertebrate faunas in many tropical and sub- of Tate's species groups of Marmosa to ge- tropical habitats. Although certain aspects of neric rank. didelphid classi®cation have remained stable As currently recognized in the systematic for many years, notably the generic taxono- literature, Marmosops includes 36 nominal my of the larger opossums, recent discover- taxa, of which 14 are considered to be valid ies have underscored our still very incom- species and 22 to be subspecies or junior plete knowledge of phylogenetic diversity synonyms (table 1). The geographic disper- and relationships among the smaller forms, sion of type localities (®g. 1) approximates especially the so-called ``marmosines'' (for- the distribution of the genus itself, which merly classi®ed as, or allied with, Marmosa ranges from Panama to southeastern Brazil. Gray, 1821). Such discoveries include highly Species of Marmosops occur in lowland rain divergent new clades (Voss et al., 2001; Voss forests, lowland dry forests, and in montane and Jansa, 2003; Voss et al., 2004) and un- (``cloud'') forests to about 3000 m above sea suspected species-level diversity concealed level. Trapping studies suggest that these by traditional synonymies (Mustrangi and small (20±120 g) opossums are predomi- Patton, 1997; Voss et al., 2001; Patton and nantly understory species that are often taken Costa, 2003). Both phenomena suggest that on the ground or a few meters above the the current classi®cation of small didelphids ground (e.g., on logs, branches, and lianas), requires critical attention. but apparently never in the canopy (Mal- The didelphid marsupial genus Marmo- colm, 1991; Patton et al., 2000; Voss et al., sops was originally named by Matschie 2001; Cunha and Vieira, 2002; Vieira and (1916) to contain 11 species, of which M. Monteiro-Filho, 2003). Although alleged to incanus (Lund, 1840) was designated as the be insectivorous or omnivorous (Emmons, type. Although Tate (1933) treated Marmo- 1997; Mustrangi and Patton, 1997; Cordero, sops as a junior synonym of Marmosa,he 2001; Voss et al., 2001), no analyses of rel- acknowledged that Matschie's taxon corre- evant dietary data have been published. In sponded to a diagnosable cluster of species fact, the natural history of Marmosops re- (the ``Noctivaga Group'') that could logically mains almost completely unstudied despite be recognized as a subgenus. In the decades the local abundance of some species in rel- that followed, however, Marmosops contin- atively accessible habitats. 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 3

TABLE 1 Valid Species and Synonyms Currently Referred to the Genus Marmosopsa 4 AMERICAN MUSEUM NOVITATES NO. 3466

Fig. 1. Type localities of nominal species referred to the genus Marmosops. Numbers are keyed to gazetteer entries in appendix 1.

Museum specimens of Marmosops are fre- sops to facilitate its recognition as a distinct quently misidenti®ed due to the absence of clade and to provide a foundation for sub- any comprehensive current reference for sequent revisionary work. We next describe identi®cations. Although progress has been a new species from Bolivia and summarize made in sorting out the taxonomy of Mar- what little is known about its ecogeographic mosops in southeastern Brazil (Mustrangi distribution on the eastern slopes of the An- and Patton, 1997), western Amazonia (Patton des. Lastly, we review the taxonomy and dis- et al., 2000), and the Guiana region (Voss et tribution of other Bolivian congeners in order al., 2001), no revisionary literature has ad- to assess the regional diversity of the genus dressed the species-level systematics of the based on museum specimens examined in the genus in other parts of its enormous geo- course of this study. graphic range, especially along the Andes. Even a super®cial familiarity with the large MATERIALS AND METHODS collections that have accumulated since Tate's (1933) revision suggests the urgent Specimens cited in our tables and text are need for testing traditionally accepted species preserved in the following collections (listed concepts with newly available material. in order of their standard institutional abbre- Herein we formally rediagnose Marmo- viations): AMNH, American Museum of 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 5

Natural History (New York); BMNH, Natu- orbits, even when the postorbital constriction ral History Museum (London); CBF, Colec- (between the temporal fossae) is narrower. cioÂn Boliviana de Fauna (La Paz); FMNH, Zygomatic Breadth (ZB): Measured at the widest Field Museum of Natural History (Chicago); point across both zygomatic arches. INPA, Instituto Nacional de Pesquisas da Palatal Length (PL): Measured from the anteriormost point of the premaxillae to the postpala- AmazoÃnia (Manaus); MNCN, Museo Na- tine torus, including the postpalatine spine (if cional de Ciencias Naturales (Madrid); present). MNRJ, Museu Nacional (Rio de Janeiro); Palatal Breadth (PB): Measured across the labial MSB, Museum of Southwestern Biology margins of the fourth molar (M4) crowns, at or (University of New Mexico, Albuquerque); near the stylar A position. MVZ, Museum of Vertebrate Zoology (Uni- Maxillary Toothrow Length (MTR): Measured versity of California, Berkeley); ROM, Royal from the anterior margin of the canine (C1) to Ontario Museum (Toronto); UMMZ, Univer- the posterior margin of the fourth molar (M4). sity of Michigan Museum of Zoology (Ann Length of Molars (LM): Measured from the an- Arbor); USNM, National Museum of Natural teriormost labial margin of M1 to the posteri- ormost point on M4. History (Washington, D.C.); ZMB, Museum Length of M1±M3 (M1±M3): Measured from the fuÈr Naturkunde der Humboldt-UniversitaÈt zu anteriormost labial margin of M1 to the poste- Berlin (Berlin); and ZMUC, Zoological Mu- riormost point on M3. seum of the University of Copenhagen (Co- Width of M4 (WM4): Measured from the labial penhagen). In addition, we examined a small margin of the crown at or near the stylar A number of uncataloged specimens (cited by position to the lingual apex of the protocone. ®eld number: LHE, Louise H. Emmons; MJS, Matthew J. Swarner) that will eventu- Except as noted below, all analyzed char- ally be transferred to the USNM. acter data were obtained from adult speci- We transcribed total length (nose to ¯eshy mens as determined by dental criteria. Fol- tail-tip, TL) and length of tail (basal ¯exure lowing Voss et al. (2001), a specimen was to ¯eshy tip, LT) from specimen labels or judged to be juvenile if dP3 is still in place; ®eld notes, and we computed head-and-body subadult if dP3 has been shed but P3 and/or length (HBL) by subtracting LT from TL. We M4 are still incompletely errupted; and adult also transcribed length of hind foot (heel to if the permanent dentition is complete. Qual- tip of longest claw, HF), length of ear (from itative character variation is described herein notch, Ear), and weight from specimen labels using terminology that is explained or ref- or ®eld notes, but we sometimes remeasured erenced by Voss and Jansa (2003) and Wible HF on ¯uid-preserved specimens to check (2003). An exception (not de®ned by those the accuracy of values recorded by the col- authors) is the pre®x ``self-'' as used in com- lector, and we used our values whenever bination with descriptors of ventral pelage large discrepancies were found. All external color, such as self-white or self-cream. This measurements are reported to the nearest mil- usage applies to hairs that have the same col- limeter, and all weights are reported to the oration from base to tip, as opposed to hairs nearest gram. that are basally gray and distally white or Craniodental measurements were taken cream (for example). with digital calipers and recorded to the nearest 0.01 mm, but values reported herein are TAXONOMIC ACCOUNTS rounded to the nearest 0.1 mm. The following Marmosops Matschie, 1916 were measured as illustrated in ®gure 2: TYPE SPECIES: Marmosops incanus (Lund, Condylo-Basal Length (CBL): Measured from the 1840) by original designation. occipital condyles to the anteriormost point of DIAGNOSIS: Species of Marmosops can be the premaxillae. Nasal Breadth (NB): Measured across the triple- distinguished from other small didelphid point sutures of the nasal, frontal, and maxillary marsupials by the following combination of bones on each side. character states: eye surrounded by mask of Least Interorbital Breadth (LIB): Measured at the blackish fur contrasting in color with paler narrowest point across the frontals between the fur of crown and cheeks; dorsal pelage uni- 6 AMERICAN MUSEUM NOVITATES NO. 3466

Fig. 2. Dorsal and ventral views of the skull, and occlusal view of the maxillary dentition of Mar- mosa murina, showing the anatomical limits of craniodental measurements de®ned in the text. formly colored, unpatterned; manus mesax- hair of each caudal-scale triplet petiolate onic (digit III longer than adjacent digits II (narrower basally than at midlength), thicker, and IV); manual claws small, not exceeding and usually more heavily pigmented than lat- ¯eshy apical pads in length; central palmar eral hairs; tail longer than the combined surface of manus smooth (not densely tuber- length of head and body, not incrassate; ven- culate); digit IV of pes longer than adjacent tral surface of tail-tip modi®ed for prehen- digit III; caudal scales in spiral series; middle sion; premaxillary rostral process usually 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 7 well developed; postorbital processes usually the three hairs of each caudal-scale triplet are absent; petrosal usually exposed on lateral subequal in thickness and similarly shaped surface of braincase through fenestra be- and pigmented in Marmosa (Voss and Jansa, tween squamosal and parietal bones; maxil- 2003: character 27). lary fenestrae absent; posterolateral palatal In dorsal cranial comparisons, these taxa foramina small, not extending lingual to M4 are distinguished by the usual absence of dis- protocones; maxillary and alisphenoid not in tinct postorbital processes of the frontal bones contact on ¯oor of orbit; extracranial course in Marmosops (®g. 4B) versus the usual pres- of mandibular nerve (V3) enclosed by an- ence of distinct postorbital processes in adults teromedial strut of alisphenoid bulla (second- of most species of Marmosa (®g. 4A). In lat- ary foramen ovale present); fenestra cochleae eral cranial view, the petrosal capsule that en- exposed laterally (not enclosed in a bony si- closes the para¯occulus and semicircular canus); I2±I5 with approximately symmetrical nals (ϭ pars canalicularis or pars mastoideus) rhomboidal crowns increasing in breadth is usually exposed through a fenestra between from front to back; P2 and P3 subequal in the squamosal and parietal bones in Marmo- height; molar dentition highly carnassialized; sops (®g. 5A), but there is no such lateral pe- centrocrista strongly V-shaped, its apex high trosal exposure in Marmosa (®g. 5B). In ven- above trigon basin; lower canine (c1) pro- tral cranial view, Marmosops is distinguished cumbent and premolariform (always with a by the invariant presence of a secondary fo- ¯attened, bladelike apex and usually with a ramen ovale (®g. 6B), which is just as con- small posterior accessory cusp); p2 taller sistently absent in Marmosa (®g. 6A). than p3; dp3 fully molariform; m3 hypoconid Marmosops and Marmosa are similar in labially salient; entoconid large and well de- most dental traits, but the morphology of the veloped on m1±m3. deciduous lower third premolar (dp3) ap- COMPARISONS WITH MARMOSA: Species of pears to provide a consistent difference. In Marmosops have long been confused with all examined juvenile specimens of Marmo- Marmosa, which they super®cially resemble sops (®g. 7, upper panel) this tooth is fully in size and external appearance, but these molariform because the trigonid includes a taxa are only distantly related (Jansa and distinct paraconid, protoconid, and metaco- Voss, 2000; Voss and Jansa, 2003) and can nid. By contrast, the dp3 of juvenile Mar- readily be distinguished by qualitative integ- mosa (®g. 7, lower panel) is not fully mo- umental and craniodental characters (table 2). lariform because the bicuspid trigonid lacks In the ®eld, two external characters are useful a distinct metaconid. for generic identi®cation. Whereas the manus COMPARISONS WITH GRACILINANUS: Museum of Marmosops is mesaxonic, with digit (d) specimens of Marmosops are sometimes mis- III distinctly longer than the adjacent digits identi®ed as Gracilinanus, apparently the re- (dII and dIV; ®g. 3A), the manus of Mar- sult of using inaccurate published keys.5 De- mosa is paraxonic, with dIII and dIV sube- spite such confusion, the two genera can be qual in length (®g. 3B). Easily determined distinguished unambiguously by several ex- from live specimens (or properly ¯uid-pre- ternal morphological differences (table 2). served material) with elastic phalangeal Unlike Marmosops, which has a mesaxonic joints that can be manipulated to align and manus with dIII longer than the other digits, straighten adjacent digits for relative length Gracilinanus (like Marmosa) has a paraxonic comparisons, this character is often hard to manus with dIII and dIV subequal in length. score from dried skins with hard, bent, twist- Also, whereas Marmosops has spirally ar- ed, or otherwise distorted ®ngers (see Voss ranged caudal scales, the caudal scales of and Jansa, 2003: character 10). A second dis- Gracilinanus are arranged in predominantly tinguishing external trait is the morphology of the caudal pelage, of which the central 5 Among other characters erroneously used in relevant hair in each caudal-scale triplet is petiolate key couplets by Hershkovitz (1992: 6) and Anderson (1997: 30), presence/absence of gular glands, plantar pad (narrower basally than at mid-length), thick- morphology, nasal proportions, and presence/absence of er, and usually more darkly pigmented than postorbital processes do not reliably distinguish species the lateral hairs in Marmosops; by contrast, of Marmosops from Gracilinanus. 8 AMERICAN MUSEUM NOVITATES NO. 3466 a Thylamys , and Marmosops, Marmosa, Gracilinanus TABLE 2 Diagnostic Morphological Comparisons among 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 9 TABLE 2 (Continued) 10 AMERICAN MUSEUM NOVITATES NO. 3466

