To resolve phylogenetic relationships among species of Marmosa we analyzed DNA sequences from one mitochondrial and three nuclear genes for every member of the nominotypical subgenus and from four species of the subgenus Micoureus. As reported in previous studies, the subgenus Marmosa was found to be paraphyletic, whereas Micoureus was recovered as a robustly supported clade. Species currently referred to the subgenus Marmosa form four strongly supported and morphologically diagnosable groups. Based on these results we recognize a total of five subgenera: Marmosa Gray, 1821 (for macrotarsus, murina, tyleriana, and waterhousei); Micoureus Lesson, 1842 (for alstoni, constantiae, demerarae, paraguayana, phaea, and regina); Stegomarmosa Pine, 1972 (for andersoni and lepida); Eomarmosa, new subgenus (for rubra); and Exulomarmosa, new subgenus (for isthmica, mexicana, robinsoni, simonsi, xerophila, and zeledoni). The best-supported hypothesis of relationships among these clades is ((Stegomarmosa (Marmosa Micoureus)) (Eomarmosa Exulomarmosa)), and our results additionally resolve many interspecific relationships within each subgenus. These clades have broadly overlapping geographic distributions, especially in western Amazonia, where the arboreal insectivorous-frugivorous niche of Marmosa is apparently partitioned among multiple sympatric congeners.
INTRODUCTION
As currently recognized (Voss and Jansa, 2009), the didelphid marsupial genus Marmosa Gray, 1821, is a robustly supported monophyletic group containing 19 species assigned to two subgenera (table 1). Recent morphology-based revisionary work (Rossi, 2005; Rossi et al., 2010) has resolved many long-standing taxonomic problems in the nominotypical subgenus, but the subgenus Micoureus Lesson, 1842, remains unrevised. Although analyses of DNA sequence data provide strong support for the monophyly of Micoureus, the same analyses suggest that the subgenus Marmosa is paraphyletic (Voss and Jansa, 2009; Gutiérrez et al., 2010; Faria et al., 2013). The nomenclatural solution to this problem has so far been elusive because sequence data are lacking for Marmosa andersoni, type species of Stegomarmosa Pine, 1972. The latter name, currently treated as a subjective junior synonym of Marmosa, is available for subgeneric usage, but its application is uncertain until the relationships of M. andersoni can be determined.
Marmosa andersoni was redescribed by Solari and Pine (2008) based on recently collected material, but they were not able to resolve the relationships of this species using the morphological data matrix of Jansa and Voss (2005). Solari (unpublished) subsequently obtained a partial cytochrome-b sequence from dry tissue of M. andersoni, and we were later successful in obtaining nuclear-gene sequences from DNA extracts of the same material. In this report, we analyze these new sequence data, which now suffice to securely place M. andersoni in a well-resolved phylogenetic context.
Additionally, we analyze sequence data from Marmosa “tobagi,” an insular form originally described by Thomas (1911). Long considered a subspecies or synonym of M. murina (e.g., by Goodwin, 1961; Creighton and Gardner, 2008), tobagi was recognized as a valid species in Rossi's (2005) still-unpublished revision of the murina complex. A recent collecting trip to Tobago provided fresh tissues for a molecular appraisal of the relationships of this nominal taxon. With these additions, our species-level molecular sampling of the nominotypical subgenus is complete, and the sequence data now in hand provide a secure basis for a revised classification of the genus Marmosa.
Materials and Methods
Source of Material: All voucher specimens and associated tissues mentioned in this report are preserved in the following collections (listed alphabetically by institutional abbreviation; in the United States except as noted): AMNH, American Museum of Natural History (New York); CM, Carnegie Museum of Natural History (Pittsburg); EBRG, Museo de la Estación Biológica de Rancho Grande (Maracay, Venezuela); FMNH, Field Museum of Natural History (Chicago); LSU, Louisiana State University, Museum of Natural Science (Baton Rouge); MNK, Museo de Historia Natural Noel Kempff Mercado (Santa Cruz, Bolivia); MSB, Museum of Southwestern Biology, University of New Mexico (Albuquerque); MUSM, Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos (Lima, Peru); MVZ, Museum of Vertebrate Zoology, University of California (Berkeley); ROM, Royal Ontario Museum (Toronto, Canada); TTU, Museum of Texas Tech University (Lubbock); USNM, National Museum of Natural History (Washington).
TABLE 1.
Currently Recognized Species of the Genus Marmosa a
Taxon Sampling and Laboratory Methods: We analyzed all the cytochrome-b (CYTB) sequences published by Gutiérrez et al. (2010), and we sequenced CYTB from nine additional specimens representing nominal taxa or geographic populations that were not included in that report (table 2). This new material includes two museum specimens of Marmosa andersoni (from which only fragments of dry skin were available), together with fresh tissue from one specimen of M. robinsoni (representing the nominal taxon M. r. luridivolta, from Tobago) and six specimens of M. “tobagi” (also from Tobago). We extracted DNA from skin fragments of M. andersoni using a phenol/chloroform/isoamyl alcohol protocol developed by K. Nelson and the late M.C. Knapp; to sequence CYTB from this material, we used laboratory protocols described by Solari (2007) with internal primers MVZ 04, MVZ 16, and MVZ 11 (Patton et al., 1996). To sequence CYTB from the Tobago specimens we used primers and laboratory protocols described by Gutiérrez et al. (2010).
