Maximum parsimony and maximum likelihood analyses

For each of the four datasets, we implemented MP analyses in PAUP* 4.0 (Swofford, 2002) using a heuristic search strategy with TBR branch swapping, 10 random-addition replicates, and with all characters unordered and equally weighted. We assessed nodal support using 5000 bootstrap pseudoreplicates with TBR branch swapping and 10 random-addition replicates. For the CYTB dataset, we first pruned out identical sequences in order to maximize computational efficiency.

We conducted ML analyses of each of the four datasets in GARLI ver. 0.96 (Zwickl, 2006) by specifying the best-fit model for each dataset as determined in MrModelTest 2.3 (Nylander, 2004) and allowing the model parameters to be estimated from the data. We did not alter the default parameters used in the program's genetic algorithm, and we set the analyses to automatically terminate after 20,000 generations with no significant change in the log-likelihood score. In order to ensure that the program had arrived at a stable log-likelihood estimate, we repeated each search five times and report only the estimate with the highest log-likelihood. We measured nodal support using 500 bootstrap replicates and the same settings used in the individual ML analyses.

Data partitioning in a Bayesian framework

Combining data from multiple genes and treating each codon position identically assumes that each site within the dataset evolves under the same nucleotide substitution model. If this assumption is invalid, as is often the case, the dataset can be partitioned into independent units (e.g., Bull et al., 1993; Nylander et al., 2004) in order to avoid misleading results (Brown and Lemmon, 2007). For each of the four datasets, we evaluated the relative fit of different data-partitioning strategies in a Bayesian framework. For the three single-gene datasets, we tested whether a three-partition (by codon position) model fit the data significantly better that a single (unpartitioned) model. For the combined-gene dataset, we evaluated the relative fit of four partitioning strategies: (1) a single partition for all genes in combination (combined: unpartitioned); (2) three partitions, one for each of the three genes (by gene); (3) three partitions, one for each of the codon positions across all genes (combined: by codon position); and (4) nine partitions, one for each codon position for each of the three genes (by gene by codon position).

Several statistical approaches have been suggested in the literature for evaluating the relative fit of different partitioning schemes, including Bayes factors (Kass and Raftery, 1995), the Akaike information criterion (AIC; Akaike, 1974), and the Bayesian information criterion (BIC; Schwarz, 1978). Although Bayes factors have been shown to be reasonably effective tools for choosing among partitioning schemes (Brown and Lemmon, 2007), the number of parameters contained in alternative models is not explicitly incorporated into the calculation of Bayes factors. The AIC and BIC both apply penalties to the addition of parameters, mitigating the problems associated with overparameterization (McGuire et al., 2007). Therefore, to account for potential overparameterization, we only ranked partitioning strategies using the AIC and BIC.

To implement this approach, we obtained log-likelihood scores and model parameter values for each of the partitioning strategies described above for the four datasets using BI as implemented in MrBayes 3.1.2 (Ronquist and Huelsenbeck, 2003). We first determined the best-fitting nucleotide substitution model for each individual partition using the AIC as implemented in MrModeltest. We specified the appropriate model for each partition in MrBayes and, for strategies with more than one partition, we unlinked all substitution parameters across partitions. We estimated branch lengths proportionally to each partition and allowed the substitution rate to vary across each partition by setting the rate prior option to “variable.” We ran two independent replicates of the Metropolis-Coupled Markov Chain Monte Carlo analysis for 5 × 10⁶ generations, sampling every 100th generation, with one cold chain and three incrementally heated chains.

Once these runs were completed, we examined a plot of the log-likelihood scores per generation using Tracer 1.4 (Rambaut and Drummond, 2007) to determine the appropriate number of generations to discard as burn-in. For all strategies and datasets, we discarded the first 500,000 generations (5,000 saved samples)—well after the log-likelihood scores stabilized—and used the remaining 45,000 samples to estimate mean log-likelihood, nodal posterior probability (BPP) values, mean branch lengths, and other parameters. In order to compare the results of different partitioning strategies, we calculated the harmonic mean log-likelihood (HML) across all parameters for each strategy in Tracer. We then used the HML values to calculate the AIC and BIC scores according to the methods of McGuire et al. (2007).

Topology tests

Previously published molecular and morphological research on Thylamys has produced several phylogenies that differ from ours. In order to evaluate these alternative hypotheses, we conducted topology tests using Shimodaira-Hasegawa (SH) likelihood-ratio tests (Shimodaira and Hasegawa, 1999). First, an ML analysis was performed in GARLI following the same protocol described above, except that constraint trees were used to infer the best tree that conformed to the alternative hypothesis in question. We then implemented the SH test in PAUP* using 1000 resampling estimated log likelihoods (RELL) replicates to compare the constrained ML tree to the best ML tree from the above analysis, allowing the program to estimate all parameters.

Parsimony ancestral state reconstruction

Species and haplogroups that comprise the Elegans Group (see Results) occur on both the eastern and western sides of the Andes. We used the combined-gene ML topology to reconstruct the ancestral distributions of Elegans Group lineages relative to the Andes. Tips were coded as occurring east of the Andes (cis-Andean), west of the Andes (trans-Andean), or both (ambiguous). We optimized ancestral areas on the topology using the parsimony criterion for ancestral state optimization of unordered characters as implemented in MacClade 4.08 (Maddison and Maddison, 2000).

Measurement data

We transcribed total length (nose to fleshy tail tip, TL) and length of tail (basal flexure to fleshy tip, LT) from specimen labels or field notes, and we computed head-and-body length (HBL) by subtracting LT from TL. We also transcribed length of hind foot (heel to tip of longest claw, HF), and length of ear (from notch, Ear), from specimen labels or field notes. Additionally, we remeasured HF to check the accuracy of values recorded by the collector, and we used our values whenever large discrepancies were found. All external measurements are reported to the nearest millimeter (mm).

Craniodental measurements were taken with digital calipers as skulls were viewed under low (6–12×) magnification and recorded to the nearest 0.01 mm, but measurement values reported herein are rounded to the nearest 0.1 mm. The following dimensions were measured as illustrated in figure 2:

Fig. 2.

Dorsal and ventral cranial views and occlusal view of the maxillary dentition of Thylamys pusillus, showing the anatomical limits of 13 craniodental measurements defined in the text.

Condylo-basal length (CBL): Measured from the occipital condyles to the anteriormost point of the premaxillae.

Nasal breadth (NB): Measured across the triple-point sutures of the nasal, frontal, and maxillary bones on each side.

Least interorbital breadth (LIB): Measured at the narrowest point across the frontals between the orbits.

Zygomatic breadth (ZB): Measured at the widest point across both zygomatic arches.

Palatal length (PL): Measured from the anteriormost point of the premaxillae to the postpalatine torus, including the postpalatine spine (if present).

Palatal breadth (PB): Measured across the labial margins of the upper fourth molar (M4) crowns, at or near the stylar A position.

Maxillary toothrow length (MTR): Measured from the anterior margin of the upper canine (C1) to the posterior margin of ipsilateral M4.

Length of molars (LM): Measured from the anteriormost labial margin of M1 to the posteriormost point on ipsilateral M4.

Length of M1–M3 (M1–M3): Measured from the anteriormost labial margin of M1 to the posteriormost point on ipsilateral M3.

Width of M3 (WM3): Measured from the labial margin of the crown at or near the stylar A position to the lingual apex of the protocone.

In addition, we took two other measurements to record taxonomic variation in bullar inflation remarked by authors and confirmed by our own visual comparisons of skulls:

Bullar width (BW): Measured from the glenoid fossa across the inflated, bubblelike part of the alisphenoid tympanic process to its posteromedial extremity.

Interbullar width (IBW): The least distance between the medial surfaces of the left and right bullae.

Qualitative characters

We examined skins, skulls, and fluid-preserved specimens for taxonomic variation in qualitative morphological characters. Surveyed characters included those previously described in the didelphid taxonomic literature (e.g., Tate, 1933; Flores et al., 2000; Solari, 2003; Voss and Jansa, 2003; Carmignotto and Monfort, 2006) and others newly reported herein. Nomenclature used to describe observed variation in integumental, osteological, and dental features follows standard anatomical usage as reviewed by Voss and Jansa (2009).

Species limits and phylogeographic structure

Nine of the monophyletic groups of Thylamys that we recovered by analyzing cytochrome-b sequence data appear to represent valid species (fig. 3, table 6). In order to simplify the presentation of these results, we anticipate our taxonomic conclusions by using Latin binomials for reciprocally monophyletic groups of sequences obtained from morphologically differentiated series of voucher specimens, and we use alphabetical designations (A, B, C) for reciprocally monophyletic groups of apparently conspecific sequences (“haplogroups”). The morphological basis for assigning names to clades is explained in a separate section of this report (see Taxonomy, below).

Fig. 3.

The tree based on ML analysis of cytochrome-b sequences for 131 ingroup terminals and six individuals representing five outgroup genera (GTR + I + Γ, ln-likelihood  =  −10723.24). Maximum parsimony (MP) and maximum likelihood (ML) bootstrap values and Bayesian posterior probabilities (BPP) are indicated at all nodes. Individual specimens of Thylamys macrurus, T. karimii, and T. velutinus are labeled with country of origin, locality number (in parentheses; see appendix 1), and an alphanumeric specimen identifier (see table 1). Other species (and haplogroups) are cartooned with black triangles, the breadth (vertical dimension) of which is proportional to the number of individuals sampled for each clade, and the length (horizontal extent) of which is proportional to genetic diversity within each clade. Clades numbered in gray circles are presented as subtrees in figures 4, 6, and 8.

TABLE 6

Uncorrected Cytochrome-b p-Distances (scaled as percent sequence divergence; below diagonal), K2P-Corrected Distances (above diagonal), and Intraspecific Distances (diagonal)^a

The first two lineages to branch within Thylamys are Brazilian endemic species—Thylamys karimii and T. velutinus (both sensu Carmignotto and Monfort, 2006)—that are represented in these data by only one and two individual(s), respectively. Also sparsely sampled is T. macrurus, a species that inhabits eastern Paraguay and the adjoining Brazilian state of Mato Grosso do Sul (Carmignotto and Monfort, 2006; Voss et al., 2009). Three clades recovered by all analyses of the CYTB data incorporate the remaining species: the Elegans Group (node 1), Thylamys pusillus (node 2), and the Venustus Group (node 3).

