A REVISION OF THE DIDELPHID MARSUPIAL GENUS MARMOSA PART 1. THE SPECIES IN TATE’S ‘MEXICANA’ AND ‘MITIS’ SECTIONS AND OTHER CLOSELY RELATED FORMS ROGE´ RIO V. ROSSI Universidade Federal de Mato Grosso Instituto de Biocieˆncias Departamento de Biologia e Zoologia Av. Fernando Correˆa da Costa s/n, Cuiaba´, MT, Brazil, CEP 78060-900 ROBERT S. VOSS Department of Mammalogy American Museum of Natural History DARRIN P. LUNDE Department of Mammalogy American Museum of Natural History BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Number 334, 83 pp., 30 figures, 8 tables Issued June 3, 2010 Copyright E American Museum of Natural History 2010 ISSN 0003-0090 CONTENTS Abstract........................................................3 Introduction . .................................................... 3 MaterialsandMethods............................................ 5 Ontogenetic Variation and Sexual Dimorphism............................. 10 Comparative Morphology ............................................ 12 External Characters .............................................. 12 Craniodental Characters ........................................... 16 Summary...................................................... 22 SpeciesAccounts.................................................. 23 Marmosa mexicana ............................................... 23 Marmosa zeledoni ................................................ 37 Marmosa isthmica ................................................ 42 Marmosa robinsoni ............................................... 48 Marmosa xerophila ............................................... 56 Marmosa simonsi ................................................ 58 Marmosa rubra .................................................. 61 Discussion . .................................................... 64 Acknowledgments.................................................65 References....................................................... 66 Appendix........................................................72 2 ABSTRACT We revise the nominal species of mouse opossums currently synonymized with Marmosa mexicana Merriam, 1897, and M. robinsoni Bangs, 1898, which include all of the trans-Andean taxa currently assigned to the nominotypical subgenus of Marmosa. In addition, we redescribe two other species that appear to be closely related to M. mexicana and M. robinsoni based on morphological or molecular citeria: M. rubra Tate, 1931, and M. xerophila Handley and Gordon, 1979. Based on first-hand examination of holotypes and other material (about 1500 specimens in total), we additionally recognize M. isthmica Goldman, 1912, and M. simonsi Thomas, 1899 (both currently synonymized with M. robinsoni), and M. zeledoni Goldman, 1917 (currently synonymized with M. mexicana), as valid species. For each of the seven species recognized as valid herein (M. mexicana, M. zeledoni, M. isthmica, M. robinsoni, M. xerophila, M. simonsi, M. rubra), we describe and illustrate diagnostic external and craniodental characters, tabulate measurement data from adult specimens, list all known examples of sympatry, and map geographic ranges based on specimens examined. The species newly recognized as valid herein, all of which occur in Central America and/or northwestern South America, substantially increase the known diversity of trans-Andean mouse opossums, but it is not currently known whether or not these represent a distinct radiation within the genus Marmosa. INTRODUCTION within Gracilinanus (see Voss et al., 2004, 2005). Species of the genus Marmosa, commonly Despite these refinements, the taxonomic known as mouse opossums (fig. 1), are long- status of Marmosa remains problematic tailed, black-masked, pouchless didelphid because all molecular phylogenies published marsupials that inhabit a wide range of to date indicate that the genus (as understood tropical and subtropical habitats from Mex- by Gardner and Creighton, 1989) is para- ico to Argentina. Marmosa was last revised phyletic with respect to Micoureus (fig. 2). by Tate (1933), who recognized several Obviously, several alternative classifications species groups that have subsequently been would be consistent with such results: either elevated to generic rank (table 1). In the (1) Micoureus could be regarded as a junior classification proposed by Gardner and synonym of Marmosa; or (2) Micoureus Creighton (1989), the species in Tate’s could be regarded as a valid subgenus of ‘‘Cinerea Group’’ were referred to the genus Marmosa; or (3) Marmosa could be restricted Micoureus Lesson, 1842; those in his ‘‘Mur- to M. murina (the type species). Of these, the ina Group’’ to Marmosa Gray, 1821; those in his ‘‘Noctivaga Group’’ to Marmosops first option would result in the loss of a useful