Fig. 3. Dorsal views of right forefeet of Marmosops incanus (A, MVZ 197629) and Marmosa robinsoni (B, AMNH 259983) illustrating generic differences in digital proportions. Species of Mar- mosops have mesaxonic forefeet in which digit III is longer than all of the other manual digits and digit V is short (about equal in length to digit I). By contrast, species of Marmosa (and Gracilinanus) have paraxonic forefeet in which digits III and IV are subequal and digit V is long. annular series (Voss and Jansa, 2003: char- with a more-or-less sharp line of transition acter 23). Lastly, the central hair of each cau- between the two colors on either side (Voss dal-scale triplet is petiolate, thicker, and usu- and Jansa, 2003: character 7). Additional ex- ally more heavily pigmented than the lateral ternal differences (table 2) include the mor- hairs in Marmosops, whereas the central hair phology of the central palmar surface of the is not conspicuously differentiated from the manus (Voss and Jansa, 2003: character 11), lateral hairs in Gracilinanus (Voss and Jansa, the occurrence of lateral carpal tubercles 2003: character 27). (Voss and Jansa, 2003: character 12), occur- Cranially, Marmosops consistently lacks rence of plantar pelage on the tarsus (Voss maxillary fenestrae, palatal openings that are and Jansa, 2003: character 16), caudal hair consistently present in Gracilinanus on each morphology (Voss and Jansa, 2003: character side of the maxillopalatine openings near the 27), and caudal incrassation (Voss and Jansa, M1/M2 commissure (®g. 8). The morpholo- 2003: character 28). Craniodental differences gy of the deciduous lower premolar (dp3) between these genera are likewise numerous, might also distinguish these taxa, but we including the presence/absence of a rostral have examined so few juvenile Gracilinanus premaxillary process (Voss and Jansa, 2003: (two specimens of G. agilis and one G. em- character 29), shape of the nasal bones (Voss iliae, all with incomplete trigonids) that the and Jansa, 2003: character 33), size and ex- diagnostic value of this trait is unclear. tent of the posterolateral palatal foramina COMPARISONS WITH THYLAMYS: Specimens (Voss and Jansa, 2003: character 41), lateral of Marmosops have occasionally been mis- exposure of the fenestra cochleae (Voss and identi®ed as Thylamys despite the large num- Jansa, 2003: character 47), and upper pre- ber of consistent differences that distinguish molar proportions (Voss and Jansa, 2003: these very dissimilar taxa. Whereas the dor- character 55). sal body pelage of Marmosops is uniformly PHYLOGENETIC STATUS AND RELATIONSHIPS: colored and unpatterned, the middorsum of Although distinguished from certain other Thylamys is distinctly darker than the ¯anks, ``marmosines'' by a relatively small number 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 11

Fig. 4. Dorsal views of the interorbital region in Marmosa murina (A, AMNH 267368), Marmosops noctivagus (B, AMNH 262402), and Gracilinanus agilis (C, MVZ 197439) illustrating generic differences in the development of postorbital processes. Distinct postorbital processes of the frontals (ppf) are normally present in fully adult specimens of most Marmosa species, but they are usually absent in Marmosops and Gracilinanus (indistinct processes are sometimes developed in old adults). Supraorbital beads (b), consisting of upturned dorsally-grooved ridges, are present in some but not all species of Marmosops. Scale bars equal 5 mm. of morphological characters, Marmosops ap- gest that Marmosops belongs to a didelphine pears to be a natural group based on recent clade that also includes Gracilinanus, Les- phylogenetic analyses in which generic todelphys, and Thylamys. The recently de- monophyly was tested rather than assumed scribed genus Chacodelphys may also belong (Jansa and Voss, 2000; Voss and Jansa, to this group (Voss et al., 2004). Although 2003). The most taxonomically complete Patton et al. (1996) recovered Marmosops as analysis to date (Voss and Jansa, 2003) in- the sister taxon of Caluromys, the cyto- cluded only ®ve species of Marmosops, but chrome-b sequences they analyzed appear to among these were members of the diminu- be substitution-saturated at this level of tax- tive M. parvidens complex as well as large onomic comparison (Jansa and Voss, 2000: Amazonian and southeastern Brazilian ®g. 12), and the relationship in question forms; generic monophyly was supported in (Marmosops ϩ Caluromys) had only trivial separate parsimony analyses of both non- bootstrap support in their parsimony results. molecular (morphological ϩ karyotypic) Palma and Spotorno's (1999) report of a sis- characters and sequences from exon 1 of the ter-group relationship between Marmosops nuclear gene encoding the Interphotoreceptor and Metachirus (supported by very large Retinoid Binding Protein (IRBP), in a like- bootstrap values in their neighbor-joining lihood analysis of the IRBP data, and in a and parsimony trees) is an unexplained parsimony analysis of a combined-data (non- anomaly that is not consistent with any other molecular ϩ IRBP) supermatrix. The only analytic results or character data of which we discrepant result concerning generic mono- are aware. phyly, Kirsch and Palma's (1995) report that Gracilinanus agilis and Marmosops doro- Marmosops creightoni, new species thea are almost indistinguishable by thermal Figure 9 elution of scnDNA heteroduplexes, is now known to be an artifact of specimen mis- HOLOTYPE: An adult male specimen in the identi®cation (see Voss and Jansa, 2003: 57). ColeccioÂn Boliviana de Fauna (CBF 6552) Several phylogenetic analyses of molecu- collected by Eric Yensen (original number lar data (e.g., Kirsch and Palma, 1995; Jansa 1705) on 5 August 1999 near the Saynani and Voss, 2000; Voss and Jansa, 2003) sug- hydroelectric generating station (ca. 16Њ07ЈS, 12 AMERICAN MUSEUM NOVITATES NO. 3466

Fig. 5. Lateral view of posterior braincase in Marmosops impavidus (A, AMNH 272709) and Mar- mosa murina (B, AMNH 272816) illustrating the presence of a fenestra (fen) that exposes the petrosal (pet) between the parietal (par) and squamosal (sq) in the former taxon. The petrosal is not laterally exposed by a fenestra between the parietal and squamosal in Marmosa.

68Њ05ЈW; 2500 m above sea level) in the val- DESCRIPTION: Marmosops creightoni is a ley of the RõÂo Zongo, Departamento La Paz, medium-sized species (see table 3 for mea- Bolivia. The holotype is a whole (unskinned) surements and weights), much larger than ¯uid-preserved specimen with an extracted members of the M. parvidens group (ca. 20± skull and mandibles. 35 g adult weight; Voss et al., 2001: tables PARATYPES: Five additional specimens 9, 10) but substantially smaller than some (listed as unidenti®ed ``Marmosa'' by An- lowland forms like M. incanus (adults of derson, 1997: 151) were collected by G.K. which can weigh as much as 120 g; Lorini Creighton in 1979 at two adjacent localities et al., 1994). The body pelage is almost en- in the valley of the RõÂo Zongo, Departmento tirely dark, the colors probably somewhat La Paz, Bolivia. Two of these (UMMZ faded in our ¯uid-preserved specimens, but 155999, 156000) were taken above the type still a rich chocolate-brown dorsally on the locality near the Cuticucho hydroelectric holotype (collected in 1999) and only a little generating station (ca. 16Њ08ЈS, 68Њ07ЈW) at more bleached on the 24-year-old paratypes. a recorded elevation of 2967 m, whereas The ventral body coloration is slightly paler three others (UMMZ 156001±156003) were brown, but there is no sharp distinction be- trapped below the type locality at a recorded tween dorsal and ventral color zones, and all elevation of 2000 m. All ®ve paratypes are of the thoracic and abdominal fur is gray- adult animals preserved as whole (unskin- based. The chin is self-white in all of the ned) ¯uids with extracted skulls and mandi- material at hand, and white fur extends pos- bles. teriorly onto the throat of one paratype DISTRIBUTION: Known only from wet mon- (UMMZ 155601). The blackish mask that tane forest in the valley of the RõÂo Zongo, surrounds the eye on each side of the head Departamento La Paz, Bolivia, between 2000 does not extend anteriorly to include the and 3000 m above sea level (®g. 10). mystacial pads (which are covered with short ETYMOLOGY: For G. Ken Creighton, col- brownish fur) or posteriorly to the base of lector of the ®rst known specimens of this the ears. The mystacial and genal vibrissae species and author of the ®rst modern anal- are blackish, whereas the submental and in- ysis of didelphid phylogeny based on mor- terramal hairs are white. The pinnae are uni- phological characters (Creighton, 1984). formly dark and appear naked (a sparse au- 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 13

Fig. 6. Ventral view of left ear region in Marmosa murina (A, AMNH 267368) and Marmosops pinheiroi (B, AMNH 267346) illustrating generic differences in the presence/absence of a secondary foramen ovale. In Marmosa, the mandibular branch of the trigeminal nerve (V3, reconstructed course shown by heavy arrow) emerges from the endocranial lumen via the foramen ovale (fo), which is bordered by the alisphenoid (als) and the petrosal (pet); the extracranial course of the nerve is unen- closed in this taxon. In Marmosops, however, the extracranial course of V3 is partially enclosed by a bony strut (st) that extends from the anteromedial surface of the tympanic wing of the alisphenoid (twa) across the transverse canal foramen (tcf); the nerve then emerges from a so-called secondary foramen ovale. Other abbreviations: bs, basisphenoid; bo, basioccipital; cc, carotid canal; ect, ectotympanic. ricular pelage is only microscopically visi- covered by epidermal scales in predominant- ble). Laid back alongside the head, the short ly spiral series. Under high magni®cation mystacial vibrissae do not extend to the tips (25±50ϫ), each caudal scale is seen to have of the pinnae. A small patch of almost-naked three hairs projecting from its posterior mar- skin on the midline of the throat suggests the gin; the central hair is visibly thicker than the presence of a weakly developed gular gland lateral hairs, but perhaps not quite twice as in all of the male specimens examined. The wide. On the dorsal surface of the tail, these metacarpals and metatarsals are covered with hairs are no more than one to two scale rows dark (brownish) hairs, but the digits of both long, but they are longer (three to four scale manus and pes are abruptly whitish. All of rows) on the ventral caudal surface. The ex- the males we examined have a knob-like gla- posed caudal epithelium is dark (grayish) brous tubercle on the lateral aspect of the above and below basally, but it is indistinctly wrist that is absent in both females. The scro- bicolored (dark above, pale below) for at tal epithelium is pigmented (mottled-grayish least half the length of that organ, and the in preservative) and sparsely covered by tail-tip is entirely white for the last 1±3 cm. brownish fur. Both female specimens have In dorsal view the skull is distinctively 4±1±4 ϭ 9 abdominal-inguinal mammae. proportioned with small orbits, narrow zy- Body fur extends onto a short (ca. 1 cm) seg- gomatic arches, and a very broad interorbital ment of the tail base, but the remainder region. The nasal bones are long and con- (Ͼ90%) of the tail is virtually naked and spicuously widened posteriorly near the max- 14 AMERICAN MUSEUM NOVITATES NO. 3466

Fig. 7. Occlusal views of right dp3 and m1 illustrating the generic difference in trigonid morphology of the deciduous tooth. Top, Marmosops impavidus (MUSM 13286) with a complete (tricuspid) dp3 trigonid. Bottom, Marmosa murina (MUSM 15297) with an incomplete (bicuspid) dp3 trigonid. Note that whereas taxonomic variation in this trait was correctly described by Voss and Jansa (2003: character 64), the accompanying plate (Voss and Jansa, 2003: ®g. 15) was mislabeled. illary/premaxillary suture. The broad inter- the alveolus of I5 in most examined speci- orbital region is due to the considerable in- mens (the suture is anterior to the alveolus ¯ation of the cranial capsule that encloses the on the left side of UMMZ 155999, and it is olfactory bulb, behind which there is a shal- behind the alveolus on the right side of low postorbital constriction; the supraorbital UMMZ 156000). The infraorbital foramen is margins are rounded, with only indistinct very large (its anterior opening approximate- beading even in the largest male specimen ly as high as P3 is tall), and the ventral de- examined.6 The braincase is in¯ated and ¯ection of the zygomatic arch beneath the smooth, without conspicuous temporal scars. orbit is shallow. The lacrimal foramina are In lateral cranial view the premaxillary/max- contained within the anterior orbital margin. illary suture is lateral to (rather than behind) The petrosal is laterally exposed through a fenestra between the squamosal and parietal 6 The terms ``bead'' and ``beading'' as traditionally used by systematic mammalogists to describe the dor- bones. In ventral cranial view the incisive fo- sally-grooved supraorbital ridges developed in some ramina extend between (but not behind) the marsupials and rodents (®g. 4B) are derived by analogy canine alveoli. The maxillopalatine fenestrae from the upturned rims used to strengthen sheet metal are long and narrow, extending on each side or other ¯at surfaces in manufacture (i.e., the last of four de®nitions listed for the noun ``bead'' in Webster's Sev- from a point opposite P3 to M3 or to the M3/ enth New Collegiate Dictionary, 1963). M4 commissure. The palatine fenestrae are 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 15

present as a cluster of larger and smaller openings between the maxillopalatine fenestrae and the postpalatine torus. The alisphenoid tympanic wings (auditory bullae) are large, almost hemispherical capsules without ventral processes or lateral constrictions. The upper canine lacks accessory cusps, and the last upper molar is not much wider (transverse dimension) than M3 and is approximately triangular in outline. MORPHOLOGICAL COMPARISONS: Marmo- sops creightoni is the only known brownish- furred species of the genus to lack whitish thoracic or abdominal markings; all others have at least some white or cream-colored midventral fur that contrasts abruptly with the darker color of the dorsum and ¯anks.7 In many species, whitish midventral fur (which may be either self-colored or gray- based) extends continuously from chin to anus, but in a few Amazonian and northern Andean forms (e.g., M. neblina; see Gardner, 1990: ®g. 2) whitish midventral markings are narrow and sometimes discontinuous. Al- though narrow whitish midventral markings might occur polymorphically in larger samples of M. creightoni, the six examples at hand suggest that most individuals have completely dark venters. Qualitative and morphometric comparisons with other Bolivian congeners (the identi®cation and taxonomy of which are dis- cussed below) reveal many additional differences (table 4). Marmosops bishopi, for example, is much smaller than M. creightoni, has dull reddish- or grayish-brown (versus rich chocolate-brown) dorsal fur, has pale- furred metapodials that do not contrast in color with the digits (versus dark-furred metapodials contrasting with whitish digits), does not have a pale tail-tip (versus pale tail- Fig. 8. Palatal morphology in Marmosops tip present), lacks palatine fenestrae (versus noctivagus (A, AMNH 262403) and Gracilinanus palatine fenestrae present), and has a small agilis (B, MVZ 197437) illustrating generic dif- posterior accessory cusp on the upper canine ferences in patterns of fenestration. Both Mar- (versus upper canine without accessory mosops and Gracilinanus have incisive foramina (if), posterolateral palatal foramina (plpf), and cusps). Also, adult males of M. bishopi usu- maxillopalatine fenestrae (mp); in addition, some ally lack a gular gland (versus gular gland species of Marmosops and all Gracilinanus have weakly developed), have large bladelike palatine fenestrae (p). Maxillary fenestrae (m), (versus knobby) lateral carpal tubercles, and however, are present only in Gracilinanus. Scale bars equal 5 mm. 7 Marmosops invictus, a small Panamanian species with blackish-gray fur, has dark underparts that are only super®cially washed with white. 16 AMERICAN MUSEUM NOVITATES NO. 3466