In addition to CYTB, we analyzed nuclear-gene sequences that we obtained from 24 specimens of Marmosa (including multiple individuals of some widespread species) and from two specimens representing outgroup taxa (table 3). Using the primers listed in appendix 1, we amplified part of exon 11 from the breast cancer type 1 susceptibility gene (BRCA1), intron 14 from the X chromosome-linked gene O-linked N-acetylglucosamine transferase (OGT), and intron 7 from the autosomal gene sodium-coupled neutral amino acid transporter 2 (SLC38). We amplified BRCA1 using the primer pair F1163a/R2151, and we subsequently reamplified and sequenced this product using primer pairs F1163a/R1780 and F1697/R2151. To obtain shorter fragments of BRCA1 from degraded DNA, we used four sets of primer pairs: F1163a/R1475, F1360/R1750, F1620/R1980, and F1850/R2151. In most cases, we amplified and sequenced the entire OGT intron using primer pair OGTF1/OGTR1; for specimens with low-quality DNA, we amplified and sequenced portions of the intron using primer pairs OGT-F1/OGT-R360, OGT-F120/OGT-R540, and OGT-F300/OGT-R1. We amplified the SLC38 intron using primer pair SLC38-F1/SLC38-R1 or primer pairs SLC38-F1/SLC38-R350 and SLC38-F250/SLC38-R1. All genes were PCR-amplified in 12.5 µL reactions using either GoTaq (Promega Corp.) or Platinum Taq (Life Technologies Corp.) with recommended reagent concentrations. Initial amplifications were performed using a four-stage touchdown protocol as described in Voss and Jansa (2009). When necessary, reamplification reactions were performed as 12.5 µL reactions using GoTaq. The resulting PCR products were sequenced using amplification primers and dye-terminator chemistry on an ABI-3730xl automated sequencer. All new sequences obtained for this study have been deposited in GenBank (CYTB: KM819039–KM819047; BRCA1 KM819017– KM819038; OGT: KM819048–KM819065; SLC 38: KM819066–KM819086).
TABLE 2.
Additional Specimens of Marmosa Sequenced for Cytochrome b
Phylogenetic Analyses: We aligned orthologous gene sequences using default settings of MUSCLE (Edgar, 2004) as implemented in Geneious v. 7.0.5 (Biomatters, Inc.). For the CYTB dataset, we determined the optimal nucleotide substitution model by comparing the likelihood scores of 55 candidate models using the Bayesian Information Criterion (BIC) as implemented in jModelTest 2.1.1 (Darriba et al., 2012). We analyzed this dataset using maximum likelihood (ML) as implemented in GARLI ver. 2.0 (Zwickl, 2006), specifying five search replicates, the optimal substitution model, and default search parameters. To estimate nodal support values, we ran 1000 bootstrap replicates also using GARLI ver. 2.0. We analyzed each of the three nuclear-gene datasets separately assuming the GTRGAMMA model on the RAxML BlackBox server (Stamitakis et al., 2008), using full maximum-likelihood optimization for the tree search and 100 rapid bootstrap searches to assess nodal support.
TABLE 3.
Specimens Sequenced for Nuclear Genes
FIG. 1.
Localities where sequenced specimens of Marmosa were collected. Numbers correspond to gazetteer entries in Gutiérrez et al. (2010: appendix) and to branch-tip labels in figure 2 of this report. Collection localities of specimens newly sequenced for this report are labeled with country of origin, next-largest political unit, and species.
We used the BIC implemented in PartitionFinder (Lanfear et al., 2012) to determine the best partitioning scheme and substitution models for two different concatenated-gene matrices: one that contained only the three nuclear genes and a second that also included CYTB from the same specimens sequenced for nuclear loci. For the first dataset, we partitioned by locus and by codon for the protein-coding gene BRCA1, and we used a single partition for each of the introns (for a total of five partitions). For the nuclear + CYTB dataset we additionally partitioned CYTB by codon for a total of eight partitions. For both datasets, we used the greedy search algorithm for model comparison. We performed partitioned ML analyses on each of the two datasets using five search replicates, optimal substitution models, and default search parameters as specified in GARLI ver. 2.0. We assessed nodal support from the multigene datasets in RAxML ver. 8.0 (Stamitakis, 2014) using 1000 bootstrap replicates with the optimal partitioning scheme and a GTRGAMMA model.
Finally, we conducted a species-tree analysis (Maddison and Knowles, 2006; Degnan and Rosenberg, 2009) as implemented by *BEAST ver. 1.7.4 (Heled and Drummond, 2010) using the 80-terminal CYTB matrix and the three 26-terminal nuclear-gene matrices. We assigned each individual to a species; specified the best-fitting model of sequence substitution and ploidy level for each locus; assumed a Yule process for the species tree with a piecewise linear, constant root model for the population size model; and modeled rates according to an uncorrelated lognormal relaxed clock with an exponential prior (mean = 1.0; standard deviation = 0.33). We ran the MCMC chain for 5 × 107 generations and assessed convergence using ESS values derived from Tracer ver. 1.5.
RESULTS
The optimal substitution model for our cytochrome-b data matrix (with 77 ingroup and three outgroup terminals) was HKY+I+Γ. The best tree resulting from maximum-likelihood (ML) analysis under this model (fig. 2) is almost identical to the ML topology obtained by Gutiérrez et al. (2010: fig. 3), with only three noteworthy differences concerning terminals new to this study. First, Marmosa andersoni was recovered as the sister taxon of M. lepida. Although these taxa are highly divergent (uncorrected pairwise distance = 13.5%; appendix 2), they form a weakly supported lineage (clade A). Second, the six specimens of M. “tobagi” were recovered as one of three geographically structured haplogroups within M. murina; average pairwise distances are 2.8% between “tobagi” and a cluster of Guianan haplotypes and 3.7% between “tobagi” and a cluster of Brazilian sequences. Lastly, our single sequence from a specimen of M. robinsoni collected on Tobago was recovered as minimally divergent (1.5%) from a conspecific sequence from mainland Venezuela.
A conspicuous pattern in these CYTB results is lack of strong support for deep structure in the tree. Although clade B (representing the subgenus Micoureus) and clade C (the cluster of forms associated with M. murina) are each strongly supported, support for clade D (the mostly trans-Andean “mexicana-robinsoni” group of Gutiérrez et al., 2010) is somewhat less convincing, and the novel pairing of M. andersoni + M. lepida is weakly supported. Only moderate ML bootstrap support was obtained for the sister-group relationship between M. rubra and clade D, and for the group that includes clades B and C.