The Elegans Group (fig. 4) consists of at least three species that collectively range from Patagonia to Peru (fig. 5). Thylamys elegans, a name that we apply in the restricted sense of Meynard et al. (2002) and Solari (2003), is represented in these data by five sequences from central Chile that form a robustly supported monophyletic group in all of our analyses. Despite the fact that T. elegans has a geographically restricted distribution, the uncorrected average sequence divergence among these haplotypes is moderately high, about 2.5% (table 6). Sequence divergence across the basal split in this species is somewhat higher (3.8%), but it is far less than the 8.6% difference that Meynard et al. (2002) reported among their Chilean specimens. Although we are unable to account for this discrepancy (we sequenced only one of the specimens that Meynard et al. included in their study, NK 27583), we note that Braun et al. (2005) reanalyzed Meynard et al.'s T. elegans sequences and obtained an average divergence value of 1.94%.

Fig. 4.

Relationships among 41 cytochrome-b sequences of the Elegans Group. This subtree shows the full details of relationships at node 1 (fig. 3) as resolved by ML analysis. Phylogenetic terminals in this tree are identified by country and state/department/province of origin, locality number (in parentheses; see appendix 1), and an alphanumeric specimen identifier (see table 1). Nodal support values are provided for all nodes recovered in common by MP, ML, and BI analyses.

Fig. 5.

Map of collecting localities for specimens of the Elegans Group sequenced for this study. Numbers are keyed to entries in the gazetteer (appendix 1).

A second species in the Elegans Group is Thylamys pallidior, a name that we apply to specimens from southern Peru, northern Chile, the Bolivian altiplano, and Argentina. The monophyly of T. pallidior is robustly supported in all of our analyses of CYTB sequence data, but average uncorrected sequence divergence is moderately high (5.0%) across the basal split in this species, which separates two reciprocally monophyletic haplogroups. The northern haplogroup (pallidior A) occurs in Peru, Bolivia, and northern Chile and comprises two well supported subgroups differing by a net uncorrected p-distance of 2.5%. The southern haplogroup (pallidior B), although restricted to Argentina, is widely distributed in that country, where it occurs from high elevations (> 4000 m) in the Andes to near sea level in the Pampas; despite this impressive ecogeographic range, genetic diversity within pallidior B is low (about 1.0%) and there is little apparent phylogeographic structure among the 22 member sequences that we analyzed.

The third species that we recognize in the Elegans Group is Thylamys tatei, represented by two individuals from west-central Peru (Ancash) that differ in sequence at only one nucleotide position. Interestingly, this genetically and morphologically distinctive taxon is part of an unresolved trichotomy with two other divergent lineages, also from western Peru, that may represent undescribed species (E. Palma, personal commun.).

Thylamys pusillus, a name that we apply to three allopatric haplogroups (fig. 6), is a lowland species from the tropical and subtropical dry forests and savannas of southeastern Bolivia, western Paraguay, and northern Argentina (fig. 7). The three haplogroups recovered in our analysis correspond to those recently reported by Teta et al. (2009), who recognized them as valid species, a difference of interpretation that we discuss elsewhere (see Taxonomy, below). Thylamys pusillus A is represented by six nonidentical sequences from the Chaco Boreal (north of the Río Pilcomayo) that exhibit an average uncorrected intraclade distance of 1.9%. Genetic variation is substantially less within pusillus B from the Chaco Austral (south of the Pilcomayo, 0.6%) and pusillus C from the Mesopotamian region of northeastern Argentina (between the Paraná and Uruguay rivers, 0.4%). Although the reciprocal monophyly of all three haplogroups is robustly supported by our MP and Bayesian analyses, we note that our geographic samples are tightly clustered with large unsampled gaps between them, and that the monophyly of pusillus A is only moderately supported by ML bootstrapping.

Fig. 6.

Relationships among 13 cytochrome-b sequences of Thylamys pusillus. This subtree shows the full details of relationships at node 2 (fig. 3) as resolved by ML analysis. Phylogenetic terminals in this tree are identified by country and state/department/province of origin, locality number (in parentheses; see appendix 1), and an alphanumeric specimen identifier (see table 1). Nodal support values are provided for all nodes recovered in common by MP, ML, and BI analyses.

Fig. 7.

Map of collecting localities for specimens of Thylamys pusillus sequenced for this study. Numbers are keyed to entries in the gazetteer (appendix 1).

The Venustus Group (fig. 8) includes two species, both of which are abundantly represented in our analyses by specimens collected on the eastern Andean slopes and foothills of Bolivia and northwestern Argentina (fig. 9). Although our analyses recovered the same basal mtDNA sequence dichotomy in this group previously reported by Braun et al. (2005), we use the name sponsorius for the species that they called cinderella (see Taxonomy, below). Thylamys sponsorius and T. venustus have partially overlapping distributions, but T. sponsorius ranges farther south into Argentina, and T. venustus ranges farther north into Bolivia. Where their latitudinal ranges overlap, T. sponsorius tends to occur at higher elevations than T. venustus, but allopatric populations of T. venustus in central Bolivia can occur at elevations as high as those at which T. sponsorius occurs further south.

Fig. 8.

Relationships among 72 cytochrome-b sequences of the Venustus Group. This subtree shows the full details of relationships at node 3 (fig. 3) as resolved by ML analysis. Phylogenetic terminals are identified by country and state/department/province of origin, locality number (in parentheses; see appendix 1), and an alphanumeric specimen identifier (see table 1). Nodal support values are provided for all nodes recovered in common by MP, ML, and BI analyses.

Fig. 9.

Map of collecting localities for specimens of the Venustus Group sequenced for this study. Numbers are keyed to entries in the gazetteer (appendix 1).

Mean intraspecific genetic diversity for Thylamys sponsorius is 1.8%, and specimens from Argentina form a strongly supported clade to the exclusion of conspecific Bolivian material, which appears as a shallowly branched paraphyletic group. By contrast, phylogeographic structure is much more apparent within T. venustus, which includes three robustly supported and spatially segregated haplogroups (fig. 10). Thylamys venustus A is represented by 12 sequences (with an average intraclade divergence value of 1.2%) from seven localities at high elevations (2450–4000 m) in central Bolivia, where voucher specimens were collected in the departments of La Paz, Chuquisaca, and Cochabamba. Thylamys venustus B is represented by 10 sequences (with an average intraclade divergence value of 1.1%) from seven localities at somewhat lower elevations (1695–2950 m) in eastern Cochabamba and western Santa Cruz. Thylamys venustus C is represented by 26 sequences (with an average intraclade divergence of only 0.6%) from 18 localities in foothill and middle elevations (349–2100 m) in western Santa Cruz southward to Argentina (Jujuy and Tucumán).

Fig. 10.

Map of collecting localities for specimens from each of the three Thylamys venustus haplogroups from Bolivia and northern Argentina (not all Argentine collecting localities of haplogroup C are shown). Numbers are keyed to entries in the gazetteer (appendix 1).

Phylogenetic relationships based on cytochrome b

In addition to resolving species-level clades and intraspecific phylogeography as described above, our analyses of cytochrome-b sequence data provide noteworthy support for some deeper nodes. Within the Elegans Group, for example, there is strong support for a sister-group relationship between T. elegans and T. pallidior, and for a group that includes T. tatei and two unnamed Peruvian lineages (fig. 4). Additionally, there is strong support for the monophyly of the genus Thylamys, and moderate to strong support for the monophyly of a group that includes all of the species of Thylamys except T. velutinus and T. karimii (fig. 3). However, other relationships of key interest are only moderately to weakly supported by these data. Among such equivocally resolved details are the problematic relationships of three Brazilian species (T. velutinus, T. karimii, and T. macrurus) and the sister-group relationship between the Elegans and Pusillus clusters.

Phylogenetic analyses of concatenated mitochondrial sequences

Because the cytochrome-b dataset was unable to convincingly resolve several key interspecific relationships, we obtained ND2 and COX2 sequences from exemplars of each of the species and haplogroups described above, choosing specimens that spanned the basal split recovered for each clade in the CYTB tree. Separate analyses of the ND2 and COX2 datasets using MP, ML, and BI methods recovered topologies that were similar to those obtained from CYTB in most respects. Importantly, relationships that were strongly supported by one gene never conflicted with relationships that were strongly supported by another. This lack of hard incongruence among datasets suggests that combining these datasets is appropriate.

Maximum parsimony, maximum likelihood, and Bayesian analyses of the combined-gene (CYTB + ND2 + COX2) dataset recovered most of the same relationships obtained from cytochrome b and provide more robust support for clades that were only weakly or moderately supported by individual-gene analyses (fig. 11). Novel features of this topology include (1) a weakly supported basal clade that unites the Brazilian species Thylamys velutinus and T. karimii, and (2) a strongly supported sister-group relationship between T. macrurus and the Venustus Group (T. sponsorius + T. venustus). Additionally, most of the supraspecific nodes with moderate to strong support from CYTB (fig. 3) are consistently strongly supported by the combined-gene analyses. Among the previously described supraspecific relationships that seem conclusively resolved in these results include the monophyly of the Elegans and Venustus groups, a sister-group relationship between the Elegans Group and Thylamys pusillus, and a clade that includes all of the species of Thylamys except T. velutinus and T. karimii. Additionally, all of the species and haplogroups delineated using CYTB were recovered with very strong support in the combined-gene analyses. The single noteworthy example of a CYTB clade with diminished support in these results is pusillus B + pusillus C, which still receives strong support from MP bootstrapping, but for which ML bootstrap support and BI posterior probabilities have substantially eroded.

Fig. 11.