Matschie, 1916; and those in his ‘‘Elegans and familiar name for a well-supported clade Group’’ to Thylamys Gray, 1843. Most of the (Micoureus), whereas the second and third species in Tate’s ‘‘Microtarsus Group’’ were alternatives would require that additional placed in a new genus, Gracilinanus Gardner subgenera or genera be resurrected from and Creighton, 1989. synonymy or described as needed to contain Recent phylogenetic research based on the other species currently referred to Mar- molecular sequence data (e.g., Patton et al., mosa. 1996; Jansa and Voss, 2000; Steiner et al., The interim solution proposed by Voss 2005) has convincingly indicated that Mar- and Jansa (2009), which we adopt herein, is mosa (sensu Tate) was polyphyletic, and the to recognize Micoureus as a subgenus of same studies have consistently supported the Marmosa, and to refer all of the species monophyly of Marmosops, Micoureus, and formerly included in Marmosa to the nomi- Thylamys as those taxa were recognized by notypical subgenus. In effect, this tactic Gardner and Creighton (1989). However, a simply moves the problem of paraphyly from new genus was proposed for ‘‘Marmosa’’ the generic level (where it affects binomial canescens by Voss and Jansa (2003), and usage) to the subgeneric level (where it does other new genera were later described for two not). Although clearly suboptimal, no further clades formerly concealed by synonymies progress in the classification of this complex 3 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 334 Fig. 1. Marmosa zeledoni, photographed at La Selva Biological Station, Heredia province, Costa Rica, in August 2005 by Marco Tschapka. Zeledon’s mouse opossum, formerly synonymized with M. mexicana, is resurrected as a valid species in this report. The local population at La Selva was previously reported as M. mexicana by Timm et al. (1989), Voss and Emmons (1996) and Sperr et al. (2009). is possible until the phylogenetic relation- mitis, M. ruatanica,andM. simonsi. All of ships among all of the included species are these were swept into synonymy by Hersh- worked out. To date, only five of the nine kovitz (1951), who alleged that the diagnostic currently recognized species in the subgenus characters mentioned by Tate were artifacts Marmosa are represented in published trees, of sexual dimorphism, age, imperfect preser- and there is reason to believe that additional vation, or clinal variation. Although Hersh- valid species may be concealed among the kovitz provided no analyses of data to putative synonyms of several geographically support these claims, his conclusions were widespread forms (table 2). Two of the latter accepted by subsequent compilers of influen- are the primary focus of this report. tial checklists (e.g., Cabrera, 1958; Hall and As currently recognized, Marmosa robin- Kelson, 1959). The nominal taxa of Tate’s soni includes 13 nominal taxa, of which Tate Mitis Section (for which M. robinsoni is the (1933) treated four as valid species belonging oldest available binomen; Cabrera, 1958) to his ‘‘Mitis Section’’: M. chapmani, M. extend from Honduras southward to Pana- 2010 ROSSI ET AL.: REVISION OF MARMOSA 5 TABLE 1 This report, which is largely based on Tate’s (1933) Groups and Sections of Marmosa and a R.V.R.’s thesis research at the Universidade Their Current Classification de Sa˜o Paulo (Rossi, 2005), summarizes our Tate’s name Current name conclusions regarding the taxonomy of spe- cies belonging to Tate’s (1933) Mitis Section Cinerea Group Marmosa (Micoureus) and his ‘‘Mexicana Section.’’ To this we Murina Group append redescriptions of Marmosa xerophila Murina Section Marmosa (Marmosa, part) Handley and Gordon, 1979, and M. rubra Mitis Section Marmosa (Marmosa, part) Tate, 1931. The former is a northern South Mexicana Section Marmosa (Marmosa, part) American form that is morphologically Canescens Section Tlacuatzin similar to M. robinsoni, whereas the latter is Noctivaga Group Marmosops (part) a western Amazonian species that consistent- ly appears as the sister taxon of M. robinsoni Microtarsus Group + M. mexicana in phylogenetic analyses of Microtarsus Section Cryptonanus, Gracilinanus molecular sequence data (e.g., Voss and Lepida Section Marmosa (Marmosa, part), Jansa, 2003, 2009; Jansa et al., 2006; Jansa Marmosops (part) and Voss, 2005). Despite these indications, it Elegans Group Chacodelphys, Thylamys is not our assumption that all of the species a After Voss and Jansa (2009). treated in this report form a natural group. Among other pertinent issues, the monophy- ly of Tate’s ‘‘sections’’ has not been tested, but distinguishing the valid species that each ma, Peru, Venezuela, and the Lesser Antilles contains is a necessary step toward a (Hall, 1981; O’Connell, 1983; Creighton