Each of our nuclear-gene datasets consists of 24 ingroup and two outgroup sequences except BRCA1, for which we lack sequence from one specimen of Marmosa murina (table 3). Separate ML analyses of each nuclear-gene dataset (results not shown) consistently recovered many relationships in common with those seen in the CYTB tree (fig. 2) including clades A, B, C, and D; a group that includes clades A, B, and C; a group that includes clade D and Marmosa rubra; a group that includes M. robinsoni and M. xerophila; and a group that includes M. isthmica, M. mexicana, and M. zeledoni. All of the nuclear loci also support the same pattern of interspecific relationships within clade B that was supported by cytochrome b. By contrast, one or more nuclear-gene trees conflict with the CYTB topology concerning relationships within clade D and/or clade C, and our analysis of OGT sequences uniquely supports a sistergroup relationship between clades A and C.
The concatenated nuclear-gene dataset consists of 2317 aligned sites, and the concatenated nuclear + CYTB dataset consists of 3463 aligned sites; both datasets include 24 ingroup and two outgroup terminals. Maximum-likelihood analysis of these two matrices under their optimal partitioning schemes and substitution models (appendix 3) resulted in highly congruent topologies (fig. 3). Although most nodes in both trees are strongly supported, several key relationships remain weakly supported: (1) alternative resolutions of Marmosa simonsi within clade D; (2) the group that includes clades B and C; and (3) alternative resolutions of M. murina, M. macrotarsus, and M. waterhousei within clade C.
FIG. 2.
Phylogeny of Marmosa inferred by maximum-likelihood analysis of CYTB sequences (lnL = -11360.78). Branch tips are sequenced specimens labeled by geographic origin and a tissue identifier (see table 2 of Gutierrez et al. [2010] and table 2 of this report); numbers in parentheses refer to localities mapped in figure 1 and georeferenced by Gutiérrez et al. (2010: appendix). Outgroup taxa are not shown. Bootstrap values below 90% are shown along branches, whereas bootstrap values ≥ 90% are indicated with filled circles at relevant nodes. For simplicity, we do not show support values for relationships among very similar sequences. Individuals marked with asterisks were newly sequenced for this report. Capital letters (A–D) indicate clades discussed in the text.
The species tree (fig. 4) convincingly resolves much of the conflict among the individual gene trees and provides improved support for some of the nodes that remain weakly supported in our concatenated-gene trees. In particular, the position of Marmosa simonsi is resolved in favor of the topology supported by OGT and CYTB, and the position of clade A is resolved in accordance with the topology supported by all the sequenced genes except OGT. In effect, the only region of the tree that remains poorly supported is the branching order among M. murina, M. waterhousei, and M. macrotarsus within clade C.
TAXONOMIC SYNTHESIS
Our molecular results provide unambiguous evidence for robust phylogenetic structure among the species currently referred to the paraphyletic subgenus Marmosa, and it now seems appropriate to propose a revised classification of the genus based on the relationships discussed above. Although taxonomic rank is biologically arbitrary, it is a matter of practical importance that nomenclature provide a stable basis for biological communication, so unnecessary changes of rank are to be avoided. The rank of subgenus has not been widely used in mammalian taxonomy, but this category usefully serves to label clades of closely related species while preserving traditional binomial usage. An alternative is to use informal nomenclature, such as Tate's (1933) “sections” and “groups,” but informal names have two disadvantages. The first is that hierarchical relationships are inapparent (for example, do sections contain groups or do groups contain sections?). The second disadvantage is that informal names are not regulated by widely accepted conventions: clade D, for example, could just as easily be called something else (the trans-Andean group, for example, or the mexicana-robinsoni complex) without any rules to mediate alternative usage. By contrast, subgenera are regulated by the ICZN (1999), so usage is constrained (and stability promoted) by typification and priority.
FIG. 3.
Phylogeny of Marmosa inferred by (A) maximum-likelihood analysis of concatenated nuclear genes (lnL = -7351.13) and (B) maximum-likelihood analysis of concatenated nuclear genes plus CYTB (lnL = -16236.59). Terminals are species or mtDNA haplogroups represented by exemplar specimens sequenced for nuclear loci (table 3). Outgroup taxa are not shown. Capital letters (A–D) indicate clades discussed in the text.
We propose recognizing five subgenera of Marmosa: four for the clades designated alphabetically in figures 2–4 and another for M. rubra. As explained below, appropriate names are already available for three of these groups, so only two new names are required. All five subgenera can be diagnosed morphologically. Below we summarize morphological character data for all of the supraspecific taxa treated in this report using anatomical terminology defined and illustrated or referenced by Voss and Jansa (2009) and Rossi et al. (2010). We also take this opportunity to provide a revised morphological description of the genus Marmosa to correct a few errors of commission and omission in Voss and Jansa (2009).
FIG. 4.
Species tree computed from four single-gene data matrices (see text). Outgroup taxa are not shown. Filled circles indicate posterior probability values of 1.0 and numbers indicate posterior probabilities < 1.0. Gray bars indicate 95% credibility intervals for relative branching times. Capital letters (A–D) indicate clades discussed in the text.
Genus Marmosa Gray, 1821
Type Species: Marmosa murina (Linnaeus, 1758).
Contents: Five subgenera, as diagnosed below.
Morphological Description: Combined length of adult head and body ca. 80–210 mm; adult weight ca. 20–170 g. Rhinarium with two ventrolateral grooves on each side of median sulcus; dark circumocular mask present; pale supraocular spot absent; dark midrostral stripe absent or indistinct; throat gland absent in some species but present in adult males of others. Dorsal pelage unpatterned, superficially brownish, reddish, or grayish, but dorsal hair bases always dark gray; dorsal guard hairs short and inconspicuous; ventral fur superficially whitish, yellowish, or orange, wholly or partly gray based (apparently never completely self-colored). Manus paraxonic (dIII = dIV); manual claws about as long as or slightly longer than fleshy apical pads of digits; dermatoglyph-bearing manual plantar pads present; central palmar epithelium smooth or sparsely covered with flattened tubercles (never densely tuberculate); carpal tubercles absent in some species, or only lateral carpal tubercles present in adult males, or both medial and lateral carpal tubercles present in adult males. Pedal digits unwebbed; dIV longer than other pedal digits; plantar surface of heel macroscopically naked. Pouch absent; mammae 3-1-3 = 7 to 7-1-7 = 15, all abdominal-inguinal; cloaca present. Tail substantially longer than combined length of head and body; slender and muscular (not incrassate); without a conspicuously furred base in some species, or tail base conspicuously furred to about the same extent dorsally as ventrally; naked caudal integument unicolored (all dark) in most species but mottled distally with white spots and/or white tipped in others; caudal scales in spiral series or in both spiral and annular series; each caudal scale usually with three subequal bristlelike hairs emerging from distal margin; ventral caudal surface always modified for prehension distally (the prehensile surface naked, consisting of a long, shallow midventral groove or sulcus and an apical pad bearing dermatoglyphs).