The maximum likelihood tree based on concatenated sequence data from cytochrome b, cytochrome c oxidase subunit II, and NADH dehydrogenase 2 from 29 specimens representing nine species of Thylamys and six individuals representing five outgroup genera (GTR + I + Γ, ln-likelihood  =  −22234.43). Phylogenetic terminals are identified by country and state/department/province of origin, locality number (in parentheses; see appendix 1), and an alphanumeric specimen identifier (see table 1). Black circles indicate nodes with 100% ML and MP bootstrap support and Bayesian posterior probabilities equal to 1.0.

Topology tests

Our combined-gene dataset provides a large sample of characters (2877 bp; table 4) for evaluating alternative hypotheses of phylogenetic relationships. Of these, only three are sufficiently explicit to merit statistical testing: (1) Palma et al. (2002) analyzed cytochrome-b sequence data from five species of Thylamys and recovered the topology (macrurus (pusillus (venustus (elegans + pallidior)))), (2) Braun et al. (2005) analyzed cytochrome-b sequence data from seven species and recovered the topology (macrurus ((venustus + sponsorius) (pusillus (pallidior (elegans + tatei))))), and (3) Carvalho et al. (2009) analyzed cytochrome-b sequence data from eight species and recovered the topology (macrurus ((sponsorius + venustus) (karimii (pusillus (pallidior (tatei + elegans)))))). All of these hypotheses can be confidently rejected using Shimodaira and Hasegawa's (1999) likelihood-ratio test with our combined-gene data (P  =  0.00 for all alternatives).

Although several possible explanations might account for the rejected topologies of Palma et al. (2002), Braun et al. (2005), and Carvalho et al. (2009), we noticed that the most striking discrepancies between their results and ours involve only two sequences. One sequence has GenBank accession number AF431926 and was said to have been obtained from a Paraguayan specimen of Thylamys macrurus (field number NK 27536). The second is AF431927, said to have been obtained from a Bolivian specimen of T. pusillus (NK 25139). Both sequences were originally obtained and analyzed by Palma et al. (2002), but Braun et al. (2005) and Carvalho et al. (2009) subsequently used AF431926 (downloaded from GenBank) in their analyses. Because we independently extracted and amplified cytochrome b from NK 27536 (T. macrurus) and NK 25139 (T. pusillus), we were able to compare our sequences with those obtained by Palma et al. (2002). Sequence alignment indicates that both of our T. macrurus sequences (from specimens NK 27536 and APC 932) are much more similar to each other (p-distance  =  0.7%) than either is to AF431926 (average p-distance  =  21.23%). A BLAST comparison of AF431926 against other sequences in GenBank reveals that its closest match to previously published data (87%) is a cytochrome-b sequence from Marmosops impavidus (U34670). A BLAST search on the purported T. pusillus sequence (AF431927) clearly demonstrates that it does not belong to a marsupial; instead, it is 95% identical to several GenBank sequences from the cricetid rodent Phyllotis xanthopygus (e.g., AY431053).

Parsimony ancestral-state reconstruction

Ancestral areas for lineages that comprise the Elegans Group (the only clade that occurs, in part, west of the Andes) were inferred across the combined-gene ML topology using the parsimony criterion (fig. 12). Results indicate three Andean dispersal events: (1) the ancestral Elegans Group lineage dispersed from east to west and diversified along the western coast of Chile and Peru, (2) the lineage leading to Thylamys pallidior B later dispersed from west to east, forming a wide-ranging clade distributed in a variety of habitats in Argentina, and (3) a subclade of T. pallidior A invaded the altiplano of Bolivia and Chile (along the crest of the Andes, and therefore of ambiguous classification with respect to our cis/trans dichotomy).

Fig. 12.

Parsimony reconstruction of ancestral distributions in relation to the Andean cordillera. Branch tips are coded as east of the Andes (cis-Andean), west of the Andes (trans-Andean), or both (ambiguous). The topology used for this reconstruction is derived from fig. 11.

Dorsal pelage pattern

Most marsupials have two distinct body pelage color zones, dorsal and ventral, with a more or less abrupt line of transition between them. In this “bicolored” phenotype, the middorsal fur may be indistinctly darker than the lateral fur, but the transition is gradual, not abrupt. By contrast, Lestodelphys and most species of Thylamys are “tricolored” (sensu Tate, 1933: 209) because a distinctly darker middorsal color zone is separated from a distinctly paler lateral color zone along a line of demarcation that extends on each side from a point on the forehead posteriorly to the rump. Most specimens with tricolored pelage also have a dark axillary spot behind the foreleg on each side.

Among the specimens of Thylamys that we examined, the tricolor pattern was consistently absent only in T. velutinus, which is uniformly brownish dorsally without any sharp pigmental difference between the middorsal and lateral fur. Our observations in this respect are consistent with those of Tate (1933: 234) and Carmignotto and Monfort (2006), who likewise noted the taxonomically unusual pelage pigmentation of this species. Although Carmignotto and Monfort (2006) suggested that the tricolor pattern is indistinct in another species, T. karimii, the specimens of T. karimii that we examined had unmistakably tricolored fur.

Ventral pelage pattern

The ventral pelage of many small didelphids consists of hairs that are either “self-colored”—having the same pale (whitish or yellowish) coloration from base to tip—or they are gray based. Gray-based hairs are, as the term implies, grayish basally, but they are abruptly paler (usually whitish or yellowish) distally. In some species of Thylamys (e.g., T. pusillus) the entire ventrum is covered with self-colored hairs, but in others (e.g., T. elegans) a median streak of self-colored ventral fur is bordered by broad lateral zones of gray-based fur; in three species (T. sponsorius, T. velutinus, and T. venustus) almost all of the ventral fur is gray based.

Manual claws

The claws of the manus are short, not extending beyond the fleshy apical pad of each digit in some species of Thylamys, but in other species the claws are much longer and extend well beyond the fleshy fingertips. Short-clawed species include T. macrurus (fig. 13B), T. pusillus, and members of the Venustus Group (T. sponsorius, T. venustus). Long-clawed species include T. karimii, T. velutinus (fig. 13A), and members of the Elegans Group (T. elegans, T. pallidior, T. tatei). Confident scoring of manual claw length is best accomplished under low magnification to ensure that claws are undamaged (neither broken nor blunted) and that fleshy fingertips (on dried skins) are not unduly distorted.

Fig. 13.

Ventral views of right manus illustrating taxonomic differences in plantar morphology and claw length (see text). In Thylamys velutinus (A, based on ZMUC 166 and 167), almost the entire plantar surface is tubercular, with vestigial dermatoglyphs (d) restricted to the apices of the plantar pads; the latter are fused together and extend inward to the center of the palm, which lacks a concave central area. The claws are long in this species, extending well beyond the fleshy finger tips. In T. macrurus (B, MZUSP 32097) dermatoglyphs are more extensive on the fingertips and on the plantar pads, the plantar pads are separate and surround a concave central palmar surface (ccps), and the claws are short, not extending much beyond the fleshy tip of each digit.

Manual plantar pads

Six fleshy pads (thenar, hypothenar, and four interdigital pads; Brown and Yalden, 1973) are present on the ventral (plantar) surface of the manus in most didelphids. In most species of Thylamys (e.g., T. macrurus; fig. 13B) these six pads are separated from one another by deep folds of elastic skin, and they surround a concave central palmar surface; the apex of each pad is covered by specialized (nontubercular) epidermis bearing dermatoglyphs (friction or papillary ridges) like those on human fingertips. Thylamys karimii and T. velutinus (fig. 13A), however, differ from other congeners in having the interdigital and carpal pads fused together, in lacking a concave central palmar surface, and in the marked reduction (T. velutinus) or absence (T. karimii) of plantar dermatoglyphs.

Pale tail tip

All species of Thylamys have a bicolored tail that is dark (brownish or grayish) above and abruptly paler (whitish) on the ventral surface. In most species, the dark dorsal coloration extends all the way to the tip of the tail, but the tail tip is pale (whitish) above and below for several millimeters in most examined specimens of T. macrurus and T. tatei, and in some specimens of T. pusillus from northeastern Argentina.

Caudal prehensile surface

In most species of Thylamys, the ventral surface of the tail is provided with a hairless prehensile surface consisting of a shallow median groove and an apical pad bearing dermatoglyphs (Carmignotto and Monfort, 2006: fig. 4c). However, in T. karimii (op. cit.: fig. 4a) and T. velutinus (op. cit.: fig. 4b) the median groove is indistinct or absent and the apical pad, although hairless, lacks dermatoglyphs.

Nasal length

Although we measured nasal length as a quantitative variable, measurement values do not capture all of the information suggested by visual inspection of skulls, so we also coded how far the nasal bones extend posteriorly with respect to a line drawn across the caudal limits of the right and left lacrimals (fig. 14). Nasals were scored as short if these bones did not extend posteriorly as far as the lacrimals, long if they terminated at about the same level as the lacrimals, and very long if they terminated well behind the lacrimals.

Fig. 14.

Dorsal view of interorbital region illustrating taxonomic differences in the posterior extension of the nasal bones (nas) with respect to the lacrimals (lac). In Thylamys elegans (A, UWBM 44446) the nasals are short, not extending as far posteriorly as the lacrimals, whereas the lacrimals of T. tatei (B, MVZ 135506) are longer, extending posteriorly behind the lacrimals. Other abbreviations: fro, frontal; max, maxillary.

Intraspecific variation in this character was often noted, but some taxonomic variation in trait frequencies are noteworthy (table 7). Whereas most species of Thylamys have long or very long nasals, T. elegans usually has short nasals, as do most examined specimens of T. velutinus. Interestingly, T. pallidior lacks an obviously modal condition for this character: all three nasal morphologies were observed in roughly equal numbers of specimens, even within haplotypes from restricted parts of the species' geographic range (i.e., the tabulated variability is not the result of pooling geographically variable samples).