Fig. 9. Dorsal, ventral, and lateral views of the skull, and lateral view of the left hemimandible of the holotype of Marmosops creightoni (CBF 6552); however, some missing details (e.g., the nasal tips, broken in the holotype) are reconstructed from UMMZ 155999 (a paratype). Note that the illustrated left hemimandible of CBF 6552 is unusual in possessing only a single mental foramen; the right hemimandible has two foramina, as do both hemimandibles of all ®ve paratypes. Scale bar equals 10 mm. 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 17

Fig. 10. Collection localities of Marmosops bishopi and M. creightoni in Bolivia. Numbers are keyed to gazetteer entries in appendix 2.

have a pale scrotum with white fur and un- the digits (versus dark-furred metapodials pigmented skin (versus scrotum dark with contrasting with whitish digits); a much nar- brown fur and mottled-grayish skin). rower interorbital region (LIB ϭ 5.6±6.3 ver- Marmosops impavidus, another Bolivian sus 6.6±7.2 mm); and much smaller bullae. species, overlaps with M. creightoni in many In addition, adult males of M. impavidus ap- measurements, although it averages smaller pear to have no trace of a gular gland (versus in most dimensions. However, M. impavidus gular gland weakly developed), and they has dull grayish- or reddish-brown (versus have a pale scrotum with white fur and un- rich chocolate-brown) dorsal fur; pale-furred pigmented skin (versus scrotum dark with metapodials that do not contrast in color with brownish fur and mottled-grayish skin). 18 AMERICAN MUSEUM NOVITATES NO. 3466

TABLE 3 Measurements (mm) and Weights (g) of Marmosops creightoni

Bolivian specimens of Marmosops nocti- digits); distinctly bi- and particolored tails, vagus overlap broadly with M. creightoni in with the terminal one-third or more com- most measured dimensions, but they usually pletely pale (versus tail indistinctly bicolored have dull reddish-brown (versus rich choco- and only the tip completely pale); a much late-brown) dorsal fur; pale-furred metapo- narrower interorbital region (LIB ϭ 5.5±6.3 dials that do not contrast in color with the mm versus 6.6±7.2 mm); and proportionately digits (versus dark-furred metapodials con- wider zygomatic arches (ZB/CBL ϫ 100 ϭ trasting with whitish digits); distinct supra- 53±54% versus 48±50%). In addition, most orbital beads that conceal the postorbital con- adult males of M. ocellatus lack a gular gland striction from dorsal view (versus supraor- (versus gular gland weakly developed), and bital beads absent or indistinct and postor- adult females have 6±1±6 ϭ 13 mammae, of bital constriction visible dorsally); and which the anteriormost two pairs are pectoral smaller, conical or ¯ask-shaped bullae (ver- (versus 4±1±4 ϭ 9 abdominal-inguinal mam- sus bullae larger and hemispherical). In ad- mae). dition, adult males of M. noctivagus have a Although we examined representative ma- pale scrotum with white fur and unpigmented terial of every species of Marmosops listed skin (versus scrotum dark with brownish fur as valid in table 1, additional pairwise com- and mottled-grayish skin), and adult females parisons are unnecessary because other cur- have 5±1±5 ϭ 11 mammae, of which the an- rently recognized species exhibit no special teriormost pair is pectoral (versus 4±1±4 ϭ similarity with M. creightoni. Indeed, most 9 abdominal-inguinal teats). extralimital (nonBolivian) species have strik- Marmosops ocellatus also overlaps broad- ingly divergent traits. Marmosops cracens, ly with M. creightoni in most measured di- M. fuscatus, and M. incanus, for example, mensions, but it has much paler grayish- have uniformly narrow nasals with subpar- brown (versus rich chocolate-brown) dorsal allel lateral margins (quite unlike the poste- fur; pale-furred metapodials that do not con- riorly expanded nasals of Bolivian forms); trast in color with the digits (versus dark- M. pinheiroi has lacrimal foramina that are furred metapodials contrasting with whitish laterally exposed anterior to the orbit (the 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 19 lacrimal foramina of Bolivian species are A branch-and-bound parsimony analysis usually concealed from lateral view inside of the nucleotide sequence data (implement- the orbit); and M. paulensis has very long ed with PAUP 4.0b10; Swofford, 1998) re- incisive foramina that extend posteriorly be- covered a single minimum-length tree with hind the canine alveoli (the incisive foramina two main clades (®g. 11), one of which con- are substantially shorter in all Bolivian taxa). sists of Marmosops creightoni, M. impavi- These and other taxonomically informative dus, M. noctivagus, and M. ocellatus in the characters merit uniform scoring for phylo- sequence (noctivagus (creightoni (impavidus genetic analysis in any comprehensive at- ϩ ocellatus))). The other clade contains tempt to revise the contents of Marmosops, members of the parvidens complex (M. bish- but such exercises are beyond the scope of opi and M. parvidens), two southeastern Bra- this preliminary report. zilian forms (M. incanus and M. paulensis), MOLECULAR COMPARISONS: The ®rst 680 and the Amazonian species M. neblina. Un- base-pairs of the mitochondrial cytochrome- fortunately, there is only weak Bremer and b gene were sequenced from the holotype of bootstrap support for several branches in this Marmosops creightoni by J.L. Patton, whose topology, notably including all of those that generosity in sharing these data allowed us resolve the relationships of creightoni with to make preliminary comparisons with ho- other congeneric forms. mologous didelphid sequences previously NATURAL HISTORY: All of our material of obtained by his laboratory (Patton et al., Marmosops creightoni was collected in the 1996; Mustrangi and Patton, 1997; Patton et valley of the RõÂo Zongo, one of the principal al., 2000). In order to assess the relationships headwaters of the RõÂo Beni on the eastern of M. creightoni with other congeneric slopes of the Cordillera Real. The Valle de forms, we assembled a small dataset repre- Zongo descends precipitously from 4800 to senting nine species of Marmosops, and we Ͻ900 m above sea level, with ¯anking slopes included Gracilinanus microtarsus as an out- that are often in excess of 60Њ. The climate group (table 5). We selected exemplar hap- is very humid, with clouds and mist each lotypes based on sequence length (the lon- morning and afternoon for most of the year. gest available sequence for each species was Between 2000 and 3000 m above sea levelÐ chosen), previous representation in published the known elevational range of M. creighto- reports (Patton et al., 1996; Mustrangi and niÐthe natural vegetation of the Valle de Patton, 1997; Patton et al., 2000), and acces- Zongo is humid montane (``cloud'') forest.8 sibility of cataloged morphological voucher Habitat information associated with speci- material in museums. Most species are rep- mens of M. creightoni taken at three locali- resented by single haplotypes in our analysis, ties in the Valle de Zongo is summarized be- but we included three sequences from geo- low, in order of their collection dates. graphically divergent samples of M. nocti- The ®ve paratypes that G.K. Creighton vagus to assess molecular variation within a collected in 1979 were taken at two localities geographically widespread taxon. along a road that descends the valley and As quanti®ed by pairwise uncorrected se- provides access to a series of hydroelectric quence divergence (table 6), Marmosops generating stations, power lines, and support creightoni is phenetically most similar to M. facilities. His ®rst collecting site was at Cu- noctivagus, from which it differs on average by about 9.7%. Interestingly, of the three an- 8 We use the nontechnical term ``cloud forest'' for alyzed haplotypes of the latter species, M. montane rainforests generally, including the formations creightoni is most divergent (10.9%) from that Grubb (1977) usefully de®ned as Lower Montane Rain Forest, Upper Montane Rain Forest, and Subalpine the geographically proximal Bolivian sample Rain Forest. Synonyms used by Bolivian botanists in- and least divergent (9.2%) from each of the clude ``Ceja de Monte de Yungas'' (corresponding to Peruvian and Brazilian specimens. All other Subalpine Rainforest; e.g., at Cuticucho) and ``Bosque pairwise interspeci®c comparisons in these HuÂmedo MontanÄoso de Yungas'' (corresponding to Up- per Montane Rainforest; e.g., at Saynani). For a critical results are Ͼ10% divergent, with the excep- review of ``cloud forest'' and its technical synonyms tion of a relatively small value (5.7%) be- from a ®eld zoologist's point of view, see Myers (1969); tween M. impavidus and M. ocellatus. for another (botanical) perspective, see Webster (1995). 20 AMERICAN MUSEUM NOVITATES NO. 3466 a Marmosops TABLE 4 Diagnostic Morphological Comparisons Among Five Bolivian Species of 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 21 TABLE 4 (Continued) 22 AMERICAN MUSEUM NOVITATES NO. 3466

TABLE 5 Exemplar Specimens Included in Molecular Analysis

ticucho (ϭ Cuti Khuchu on recent topo- arti®cial clearings under power lines . . . We are very graphic maps), which he recorded as having near treeline here; as one looks up the valley the trees thin out and disappear at a fairly distinct ``line'' about an elevation of 2967 m above sea level (but 300 m above camp. Looking down the valley the see appendix 2: locality 16). The site is de- trees get noticeably higher . . . [3 October 1979] scribed in his ®eldnotes as follows: Both specimens of Marmosops creightoni Since our arrival [six days ago] we have not seen the from Cuticucho were taken in the cloud for- sun for more than an hour a day. Clouds move up the valley in the morning at around 8:30AM and form est with snap traps: one (UMMZ 155999) at just below our camp. Intermittent rain and mist are the base of a tree and the second (UMMZ the order of the day. The temperature here stays rel- 156000) on a low branch. Creighton also col- atively constant, between 55 and 65Њ F [13±18Њ C]. lected ®ve species of sigmodontine rodents The forest here is very dampÐmany lichens, moss and ferns on branches of trees. We have trapped in at this site, of which Akodon mimus, Oxy- three habitats here: cloud forest (relatively undis- mycterus nigrifrons, and two undescribed turbed), 2nd growth forestÐwith much bamboo; and forms of Thomasomys (``sp. 1'' and ``sp. 3''

TABLE 6 Uncorrected Percent Sequence Divergence Among Cytochrome-b Haplotypes from Nine Species of Marmosops 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 23

Fig. 11. The unique minimum-length tree recovered by a branch-and-bound parsimony analysis of cytochrome-b nucleotide sequences representing 11 ingroup haplotypes from 9 Marmosops species and 1 outgroup haplotype from Gracilinanus microtarsus (used to root the tree, not shown). Bremer support and bootstrap values (Ͼ50%) are provided above and below each branch, respectively. The bootstrap values were obtained from branch-and-bound analyses of 1000 pseudoreplicated datasets. Tree length (including autapomorphies) ϭ 689 steps, consistency index (excluding autapomorphies) ϭ 0.48, retention index ϭ 0.44. of Pacheco, 2003) were trapped in the cloud near arti®cial clearings. Little undergrowth beneath forest; Oryzomys levipes was trapped in near- the bamboo except for a few ferns. [6 October 1979] by secondary growth. Additional species of The ®rst four nights of trapping (5±8 Octo- small mammals subsequently trapped near ber) at this site yielded only rodents (the Cuticucho by other researchers include Gra- same species as those found in the cloud for- cilinanus aceramarcae (see Salazar-Bravo et est near Cuticucho plus Microryzomys minu- al., 2002), Akodon cf. boliviensis, and an un- tus and Thomasomys cf. oreas), but two described form of Oligoryzomys (``sp. B'' of specimens of Marmosops creightoni were the ¯avescens group; Carleton and Musser, taken on 9 October and another on 10 Oc- 1989). tober. Creighton's notes for these dates indi- Farther down the valley G.K. Creighton cate that the marsupials were trapped in ``hu- described his camp at a locality he called mid, dark, mossy forest'' on horizontal ``Cement Mine'' (not identi®able on our to- branches near large trees ``covered with pographic maps), with a recorded elevation moss, ferns, bromeliads & vines''. Appar- of 2000 m, as being much warmer than at ently, they were ``not caught coming to the Cuticucho and with taller forest. However, bait but, rather, blundering into the traps as Much of the forest has been cut or burned at one time they tiptoe along their accustomed routes.'' or another in the past 30 years, hence most . . . is The area where all three Marmosops were 2nd growth except in the higher valleys intersecting trapped included ``about the only places this one at more or less right angles. Lichens & moss where high, undisturbed natural forest is ac- on trees are solid in places but not as ``lush'' as near Cuticucho. There are more broadleaf trees and ``pa- cessible enough to set a trapline'', the only paya-like'' trees here. Also large (ϳ10 m) tree ferns. other primary forest being restricted to very Bamboo undergrowth is extensive and very dense steep slopes above the cultivated valley ¯oor. 24 AMERICAN MUSEUM NOVITATES NO. 3466