Premaxillary rostral process present in most species (absent only in M. xerophila). Nasals long, extending anteriorly beyond I1 (concealing nasal orifice in dorsal view), and conspicuously widened posteriorly near maxillary-frontal suture. Maxillary turbinais elaborately branched. Lacrimal foramina usually two on each side, exposed to lateral view on or near anterior orbital margin. Supraorbital margins with distinct beads or prominent crests; postorbital processes usually present in mature adults but substantially larger in some species than in others (and absent or indistinct in some). Left and right frontals and parietals separated by persistent median sutures. Parietal and alisphenoid in contact on lateral braincase (no frontal-squamosal contact). Sagittal crest usually absent (occasionally present in old adult males of M. robinsoni and M. xerophila). Petrosal usually not exposed laterally through fenestra in parietal-squamosal suture (fenestra absent in almost all examined specimens). Parietal-mastoid contact present (interparietal does not contact squamosal).
Maxillopalatine fenestrae present; palatine fenestrae consistently absent in some species but consistently present in others; maxillary fenestrae absent; posterolateral palatal foramina small, never extending anteriorly between M4 protocones; posterior palatal morphology conforms to Didelphis morphotype (with well-developed lateral corners, the choanae constricted behind). Maxillary and alisphenoid not in contact on floor of orbit (separated by palatine). Transverse canal foramen present. Alisphenoid tympanic process usually without anteromedial process or posteromedial lamina enclosing the maxillary nerve (secondary foramen ovale almost always absent), and not in contact with rostral tympanic process of petrosal. Anterior limb of ectotympanic suspended directly from basicranium. Stapes usually triangular, with large obturator foramen (microperforate and imperforate stapes occur as rare variants in several species). Fenestra cochleae exposed in most species (but concealed in a sinus formed by the caudal and rostral tympanic processes of the petrosal in M. rubra). Paroccipital process small, rounded, adnate to petrosal. Dorsal margin of foramen magnum bordered by supraoccipital and exoccipitals (incisura occipitalis present).
Two mental foramina usually present on lateral surface of each hemimandible (one foramen or three foramina occur as rare, usually unilateral variants in some species); angular process acute and strongly inflected.
Unworn crowns of 12–15 symmetrically rhomboidal (“premolariform”), with subequal anterior and posterior cutting edges, and increasing in length (mesio-distal dimension) from 12 to 15. Upper canine (C1) alveolus in premaxillary-maxillary suture; C1 usually simple, without accessory cusps (but posterior accessory cusp consistently present in M. lepida). First upper premolar (P1) smaller than posterior premolars but well formed and not vestigial; second and third upper premolars (P2 and P3) subequal in height; P3 with posterior cutting edge only; upper milk premolar (dP3) large and molariform. Molars moderately carnassialized (postmetacristae are visibly longer than postprotocristae); relative widths usually Ml < M2 < M3 < M4; centrocrista strongly inflected labially on M1–M3; ectotlexus indistinct or absent on Ml, shallow but usually distinct on M2, and consistently deep on M3; anterolabial cingulum continuous with preprotocrista (complete anterior cingulum present) on M3. Last upper tooth to erupt is P3.
Lower incisors (i1–i4) with distinct lingual cusps. Unworn lower canine (c1) usually semiprocumbent, with flattened bladelike apex, with or without distinct posterior accessory cusp. Second lower premolar (p2) taller than p3; lower milk premolar (dp3) trigonid incomplete (bicuspid). Hypoconid labially salient on m3; hypoconulid twinned with entoconid on m1–m3; entoconid much taller than hypoconulid on m1–m3.
TABLE 4.
Diagnostic Characters and Geographic Distribution of Subgenera Recognized in this Report
Eomarmosa, new subgenus
Type Species: Marmosa rubra Tate, 1931.
Contents: rubra Tate, 1931.
Diagnosis: Guiar gland absent; manual claws small (not extending beyond fleshy apical pads of fingers); medial and lateral carpal tubercles present in large adult males (e.g., AMNH 71950); dorsal tail scales rhomboidal and arranged in predominantly spiral series; ventral prehensile surface of tail not densely fringed by long hairs. Postorbital processes absent, indistinct, or very small; palatine fenestrae absent; fenestra cochleae concealed; M2 preparacrista attaches to stylar cusp A.
Comparisons: Eomarmosa differs from other subgenera of Marmosa in petrosal and dental characters. Specifically, the fenestra cochleae of Eomarmosa is concealed within a sinus formed by the caudal and rostral tympanic processes of the petrosal, whereas the fenestra cochleae in other subgenera is normally exposed because the rostral and caudal tympanic processes of the petrosal are widely separated. Additionally, the preparacrista on M2 of Eomarmosa attaches to stylar cusp A (or to the corresponding anterolabial corner of the stylar shelf when a discrete cusp is missing), whereas the M2 preparacrista of other subgenera usually passes straight labially to terminate at or near stylar cusp B.