TABLE 7

Variation in Nasal Morphology among Species of Thylamys^a

Lacrimal foramina

All examined specimens of Thylamys have two lacrimal foramina on each side, but there is taxonomic variation in the position of these paired openings. In some species the lacrimal foramina are more or less concealed from lateral view because they open inside the anterior orbital margin (fig. 15B), but in others the foramina are laterally exposed because they are outside the orbit (fig. 15A). We also recognize an intermediate condition in which the anteroventral foramen is exposed and the posterodorsal foramen is concealed (fig. 15C). Most species exhibit variation in this character, but there are potentially useful taxonomic differences (table 8). For example, the lacrimal foramina are consistently concealed in T. tatei, but one or both foramina are exposed in most examined specimens of T. elegans, and both foramina are prominently exposed in most specimens of T. velutinus.

Fig. 15.

Left lateral cranial views illustrating taxonomic differences in the position of the infraorbital foramen (iof) above the tooth row, in the position of the lacrimal foramina (lf) with respect to the orbit, and in depth of the nasolabial fossa (nlf). In Thylamys velutinus (A, OMNH 22284) the infraorbital foramen is usually above P3 or above the P3/M1 commissure, the lacrimal foramina are laterally exposed anterior to the orbit, and the nasolabial fossa is deeply excavated. In T. pusillus (B, AMNH 260025) the infraorbital foramen is usually above P3, the lacrimal foramina are often concealed inside the orbit, and the nasolabial fossa is shallow. In Thylamys pallidior (C, AMNH 262406) the infraorbital foramen is usually above M1, the lacrimal foramina are often partially exposed, and the nasolabial fossa is shallow.

TABLE 8

Exposure of the Lacrimal Foramina among Species of Thylamys^a

Infraorbital foramen

In most didelphids, and in many species of Thylamys, the infraorbital foramen opens above P3 (fig. 15B) or above the commissure between P3 and M1 (fig. 15A). One or both of these taxonomically widespread traits were observed in all examined specimens of T. karimii, T. macrurus, T. pusillus, T. tatei, and T. velutinus (table 9). By contrast, the infraorbital foramen opens above M1 in most examined specimens of T. elegans and T. pallidior (fig. 15C), and this trait was also observed in a few specimens of the Venustus Group (T. sponsorius and T. venustus).

TABLE 9

Variation in the Position of the Infraorbital Foramen among Species of Thylamys^a

Nasolabial fossa

The concave lateral surface of the jugal just below the orbit is the site of origin for M. maxillonasolabialis (also known as M. zygomaticus; Coues, 1872). The nasolabial fossa2 is shallow, inconspicuous, and lacks sharply defined borders in most species of Thylamys—for example, in T. pusillus (fig. 15B) and T. pallidior (fig. 15C). By contrast, the fossa is deeply excavated (with sharply defined dorsal, ventral, and caudal margins) and dorsoventrally narrower in T. velutinus (fig. 15A).

Supraorbital margins

The dorsolateral contours of the frontal bones between the orbitotemporal fossae are smoothly rounded or squared in most species of Thylamys, but some species have supraorbital “beads” (upturned lateral margins flanked by shallow grooves; see Voss and Jansa, 2009: fig. 11B). We observed supraorbital beads on all examined adult specimens of T. karimii and most examined adult specimens of T. macrurus and T. pusillus, but beading occurs infrequently in other species (table 10). Beading is clearly subject to ontogenetic variation because juveniles and subadults seldom have beads, even in species that usually have them as adults, and they tend to be better developed among old adult specimens than among conspecific young adults. In some large adult male specimens of T. karimii, T. macrurus, and T. pusillus, supraorbital beads project laterally to form small postorbital processes.

TABLE 10

Variation in Interorbital Morphology among Species of Thylamys^a

We did not tabulate “rounded” and “squared” as alternative states because we often observed intermediate conditions that were hard to classify. Most young animals (e.g., juveniles, subadults, young adults) have rounded supraorbital margins, whereas squared supraorbital margins normally occur in older animals. Such age-correlated conspecific differences were sometimes interpreted taxonomically by Tate (1933) who, for example, characterized “Marmosa janetta” (all examined specimens of which were old adults) as having sharp supraorbital margins.

Maxillary fenestrae

Most didelphids have several sets of bilaterally paired palatal perforations, some of which (foramina) transmit nerves, blood vessels, or other soft-tissue structures, whereas others (fenestrae) do not transmit anything. Incisive foramina, maxillopalatine fenestrae, palatine fenestrae, and posterolateral palatal foramina (fig. 16) are invariably present in Thylamys. By contrast, maxillary fenestrae—perforations in the maxillary bone lateral to the maxillopalatine openings—are normally present in some species but not in others.

Fig. 16.

Palatal morphology of Thylamys venustus (A, AMNH 261254) and T. pallidior (B, UMMZ 155831). Maxillary fenestrae (m) are usually present in T. venustus, but they are almost always absent in T. pallidior. Other abbreviations: if, incisive foramen; mp, maxillopalatine fenestra; p, palatine fenestra; plpf, posterolateral palatal foramen.

Maxillary fenestrae are consistently present in examined specimens of Thylamys karimii, T. macrurus, T. sponsorius, and T. velutinus, but they are usually also present in T. pusillus and T. venustus (table 11). By contrast, maxillary fenestrae are consistently absent in examined specimens of T. elegans, and they are usually absent in T. pallidior and T. tatei. In most examined specimens of all species, maxillary fenestrae are either bilaterally present or bilaterally absent; unilateral presence of maxillary fenestrae is rare.

TABLE 11

Variation in the Occurrence of Maxillary Fenestrae among Species of Thylamys^a

Upper second incisor

In most species of Thylamys the crowns of I2–I5 increase in size from front to back, a progression that is most easily seen in lateral view. In specimens conforming to this morphology, the crown of I2 is visibly smaller than or subequal to the crown of I3 (fig. 17B). However, in all examined specimens of T. tatei with unworn or lightly worn teeth, I2 is visibly larger that I3 (fig. 17A), an apparently unique trait of this species.

Fig. 17.

Labial view of left upper incisor row of Thylamys tatei (A, MVZ 135507), with I2 larger than I3; and T. elegans (B, AMNH 97755), with I2 smaller than I3. Abbreviations: pre, premaxillary; max, maxillary.

Stylar cusp C

The stylar shelf of didelphid upper molars is provided with a series of five or six small cusps along the labial margin. By convention, these cusps are given alphabetic designations based on positional criteria (Clemens, 1966). The most consistently recognizable are stylar cusp B (labial to the paracone) and stylar cusp D (labial to the metacone). One or two small cusps that are posterolabial to the paracone and anterolabial to the metacone correspond to stylar cusp(s) C. In some species of Thylamys, there is no distinct cusp in the C position, and in these taxa the ectoflexus (a labial indentation in the stylar shelf) is deeper and more distinct than it is in taxa with stylar cusp C. The presence of stylar cusp C contributes to the “serrated” appearance of the stylar shelf as noted by authors (Solari, 2003; Carmignoto and Monfort, 2006). Because stylar cusp C may be present or absent at different dental loci, we scored this character separately on M1 and M2 (table 12); due to frequent polymorphism and lack of useful taxonomic differences, we did not score this trait on M3. Minimally worn (juvenile, subadult, or young adult) teeth are necessary for confident scoring of this character.

TABLE 12

Variation in the Occurrence of Stylar Cusp C on M1 and M2 among Species of Thylamys^a

Metaconule: The metaconule, a small but taxonomically important occlusal element of the primitive tribosphenic molar bauplan (Simpson, 1936), is absent or indistinct in most didelphids (Voss and Jansa, 2009), as it is in most species of Thylamys. In species conforming with this widespread opossum phenotype (including T. elegans, T. karimii, T. pallidior, T. sponsorius, T. tatei, and T. venustus; table 13), the metaconule appears (if at all) only as a small chevron on the postprotocrista that is just barely visible when the newly erupted or lightly worn teeth of juveniles and subadults are viewed under high magnification; presumably because such diminutive structures are quickly obliterated by wear, adult teeth seldom retain any clearly identifiable vestige of the metaconule. By contrast, the metaconule is often substantially larger and ontogenetically persistent in T. pusillus and T. velutinus, and a well-developed metaconule was likewise observed in the single specimen of T. macrurus that we were able to score for this character.

TABLE 13

Variation in the Occurrence of Metaconules among Species of Thylamys^a

Type species

Thylamys karimii.

Contents

Thylamys karimii and T. velutinus.

Diagnosis

Members of the subgenus Xerodelphys can be distinguished from other congeners (herein referred to the nominotypical subgenus, see below) by their reduction or loss of plantar dermatoglyphs, lack of a concave central palmar surface, tails that are shorter than the combined length of head and body, and absence of distinct modifications for caudal prehension (table 15).

TABLE 15

Morphological Comparisons among Thylamys (Xerodelphys) karimii, T. (X.) velutinus, and Members of the Subgenus Thylamys

Remarks

Although the monophyly of Xerodelphys is only weakly supported by phylogenetic analyses of our concatenated-gene dataset (fig. 11), the absence of a concave central palmar surface (resulting from fusion of plantar pads on the manus) and the reduction or loss of plantar dermatoglyphs are unique in the family Didelphidae and provide supporting evidence that these two species form a clade. Sequence data from nuclear loci will presumably allow future tests of this hypothesis.

The morphological distinctness of the taxa we refer to Xerodelphys was previously recognized by Solari (2003), who, however, regarded karimii as a synonym of velutinus. Carmignotto and Monfort (2006) were the first to clearly identify the diagnostic traits that distinguish T. karimii and T. velutinus from each other and from congeneric species that we refer to the nominotypical subgenus, but they prudently refrained from naming a new genus-group taxon in the absence of a supporting phylogenetic analysis. We commend their restraint and credit them with the morphological observations on which our subgeneric diagnosis is based.

Thylamys karimii (Petter, 1968)

Synonyms

None.

Distribution

According to Carmignotto and Monfort (2006), who examined much more material than we have seen, Thylamys karimii inhabits open (nonforest) Cerrado and Caatinga landscapes in the Brazilian states of Bahia, Goiás, Mato Grosso, Minas Gerais, Pernambuco, Piauí, Rondônia, and Tocantins; it has also been collected in the Distrito Federal. Although no Bolivian specimens are known, T. karimii could be expected to occur on the Serranía de Huanchaca in Santa Cruz department, where other Cerrado endemics have recently been discovered (Emmons et al., 2006). The geographic range of T. karimii overlaps that of T. velutinus (see Carmignotto and Monfort, 2006: fig. 7), but these species have yet to be collected sympatrically.