The holotype of Marmosops creightoni TABLE 7 was collected by EY near the Saynani hy- Measurements (mm) and Weights (g) droelectric generating station at 2500 m on of the Brazilian Holotype and Referred the southeastern side of the valley. The ob- Bolivian Material of Marmosops bishopi viously disturbed forest at this site consisted of patches with a closed canopy but also open areas with isolated trees. The trees were laden with epiphytes, and the ground cover consisted of vines, small bamboo, grasses, moss, and ferns. The holotype was taken on the ground, in a snap trap set in dense grass under a small tree. All of the other small mammals trapped at Saynani by EY and other researchers are sigmodontine rodents, including Akodon mimus, Chibchanomys sp., Microryzomys minutus, Oryzomys levipes, Oxymycterus nigrifrons, and the same undescribed forms of Oligoryzomys and Thoma- somys that occur at Cuticucho. In summary, Marmosops creightoni has been collected in both primary cloud forest and disturbed vegetation in the Valle de Zon- go, where it occurs sympatrically with Gra- cilinanus aceramarcae and several species of sigmodontine rodents between 2000 and 3000 m above sea level. A different species of Marmosops, however, inhabits lower ele- vations in the Zongo valley. Two specimens, low, we explain the morphological and other one from El Vertigo at 1800 m (CBF 4003) criteria that were used to assign names, and and another from Kahua at 1150 m (CBF we note con¯icting identi®cations of the 3880), are both referable to M. noctivagus as same material reported by Anderson (1997), the most authoritative available reference on we recognize that species below. Future col- 9 lecting in the elevational interval between Bolivian mammalogy. 2000 and 1000 m would be helpful in pin- pointing just where these congeneric species Marmosops bishopi occur sympatrically or parapatrically along This appears to be the appropriate name the forested eastern slopes of the Cordillera for Bolivian specimens of the Marmosops Real. parvidens group, of which we have examined eight examples from widely scattered NOTES ON THE TAXONOMY AND localities in Beni, Cochabamba, La Paz, and DISTRIBUTION OF OTHER Santa Cruz (®g. 10). These are all very small BOLIVIAN FORMS animals (table 7) with much shorter molar series (LM Յ 5.8 mm) than any other Boli- In the course of searching museum collec- vian congener. The skulls have rounded su- tions for additional examples of Marmosops praorbital margins that lack distinct beading, creightoni, we examined many other Boli- shallow postorbital constrictions are dorsally vian specimens of Marmosops and sorted visible, there are no palatine fenestrae, and them into phenotypically distinctive groups that we provisionally recognize as species. 9 Because Anderson (1997) seldom cited museum cat- The following accounts summarize our con- alog numbers, we consulted his card ®le of examined specimens (preserved in the AMNH Department of clusions about the taxonomy and distribution Mammalogy archives) to determine the empirical basis of these forms based on currently available for mapped collection localities of Bolivian Marmosops information. For each species recognized be- (Anderson, 1997: ®gs. 493±495). 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 25 the upper canines have small posterior ac- Marmosops impavidus cessory cusps. Although one adult male (MSB 55843) has an almost-naked patch of The application of the name impavidus is skin and a streak of red-stained fur on the problematic because no type material is midline of the throat, two others (CBF 7531, known to exist, and because Tschudi's (1845) LHE 1541) have no trace of a gular gland. original description is diagnostically inade- quate (Tate, 1933: 25, 182). Current usage All three adult males have well-developed, dates from Cabrera (1958), who used the bi- blade-like lateral carpal tubercles, and the nomen M. impavidus for the taxon that Tate scrotum is pale, with white fur and unpig- (1933) recognized as M. caucae. Although mented skin. None of the females we ex- the lost type material is assumed to have amined had well-developed mammae, so it is been collected in the eastern foothills of the not known whether or not this taxon has pec- Peruvian Andes, no exact location was ex- toral teats. plicitly stated in the original description. For There is noteworthy pelage color variation the purposes of this report, we assume that a among the material at hand. All of the spec- small series of specimens (AMNH 230017, imens we examined from Beni, Cochabamba, 230018, 230020, 230022±230027) collected La Paz, and western Santa Cruz have dull near San Pablo (ca. 10Њ27ЈS, 74Њ52ЈW; 275 reddish-brown dorsal fur, self-whitish ventral m above sea level) in the Peruvian depart- fur bordered by more-or-less distinct lateral ment of Pasco represents the typical mor- zones of gray-based hairs, and unicolored phology. tails. By contrast, the single available skin This a medium-size species (table 8), larg- from eastern Santa Cruz (LHE 1541)10 has er than Marmosops bishopi but averaging pale grayish-brown dorsal fur, self-cream smaller than typical specimens of M. nocti- ventral fur (without well- de®ned lateral vagus (see Patton et al. [2000] and below). zones of gray-based hairs), and an indistinct- The dorsal pelage is dull grayish-brown to ly bicolored tail. Despite these pigmental dif- reddish-brown, with drab (grayish) hues pre- ferences, specimens from eastern Santa Cruz dominating among younger specimens and do not differ from the rest of our Bolivian warmer (reddish) tones predominating material in other respects (e.g., measure- among older adults. The ventral pelage of ments and qualitative craniodental traits), most Peruvian and western Brazilian speci- and partial cytochrome-b sequences from mens consists of a median strip of self-white representative exemplars (LHE 1541 and or cream-colored fur from chin to anus, bor- AMNH 268938) differ by only about 3.8% dered by lateral zones of gray-based fur be- (J.L. Patton, personal commun.). tween the fore- and hindlimbs. However, in Specimens that we identify as Marmosops the Peruvian material at hand (which in- bishopi were included among the material cludes specimens from Ucayali and Loreto in that Anderson (1997) reported as M. impav- addition to the Pasco sample mentioned idus (e.g., UMMZ 156014), M. noctivagus above) the relative widths of the median self- keaysi (USNM 579249), and M. parvidens colored and lateral gray-based zones exhibit (AMNH 268938). considerable variation. The metapodials are BOLIVIAN SPECIMENS EXAMINED: Beni, 1kmE covered with pale fur that does not contrast La Embocada (UMMZ 156014); Cochabamba, abruptly in color with the digits. Tails are Cordillera de Mosetenes (CBF 7531); La Paz, sometimes indistinctly bicolored (paler be- Alto RõÂo Madidi (USNM 579249), La Reserva low than above; e.g., AMNH 230017) and (AMNH 268938); Santa Cruz, El Refugio (LHE particolored (paler distally than proximally; 1541, 1553, 1556), San Rafael de AmboroÂ (MSB e.g., AMNH 230027), but other individuals 55843). have almost unicolored tails (e.g., AMNH 76531). Adult males have knob-like lateral carpal tubercles, but none that we examined 10 Unfortunately, both of the other specimens we examined from eastern Santa Cruz (LHE 1553, 1556) con- has a gular gland; the scrotal fur is white and sist of skulls extracted from rotted specimens found in the scrotal epithelium is unpigmented. One pitfall traps. lactating female from Loreto (AMNH 26 AMERICAN MUSEUM NOVITATES NO. 3466

TABLE 8 Measurements (mm) and Weights (g) of Peruvian and Bolivian Specimens of Marmosops impavidus

272709) has 3±1±3 ϭ 7 mammae, but anoth- a juvenile whose broken skull (originally er from the same locality (AMNH 273151) miscataloged as 38711) exhibits incipient su- has 4±1±4 ϭ 9; all of the teats in both in- praorbital beading, very short canines, and dividuals are abdominal/inguinal. small molars (LM ϭ 6.3 mm), traits that In craniodental morphology, Marmosops closely match Tate's (1933) description of impavidus can be distinguished from other yungasensisÐa taxon that we consider to be Bolivian congeners by the absence of supra- a junior synonym of M. noctivagus (see be- orbital beads (usually distinct in M. noctiva- low). Two other Bolivian specimens explic- gus); presence of a shallow but usually dis- itly identi®ed by Anderson as M. impavidus tinct postorbital constriction (usually absent in are UMMZ 126680 (from 20 km NNE Car- M. noctivagus); presence of palatine fenestrae anavi, in La Paz department) and UMMZ (absent in M. bishopi); absence of accessory 156014 (from 1 km E La Embocada, in Beni upper canine cusps (present in M. bishopi); department). The former is a subadult whose and very small, usually conical bullae. lack of supraorbital beading is possibly due The sole Bolivian specimen that we identify as Marmosops impavidus is MSB 57002, to immaturity, but whose lack of a postorbital which was collected sympatrically with M. constriction more closely resembles the ma- noctivagus at Palmira in the lowlands of Pan- terial we refer to M. noctivagus (see below). do department (®g. 12). This specimen has UMMZ 156014, however, is an example of redder dorsal fur and narrower nasals than do M. bishopi (see above). The only additional most Peruvian specimens, but its morpholo- material that Anderson referred to M. impav- gy is otherwise a good match with topotyp- idus consists of MSB 57000 (from Indepen- ical (Peruvian) material. Other Bolivian dencia in Pando department) and MSB specimens previously referred to this species, 58508 (from 2 km S Caranda in Santa Cruz however, appear to represent different taxa. department); both are examples of ``Graci- Anderson (1997: 153) reported two spec- linanus'' unduaviensis, a taxon that is cur- imens identi®ed as Marmosops impavidus rently treated (e.g., by Gardner, 1993) as a from Incachaca in Cochabamba department, junior synonym of G. agilis. but cited only one by catalog number BOLIVIAN SPECIMENS EXAMINED: Pando, Palmira (AMNH 38718); the specimen in question is (MSB 57002). 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 27

Fig. 12. Collection localities of Marmosops impavidus, M. noctivagus, and M. ocellatus in Bolivia. Numbers are keyed to gazetteer entries in appendix 2.

Marmosops noctivagus Tarma in the department of JunõÂn are more- This appears to be the appropriate name or-less topotypical, and we have a large se- for the most-frequently collected species of ries of very similar specimens from Loreto. Marmosops in north-central Bolivia (®g. 12); One Bolivian specimen from La Paz (USNM we consider albiventris, dorothea, keaysi, 579250) and three from Pando (AMNH and yungasensis to be synonyms, as ex- 262402±262404) closely resemble this low- plained below. The name noctivagus is based land Peruvian material and could be referred on Peruvian type material that we have not to the nominotypical subspecies (M. n. noc- examined, but a series of AMNH specimens tivagus) if a trinomial nomenclature were from the Andean piedmont (Ͻ1000 m) near needed. This material comprises relatively 28 AMERICAN MUSEUM NOVITATES NO. 3466