Other characters distinguish Eomarmosa from some, but not all, other congeners. For example, Eomarmosa differs from Exulomarmosa by its lack of a guiar gland—consistently present as a hairless patch of (often greasy) skin in adult male specimens of the latter subgenus—and by the spiral arrangement of its caudal scales (the arrangement of tail scales of Exulomarmosa usually includes at least some in annular series). Eomarmosa differs from members of the subgenus Marmosa by the presence of well-developed medial and lateral carpal tubercles in large adult males (sexually dimorphic carpal tubercles are consistently absent in Marmosa). Eomarmosa has small manual claws, the tips of which do not extend distally beyond the fleshy apical pads of the fingers, whereas the manual claws of Micoureus are larger, stronger, and (unless blunted) usually extend slightly beyond the fleshy apical pads of the fingers. Lastly, Eomarmosa lacks the dense fringes of long silvery hairs that border the caudal prehensile surface in both species of Stegomarmosa.
Etymology: From the ancient Greek 
(“dawn”), by metonymy, for the reddish fur of the single included species.
Remarks: A detailed description of Marmosa (Eomarmosa) rubra was provided by Rossi et al. (2010), who also mapped the collection localities of all known specimens. Based on those data, the geographic range of M. rubra appears to be restricted to western Amazonia, where it is known to occur from southeastern Colombia to southeastern Peru (Rossi et al., 2010: fig. 30). This taxon occurs sympatrically—but perhaps not syntopically—with members of the subgenera Marmosa and Micoureus, (e.g., at Boca Río Curaray, Departamento Loreto, Peru; Tate, 1933) and with Stegomarmosa (e.g., at Hacienda Villa Carmen, Departamento Cusco, Peru; Pine, 1972).
Exulomarmosa , new subgenus
Type Species: Marmosa robinsoni Bangs, 1898.
Contents: isthmica Goldman, 1912 (including mimetra Thomas, 1921); mexicana Merriam, 1897 (including mayensis Osgood, 1913; ruatanica Goldman, 1911; and savannarum Goldman, 1917); robinsoni Bangs, 1898 (including casta Thomas, 1911; chapmani J.A. Allen, 1900; fulviventer Bangs, 1901; grenadae Thomas, 1911; luridavolta Goodwin, 1961; mitis Bangs, 1898; nesaea Thomas, 1911; and pallidiventris Osgood, 1912); simonsi Thomas, 1899; xerophila Handley and Gordon, 1979; and zeledoni Goldman, 1917.
Diagnosis: Guiar gland consistently present and well developed in adult males; manual claws small (not extending beyond fleshy apical pads of fingers); medial and lateral carpal tubercles present in large adult males 6 ; dorsal tail scales rhomboidal or oblong, usually in both spiral and annular series on most specimens, but one or the other pattern sometimes predominating; ventral prehensile surface of tail not densely fringed with long hairs. Postorbital processes usually absent, indistinct, or very small in some species (e.g., Marmosa zeledoni) but consistently large and well developed in adults of other species (e.g., M. simonsi); palatine fenestrae consistently absent in some species (e.g., M. isthmica) but consistently present in others (e.g., M. xerophila); fenestra cochleae usually exposed (but partially concealed in a few examined specimens of M. robinsoni, M. simonsi, and M. xerophila); M2 preparacrista attaches to or terminates near stylar cusp B.
Comparisons: Comparisons of Exulomarmosa with Eomarmosa have already been provided (see above). Species of Exulomarmosa differ consistently from members of the subgenus Marmosa in sexually dimorphic features of the carpal (wrist) region: whereas large adult male specimens of Exulomarmosa have well-developed lateral and medial carpal tubercles, neither sex has large carpal tubercles in the subgenus Marmosa. Additionally, guiar glands are consistently well developed (especially in adult males) in Exulomarmosa, but guiar glands are absent in most species of Marmosa (appearing polymorphically only in M. waterhousei; Rossi, 2005). Exulomarmosa also differs from Micoureus in possessing guiar glands, and these subgenera further differ in manual claw morphology (small in Exulomarmosa, large in Micoureus). In Exulomarmosa the caudal prehensile surface lacks the dense lateral fringes of long hairs that occur in both species of Stegomarmosa.
Remarks: Morphological diagnoses of the species in this subgenus were provided by Rossi et al. (2010), who also discussed relevant synonymies. At least two species (Marmosa mexicana and M. robinsoni) contain highly divergent mtDNA haplotype groups that might represent cryptic taxa (Gutiérrez et al., 2010; Gutiérrez et al., in press).
Species of Exulomarmosa are mostly trans-Andean (sensu Haffer, 1967), but the geographic range of Marmosa robinsoni extends east of the Andes into the dry coastal forests of northern Venezuela, several Caribbean islands (Isla Margarita, Trinidad, Tobago, Grenada), and the llanos (Rossi et al., 2010: fig. 25). Exulomarmosa and Micoureus have broadly overlapping geographic distributions in Central America and northwestern South America, where they have been collected together at numerous localities. 7 Although the distributions of Exulomarmosa and the subgenus Marmosa narrowly overlap in Colombia and northernmost Venezuela, they are seldom collected sympatrically at mainland localities. By contrast, M. (Ma.) murina and M. (Ex.) robinsoni occur syntopically in abandoned cacao plantations on Tobago (e.g., near Charlotteville at 11°19′N, 60°33′W; Voss, 1991).
Etymology: From the Latin exul (“a banished or exiled person”), in reference to the isolation of this trans-Andean clade (Gutiérrez et al., 2010) from the cis-Andean distribution of most of the rest of the genus.
Subgenus Marmosa Gray, 1821
Type Species: Marmosa murina (Linnaeus, 1758).
Contents: macrotarsus Wagner, 1842 (including madeirensis Cabrera, 1913; musicola Osgood, 1913; and quichua Thomas, 1899); murina Linnaeus, 1758 (including chloe Thomas, 1907; dorsigera Linnaeus, 1758; duidae Tate, 1931; guianensis Kerr, 1792; klagesi J.A. Allen, 1900; meridionalis Miranda-Ribeiro, 1936; moreirae Miranda-Ribeiro, 1936; muscula Cabanis, 1848; parata Thomas, 1911; roraimae Tate, 1931; and tobagi Thomas, 1911); tyleriana Tate, 1931 (including phelpsi Tate, 1939); and waterhousei Tomes, 1860 (including bombascarae Anthony, 1922; and maranii Thomas, 1924).