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present) in all specimens examined by us, but perhaps less distinctly so in other material (Carmignotto and Monfort, 2006); ventral pelage entirely self-white, usually without lateral zones of gray-based hairs; plantar pads of manus fused together (concave central palmar surface absent) and uniformly covered with small tubercles (plantar dermatoglyphs absent); manual claws long, extending well beyond fleshy apical pads of digits; tail much shorter than combined length of head and body (LT/HBL × 100  =  77%; N  =  32 [based on measurement data from Carmignotto and Monfort, 2006: table 1]), without pale tip; ventral prehensile surface of tail tip absent. Nasals usually very long (extending posteriorly beyond lacrimals); lacrimal foramina partially exposed to lateral view on orbital margin in most specimens; infraorbital foramen above P3; nasolabial fossa usually shallow (deeply excavated in a few specimens); supraorbital margins beaded; maxillary fenestrae present; crown of second upper incisor (I2) smaller than or subequal to crown of I3; stylar cusp C often present on M1 and M2; metaconule indistinct or absent on M3.

Comparisons

Thylamys karimii can be distinguished unambiguously from T. velutinus, the only other member of the subgenus Xerodelphys, by its self-white ventral pelage (the ventral pelage is entirely gray based in velutinus), and complete lack of plantar dermatoglyphs (vestigial dermatoglyphs are present at the apex of each plantar pad in velutinus). Cranial character differences are difficult to evaluate with the small samples at hand, but the nasal bones appear to be absolutely longer on average (NL  =  11.6–13.4 mm; N  =  5) in karimii than in velutinus (LN  =  11.3–11.7 mm; N  =  3) and they usually project posteriorly well behind the lacrimals (the nasals do not extend posteriorly beyond the lacrimals in any examined specimen of velutinus; table 7). Although we observed supraorbital beads in karimii that we did not see in velutinus, Carmignotto and Monfort (2006) reported beaded interorbitals in specimens of velutinus that we have not examined. Whereas metaconules were indistinct or absent in all three specimens of karimii that we were able to score confidently for this character, distinct metaconules were present in all five scored specimens of velutinus (table 13).

Comparisons of measurement data between larger samples of Thylamys karimii and T. velutinus than we were able to examine are provided in Carmignotto and Monfort's (2006) exemplary study. Briefly, T. karimii is larger than T. velutinus in most measured dimensions, but it has absolutely and relatively smaller canines, a difference that is conspicuous in side-by-side comparisons. Despite broad overlap in most univariate morphometric comparisons, T. karimii and T. velutinus can be separated into nonoverlapping clusters by canonical discriminant functions analysis (Carmignotto and Monfort, 2006: fig. 8).

Remarks

Originally described as a valid species by Petter (1968), karimii was subsequently listed as a synonym of Thylamys pusillus by Gardner (1993), whereas Palma (1995) and Solari (2003) treated it as a synonym of T. velutinus. Carmignotto and Monfort (2006) effectively refuted these synonymies by identifying most of the characters that we now recognize as distinguishing karimii from other species of Thylamys (table 15).

Three specimens of small didelphids collected near Exu in the Brazilian state of Ceará were misidentified as Marmosa karimii by Mares et al. (1981) and Streilein (1982). The single specimen from this series that we examined (CM 80015, from 0.5 km S Exu) is an old adult male Gracilinanus agilis, whereas two others examined by A.P. Carmignotto (in litt., 8 March 2010) are Cryptonanus agricolai: one is MZUSP 16610 (an immature female from Fazenda Guaraní, 2.9 km N Exu), and the other is MZUSP 16961 (an adult male from Escola Agricola, 0.7 km S Exu). Behavioral observations based on these specimens were erroneously attributed to Thylamys karimii by Creighton and Gardner (2008) despite Palma's (1995) well-founded doubts about their correct identification.

Specimens examined (N  =  12)

Brazil—Mato Grosso, Chapada dos Guimarães (ANSP 4632), 264 km N Xavantina (BMNH 76.632–76.639; USNM 393536–393538).

Thylamys velutinus (Wagner, 1842)

Synonyms

pimelurus Reinhardt, 1851.

Distribution

According to Carmignotto and Monfort (2006), Thylamys velutinus, which is only known from a few localities, occurs in Cerrado habitats in the Brazilian states of Goiás, Minas Gerais, and São Paulo; it has also been collected in the Distrito Federal. This species is not known to occur sympatrically with any other congener, although it could be expected to coexist with T. karimii throughout much of its geographic range.

Morphological diagnosis

Body pelage bicolored (abrupt line of transition from darker middorsal to paler lateral coloration absent); ventral pelage completely gray based (superficially buffy or whitish) except on chin; plantar pads of manus fused together (concave central palmar surface absent) and almost completely covered with small tubercles (small plantar dermatoglyphs are present at the apex of each pad); manual claws long, extending well beyond fleshy apical pads of digits; tail much shorter than combined length of head and body (LT/HBL × 100  =  79%; N  =  6 [based on measurement data from Carmignotto and Monfort, 2006: table 1]), without pale tip; ventral prehensile surface of tail tip indistinct or absent. Nasal bones usually short, not extending posteriorly behind lacrimals in any examined specimen; lacrimal foramina usually exposed to lateral view anterior to orbit; infraorbital foramen above P3 or above P3/M1 commissure; nasolabial fossa deeply excavated; supraorbital margins rounded or squared, not beaded in any examined specimen; maxillary fenestrae present; crown of second upper incisor (I2) subequal to crown of I3; stylar cusp C consistently present on M1 and M2; metaconule present on M3.

Comparisons

Qualitative and morphometric comparisons with Thylamys karimii, the only other member of the subgenus Xerodelphys, were provided by Carmignotto and Monfort (2006). The most consistently useful of these are summarized in the preceding species account.

Remarks

Although velutinus has always been recognized as a valid species, other forms have been regarded as junior synonyms by authors. Of these, only pimelurus is clearly conspecific (Tate, 1933; Carmignotto and Monfort, 2006). By contrast, formosa Shamel, 1930, listed as a subspecies of velutinus by Cabrera (1958), is a morphologically distinctive taxon of still-uncertain phylogenetic relationships; it is currently referred to the monotypic genus Chacodelphys (see Voss et al., 2004; Voss and Jansa, 2009).

Specimens examined (N  =  7)

Brazil—Distrito Federal, Brasilia (OMNH 37216), 25 km S Brasilia (OMNH 22284); Minas Gerais, Lagoa Santa (UZM 164 [holotype of pimelurus], 165–168).

Subgenus Thylamys Gray, 1843

Type species

Thylamys elegans (Waterhouse, 1839).

Contents

Seven species, as recognized below.

Remarks

Monophyly of the subgenus Thylamys is strongly supported by phylogenetic analyses of mtDNA sequence data (fig. 11). Although none of the morphological traits that distinguish this subgenus from Xerodelphys (table 15) optimize as synapomorphies on the best current estimate of thylamyine relationships (Voss and Jansa, 2009: fig. 36), resolution of the still-uncertain relationships of Chacodelphys could change this situation. Because Chacodelphys, Lestodelphys, and Xerodelphys are all short-tailed taxa that lack a distinct caudal prehensile surface, phylogenies that include the nested clades (Chacodelphys (Lestodelphys (Xerodelphys (Thylamys)))) could constrain character optimizations such that a relatively long tail (≥ HBL) and well-developed caudal prehensile surface would be synapomorphic for the subgenus Thylamys.

Thylamys macrurus (Olfers, 1818)

Synonyms

griseus Desmarest, 1827; marmotus Oken, 1816 (unavailable).

Distribution

Thylamys macrurus is only known from eastern Paraguay and the adjacent Brazilian state of Mato Grosso do Sul. Published range maps based on correctly identified material of this species are in Carmignotto and Monfort (2006: fig. 7), Creighton and Gardner (2008: map 50), and Voss et al. (2009: fig. 3). Additional Brazilian records that we assume to be valid were reported by Cáceres et al. (2007). As noted by Voss et al. (2009), a Bolivian record (in Anderson, 1997) and a record from NE Argentina (mapped by Brown, 2004) are based on misidentifications. Voss et al. (2009: 417–418) discussed the problematic ecological interpretation of capture records for this species, which has been taken in arborescent Cerrado habitats in Brazil but in second-growth moist forest in Paraguay. This species is not known to occur sympatrically with any other congener.

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present); ventral pelage self-white (or -whitish), usually with narrow lateral zones of gray-based fur; plantar pads of manus separate, surrounding concave central palmar surface, and with well-developed dermatoglyphs; manual claws short, not extending much if at all beyond fleshy apical pads of digits; tail much longer than combined length of head and body (LT/HBL × 100  =  123%; N  =  5), with pale tip (terminal 10–50 mm whitish above and below); ventral prehensile surface of tail tip well developed. Nasal bones long (extending posteriorly as far as lacrimals) or very long (extending posteriorly beyond lacrimals); lacrimal foramina partially exposed on orbital margin or concealed inside orbit; infraorbital foramen above P3 or above P3/M1 commissure; nasolabial fossa shallow; supraorbital margins usually beaded, the beads sometimes developed as postorbital processes in old adult males; maxillary fenestrae present; crown of second upper incisor (I2) smaller than or subequal to crown of I3; stylar cusp C almost always present on M1 and M2; metaconule distinct on M3 (but only one specimen confidently scored for this trait).