TABLE 9 Measurements (mm) and Weights (g) of Lowland Specimens of Marmosops noctivagus from Peru and Bolivia

large specimens (table 9) with dull reddish- Cochabamba, and La Paz. The specimens in brown dorsal fur, mostly self-whitish ventral question include the type material of doro- fur, pale-furred metapodials that do not con- thea and yungasensis, but an older name for trast abruptly in color with the digits, and the same phenotype is keaysi (from southern more-or-less unicolored (all-dark) tails. All Peru; appendix 1).11 Measurements of rep- of the adult males we examined have well- resentative examples (including relevant type developed gular glands, knobby (not blade- material; table 10) indicate that these high- like) lateral carpal tubercles, and pale scro- land specimens are about the same size as tums with whitish fur and unpigmented skin. typical (lowland) noctivagus, but they have Although we have not seen ¯uid-preserved (on average) slightly broader interorbital re- parous adult females from Bolivia, two Pe- gions, shorter molar rows, and narrower ruvian specimens (AMNH 272782, 273051) fourth upper molars. In addition, the dorsal each have 5±1±5 ϭ 11 mammae, of which fur tends to be slightly longer, the palatal fe- the anteriormost pair is pectoral. Associated nestrae to be somewhat larger, and the bullae skulls have distinct supraorbital beads (best to be more in¯ated in highland examples developed in fully adult individuals; ®g. 4B) than in lowland specimens. Also, whereas and lack dorsally visible postorbital constrictions; palatine fenestrae are present, and the 11 Although Anderson (1997: 152±154) reported auditory bullae are small and conical. The ``Marmosops dorothea'' and ``Marmosops noctivagus upper canine lacks accessory cusps, and M4 keaysi'' as co-occuring at Chijchijpa in the department is very wide. of La Paz, all of the specimens in question (at AMNH and MSB) are taxonomically indistinguishable. Ander- The remaining Bolivian material that we son's records of these taxa from Chijchijpa therefore rep- refer to Marmosops noctivagus is from the resent alternative identi®cations (apparently copied from foothills and lower montane regions of Beni, skin labels), not sympatry. 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 29 most lowland specimens have more-or-less lotype of keaysi is an exceptionally large unicolored (all dark) tails, tails of highland male with a very broad interorbit that is not specimens are usually indistinctly bicolored matched by any paratype or topotype that we (dark above, pale below) and particolored examined. By contrast, the holotype of do- (paler distally than basally). Despite such el- rothea has an unusually narrow interorbit, evation-correlated phenotypic variation, cy- and the holotype of yungasensis has a con- tochrome-b sequences analyzed by Mustran- spicuously short molar toothrow. Unfortu- gi and Patton (1997) and by Patton et al. nately, there are no large series from any lo- (2000) indicate only minor divergence (about cality to serve as a relevant standard by 3%) among lowland and highland haplotypes which such morphometric variation might be of noctivagus-like Marmosops, including assessed. Just among the 11 specimens of specimens from the Amazonian lowlands of ``yungasensis'' from Pitiguaya, however, in- western Brazil (e.g., INPA 2931), the An- terorbital breadth varies from 5.6 to 6.4 mm, dean slopes of southeastern Peru (e.g., MVZ and length of the molar series varies from 6.3 171408), and northern Bolivia (e.g., AMNH to 7.1 mm. Therefore, none of the exception- 268936). The nominal taxon that Tate (1931) al values for these dimensions in table 10 described as albiventris, currently synony- would appear inconsistent with the hypoth- mized with M. impavidus, is morphologically esis that the taxa represent one geographi- indistinguishable from M. noctivagus and has cally variable species. similar mtDNA sequences (Mustrangi and Most of the specimens that we refer to Patton, 1997).12 Marmosops noctivagus were identi®ed by Traditional taxonomic distinctions among Anderson (1997) as M. dorothea or M. noc- forms that we consider conspeci®c with Mar- tivagus keaysi, but some were identi®ed as mosops noctivagus are almost entirely based M. impavidus as explained in the preceding on coat color. Tate (1933: 153±161), for ex- account, and three others (UMMZ 155829, ample, described the dorsal fur as ``varying 156006, 156007) were identi®ed as Gracili- from natal brown . . . to Mars brown'' in nanus agilis unduaviensis. typical noctivagus, by contrast with ``bone BOLIVIAN SPECIMENS EXAMINED: Beni, 1kmE brown . . . or warm sepia'' in keaysi and La Embocada (UMMZ 156008, 156013), Rurren- ``pecan brown'' in dorothea. These chromat- abaque (AMNH 247651); Cochabamba, Incacha- ic differences are real, but similarities among ca (AMNH 38718), Cordillera de Mosetenes these nominal taxa in other morphological (CBF 7527, 7560, 7573, 7577), Tablas Monte (AMNH 275451±275458; MSB 70276±70280, characters, the existence of intermediate pel- 87080); La Paz, 20 km NNE Caranavi (UMMZ age hues in larger series of noctivagus-like 126680), Chijchijpa (AMNH 268936, 268937, eastern-slope specimens obtained in recent 275459; MSB 57004, 68333, 68334, 87092, years, and the modest range of mtDNA var- 87110), El Vertigo (CBF 4003), 13 km SW Ixia- iation among geographically widespread mas (USNM 579250), Kahua (CBF 3880), La Re- samples suggest that such color contrasts are serva (MSB 68335), Pitiguaya (AMNH 72550, not taxonomically signi®cant. 72558±72562, 72564, 72566; UMMZ 155829, Paradoxically, several holotypes in this 156006, 156007), RõÂo Solacama (BMNH complex exhibit atypical craniodental mea- 1.2.1.36, 1.6.7.79), SerranõÂa Bella Vista (AMNH surements, a factor that may have contributed 275460), Ticunhuaya (AMNH 72567); Pando, Palmira (AMNH 262402, 262403), Santa Rosa to the traditional view that these specimens (AMNH 262404). represent distinct taxa. For example, the ho-

12 We examined the holotype of Marmosa caucae al- Marmosops ocellatus biventris (USNM 194378) and the specimens that Mus- trangi and Patton (1997) identi®ed as Marmosops albi- Judging from the large numbers of speci- ventris (MVZ 171408, 173930±173935). All of this ma- mens in museum collections, Marmosops terial ®ts our working diagnosis of M. noctivagus (table ocellatus is the commonest (or most fre- 4) and exhibits no close similarity with M. impavidus. quently trapped) small marsupial throughout In particular, the ventral fur is mostly self-white, the supraorbital margins are distinctly beaded in mature adults, most of Santa Cruz department, from the and all of the adult male skins exhibit well-developed base of the Andes to the Brazilian frontier gular glands. (®g. 12). Currently considered to be a junior 30 AMERICAN MUSEUM NOVITATES NO. 3466

TABLE 10 Measurements (mm) and Weights (g) of Highland Specimens of Marmosops noctivagus, Including the Type Material of Nominal Taxa Originally Described as keaysi, dorothea, and yungasensis

synonym of dorothea (ϭ M. noctivagus in whitish fur and unpigmented skin). Three our usage; see above), M. ocellatus is an un- parous adult female specimens prepared in ambiguously diagnosable taxon that more the ®eld by L.H. Emmons (LHE 1468, 1596, closely resembles M. impavidus in morphol- 1808) each had 6±1±6 ϭ 13 mammae, of ogy and mtDNA sequences (tables 4, 6). which the anteriormost two pairs appear to Externally, Marmosops ocellatus is dis- be pectoral in digital images that we exam- tinctively paler than other Bolivian species, ined. The supraorbital margins of the skull with sandy grayish-brown dorsal fur and are rounded, lacking distinct beads even in self-cream underparts that lack distinct lat- the largest specimens, and a shallow postor- eral zones of gray-based hairs. The metapo- bital constriction is dorsally visible. Palatine dials are covered dorsally with pale (whitish) fenestrae are consistently present, the upper hairs that do not contrast abruptly in color canine lacks accessory cusps, and the audi- with the digits. The tail is distinctly bicolored tory bullae are moderately large. (dark above, pale below) and particolored Marmosops ocellatus differs from M. im- (paler distally than proximally), such that the pavidus by its paler and grayer dorsal col- distal one-third or more of the organ is com- oration, absence of distinct lateral zones of pletely pale in most specimens. Most exam- gray-based ventral hairs, possession of pec- ined males have no trace of a gular gland, toral mammae, and more distinctly bi- and the only exception being the holotype (an un- particolored tail. In addition, the palate is usually large specimen; table 11), in which more extensively fenestrated and the audito- the gland appears to be present. The lateral ry bullae are much larger in M. ocellatus carpal tubercles of adult males are bulbous than in M. impavidus. Although the differ- or knoblike, and the scrotum is pale (with ence between exemplar sequences of these 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 31

TABLE 11 Measurements (mm) and Weights (g) of Selected Specimens of Marmosops ocellatusa

taxa (about 5.7%; table 6) is not as large as Miguel RincoÂn (AMNH 260029), 10 km N San the differences observed among other species RamoÂn (AMNH 261266, 261267), 15 km S Santa recognized by us, the phenotypic distinctive- Cruz (MSB 58510±58514, 59886, 67020), 27 km ness of M. ocellatus and its apparently dis- SE Santa Cruz (59884, 59885), Santa Rosita crete geographic distribution suggest that it (USNM 390022), Tita (MSB 55071). represents a unique evolutionary lineage that should be known by its own name. Unidenti®ed Marmosops Most of the material that we refer to Mar- Among the Bolivian specimens we ex- mosops ocellatus was identi®ed by Anderson amined are three that we are unable to con- (1997) as M. dorothea, but two specimens ®dently identify with any of the taxa recog- (MSB 67020, 87094) were identi®ed as Gra- cilinanus agilis buenavistae, and another nized in this report. Two (AMNH 275461, (AMNH 263549) was identi®ed as Thylamys MSB 67021) were collected 7 km SE Ari- macrurus. ruma at an elevation of 1750 m on the lower BOLIVIAN SPECIMENS EXAMINED: Santa Cruz, 6 Andean slopes of western Santa Cruz de- km by road W AscencioÂn (AMNH 261265), As- partment. Unlike Marmosops ocellatus, erradero Pontons (CBF 6163), Ayacucho (USNM which occurs in the adjacent lowlands of 390571, 390572), Buenavista (BMNH 26.1.5.25 Santa Cruz, these specimens have darker dor- [holotype], 28.2.9.87, 28.2.9.90), 4.5 km N and sal fur, paler (self-white) underparts, and lack 1.5 km E Cerro AmboroÂ (MSB 55844), El Refu- distinct postorbital constrictions. They differ gio (LHE 1569, 1573, 1577, 1596), 3.5 km W from M. impavidus by their grayer dorsal fur, EstacioÂn El Pailon (AMNH 260026±260028; by lacking lateral zones of gray-based ventral MSB 55070), Hacienda el Pelicano (AMNH fur, and by lacking distinct postorbital con- 275462), 7 km E and 3 km N Ingeniero Mora (AMNH 247652), Lago CaimaÂn (USNM 581979), strictions. They differ from all of the forms 2 km SW Las Cruces (AMNH 263549; MSB herein referred to M. noctivagus by lacking 63274), Mangabalito (MJS 005), 3 km SE Mon- interorbital beads. They differ from M. bish- tero (MSB 87094), Palmar (USNM 390569), San opi by having grayer dorsal pelage, well-de- 32 AMERICAN MUSEUM NOVITATES NO. 3466

TABLE 12 Geographic and Elevational Distribution of Small Nonvolant Cloud-Forest Mammals Collected Sympatrically with Marmosops creightoni

veloped palatine fenestrae, much larger teeth the Amazonian (``Acre y Madre de Dios'') (LM ϭ 6.7 mm), and by lacking accessory and ``Yungas'' biogeographic provinces of cusps on C1. Navarro and Maldonado (2002). This species The other unidenti®ed specimen is MSB is also widespread in eastern Peru and west- 63275, a young adult female from the RõÂo ern Brazil, where it has been collected in LimoÂn at 1300 m on the lower Andean similar habitats (Voss and Emmons, 1996; slopes of northeastern Chuquisaca. It has Patton et al., 2000). Future collecting in Bo- browner and somewhat paler dorsal fur than livia will probably show that M. noctivagus does the RõÂo Ariruma material, and self- occurs throughout Pando department, in the cream ventral fur with broad lateral bands of forested parts of La Paz up to about 2000 m gray-based hairs (versus underparts entirely on the eastern Andean slopes, along the for- self-white). However, the crania of all three ested Andean foothills of southernmost Beni, specimens are not dissimilar, and they differ and on the Andean piedmont and lower-mon- from those of recognized Bolivian taxa by tane slopes of Cochabamba to at least 18ЊS. essentially the same characters. The other Bolivian species for which we have examined enough material to character- DISCUSSION ize its ecogeographic distribution (a total of Bolivia encompasses a diversity of land- 37 individuals from 20 localities) is Mar- scapes with different vegetation types that mosops ocellatus, which seems to be wide- might be expected to support distinct faunas. spread in the ``Cerrado'' biogeographic prov- As currently recognized in the literature, ince of Navarro and Maldonado (2002). however, Bolivian species of Marmosops Within this region, however, M. ocellatus has have broadly overlapping distributions that only been trapped in closed-canopy forests, do not coincide with any obvious biogeo- not in the savanna woodlands or shrubby graphic regions or macrohabitats. Although habitats that characterize Cerrado sensu stric- we have not examined all of the material rep- to (L.H. Emmons, in litt.). We expect that resented in published range maps, the spec- future collecting will show that M. ocellatus imens identi®ed in this report suggest that occurs throughout those parts of Santa Cruz some species have much more geographical- receiving about 1000±1500 mm of annual ly restricted and ecologically comprehensible rainfall and in similar habitats of southwest- distributions than do those illustrated by An- ern Brazil (Mato Grosso). derson (1997) and Brown (2004). Although currently known from only three Specimens that ®t our working diagnosis adjacent localities in La Paz, Marmosops of Marmosops noctivagus (a total of 53 in- creightoni is almost certainly more wide- dividuals from 17 localities) are from the spread. Based on the known geographic lowland rain forests and lower-montane ranges of other small nonvolant cloud-forest cloud forests of northern and central Bolivia, mammals that occur in the same elevational a region that approximately corresponds to interval in the Valle de Zongo (table 12), we 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 33