Diagnosis: Guiar gland consistently absent in some species (e.g., Marmosa tyleriana), usually present in adult males of other species (e.g., M. murina), and apparently polymorphic in some (e.g., M. waterhousei); manual claws small (not extending beyond fleshy apical pads of fingers); medial and lateral carpal tubercles consistently absent; dorsal tail scales rhomboidal and arranged in spiral series (except in M. tyleriana, in which the scale arrangement is sometimes predominantly annular); ventral prehensile surface of tail not densely fringed with long hairs. Postorbital processes usually well developed in large adults (but absent in M. tyleriana); palatine fenestrae usually absent in most species (but consistently present in M. tyleriana); fenestra cochleae exposed; M2 preparacrista consistently attaches to or terminates near stylar cusp B.
Comparisons: Members of the nominotypical subgenus uniquely lack sexually dimorphic carpal tubercles. By contrast, large adult males belonging to species in other subgenera have well-developed medial and lateral carpal tubercles that are absent in females and less well developed in smaller (presumably younger) males. Additionally, species of Marmosa (Marmosa) consistently differ from species in the subgenus Micoureus by their smaller manual claws, and from species of Stegomarmosa by lacking dense fringes of long hairs along the margins of the caudal prehensile surface. Comparisons with Eomarmosa and Exulomarmosa were summarized in the preceding accounts.
Remarks: Our synonymies for species in the nominotypical subgenus of Marmosa follow Rossi (2005), except that we treat tobagi (recognized as a valid species by Rossi, 2005) as a subjective synonym of murina for reasons explained earlier in this report. Diagnostic morphological comparisons among macrotarsus, murina, and waterhousei were provided by Rossi (2005), and characters distinguishing murina from waterhousei were discussed by Gutiérrez et al. (2011).
Species of Marmosa (Marmosa) occur in lowland moist forests from northern Venezuela to eastern Bolivia and south-central Brazil (Mato Grosso), and along the Atlantic coast of Brazil southward to Espírito Santo. Apparently they are everywhere sympatric with members of the subgenus Micoureus (e.g., at all the cis-Andean inventory sites analyzed by Voss and Emmons, 1996: appendices 4–11). Additionally, species of the subgenus Marmosa are broadly sympatric with Stegomarmosa in northeastern and western Amazonia—for example, at Huampami (4°28′S, 78°10′W) in Amazonas department, Peru (specimens at MVZ)—and they are also sympatric with Eomarmosa from southeastern Colombia to southeastern Peru (see above).
Subgenus Micoureus Lesson, 1842
Type Species: Didelphis cinerea Temminck, 1824, by subsequent designation (Thomas, 1888). According to Gardner and Creighton (2008: 79), cinerea Temminck, 1824, is an unavailable name (preoccupied by cinerea Goldfuss, 1812) for the southeastern Brazilian species currently known as Marmosa (Micoureus) paraguayana Tate, 1931.
Contents: alstoni J.A. Allen, 1900 (including nicaraguae Thomas, 1905); constantiae Thomas, 1904 (including budini Thomas, 1920); demerarae Thomas, 1905 (including areniticola Tate, 1931; domina Thomas, 1920; esmeraldae Tate, 1931; limae Thomas, 1920; meridae Tate, 1931); paraguayana Tate, 1931 (including cinerea Temminck, 1824 [preoccupied]); phaea Thomas, 1899 (including perplexa Anthony, 1922); regina Thomas, 1898 (including germana Thomas, 1904; mapiriensis Tate, 1931; parda Tate, 1931; rapposa Thomas, 1899; rutteri Thomas, 1924).
Diagnosis: Guiar gland consistently absent; manual claws relatively larger than in other congeners (usually extending just beyond fleshy apical pads of fingers); medial and lateral carpal tubercles present in large adult males; caudal scales usually rhomboidal in spiral series; ventral prehensile surface of tail not densely fringed with long hairs. Postorbital processes usually well developed in large adults; palatine fenestrae absent in most species (but consistently present in some); fenestra cochleae usually exposed (partially concealed in some specimens of at least two species); M2 preparacrista usually attaches to or terminates near stylar cusp B.
Comparisons: Members of the subgenus Micoureus tend to have slightly larger manual claws than species in other subgenera, but share no other unique trait (see Remarks, below). Instead, they are distinguished by a unique combination of traits, including those previously discussed in comparisons with Eomarmosa, Exulomarmosa, and the nominotypical subgenus (above). They can be distinguished from species of Stegomarmosa by the lack of dense fringes of long hairs along the caudal prehensile surface.
Remarks: Although species of Micoureus are commonly known as “woolly mouse opossums” and are said to differ from other species of Marmosa in pelage length and texture (e.g., by Gardner, 2008: 2), we have not found pelage length or texture to be consistently useful as diagnostic traits. To be sure, some species of Micoureus (e.g., M. demerarae) have longer and woollier pelts than sympatric members of the nominotypical subgenus (e.g., M. murina). However, other species of Micoureus (e.g., lowland populations of the yellowish eastern Bolivian form currently known as M. constantiae) have short fur, some highland forms of Marmosa (e.g., M. tyleriana) have long fur, and we are unable to define “woolliness” by any satisfactory (nonsubjective) criterion. Species of Micoureus are also said to be larger than other marmosines (Gardner, 2008), but small species of Micoureus—such as the southwestern Ecuadorean taxon that Anthony (1922) described as M. perplexa—exhibit broad morphometric overlap with large species in other subgenera (such as M. isthmica). Other traits that have sometimes been used to diagnose Micoureus include the prominent extension of body fur onto the base of the tail, and the presence of a white tail-tip. Although it is true that species in other subgenera do not commonly exhibit either of these traits, neither is consistently exhibited by all of the phenotypes that seem to belong in Micoureus based on molecular analyses. Despite the fact that it was once ranked as a genus, Micoureus is, paradoxically, one of the least morphologically distinctive clades within the genus Marmosa.