Comparisons

Thylamys macrurus has been described as the largest species in the genus (Solari, 2003; Carmignotto and Monfort, 2006), but it exhibits broad morphometric overlap with some species in the Elegans and Venustus groups, from which it is more readily distinguished by qualitative external and craniodental characters (table 16). Within the nominotypical subgenus, macrurus is qualitatively similar to pusillus (both species usually have beaded supraorbital margins, maxillary fenestrae, stylar cusp C on M1 and M2, and a distinct metaconule on M3), but macrurus is much larger (e.g., with nonoverlapping molar measurements; table 16), usually has narrow lateral zones of gray-based ventral fur (the ventral pelage is entirely self-white in pusillus), a pale tail tip (usually absent in pusillus), and shorter posterolateral palatal foramina (see Carmignotto and Monfort, 2006: fig. 6).

TABLE 16

Morphological Comparisons among Species and Species Groups of Thylamys (subgenus Thylamys)^a

Remarks

The complex nomenclatural history of this species was reviewed by Voss et al. (2009), who designated a neotype and tabulated measurement data from most of the known adult and subadult specimens. A detailed morphological description based on freshly collected Brazilian material was provided by Carmignotto and Monfort (2006), who also tabulated measurement data and illustrated craniodental traits not depicted in this report.

Specimens examined (N  =  11)

Brazil—Mato Grosso do Sul, Campo Grande (MZUSP 3782), Fazenda Califórnia (MZUSP 32094, 32095, 32096), Fazenda Santa Terezinha (MZUSP 32097). Paraguay—Amambay, 28 km SW Pedro Juan Caballero (UMMZ 125243 [neotype], 125259, 125260); Central, Asunción (BMNH 99.11.17.1); Concepción, 7 km NE Concepción (MSB 70700); Paraguarí, Sapucay (BMNH 3.4.7.21).

Thylamys pusillus (Desmarest, 1804)

Synonyms

bruchi Thomas, 1921; citellus Thomas, 1912; nanus Oken, 1816 (unavailable); nanus Olfers, 1818; pulchellus Cabrera, 1934; verax Thomas, 1921.

Distribution

This lowland species (all examined specimens are from <1000 m above sea level) occurs in southeastern Bolivia, western Paraguay, and northern Argentina. Based on material that we examined, the distribution of Thylamys pusillus is almost coextensive with the tropical and subtropical dry forests and savannas collectively known as the Chaco, but a few collection localities (in the Argentine provinces of Mendoza, Misiones, and San Luis) are just outside the limits of the Chaco as that biome is conventionally recognized by authors (e.g., Short, 1975). Records of this species from Patagonian habitats (Birney et al., 1996) are based on misidentifications, as are Anderson's (1997) records from high elevations in the Bolivian department of Chuquisaca (see Remarks, below) and, apparently, a single published record from Uruguay (González et al., 2000; see Teta et al., 2009). Brown's (2004: fig. 88) map of the range of T. pusillus includes numerous extralimital records, all of which are apparently based on misidentifications or erroneous synonymies.3

The known geographic range of Thylamys pusillus overlaps that of T. pallidior in northern Argentina, where the two species have been collected sympatrically in the Reserva Biósfera de Ñacuñán (ca. 34°S, 68°W) in Mendoza province. The range of T. pusillus also overlaps that of T. venustus in southeastern Bolivia, where these species have been collected sympatrically near Carandaytí (gazetteer locality 44) in Chuquisaca department, and near Villa Montes (21°15′S, 63°30′W) in Tarija department.

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present); ventral pelage entirely self-white (yellowish in some museum specimens); plantar pads of manus separate, surrounding concave central palmar surface, and with well-developed dermatoglyphs; manual claws short, not extending much if at all beyond fleshy apical pads of digits; tail longer than combined length of head and body (LT/HBL × 100  =  116%; N  =  30), without pale tip; ventral prehensile surface of tail tip well developed. Nasal bones long (extending posteriorly as far as lacrimals) or very long (extending beyond lacrimals); lacrimal foramina usually concealed inside orbit or partially exposed on orbital rim; infraorbital foramen usually above P3; nasolabial fossa shallow; supraorbital margins often beaded, the beads sometimes laterally expanded as postorbital processes in old male specimens; maxillary fenestrae usually present; crown of second upper incisor (I2) consistently smaller than or subequal to crown of I3; stylar cusp C usually present on M1 and M2; metaconule usually well developed on M3.

Comparisons

Thylamys pusillus can be distinguished from other members of the nominotypical subgenus by a unique combination of qualitative external and craniodental characters (table 16). Additionally, pusillus is smaller than most other species in the subgenus Thylamys (LM  =  4.8–5.8 mm; N  =  58), and the skull is unusually wide relative to its length, a distinctive proportion that is conspicuous in side-by-side cranial comparisons (Voss et al., 2009: fig. 5).

Thylamys pusillus has previously been confused with two other small congeners with self-white ventral pelage, T. karimii and T. pallidior. Diagnostic morphological differences between T. pusillus and T. karimii, first clearly stated by Carmignotto and Monfort (2006), are effectively summarized in our diagnosis of the subgenus Xerodelphys (above) and need not be repeated here. By contrast, morphological comparisons between T. pusillus and T. pallidior, although discussed in substantive detail by Voss et al. (2009), merit additional comment based on the larger samples of both species examined for this study.

Thylamys pusillus and T. pallidior are most reliably distinguished externally by claw length: whereas the manual claws of pusillus are short, not extending much (if at all) beyond the fleshy apical pads of the digits, those of pallidior are substantially longer, projecting well beyond the fleshy fingertips. Additionally, pusillus is, on average, shorter furred than pallidior. Among 53 measured adult specimens of pusillus, the middorsal fur ranges in length from 6–11 mm, but most specimens (79%) have middorsal fur that is only 7–9 mm long. By contrast, among 44 measured adult specimens of pallidior, the middorsal fur ranges in length from 10–15 mm, but most specimens (82%) have fur that is 11–13 mm long. Although a couple of millimeters' difference in average fur length does not sound like much, the contrast is visibly and tactilely apparent with representative specimens of both species in hand. A third external difference that might be useful for field identification concerns ventral pelage color, which is completely self-white in almost all examined specimens of pusillus, whereas narrow lateral zones of gray-based hairs are often present in pallidior.

Tables 7–13 document modal differences between Thylamys pusillus and T. pallidior in several craniodental characters, notably including infraorbital foramen position (usually over P3 in pusillus, usually over M1 in pallidior), maxillary fenestrae (usually present in pusillus, usually absent in pallidior), stylar cusp C (usually present in pusillus, usually absent in pallidior), and metaconules (usually well developed in pusillus, usually indistinct or absent in pallidior). These species are additionally distinguished by cranial proportions described and illustrated by Voss et al. (2009: 421, fig. 5).

Although no single qualitative craniodental character unambiguously distinguishes Thylamys pusillus from T. pallidior (contra Voss et al. [2009], who examined smaller series of both species), specimens that are atypical in one character are seldom atypical in others. For example, USNM 390033, a Paraguayan specimen that we identify as pusillus despite its lack of maxillary fenestrae, exhibits most of the other distinctive characteristics of the species, including completely self-white ventral pelage, short claws, weakly beaded supraorbital margins, infraorbital foramen above P3, well developed stylar cusp C on M1 and M2, and a well-developed metaconule on M3. Thus, specimens can be reliably identified to species using a combination of traits.

Remarks

Our analyses of cytochrome-b sequence data recovered several robustly supported haplogroups that we regard as conspecific despite the presence of noteworthy geographic size variation (Teta et al., 2009; Voss et al., 2009). Because the collection localities of sequenced specimens of Thylamys pusillus are spatially clustered with large intervening areas from which no sequence data are available (fig. 7), geographic sampling bias is a possible explanation for the seemingly distinct groups in our cytochrome-b analyses. Nevertheless, sequence divergence among these haplogroups is impressive: 6.8% between haplogroup A and haplogroup B, 7.6% between haplogroup A and haplogroup C, and 5.8% between haplogroup B and haplogroup C. If these really are conspecific mitochondrial lineages, they are old ones. In the event that diagnostic morphological criteria are eventually found to reliably distinguish these haplogroups, we discuss the application of available names in the paragraphs that follow.

Haplogroup A, represented in our analyses by samples from southeastern Bolivia and western Paraguay, corresponds to Thylamys pusillus in the strict sense of Voss et al. (2009), who designated a neotype from the Paraguayan department of Boquerón; nanus is an objective junior synonym (based on the same type material); and verax, a subjective junior synonym, is also based on a western Paraguayan type (table 14). Voss et al. (2009) conjectured that typical pusillus might be restricted to the tropical and subtropical dry forests north of the Río Pilcomayo, from which measured specimens seem to be larger (with upper molar tooth rows > 5 mm) than specimens collected south of the Pilcomayo—in Argentina—which are mostly smaller (with tooth rows < 5 mm). However, a previously unexamined USNM series from Fortín Guachalla (22°27′S, 62°20′W) on the north bank of the Pilcomayo is geographically intermediate to the material examined by Voss et al. (2009), and it is also morphologically intermediate (with LM  =  4.9–5.3 mm), suggesting that geographic variation is clinal rather than abrupt, and that the Pilcomayo does not separate morphologically distinctive forms.

Haplogroup B, represented by geographically adjacent samples from the Argentine provinces of Catamarca and Santiago del Estero, corresponds to the small-bodied phenotype that Voss et al. (2009) associated with bruchi, a name that has been variously treated as a junior synonym of Thylamys pusillus (e.g., by Voss and Jansa, 2009) or T. pallidior (e.g., by Gardner, 2005). Although the type locality of bruchi is in the Argentine province of San Luis, from which no additional pusillus-like material is known to have been collected, the subadult holotype (BMNH 21.4.21.8) has all the distinguishing characteristics of pusillus, including short claws, an infraorbital foramen that is dorsal to P3, distinct maxillary fenestrae, a well-developed stylar cusp C on M1 and M2, and a well-developed metaconule on M3. The name pulchellus (based on a specimen from Santiago del Estero; table 14) also applies to this haplogroup according to Teta et al. (2009), but we have not examined the type.