TABLE 13 opi is an unusually eurytopic species, or the Records of Sympatry Among Bolivian material that we consider conspeci®c may Species of Marmosops consist of two or more morphologically (Based on material examined for this report) cryptic taxa. Some of the obvious gaps in our distribution maps of Bolivian Marmosops are certainly artifactual, but others are probably not. Despite considerable collecting effort in the seasonally ¯ooded savannas and gallery forests of central and northern Beni, for example, no specimens of Marmosops have been collected there. The genus is likewise un- known from substantial collections made in the arid thorn-scrub of southern Santa Cruz, eastern Chuquisaca, and eastern Tarija, nor have we seen any material among the many hundreds of small mammals collected in the predict that M. creightoni will eventually be altiplano habitats of southwestern Bolivia. found in wet montane forests from central By contrast, the eastern-slope lower-montane Bolivia to southeastern Peru. Whether or not forests of Chuquisaca and Tarija remain very M. creightoni is really co-distributed with poorly sampled by mammalogists, and it is any of the taxa listed in table 12 remains to possible that the genus is more widely dis- be demonstrated by more intensive collect- tributed there than current records suggest. ing, but it is noteworthy that this assemblage of species is largely restricted to humid east- ACKNOWLEDGMENTS ern-Andean slopes with few convenient points of access for inventory ®eldwork ex- We thank the curators and collection sup- cept along roads where colonization has al- port staffs that hosted our visits and pro- ready caused extensive deforestation. There- cessed the loans that made this study possi- fore, it may be a long time before signi®cant ble, especially Paula Jenkins (BMNH); Julie- range extensions of M. creightoni are report- ta Vargas (CBF); Jorge Salazar-Bravo, Bill ed. Gannon, and Joe Cook (MSB); Phil Myers The material that we refer to Marmosops and Steve Hinshaw (UMMZ); and Al Gard- bishopi (a total of eight individuals from six ner, Linda Gordon, and Helen Kafka localities) is an interesting exception to the (USNM). Bolivian ®eldwork was partially ecogeographically restricted distributions of funded by a grant from the American Society other Bolivian forms. Instead, specimens of of Mammalogists (to TT); and by the Na- M. bishopi have been taken in a diversity of tional Geographic Society, the Weeden Foun- landscapes, where they account for most of dation, and BIOPAT (through grants to M. the known examples of sympatry document- Kessler and S. Herzog). Logistical support in ed by congeneric material examined for this the Valle de Zongo was graciously provided report (table 13). Because strikingly different by the CompanÄõÂa Boliviana de EnergõÂa EleÂc- macrohabitats can sometimes occur in close trica (COBEE). Cristian Olivo helped EY proximity, we consulted ®eld notes associ- with ®eldwork in the Saynani area. Permits ated with Bolivian specimens of M. bishopi to collect in Bolivia were issued by the Di- seeking some common ecological circum- reccioÂn General de Biodiversidad, Ministerio stance in which they might have been taken, de Desarrollo Sostenible y Medio Ambiente. but we found none: captures have been re- Louise Emmons provided useful observa- corded in primary lowland rainforest tions and digital images from her recent (UMMZ 156014, USNM 579249), lowland ®eldwork in eastern Santa Cruz and allowed dry forest (LHE 1541), seasonally ¯ooded us to examine uncataloged material that she evergreen gallery forest (LHE 1553, 1556), collected there. At the AMNH, Neil Duncan and cloud forest (CBF 7531). Either M. bish- expertly extracted and cleaned skulls from 34 AMERICAN MUSEUM NOVITATES NO. 3466

¯uid-preserved specimens, and Pat Brunauer Cabrera, A. 1913. Dos mamõÂferos nuevos de la provided essential bibliographic support. Jim fauna neotropical. Trabajos del Museo de Cien- Patton set new standards for professional cias Naturales (Serie ZooloÂgica) 9: 1±15 ϩ 1 generosity in permitting us unrestricted ac- pl. cess to the cytochrome-b sequences obtained Cabrera, A. 1958[``1957'']. CataÂlogo de los mam- Âferosõ de AmeÂrica del Sur [part 1]. Revista del by his lab, and Marcelo Weksler helped us Museo Argentino de Ciencias Naturales ``Ber- analyze them. We gratefully acknowledge a nardino Rivadavia'' (Ciencias ZooloÂgicas) 4: familiar debt to Pat Wynne, whose artistic 1±307. skills did so much to enhance our descrip- Carleton, M.D., and G.G. Musser. 1989. System- tions of didelphid morphology, and we thank atic studies of oryzomyine rodents (Muridae, Rob Anderson for his technical assistance in Sigmodontinae): a synopsis of Microryzomys. the production of ®gures 10 and 12. Our Bulletin of the American Museum of Natural manuscript was substantially improved by History 191: 1±83. helpful suggestions provided by Rob Ander- Chubb, C. 1919. Notes on collections of birds in son, Louise Emmons, David Flores, Al Gard- the British Museum, from Ecuador, Peru, Bo- livia, and Argentina. Part 1. Tinamidae±Ralli- ner, and Rogerio Rossi, who generously read dae. Ibis 11: 1±55. a preliminary draft when they had many oth- Cordero R., G.A. 2001. Ecological data on Mar- er things to do: we thank them for their time, mosops fuscatus in a lowland tropical forest in effort, and expertise. northern Venezuela. Mammalia 65: 228±231. Creighton, G.K. 1984. Systematic studies on REFERENCES opossums (Didelphidae) and rodents (Criceti- dae). Ph.D. diss., University of Michigan. Ann Allen, J.A. 1900. Descriptions of new American Arbor, MI: University Micro®lms. marsupials. Bulletin of the American Museum Cunha, A.A., and M.V. Vieira. 2002. Support di- of Natural History 13: 191±199. ameter, incline, and vertical movements of four Allen, J.A. 1901. On a further collection of mam- didelphid marsupials in the Atlantic forest of mals from southeastern Peru, collected by Mr. Brazil. Journal of Zoology (London) 258: 419± H.H. Keays, with descriptions of new species. 426. Bulletin of the American Museum of Natural DMA. 1992. Gazetteer of Bolivia, 2nd ed. Wash- History 14: 41±46. ington, DC: Defense Mapping Agency. Allen, J.A., and F.M. Chapman. 1897. On a sec- Emmons, L.H. 1997. Neotropical rainforest mam- ond collection of mammals from the island of mals, 2nd ed. Chicago: University of Chicago Trinidad, with descriptions of new species, and Press. a note on some mammals from the island of Fairchild, G.B., and C.O. Handley, Jr. 1966. Gaz- Dominica, W.I. Bulletin of the American Mu- etteer of collecting localities in Panama. In R.L. seum of Natural History 9: 13±30. Wenzel and V.J. Tipton (editors), Ectoparasites Anderson, R.P. 1999. Preliminary review of the of Panama: 9±22, folding map.Chicago: Field systematics and biogeography of the spiny Museum of Natural History. pocket mice (Heteromys) of Colombia. Revista Flores, D.A., M.M. DõÂaz, and R.M. BaÂrquez. de la Academia Colombiana de Ciencias Ex- 2000. Mouse opossums (Didelphimorphia, Di- actas, FõÂsicas y Naturales 23(suplemento espe- delphidae) of northwestern Argentina: system- cial): 613±630. atics and distribution. Zeitschrift fuÈr SaÈugetier- Anderson, S. 1997. Mammals of Bolivia, taxon- kunde 65: 321±339. omy and distribution. Bulletin of the American Gardner, A.L. 1990 [``1989'']. Two new mammals Museum of Natural History 231: 1±652. from southern Venezuela and comments on the Anthony, H.E. 1922. Preliminary report on Ecua- af®nities of the highland fauna of Cerro de la dorean mammals. No. 2. American Museum Neblina. In K.H. Redford and J.F. Eisenberg Novitates 32: 1±6, map. (editors), Advances in Neotropical mammalo- Brown, B.E. 2004. Atlas of New World marsu- gy: 411±424. Gainesville, FL: Sandhill Crane pials. Fieldiana Zoology (new series) 102: i±vii, Press. [According to A.L. Gardner (personal 1±308. commun.), this volume was published on 6 Burmeister, H. 1856. ErlaÈuterung zur Fauna Bras- February 1990] iliens, enthaltend Abbildungen und ausfuÈhrli- Gardner, A.L. 1993. Order Didelphimorphia. In che Beschreibungen neuer oder ungenuÈgend D.E. Wilson and D.M. Reeder (editors), Mam- bekannter Thier-Arten. Berlin: Druck und Ver- mal species of the world, 2nd ed: 15±23.Wash- lag von Georg Reimer. ington, DC: Smithsonian Institution Press. 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 35

Gardner, A.L., and G.K. Creighton. 1989. A new turvidenskabelige og Matematiske Afhandlin- generic name for Tate's microtarsus group of ger 8: 219±272 ϩ pls. xiv±xxiv. [Preprint 56 South American mouse opossums (Marsupialia: pp., repaginated; not seen. Journal issue pagi- Didelphidae). Proceedings of the Biological So- nated as above published 1841] ciety of Washington 102: 3±7. Malcolm, J.R. 1991. Comparative abundances of Goldman, E.A. 1912. New mammals from eastern Neotropical small mammals by trap height. Panama. Smithsonian Miscellaneous Collec- Journal of Mammalogy 72: 188±192. tions 60: 1±18. Matschie, P. 1916. Bemerkungen uÈber die Gattung Goodwin, G.G., and A.M. Greenhall. 1961. A re- Didelphis L. Sitzungsberichte der Gesellschaft view of the bats of Trinidad and Tobago. Bul- Naturforschender Freunde, Berlin 1916: 259± letin of the American Museum of Natural His- 272. tory 122: 187±302 ϩ 40 pls. Miller, G.S. 1913. Five new mammals from trop- Grubb, P.J. 1977. Control of forest growth and ical America. Proceedings of the Biological So- distribution on wet tropical mountains: with ciety of Washington 26: 31±34. special reference to mineral nutrition. Annual Miranda-Ribeiro, A. de. 1936. Didelphia ou Review of Ecology and Systematics 8: 83±107. Mammalia-Ovovivipara. Revista do Museu Handley, C.O., Jr., and L. Gordon. 1979. New Paulista 20: 245±424. species of mammals from northern South Mustrangi, M.A., and J.L. Patton. 1997. Phylo- America: mouse possums, genus Marmosa geography and systematics of the slender Gray. In J.F. Eisenberg (editor), Vertebrate ecol- mouse opossum Marmosops (Marsupialia, Di- ogy in the northern Neotropics: 65±72. Wash- delphidae). University of California Publica- ington, D.C.: Smithsonian Institution Press. tions in Zoology 130: i±x, 1±86. Hershkovitz, P. 1992. The South American gracile Myers, C.W. 1969. The ecological geography of mouse opossums, genus Gracilinanus Gardner cloud forest in Panama. American Museum and Creighton, 1989 (Marmosidae, Marsupia- Novitates 2396: 1±52. lia): a taxonomic review with notes on general Myers, P., J.L. Patton, and M.F. Smith. 1990. A morphology and relationships. Fieldiana Zool- review of the Boliviensis Group of Akodon ogy, new series 39: 1±56. (Muridae: Sigmodontinae), with emphasis on IGM. 1967. Zongo, 1: 50,000, Hoja 5945 I, Serie Peru and Bolivia. Miscellaneous Publications H731 [topographic map]. La Paz: Instituto Museum of Zoology University of Michigan GeograÂ®co Militar. 177: i±iv, 1±104. Jansa, S.A., and R.S. Voss. 2000. Phylogenetic Navarro, G., and M. Maldonado. 2002. GeografõÂa studies on didelphid marsupials I. Introduction ecoloÂgica de Bolivia, vegetacioÂn y ambientes and preliminary results from nuclear IRBP acuaÂticos. Cochabamba: Centro de EcologõÂa Si- gene sequences. Journal of Mammalian Evo- moÂn I. PatinÄo. lution 7: 43±77. Oliveira, J. de. 1998. Morphometric assessment of Killeen, T.J., and T.S. Schulenberg (editors). 1998. species groups in the South American rodent A biological assessment of Parque Nacional genus Oxymycterus (Sigmodontinae), with tax- Noel Kempff Mercado, Bolivia. RAP Working onomic notes based on the analysis of type ma- Papers 10: 1±372. terial. Unpublished Ph.D. diss., Texas Tech Kirsch, J.A.W., and R.E. Palma. 1995. DNA/DNA University. hybridization studies of carnivorous marsupi- Osgood, W.H. 1913. New Peruvian mammals. als. V. A further estimate of relationships Field Museum of Natural History Zoological among opossums (Marsupialia: Didelphidae). Series 10: 93±100. Mammalia 59: 403±425. Osgood, W.H. 1915. New mammals from Brazil Langguth, A., V.L.A.G. Limeira, and S. Franco. and Peru. Field Museum of Natural History 1997. Novo cataÂlogo do material-tipo da cole- Zoological Series 10: 187±198. cËaÄo de mamõÂferos do Museu Nacional. Publi- Pacheco, V.R. 2003. Phylogenetic analyses of the cacËoÄes Avulsas do Museu Nacional 70: 1±29. Thomasomyini (Muroidea: Sigmodontinae) Lorini, M.L., J.A. de Oliveira, and V.G. Persson. based on morphological data. Unpublished 1994. Annual age structure and reproductive Ph.D. diss., City University of New York. patterns in Marmosa incana (Lund, 1840) (Di- Palma, R. E., and A. E. Spotorno. 1999. Molec- delphidae, Marsupialia). Zeitschrift fuÈr SaÈuge- ular systematics of marsupials based on the tierkunde 59: 65±73. rRNA 12S mitochondrial gene: the phylogeny Lund, P.W. 1840. [Preprint of] Blik paa Brasiliens of Didelphimorphia and of the living fossil mi- Dyreverden foÈr sidste Jordomvaeltning. Tredie crobiotheriid Dromiciops gliroides. Molecular Afhandling: Fortsaettelse af Pattedyrene. Kon- Phylogenetics and Evolution 13: 525±535. gelige Danske Videnskabernes Selskabs Na- Patton, J.L., and L.P. Costa. 2003. Molecular phy- 36 AMERICAN MUSEUM NOVITATES NO. 3466