Species of Micoureus occur in moist lowland and montane forests from Belize southward to Paraguay and northern Argentina. The contents of this subgenus are unrevised, and current taxonomic usage is impossible to reconcile either with DNA sequencing results or with our examination of relevant type material (Voss and Jansa, in prep.). Therefore, the species-level taxonomy used in this report uncritically follows Gardner and Creighton (2008) pending analyses of relevant morphological and molecular data.
Species of Micoureus occur sympatrically with members of every other subgenus, including Eomarmosa, Exulomarmosa, and Marmosa (see above). Micoureus and Stegomarmosa are probably sympatric throughout the Guiana Region (e.g., at Nouragues, French Guiana; Guillemin et al., 2001) and western Amazonia. At least two species of Micoureus occur sympatrically (but perhaps not syntopically) throughout most, if not all, of western Amazonia (e.g., along the Rio Juruá; Patton et al., 2000). This subgenus includes the largest members of the genus, with some specimens of several species weighing > 150 grams, but other species of Micoureus are substantially smaller (with adult weights consistently < 100 g).
Subgenus Stegomarmosa Pine, 1972
Type species: Marmosa andersoni Pine, 1972.
Contents: andersoni Pine, 1972; and lepida Thomas, 1888.
Diagnosis: Guiar gland present (andersoni) or absent (lepida); manual claws small (not extending beyond fleshy apical pads of fingers); medial carpal tubercle present (in the type and only examined adult male specimen of andersoni) or absent (lepida); lateral carpal tubercle present; caudal scales in both spiral and annular series; prehensile caudal surface densely fringed with long hairs; postorbital processes well developed in mature adult specimens; palatine vacuities present (andersoni) or absent (lepida); fenestra cochleae exposed; M2 preparacrista attaches to or terminates near stylar cusp B.
Comparisons: Comparisons of Stegomarmosa with other subgenera have already been summarized in the preceding accounts.
Remarks: Marmosa andersoni and M. lepida differ in several characters, but they uniquely share twin rows of long, silvery hairs that flank the median prehensile sulcus on the underside of the distal part of the tail. Additionally, mature adults of these species resemble one another— and differ from most other congeners—by having a very long rostral process of the premaxillae; large, triangular postorbital processes; and distinctly reddish dorsal fur. Both species occur in western Amazonia, although the range of Marmosa lepida also extends eastward into the Guiana Region (NE Amazonia). Examples of sympatry between Stegomarmosa and other subgenera of Marmosa have already been mentioned in the preceding accounts.
Several discrepancies between our diagnosis of Stegomarmosa and published descriptions of Marmosa andersoni merit discussion. (1) Solari and Pine (2008: 56) suggested that “The normal condition of the palate may be completely without fenestration” in M. andersoni, but all the adult specimens we examined (FMNH 84252; MUSM 14154, 14155) have well-developed maxillopalatine and palatine fenestrae. 8 (2) Voss and Jansa (2003: 64) described the caudal scales of FMNH 84252 as arranged in spiral series, but our reexamination of this specimen and other newly available material convinces us that Solari and Pine (2008) were correct in describing the arrangement as spiral and annular. (3) Rossi (2005: 80) described the medial carpal tubercle as absent, but close inspection reveals that a small medial tubercle is present on the wrist of FMNH 84252, the only known adult male specimen. Because FMNH 84252 is a young adult (with fully erupted P3 but unworn posterior molars), the tubercle is probably larger on older males.
NOTES ON SYMPATRIC DIVERSITY
Based on a review of available ecobehavioral information (Voss and Jansa, in prep.), the genus Marmosa appears to be the dominant clade of small insectivorous-frugivorous arboreal opossums throughout most of the forested Neotropical lowlands. 9 Members of other sympatric arboreal clades are larger and perhaps more frugivorous (e.g., Caluromys spp.), whereas sympatric clades that overlap Marmosa in size and appear to have similar diets are apparently scansorial (e.g., Marmosops). In dry forests south of Amazonia, Marmosa seems to be replaced ecologically by Gracilinanus. However, species of Marmosa occupy dry forests where Gracilinanus does not occur—for example, M. mexicana in Central America, M. xerophila in northern Venezuela, and M. simonsi, in western Ecuador (Rossi et al., 2010).
Interestingly, two or more species of Marmosa occur sympatrically at many rainforested localities. Based on geographic range overlap (see maps in Rossi, 2005), most western Amazonian sites probably have one species each of the subgenera Marmosa (either M. waterhousei or M. macrotarsus) and Stegomarmosa (usually M. lepida) and two species each of Micoureus (usually one each of the taxa currently known as M. regina and M. demerarae), and some sites close to the base of the Andes could have additional species of Eomarmosa and Stegomarmosa. Therefore, some western Amazonian sites might have as many as six sympatric species of Marmosa, although no more than four seem to have been collected to date at any one site.
These observations suggest that the adaptive zone occupied by small, arboreal, insectivorous-frugivorous opossums can be subdivided, and it would be interesting to know whether the clades recognized as subgenera in this report are ecobehaviorally distinctive. Some field observations (e.g., Charles-Dominique et al., 1981; Malcolm, 1990) suggest that Marmosa (Micoureus) demerarae, for example, is active in the forest canopy and subcanopy, whereas sympatric Marmosa (Marmosa) murina is primarily active in understory vegetation. In western Amazonia, M. (Mi.) demerarae appears to be more common in terra firme forest, whereas sympatric M. (Mi.) regina prefers seasonally flooded forest (Patton et al., 2000). In central Panama, M. (Exulomarmosa) isthmica is more abundant in secondary vegetation than in undisturbed old-growth forest (Enders, 1935). Such observations, few as they are, suggest that both vertical and horizontal habitat segregation may be important correlates of sympatry in this widespread genus of small opossums.