Voss et al. (2009) thought that bruchi might be distinct from pusillus based on the smaller size of specimens collected south of the Pilcomayo by comparison with specimens collected in Paraguay and Bolivia, and because they observed correlated geographic variation in the frequency of stylar cusp C on M3. As discussed above, however, newly examined material suggests that geographic size variation in this complex is clinal and does not provide a satisfactory basis for identifying unsequenced specimens. Representative voucher specimens of haplogroup B are, in fact, smaller than vouchers of haplogroup A (table 17), but observed metrical differences are small, and stylar cusp C occurs polymorphically on M3 in voucher material of both haplogroups. Because close scrutiny has not revealed additional phenotypic differences between specimens that might be referred to bruchi on the one hand and to pusillus on the other, it seems appropriate to adopt the taxonomically conservative view that these are conspecific forms.

TABLE 17

Measurements (mm) of Sequenced Adult Specimens of Thylamys pusillus^a

Haplogroup C represents the nominal taxon citellus, an identification that is convincingly established by sequence data that we obtained from a century-old paratype (BMNH 1898.8.19.12; table 1).4 This haplogroup occurs in the so-called Mesopotamian region of Argentina (between the Paraná and Uruguay rivers), where our samples were collected in the provinces of Corrientes, Entre Ríos, and Misiones. Measured voucher material and other (unsequenced) Mesopotamian specimens are larger than bruchi (e.g., with LM  =  5.4–5.8 mm), but they overlap broadly with typical pusillus (from the Chaco Boreal) in all measured craniodental dimensions and lack any distinguishing qualitative trait. Whereas several specimens from Corrientes (e.g., BMNH 98.8.19.9, 98.8.19.11, 98.8.19.12) have white-tipped tails, for example, this marking was not observed in material that we examined from Misiones.

Specimens from high elevations (> 2000 m) in the Andes of Chuquisaca department, Bolivia, that Anderson (1997: 164) identified as Thylamys pusillus all appear to be examples of T. venustus. Relevant specimens that we examined include sequenced material from “9 km by road N Padilla,” “11 km N and 16 km W Padilla,” “12 km N and 11 km E Tarabuco,” “2 km N Tarabuco,” and “4 km N Tarabuco” (see Specimens Examined in our account for T. venustus, below). Although some material from these localities remains unexamined, the lack of any verified high-elevation records of T. pusillus and the lack of accurately diagnostic couplets in Anderson's (1997: 29) key to Thylamys lead us to believe that none of his Andean material represents this species.

Specimens of Thylamys pallidior have sometimes been misidentified as T. pusillus, most recently by Birney et al. (1996), who reported finding two distinct phenotypes in material they collected in “Monte” and “Patagonian” habitats of Chubut province, Argentina. However, we examined Birney at al.'s material (in the MMNH) and found no morphological differences between their Monte and Patagonian series. The morphometric variation among adult MMNH specimens from Chubut is well within the normal range of intraspecific variation in Thylamys, and both series conform qualitatively to our diagnosis of T. pallidior.5 No specimen in either series resembles T. pusillus. Unfortunately, tissue samples from Birney at al.'s Chubut material (which may still exist in a freezer at Texas Tech University; C.J. Phillips, personal commun.) could not be obtained for our sequencing study.

Specimens examined (N  =  84)

Argentina—Catamarca, Bella Vista (OMNH 32562, 32563), Chumbicha (OMNH 23483), 5.2 km NW Chumbicha (OMNH 32561); Corrientes, Estancia Corona, near Goya (BMNH 94.6.30.1), Goya (BMNH 98.8.19.9 [holotype of citellus], 98.8.19.10–98.8.19.12); Entre Ríos, La Paz (BMNH 23.12.12.16); Mendoza, Ñacuñán Reserve (UWBM 72205); Misiones, Dos de Mayo (ZSM 1966/70–1966/74); San Luis, Alto Pencoso (BMNH 21.4.21.8 [holotype of bruchi]); Santiago del Estero, 6 km S and 2 km E Pampas de los Guanacos (OMNH 23479); Tucumán, Tapia (USNM 236332). Bolivia—Chuquisaca, 3.8 km by road E Carandaytí (AMNH 261268; MSB 55846); Santa Cruz, 53 km E of Boyuibe (AMNH 275441; MSB 87105), Tita (AMNH 260025); Tarija, Estancia Bolívar (AMNH 275440, 275442, 275445, 275446; MSB 67016–67018, 87103, 87104), 8 km S [and] 10 km E Villa Montes (AMNH 246442–246444, 246446–246449, 246452). Paraguay—Alto Paraguay, Destacamento Militar Gabino Mendoza (TTU-TK 65601, 65632, 65635), Fortín Pikyrenda (TTU-TK 65592, 65612), Palmar de las Islas (TTU-TK 65458, 65463); Boquerón, Estancia “El 43” (TTU-TK 60217, 60227), Estancia Toledo (FMNH 164097), Experimental Farm (FMNH 164095, 164096), Fortín Guachalla (FMNH 54369; USNM 390027–390033), Orloff (FMNH 63862), Parque Cué (TTU-TK 63360, 63367), Parque Nacional Teniente Agripino Enciso (TTU-TK 65031, 65104, 65215, 66463, 66468, 66469, 66476; USNM 555660), Schmidt Ranch (FMNH 164086), 410 km NW Villa Hayes (MVZ 144312, 144313), 460 km NW Villa Hayes (MVZ 144311 [neotype of pusillus and nanus]); Chaco, 50 km WNW Fortín Madrejon (UMMZ 124676); Nueva Asunción, 1 km SW km 620 [of] Trans-Chaco Road (UMMZ 176357); 19 km by road WSW km 588 [of] Trans-Chaco Road (UMMZ 176358, TWN 240, 275, 390). Presidente Hayes, Misión Central (BMNH 20.12.18.34 [holotype of verax]), 295 km NW Villa Hayes (MVZ 144310).

Synonyms

?soricinus Philippi, 1894.

Distribution

As restricted by authors (Meynard et al., 2002; Solari, 2003; Braun et al., 2005; Creighton and Gardner, 2008) and as recognized in this report, Thylamys elegans is endemic to central Chile, where its distributional limits remain to be convincingly documented. Examined specimens that we refer to this species were collected between about 30° and 36°S (in the regions of Coquimbo, Valparaíso, Metropolitana, and Maule), from near sea level to about 1000 m. As noted by Palma (1995), the specimens reported from much lower latitudes and/or higher elevations in northern Chile (Atacama and Tarapacá) by Pine et al. (1979) are referable to T. pallidior. Although Creighton and Gardner (2008: 110) did not map any northern Chilean localities for T. elegans, the elevational range that they attributed to this species (“from sea level to over 3500 m”) was presumably based on Pine et al.'s (1979) misidentified material. The type of soricinus is said to have been collected near Valdivia (ca. 40°S), but no additional specimens have apparently been collected so far to the south (Palma, 1997).

Although the geographic range of T. elegans is not known to overlap that of any other congeneric species, newly identified material (see Remarks, below) indicates that the the range of T. pallidior closely approaches it in the Coquimbo region, where the two species might eventually be found to occur sympatrically.

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present); ventral pelage narrowly self-whitish or -yellowish from chin to anus, with broad lateral zones of gray-based hairs (in specimens from Coquimbo, Metropolitana, and Valparaíso), or almost entirely gray-based buffy with or without self-whitish midpectoral markings (in specimens from Maule); plantar pads of manus separate, surrounding a concave central palmar surface, and with well developed dermatoglyphs; manual claws long, extending well beyond fleshy apical pads of digits; tail usually longer than combined length of head and body (LT/HBL × 100  =  108%; N  =  27), without pale tip; prehensile ventral surface of tail tip well developed. Nasal bones usually short (not extending posteriorly as far as the lacrimals); lacrimal foramina usually exposed anterior to orbit or partially exposed on orbital rim; infraorbital foramen above P3/M1 commissure or above M1; nasolabial fossa shallow; supraorbital margins usually rounded or squared; maxillary fenestrae consistently absent; crown of second upper incisor (I2) smaller than or subequal to crown of I3; stylar cusp C almost always absent on M1 and M2; metaconule absent or indistinct on M3.

Comparisons

Thylamys elegans can be distinguished unambiguously from T. pallidior by its much broader lateral zones of gray-based ventral fur, which are equal to or wider than the median streak of self-colored fur (versus much narrower than the self-colored zone in pallidior). Additionally, the self-colored ventral fur of elegans is usually off-white (“cream”) or even yellowish, whereas the self-colored ventral fur of pallidior is snowy white. Other characters contribute to the phenotypic distinctness of these species, although none provides a sufficient basis for identification by itself: elegans has, on average, larger hind feet (14–17 mm) than pallidior (12–15 mm), longer manual and pedal digits, shorter nasals (table 7), and less inflated auditory bullae. Digital lengths are difficult to measure (we did not attempt to do so), but the longer fingers and toes of elegans are apparent in most side-by-side comparisons. Bullar size is indexed by the ratio of bullar width to interbullar width (BW/IBW; see fig. 2), a proportion that averages 0.85 ± 0.05 SD (N  =  23) in elegans and 0.98 ± 0.07 SD (N  =  44) in pallidior.

Thylamys elegans externally resembles T. tatei in having wide lateral zones of gray-based ventral fur, but these species can be distinguished unambiguously by incisor morphology. Whereas the crown of I2 is consistently smaller than or subequal to the crown of I3 in elegans, the crown of I2 is consistently larger than the crown of I3 in tatei (fig. 17). Additionally, the nasals of elegans are usually shorter than those of tatei (table 7); one or both lacrimal foramina are usually exposed in elegans, whereas both foramina are concealed in tatei (table 8); and the infraorbital foramen of elegans is usually positioned over M1, whereas the foramen is often over P3 in tatei (table 9). Among the integumental differences that Solari (2003) observed between elegans and tatei, the usual presence of a pale tail tip in tatei and its absence in elegans is the most consistently useful; we did not observe conspicuous differences between these taxa in fur length (10–12 mm middorsally in most examined specimens of both species), dorsal pelage pigmentation, or facial markings.