logeography and species limits in rainforest di- Swofford, D.L. 1998. PAUP*, Phylogenetic anal- delphid marsupials of South America. In M.E. ysis using parsimony, beta version 4.0b10. Sun- Jones, C.R. Dickman, and M. Archer (editors), derland, MA: Sinauer Associates. Predators with pouches: the biology of carniv- Tate, G.H.H. 1931. Brief diagnoses of twenty-six orous marsupials: 63±81.Melbourne: CSIRO apparently new forms of Marmosa (Marsupia- Press. lia) from South America. American Museum Patton, J.L., S.F. dos Reis, and M.N.F. da Silva. Novitates 493: 1±14. 1996. Relationships among didelphid marsupi- Tate, G.H.H. 1933. A systematic revision of the als based on sequence variation in the mito- marsupial genus Marmosa with a discussion of chondrial cytochrome b gene. Journal of Mam- the adaptive radiation of the murine opossums malian Evolution 3: 3±29. (Marmosa). Bulletin of the American Museum Patton, J.L., M.N.F. da Silva, and J.R. Malcolm. of Natural History 66: 1±250 ϩ 26 pls. 2000. Mammals of the Rio JuruaÂ and the evo- Thomas, O. 1896. On new small mammals from lutionary and ecological diversi®cation of the Neotropical region. Annals and Magazine Amazonia. Bulletin of the American Museum of Natural History, Series 6, 18: 301±314. of Natural History 244: 1±306. Thomas, O. 1900. Descriptions of new Neotropi- Paynter, R.A., Jr. 1982. Ornithological gazetteer cal mammals. Annals and Magazine of Natural of Venezuela. Cambridge, MA: Museum of History, Series 7, 5: 217±222. Comparative Zoology (Harvard University). Thomas, O. 1911. New mammals from tropical Paynter, R.A., Jr. 1992. Ornithological gazetteer South America. Annals and Magazine of Nat- of Bolivia, 2nd ed. Cambridge, MA: Museum ural History, Series 8, 7: 513±517. of Comparative Zoology (Harvard University). Thomas, O. 1913. New mammals from South Paynter, R.A., Jr. 1993. Ornithological gazetteer America. Annals and Magazine of Natural His- of Ecuador, 2nd ed. Cambridge, MA: Museum tory, Series 8, 12: 567±574. of Comparative Zoology (Harvard University). Thomas, O. 1920. On mammals from the lower Paynter, R.A., Jr. 1997. Ornithological gazetteer Amazons in the Goeldi Museum, Para. Annals of Colombia, 2nd ed. Cambridge, MA: Muse- and Magazine of Natural History, Series 9, 6: um of Comparative Zoology (Harvard Univer- 266±283. sity). Thomas, O. 1924. New South American mam- Paynter, R.A., Jr., and M.L. Traylor, Jr. 1991. Or- mals. Annals and Magazine of Natural History, nithological gazetteer of Brazil [2 vols.]. Cam- Series 9, 13: 234±237. bridge, MA: Museum of Comparative Zoology Thomas, O. 1927a. The Godman-Thomas Expe- (Harvard University). dition to Peru.ÐV. On mammals collected by Pine, R.H. 1981. Reviews of the mouse opossums Mr. R.W. Hendee in the Province of San Mar- Marmosa parvidens Tate and Marmosa invicta tin, N. Peru, mostly at Yurac Yacu. Annals and Goldman (Mammalia: Marsupialia: Didelphi- Magazine of Natural History, Series 9, 19: dae) with description of a new species. Mam- 361±375. malia 45: 55±70. Thomas, O. 1927b. The Godman-Thomas Expe- Reig, O.A., J.A.W. Kirsch, and L.G. Marshall. dition to Peru.ÐVI. On mammals from the up- 1985. New conclusions on the relationships of per Huallaga and neighboring highlands. An- the opossum-like marsupials, with an annotated nals and Magazine of Natural History, Series 9, classi®cation of the Didelphimorphia. Amegh- 20: 594±608. iniana 21: 335±343. Tschudi, J. J. von. 1844±1846. Untersuchungen Salazar-Bravo, J., E. Yensen, T. Tarifa, and T.L. uÈber die Fauna Peruana. St. Gallen: Scheitlin Yates. 2002. Distributional records of Bolivian und Zollikofer. [Published in parts; pp. 133± Mammals. MastozoologõÂa Neotropical 9: 70± 188, in which the descriptions of Marmosops 78. impavidus and M. noctivagus occur, appeared Sherborn, C.D. 1922. Index animalium [section 2, in 1845 according to Sherborn, 1922: cxxiv] part 1]. London: Bitish Museum (Natural His- Vieira, E.M., and E.L.A. Monteiro-Filho. 2003. tory). Vertical strati®cation of small mammals in the Stephens, L., and M.L. Traylor, Jr. 1983. Ornitho- Atlantic rainforest of south-eastern Brazil. Jour- logical gazetteer of Peru. Cambridge, MA: Mu- nal of Tropical Ecology 19: 501±507. seum of Comparative Zoology (Harvard Uni- Voss, R.S., and L.H. Emmons. 1996. Mammalian versity). diversity in Neotropical lowland rainforests: a Stephens, L., and M.L. Traylor, Jr. 1985. Ornitho- preliminary assessment. Bulletin of the Amer- logical gazetteer of the Guianas. Cambridge, ican Museum of Natural History 230: 1±115. MA: Museum of Comparative Zoology (Har- Voss, R.S., and S.A. Jansa. 2003. Phylogenetic vard University). studies on didelphid marsupials II. Nonmolec- 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 37

ular data and new IRBP sequences: separate phylogenetic puzzle from the Chaco of northern and combined analyses of didelphine relation- Argentina. American Museum Novitates 3442: ships with denser taxon sampling. Bulletin of 1±18. the American Museum of Natural History 276: Webster, G.L. 1995. The panorama of Neotropical 1±82. cloud forests. In S.P. Churchill, H. Balslev, E. Voss, R.S., D.P. Lunde, and N.B. Simmons. 2001. Forero, and J.L. Luteyn (editors), Biodiversity The mammals of Paracou, French Guiana: a and conservation of Neotropical montane for- Neotropical rainforest fauna. Part 2. Nonvolant ests: 53±77.New York: New York Botanical species. Bulletin of the American Museum of Garden. Natural History 263: 1±236. Wible, J.R. 2003. On the cranial osteology of the Voss, R.S., A.L. Gardner, and S.A. Jansa. 2004. short-tailed opossum Monodelphis brevicauda- On the relationships of ``Marmosa'' formosa ta (Didelphidae, Marsupialia). Annals of the Shamel, 1930 (Marsupialia: Didelphidae), a Carnegie Museum 72: 137±202.

APPENDIX 1

TYPE LOCALITIES OF NOMINAL SPECIES BRAZIL OF MARMOSOPS 5. AmapaÂ, Rio Amapari, Serra do Navio Below we list the type localities of all nominal [0Њ59ЈN, 52Њ03ЈW; ca. 100 m (Paynter and taxa currently referred to the genus Marmosops, Traylor, 1991)]. Type locality of pinheiroi. including the new species described herein. Itali- 6. Amazonas, HuitanaaÄ[7Њ40ЈS, 65Њ46ЈW; ca. cized place names are those of the largest political 50 m (Paynter and Traylor, 1991)]. Type division (state/department/province) within each locality of purui, misspelled ``Hyutana- country. Unless reported by the original describer, ham'' by Miller (1913). geographic coordinates and elevation above sea 7. Bahia, LamaraÄo [11Њ45ЈS, 38Њ55ЈW; 291 m level are provided in square brackets with a cited (Paynter and Traylor, 1991)]. Type locality secondary source for these data. Numbered local- of bahiensis. ities are mapped in ®gure 1. 8. Mato Grosso, Aripuanan [7Њ35ЈS, 60Њ26ЈW; no elevation (Paynter and Traylor, 1991)]. BOLIVIA Type locality of stollei. The coordinates are those at the con¯uence of the Rio AripuanaÄ 1. La Paz, RõÂo Solacama [ca. 16 16 S, Њ Ј and the Rio Roosevelt, where the collector 67Њ24ЈW; 2300 m (Paynter, 1992)]. Type of the holotype (E. Stolle) is known to have locality of dorothea. Thomas (1911: 517) worked (Paynter and Traylor, 1991: 37). misspelled the type locality ``Solocame'' 9. Mato Grosso, 264 km N (by road) Xav- and stated that the type was collected by antina, Serra do Roncador (12 51 S, P.O. Simons on 26 January 1901. The co- Њ Ј ordinates given above are those of Choca- 51Њ46ЈW; ca. 533 m). Type locality of bish- chaca on the RõÂo Tamampaya, where Si- opi. mons is known to have camped at 1200 m 10. Minas Gerais, Rio das Velhas, near Lagoa from 10 to 29 January 1901 (Chubb, 1919: Santa [19Њ38ЈS, 43Њ53ЈW; 760 m (Paynter 5). Presumably, Simons was running trap- and Traylor, 1991)]. Type locality of in- lines from his base on the Tamampaya up canus. the valley of the nearby Solacama, but no ±. Minas Gerais,? Porto Alegre [not located]. documentary evidence is apparently avail- Type locality of scapulatus. Burmeister able to support this conjecture. (1856) only said that the type was collected 2. La Paz, Saynani hydroelectric plant in Minas Gerais, but Matschie (1916) gave (16Њ07ЈS, 68Њ05ЈW; 2500 m) on southeast the type locality as Porto Alegre, possibly side of the Zongo Valley. Type locality of a fazenda no longer in existence or too creightoni. small to appear in maps and gazetteers. 3. La Paz, RõÂo Unduavi, Pitiguaya, La Florida 11. ParaÂ, Nova Area Experimental, Utinga, [ca. 16Њ21ЈS, 67Њ47ЈW; 1785 m (Anderson, BeleÂm [1Њ27ЈS, 48Њ29ЈW; ca. 10 m (Paynter 1997)]. Type locality of yungasensis. and Traylor, 1991)]. Type locality of wood- 4. Santa Cruz, Buenavista [17Њ27ЈS, 63Њ40ЈW; alli. ca. 400 m (Paynter, 1992)]. Type locality 12. ParaÂ, Rio Tapajos, Vila Braga [4Њ25ЈS, of ocellatus. 56Њ17ЈW; ca. 100 m (Paynter and Traylor, 38 AMERICAN MUSEUM NOVITATES NO. 3466

1991)]. Type locality of collega, misspelled 25. Cusco, Torontoy ([13Њ10ЈS, 72Њ30ЈW; Ste- ``Villa Braga'' by Thomas (1920: 281). phens and Traylor, 1983], 2438 m). Type 13. Rio de Janeiro, TeresoÂpolis [22Њ26ЈS, locality of albiventris. 42Њ59ЈW; 902 m (Paynter and Traylor, 26. JunõÂn, MontanÄa de Vitoc, near Chancha- 1991)]. Type locality of paulensis, spelled mayo [ca. 11Њ03ЈS, 75Њ19ЈW; no elevation, ``Therezopolis'' by Tate (1931: 8). but presumably Ͼ500 m (Stephens and Traylor, 1983)]. Alleged type locality of COLOMBIA impavidus and noctivagus. Although 14. 14. Antioquia, 9 km S Valdivia ([7Њ06ЈN, Tschudi (1845) did not state exactly where 75Њ28ЈW; Anderson, 1999], elev. 1400 m). his material was collected, Tate (1933: Type locality of handleyi. 152±153) inferred that the type of nocti- 15. Cundinamarca, FusagasugaÂ[4Њ21ЈN, vagus was collected here based on a pub- 74Њ22ЈW; 1746 m (Paynter, 1997)]. Type lished itinerary of Tschudi's collecting ac- locality of perfuscus. tivities. Cabrera (1958: 16) assumed that 16. Valle del Cauca, RõÂo Cauquita, near Cali the type of impavidus came from the same ([3Њ27ЈN, 76Њ31ЈW; Paynter, 1997], 1000 place. m). Type locality of caucae. Thomas 27. JunõÂn, Utcuyacu ([ca. 11Њ12ЈS, 75Њ28ЈW; (1900: 222) said the type was collected at Stephens and Traylor, 1983], 1463 m). ``Rio Caqueta, a tributary of the Cauca, Type locality of juninensis. near Cali'', but see Tate (1933: 178). 28. Loreto, Yurimaguas ([5Њ54ЈS, 76Њ05ЈW; Stephens and Traylor, 1983], 183 m). Type ECUADOR locality of neglectus. 29. Puno, Inca Mines ([ca. 13Њ51ЈS, 69Њ41ЈW; 17. El Oro, Portovelo ([3Њ43ЈS, 79Њ39ЈW; Stephens and Traylor, 1983], 1829 m). Paynter, 1993], 610 m). Type locality of Type locality of keaysi. Allen (1900) stated oroensis. that the type was collected at Juliaca, but 18. Loja, Celica ([4Њ07ЈS, 79Њ59ЈW; Paynter, later corrected himself (Allen, 1901). 1993], 2103 m). Type locality of celicae. 30. San MartõÂn, Yurac Yacu (ca. 762 m), about 19. Napo, Coca [ϭ Puerto Francisco de Orel- 20 miles WNW of Moyobamba [6Њ03ЈS, lana; 0Њ28ЈS, 76Њ58ЈW, ca. 260 m (Paynter, 76Њ58ЈW; Stephens and Traylor, 1983]. 1993)]. Type locality of politus. Type locality of lugendus. 20. Pichincha, Mindo ([0Њ02ЈS, 78Њ48ЈW; 31. Ucayali,RõÂo Ucayali, Lagarto [10Њ40ЈS, Paynter, 1993], 1280 m). Type locality of 73Њ54ЈW; Ͻ500 m (Stephens and Traylor, sobrinus. 1983)]. Type locality of ucayalensis. GUYANA TRINIDAD AND TOBAGO 21. Demerara-Mahaica, Hyde Park [6Њ30ЈN, 58Њ16ЈW; ca. 100 m (Stephens and Traylor, 32. Trinidad, Caparo [10Њ27ЈN, 61Њ20ЈW; 46 m 1985)]. Type locality of parvidens. (Goodwin and Greenhall, 1961)]. Type locality of carri. PANAMA 22. DarieÂn, Cana ([7Њ47ЈN, 77Њ42ЈW; Fairchild VENEZUELA and Handley, 1966], 610 m). Type locality 33. Amazonas, Camp VII (0Њ51ЈN, 65Њ58ЈW; of invictus. 1800 m), Cerro de la Neblina. Type locality of neblina. PERU 34. FalcoÂn, 14 km ENE Mirimire, near La Pas- 23. Amazonas, Tambo Carrizal ([ca. 6Њ47ЈS, tora (11Њ12ЈN, 68Њ37ЈW; 150 m). Type lo- 77Њ52ЈW; Stephens and Traylor, 1983], ca. cality of cracens. 2743 m). Type locality of leucastrus. 35. MeÂrida,RõÂo Albarregas ([probably near 24. Amazonas, Tambo Ventija ([6Њ11ЈS, MeÂrida at 8Њ36ЈN, 71Њ08ЈW; Paynter, 77Њ33ЈW; Stephens and Traylor, 1983], ca. 1982], 1630 m). Type locality of fuscatus, 2743 m), 10 mi E Molinopampa. Type lo- misspelled ``Abbaregas'' by Thomas cality of madescens. (1896: 314). 2004 VOSS ET AL.: BOLIVIAN MARMOSOPS 39