ACKNOWLEDGMENTS
We thank Angela Ramsey and her helpful staff at the Department of Natural Resources and the Environment (in Scarborough) for permission to collect specimens and tissues of Marmosa murina and M. robinsoni on Tobago (where Nancy Simmons and Keith Barker cheerfully pitched in so that R.S.V. and S.A.J. could work without interruption). Venezuelan material of M. robinsoni and M. xerophila resulted from fieldwork in which E.E.G. collaborated with R.P. Anderson, J. Ochoa-G., and Marisol Aguilera. The latter made possible collection of these samples and their subsequent importation into the United States via a “Contrato de Acceso a Recursos Genéticos” between the Universidad Simón Bolivar and the Ministerio del Poder Popular para el Ambiente. The ever-patient curators and staff at the USNM (especially Kris Helgen, Darrin Lunde, and Esther Langan) and the FMNH (Larry Heaney and Bill Stanley) hosted R.S.V.'s visits to their collections, as did the wonderful folks at the MUSM (Victor Pacheco and Giüseppy Calizaya). Additionally, we thank Victor Pacheco for allowing S.S. to sample MUSM specimens of M. andersoni. Alfred L. Gardner and Diego Astúa de Moraes provided useful comments that helped improve our manuscript. Lastly, and most importantly, we are grateful to Lorissa Fujishin for expert assistance and long hours in the lab sequencing nuclear loci for this report.
This work was funded, in part, by awards to R.S.V. and S.A.J. from the National Science Foundation (DEB-743039 and -743062, respectively) and by awards to E.E.G. from the AMNH (Theodore Roosevelt Memorial Fund), the American Society of Mammalogists (Grants-in-Aid of Research), the City College of the City University of New York (Graduate Student Award), the Graduate Center of the City University of New York (Science Fellowship, and Sue Rosenberg Zalk Student Travel and Research Fund), and the Smithsonian Institution (Buck Postdoctoral Fellowship Program).
REFERENCES
Appendices
APPENDIX 4 Exemplar Series of Morphological Specimens Examined
Marmosa (Eomarmosa) rubra (N = 11): AMNH 68128, 68129, 71950, 71952, 71974, 71976; MVZ 153280, 154759, 154765; USNM 274577, 274578.
Marmosa (Exulomarmosa) isthmica (N = 9): USNM 306415-306422, 456814.
Marmosa (Exulomarmosa) mexicana “A” (N = 9): AMNH 277680-277682; ROM 95795, 96090, 96318, 96968, 99608, 99776.
Marmosa (Exulomarmosa) mexicana “B” (N = 4): AMNH 79250; ROM 98459; USNM 569858, 570071.
Marmosa (Exulomarmosa) robinsoni (N = 9): USNM 372938-372940, 372942-372944, 372947, 385068, 385069.
Marmosa (Exulomarmosa) simonsi (N = 15): AMNH 61195, 63350, 63404, 63407, 63413, 64525, 64532, 66873, 66883; USNM 121155, 121156, 121158, 461643, 513423, 551640.
Marmosa (Exulomarmosa) xerophila (N = 9): USNM 443920, 443929, 443840, 443845, 443847, 443920, 443924, 443947, 443951.
Marmosa (Exulomarmosa) zeledoni (N = 11): AMNH 28283, 29541, 269997; USNM 361194-361201.
Marmosa (Marmosa) macrotarsus (N = 11): AMNH 268214, 272816, 273062, 273063, 273178, 273188; MUSM 13283, 15293, 15294, 15296, 15297.
Marmosa (Marmosa) murina (N = 13): AMNH 266416, 266417, 267368, 267816; USNM 393480-393482, 393484, 393485, 578006-578008, 588119.
Marmosa (Marmosa) tyleriana (N = 8): AMNH 130501, 130503, 130504, 130507, 130509-130511,130559.
Marmosa (Marmosa) waterhousei (N = 8): AMNH 47186, 68127; TTU 98654, 98717, 98934, 100922, 101098, 101153.
Marmosa (Micoureus) alstoni (N = 10): AMNH 131732, 131737, 136863, 136865, 136866, 137288, 137997, 137999, 139280, 139780.
Marmosa (Micoureus) constantiae (N = 9): AMNH 209159, 209160, 210398-210400, 275463, 275465-275467.
Marmosa (Micoureus) demerarae (N = 12): AMNH 266427-266434, 267369-267371, 267818.
Marmosa (Micoureus) paraguayana (N = 7): FMNH 141586, 211415, 211416; MVZ 182063-182065; UMMZ 134551).
Marmosa (Micoureus) phaea (N = 8): FMNH 88543, 88545, 89364, 90101, 90102; Rom 57242, 57243, 57245).
Marmosa (Micoureus) regina (N = 5): MVZ 190323, 190324, 190328, 190332, 190333.
Marmosa (Stegomarmosa) andersoni (N = 4): FMNH 84252; MUSM 14154, 14155; USNM 582777.
Marmosa (Stegomarmosa) lepida (N = 7): AMNH 78001, 98656, 182937, 273186; FMNH 140824; USNM 461467, 461468
Notes
[1] 6 Exemplar adult male specimens in which these tubercles are well developed include AMNH 12454 (Marmosa mexicana), AMNH 37890 (M. isthmica), AMNH 66852 (M. simonsi), AMNH 69939 (M. robinsoni), USNM 443920 (M. xerophila), and AMNH 147759 (M. zeledoni). The same specimens also exhibit the well-developed guiar gland that characterizes this subgenus.
[2] 7 For example, at Armila (8°40′N, 77°27′W; Rossi et al., 2010) in the Panamanian province of San Blas, where Marmosa (Exulomarmosa) isthmica was collected sympatrically with a small species of Micoureus said to resemble M. phaea (see Handley, 1966). At Curupao (10°31′N, 66°38′W; Handley, 1976) in the Venezuelan state of Miranda several specimens of M. (Ex.) robinsoni were collected sympatrically with specimens of a large form resembling M. (Mi.) demerarae. Representative specimens from both localities are at the USNM.
[3] 8 The palatine fenestrae are covered by a membrane on the incompletely cleaned skull of MUSM 14154, which accounts for their indistinct appearance in Solari and Pine's (2008: fig. 2) photograph of that specimen.