APPENDIX 2

GAZETTEER OF EXAMINED SPECIMENS OF derson, H. Gomez, and R. Tejada, 6±8 July BOLIVIAN MARMOSOPS 1992). 10. La Paz, 13 km SW Ixiamas (13Њ53ЈS, This gazetteer includes all localities from which 68Њ15ЈW; 400 m): Marmosops noctivagus we personally examined Bolivian specimens of (coll. L.H. Emmons, 2 June 1990). Marmosops. Italicized place names are those of 11. La Paz, La Reserva (15Њ43ЈS, 67Њ31ЈW; currently recognized departments; boldface type 940 m): Marmosops bishopi (coll. T. Tiu- identi®es collection localities as they appear in the rina, 22 July 1992) and M. noctivagus text of this report. Unless recorded by the collec- (coll. AMNH/MSB party, 28 July 1992). tor, geographic coordinates and elevation above 12. La Paz, RõÂo Unduavi, Pitiguaya, La Flor- sea level are provided in square brackets with a ida [ca. 16Њ21ЈS, 67Њ47ЈW; 1785 m (An- cited secondary source for these data. The species derson, 1997)]: Marmosops noctivagus name(s), name(s) of the collector(s), and date(s) (coll. G.H.H. Tate, 9±14 May 1926; G.K. of collection are separated from each locality en- Creighton, 30 November 1979). try by a colon. Numbered localities are plotted in 13. La Paz, RõÂo Solacama [ca. 16Њ16ЈS, ®gures 10 and 12. 67Њ24ЈW; 2300 m (Paynter, 1992)]: Mar- 1. Beni, 1 km E La Embocada, Estancia la mosops noctivagus (coll. P.O. Simons, 26 CabanÄa [15Њ03ЈS, 66Њ58ЈW; 600 m (Ander- June 1901). See appendix 1 for comments son, 1997)]: Marmosops bishopi and M. on this locality. noctivagus (coll. G.K. Creighton, 10±13 14. La Paz, SerranõÂa Bella Vista [15Њ42ЈS, November 1979). 67Њ29ЈW; 1300±1525 m (Anderson, 2. Beni, Rurrenabaque (14Њ28ЈS, 67Њ34ЈW; 1997)]: Marmosops noctivagus (coll. S. 227 m): Marmosops noctivagus (coll. R.H. Anderson, 24 July 1992). Pine, 1 September 1980). 15. La Paz, Ticunhuaya [15Њ28ЈS, 68Њ18ЈW; 3. Chuquisaca, RõÂo LimoÂn (19Њ33ЈS, 1540 m (Anderson, 1997)]: Marmosops 64Њ08ЈW; 1300 m): Marmosops sp. (coll. noctivagus (coll. G.H.H. Tate, 23 April R.L. Cuellar, 5 August 1990). 1926). 4. Cochabamba, Incachaca [17Њ14ЈS, 16. La Paz, [Valle de Zongo,] Cuticucho ([ϭ 65Њ41ЈW; 2347 m (Anderson, 1997)]: Mar- Cuti Khuchu at 16Њ08ЈS, 68Њ07ЈW; IGM, mosops noctivagus (coll. L.E. Miller and 1967]; ``2967 m''): Marmosops creightoni H.S. Boyle, 15 May 1915). (coll. G.K. Creighton, 29 September±1 Oc- 5. Cochabamba, Cordillera de Mosetenes tober 1979). There is some question about (16Њ14ЈS, 66Њ25ЈW; 1300±1600 m): Mar- exactly where Creighton camped at Cuti- mosops bishopi (coll. E. Yensen, 3 Septem- cucho. His ®eld notes say 2967 m, but this ber 2003) and M. noctivagus (coll. D. Em- is oddly similar to the posted elevation on bert, F. Guerra, and T. Tarifa, 2±13 Septem- road signs that now give the elevation of ber 2003). Cuticucho as 2697, so Creighton's ®gure 6. Cochabamba, Tablas Monte, including might have resulted from an accidental 4.4 km by road N (17Њ04ЈS, 66Њ01ЈW; transposition of two numerals. Modern to- 1833±2100 m) and 9.5 km by road NE pographic maps (IGM, 1967) give the el- (17Њ02ЈS, 65Њ59ЈW; 1500 m): Marmosops evation at the road adjacent to the Cuticu- noctivagus (coll. M.L. Campbell, S.L. cho power plant as 2800 m. Gardner, S.C. Peurach, and C.T. Seaton, 17. La Paz, [Valle de Zongo,] El Vertigo 13±17 July 1993). (16Њ05ЈS, 68Њ02ЈW, 1800 m): Marmosops 7. La Paz, Iturraldi, Alto RõÂo Madidi [ϭ noctivagus (coll. F. Guerra, 15 August ``Moira Camp'' or ``Moire'' in Iturralde 1993). province; 13Њ35ЈS, 68Њ46ЈW; 270 m (An- 18. La Paz, [Valle de Zongo,] Kahua (ca. derson, 1997)]: Marmosops bishopi (coll. 16Њ02ЈS, 67Њ59ЈW, 1150 m): Marmosops L.H. Emmons, 27 May 1990). noctivagus (coll. T. Tarifa and M. Tarifa, 8. La Paz, 20 km NNE Caranavi [15Њ42ЈS, 23 January 1995). An alternative spelling 67Њ35ЈW; 610 m (Anderson, 1997)]: Mar- on modern topographic maps is ``Khaua''. mosops noctivagus (coll. G.K. Creighton, 8 19. La Paz, [Valle de Zongo,] Saynani May 1978). (16Њ07ЈS, 68Њ05ЈW; 2500 m): Marmosops 9. La Paz, Chijchijpa, 20 km by road from creightoni (coll. E. Yensen, 5 August Coroico (16Њ09ЈS, 67Њ45ЈW; 1114±1400 1999). m): Marmosops noctivagus (coll. S. An- 20. La Paz, [Valle de Zongo,] Cement Mine 40 AMERICAN MUSEUM NOVITATES NO. 3466

([?ca. 16Њ05ЈS, 68Њ03ЈW], 2000 m): Mar- Kempff Mercado: Marmosops ocellatus mosops creightoni (coll. G.K. Creighton, (coll. E. Cuellar, 27 September 1995). 4±11 October 1979). 33. Santa Cruz, 2 km SW Las Cruces 21. Pando, Palmira (11Њ42ЈS, 67Њ56ЈW; 170 (17Њ47ЈS, 63Њ22ЈW; 480 m): Marmosops m): Marmosops impavidus and M. nocti- ocellatus (coll. S. Anderson, 5 July 1990). vagus (coll. L.A. Ruedas, 4 August 1986). 34. Santa Cruz, Mangabalito [13Њ55ЈS, 22. Pando, Santa Rosa (12Њ13ЈS, 68Њ24ЈW; 60Њ30ЈW; ca. 200 m (DMA, 1992)], Parque 180 m): Marmosops noctivagus (coll. L.A. Nacional Noel Kempff Mercado: Marmo- Ruedas, 1 August 1986). sops ocellatus (coll. M.J. Swarner, 4 Oc- 23. Santa Cruz, 6 km by road W AscencioÂn tober 2000). Spelled ``Mangabalsinho'' in (15Њ43ЈS, 63Њ09ЈW; 240 m): Marmosops DMA (1992); mapped by Killeen and ocellatus (coll. I.D. Mercado T., August Schulenberg (1998: ®g. 1) closer to 1985). 13Њ50ЈS. 24. Santa Cruz, Aserradero Pontons 35. Santa Cruz, 3 km SE Montero, 1kmN (17Њ05ЈS, 59Њ34ЈW; no elevation): Mar- Villa Copacabana (17Њ23ЈS, 63Њ15ЈW; 250 mosops ocellatus (coll. H. Aranibar and J. m): Marmosops ocellatus (coll. AMNH/ Vargas, 19 April 1999). MSB party, 25 June 1991). 25. Santa Cruz, Ayacucho [17Њ51ЈS, 63Њ20ЈW 36. Santa Cruz, Palmar [17Њ51ЈS, 63Њ09ЈW; no (Anderson, 1997); no elevation]: Marmo- elevation (Anderson, 1997)]: Marmosops sops ocellatus (coll. F. Bercerra and W.K. ocellatus (coll. D.R. Hadden, 13 August Kerr, 24 August 1966). 1966). 26. Santa Cruz, Buenavista ([17Њ27ЈS, 37. Santa Cruz, RõÂo Ariruma, 7 km by road 63Њ40ЈW; Paynter, 1992], 450±500 m): SE Ariruma (18Њ20ЈS, 64Њ13ЈW; 1750 m): Marmosops ocellatus (coll. J. Steinbach, 6 Marmosops sp. (coll. J.A. Cook, 5 June August 1924). 1991). 27. Santa Cruz, 4.5 km N and 1.5 km E Cer- 38. Santa Cruz, San Miguel RincoÂn (17Њ23ЈS, ro AmboroÂ, RõÂo Pitasama (17Њ44ЈS, 63Њ32ЈW; 300 m): Marmosops ocellatus 63Њ39ЈW; 620 m): Marmosops ocellatus (coll. N. Olds, 12 August 1984). (coll. B.R. Riddle, 15 August 1985). 39. Santa Cruz, San Rafael de AmboroÂ 28. Santa Cruz, El Refugio (14Њ46ЈS, 61Њ02ЈW; (17Њ36ЈS, 63Њ36ЈW; 400 m): Marmosops ca. 200 m), Parque Nacional Noel Kempff bishopi (coll. B.R. Riddle, 31 July 1985). Mercado: Marmosops bishopi (coll. L.H. 40. Santa Cruz, 10 km N San RamoÂn Emmons, 6±15 November 1998) and M. (16Њ36ЈS, 62Њ42ЈW; 250 m): Marmosops ocellatus (coll. L.H. Emmons; 31 August± ocellatus (coll. S. Anderson and N. Olds, 7 September 1999, 13 September 2000). 7 August 1985). 29. Santa Cruz, 3.5 km W EstacioÂn El PailoÂn 41. Santa Cruz, 15 km S Santa Cruz (17Њ39ЈS, 62Њ45W; 300 m): Marmosops (17Њ53ЈS, 63Њ07ЈW; 400 m): Marmosops ocellatus (coll. S. Anderson, 22 September ocellatus (coll. MSB parties, 22±25 July 1984). 1987 and 16 June 1988; AMNH/MSB par- 30. Santa Cruz, Hacienda El Pelicano, 3km ty, 16 June 1991). N Zanja Honda (18Њ16ЈS, 63Њ11ЈW; 500 42. Santa Cruz, 27 km SE Santa Cruz m): Marmosops ocellatus (coll. M. Camp- (17Њ58ЈS, 63Њ03ЈW; 365 m): Marmosops bell, 4 July 1991). ocellatus (coll. MSB party, 28 May 1988). 31. Santa Cruz, 7 km E and 3 km N Ingeni- 43. Santa Cruz, Santa Rosita [17Њ31ЈS, ero Mora ([18Њ10ЈS, 63Њ16ЈW; Anderson, 63Њ11ЈW; no elev. (Anderson, 1997)], 3 km 1997] 488±579 m): Marmosops ocellatus S Warnes: Marmosops ocellatus (coll. (coll. P.A. Remsik, 10 August 1980). M.L. Kuns, 29 July 1965). 32. Santa Cruz, Lago CaimaÂn [13Њ36ЈS, 44. Santa Cruz, Tita (18Њ25ЈS, 62Њ10ЈW; 60Њ55ЈW; ca. 200 m (Killeen and Schulen- 300m): Marmosops ocellatus (coll. berg, 1998)], Parque Nacional Noel AMNH/MSB party, 19 October 1984).