Remarks

The holotype of coquimbensis (FMNH 22302), a taxon originally described from the Coquimbo region of Chile as a subspecies of elegans by Tate (1931), is unambiguously referable to Thylamys pallidior. All of the distinctive traits of coquimbensis mentioned by Tate (1931, 1933)—including its pale dorsal coloration, mostly white ventral fur, small feet, and large bullae—are consistent with the diagnostic traits that we attribute to T. pallidior. Side-by-side comparisons of FMNH 22302 with examples of typical T. elegans (which also occurs in Coquimbo; e.g., FMNH 119487) are likewise consistent with the suite of differences that distinguish these species (see Comparisons, above). The occurrence of both species in Coquimbo suggests that they may occur sympatrically there, perhaps somewhere in between Paiguano (30°01′S, 70°32′W; the type locality of coquimbensis) and Parque Nacional Fray Jorge (30°30′S, 71°30′W; our northernmost record of T. elegans).

We have not examined the type of soricinus, whose widely assumed synonymy with elegans still needs to be confirmed. The type of soricinus, a mounted skin with the skull inside, was last seen by Osgood (1943), who noted that it differed from typical elegans by lacking the continuous median streak of self-whitish fur seen in the latter form; instead, he described the ventral surface of the type of soricinus as entirely covered by gray-based buffy hairs. The only Chilean specimens we examined that fit this description are a small series from Maule (USNM 541587–541591) from which, unfortunately, we have no sequence data. Pine et al. (1979) previously examined this series and concluded that it represents the soricinus phenotype. We agree, but because the Maule specimens are indistinguishable from typical elegans in measurements and qualitative craniodental traits, no taxonomic distinction seems warranted at this time.

Measurements of sequenced adult specimens of Thylamys elegans are provided in table 19.

TABLE 19

Measurements (mm) of Sequenced Adult Specimens of Thylamys elegans

Specimens examined (N  =  29)

Chile—Coquimbo, 10 km N Puente Los Molles (FMNH 119487), Fray Jorge National Park (MSB 70588, 87095, 87096, 87097); Maule, confluence of Río Maule and Río Claro (USNM 541587), Siete Tasas (USNM 541588–541591); Metropolitana, 7 road km SW Camino Rinconada (UWBM 49000, 49006, 49007, 49011, 49014), 12 road km SW Camino Rinconada (UWBM 44443–44446), Fundo El Pangue (UWBM 49059), Rinconada de Maipu (MSB 133097), San Carlos de Apoquindo (MSB 133104); Valparaíso, Las Hijuelas (AMNH 97752, 97753), Olmué (AMNH 97755), Palmas de Ocoa (MSB 87098), Papudo (AMNH 97754), Quebrada del Tigre (MSB 133095).

Thylamys pallidior (Thomas, 1902)

Synonyms

coquimbensis Tate, 1931; fenestrae Marelli, 1932.

Distribution

As recognized in this report, Thylamys pallidior occurs from eastern Arequipa and Tacna departments in southwestern Peru eastward and southward into Bolivia (Oruro, Chuquisaca, Tarija) and northern Chile (Antofagasta, Atacama, Coquimbo, Tarapacá) to Argentina. In Argentina, Thylamys pallidior ranges from Jujuy province southward to Chubut (Flores et al., 2007). Specimens that we examined document an altitudinal range from near sea level to 3750 m. Misidentified material that has been reported from extralimital localities is discussed under Remarks (below).

The known geographic range of Thylamys pallidior overlaps that of T. sponsorius in northwestern Argentina, where the two species have been collected sympatrically near Barcena (gazetteer locality 11) in Jujuy province. The range of T. pallidior also overlaps that of T. pusillus, as described in the account of the latter species (above). Although the ranges of T. pallidior and T. venustus also overlap in northwestern Argentina, these species have apparently not been collected sympatrically. Sympatry between T. pallidior and T. elegans is to be expected in the Coquimbo region of Chile, where their known geographic ranges are closely juxtaposed (see Remarks for T. elegans, above).

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present); ventral pelage entirely self-white or with narrow lateral zones of gray-based hairs; plantar pads of manus separate, surrounding concave central palmar surface, and with well-developed dermatoglyphs; manual claws long, extending well beyond the fleshy apical pads of digits; tail longer than combined length of head and body (LT/HBL × 100  =  112%; N  =  41), without pale tip; prehensile ventral surface of tail tip well developed. Nasal bones variable in length (without any clearly modal condition; table 7); lacrimal foramina often partially exposed on orbital rim; infraorbital foramen usually above M1 but sometimes above P3/M1 commissure (apparently never above P3); nasolabial fossa shallow; supraorbital margins usually rounded or squared; maxillary fenestrae almost always absent; crown of second upper incisor (I2) consistently smaller than crown of I3; stylar cusp C almost always absent on M1 and M2; metaconule usually absent on M3.

Comparisons

Comparisons of Thylamys pallidior with T. pusillus and T. elegans are summarized in the preceding species accounts. Thylamys pallidior differs from the only other currently recognized species in the Elegans Group, T. tatei, in size and qualitative characters. Although pallidior is, on average, smaller than tatei in all measured dimensions, age-invariant molar measurements provide the best univariate index for this comparison: mean length of the upper molar row (LM, see fig. 2) is 5.7 ± 0.2 mm S.D. (observed range: 5.3–6.0 mm; N  =  51) in pallidior and 6.2 ± 0.2 mm (observed range: 5.9–6.4 mm; N  =  11) in tatei. Qualitatively, these species are most readily distinguished by their ventral pelage (almost completely self-white in pallidior, with broad lateral zones of gray-based hairs in tatei) and by upper incisor morphology (the crown of I2 is consistently smaller than or about the same size as the crown of I3 in pallidior, whereas the crown of I2 is consistently larger than the crown of I3 in tatei; fig. 17).

Remarks

As previously reported by Braun et al. (2005), cytochrome-b sequence variation in this species is geographically structured. In particular, our results confirm those authors' discovery of two allopatric haplotype groups with distributions that are juxtaposed at or near the border between Bolivia and Argentina (fig. 5). Although this basal dichotomy is moderately well supported by nodal statistics (fig. 4), we are unable to distinguish representative voucher specimens by any morphological criterion (univariate metrical comparisons are summarized in table 20), and we therefore treat these haplogroups as conspecific. Nevertheless, names are available for each of them should future research justify their recognition as distinct taxa. In that event, pallidior (based on a Bolivian specimen; table 14) unambiguously applies to haplogroup A, and fenestrae (based on an Argentinian specimen) could be used for haplogroup B. According to Martin (2009), however, fenestrae is a distinct species restricted to the Argentinian provinces of Buenos Aires, Córdoba, and La Pampa. Our only tissue samples from these areas (AC 47, LTU 77) are vouchered by morphological specimens that we have not examined, but the sequences we obtained from them are intermingled in our tree (fig. 4) with sequences from specimens collected in other provinces. Although we have not examined any of the morphological material that Martin (2009) referred to fenestrae, our results suggest that the application of this name merits closer scrutiny.

TABLE 20

Measurements (mm) of Sequenced Adult Specimens of Thylamys pallidior^a

Specimens of other Thylamys species have sometimes been misidentified as T. pallidior. For example, Anderson's (1997: 29) key does not reliably distinguish T. pallidior from other Bolivian congeners, and his range maps include several records that fall outside the range of that species as we recognize it. Although we were not able to see every Bolivian specimen that Anderson identified as T. pallidior, all of those that we examined from the departments of Cochabamba (e.g., AMNH 275427, MSB 87109), La Paz (e.g., AMNH 248704), and Santa Cruz (e.g., AMNH 260030) are unambiguously referable to T. venustus (see below).

Among the Peruvian specimens that Solari (2003) referred to Thylamys pallidior (and which were mapped as such by Brown, 2004: fig. 87) are several that appear to represent undescribed taxa (table 1). Some of these (e.g., MVZ 116614; from 3 mi W Atico in Arequipa department; fig. 5, locality 98) are morphologically similar to pallidior (with mostly self-white underparts), whereas others (e.g., MVZ 137585; from Lima department; fig. 5, locality 103) differ from typical pallidior by having broad lateral zones of gray-based hairs bordering a narrower median streak of self-white fur. Cytochrome-b sequences that we amplified from museum skins of these and other unidentified specimens from southwestern Peru are highly divergent from pallidior and form a monophyletic group with two sequences of T. tatei (fig. 4). Indeed, one of these specimens (MVZ 119913; also from Lima department; fig. 5, locality 102) resembles tatei and differs from other (more pallidior-like) Peruvian specimens by having a white-tipped tail and very long nasal bones. Although these results are hard to interpret taxonomically with the sparse material at hand, they do suggest that the systematics of Peruvian Thylamys is likely to reward further study. Other researchers are currently working on this problem using freshly collected material (E. Palma, personal commun.), and we await their results with the greatest interest.

Specimens examined (N  =  69)

Argentina—Catamarca, 17 km N Barranca Larga (OMNH 29964), 34.6 km by road W Fiambala (OMNH 34903), 7 km SW Los Morteros (OMNH 32556); Chubut, ca. 208 km W Dolavon (MMNH 15709, 15712–15714), Estancia La Escondida (MMNH 17323, 17324, 17366), Istmoameghino (MMNH 15722); Jujuy, 9 km NW Barcena (OMNH 29963), 11 km E Humahuaca and 2 km E Pucará on road to Cianzo (OMNH 29957), 8.2 km S Sey (OMNH 34911); Mendoza, Ñacuñán Reserve (UWBM 72195), 49.2 road km N Mendoza (UWBM 72224); 3 km W Refugio Militar General Alvarado (OMNH 23482), Salinas del Diamante RR Station (OMNH 23480); Neuquén, 16 km SE La Rinconada (MVZ 163772); Río Negro, General Roca (USNM 236331); Salta, 16 km S and 1.8 km W Barrancas along Río de las Burras (OMNH 34908, 34909), 17 km NW Cachi (OMNH 32559), Los Sauces (OMNH 32544); San Juan, Castaño Nuevo (OMNH 23485), 8 km W Complejo Astronómico El Leoncito (OMNH 32571), Quebrada de los Flores (OMNH 29961, 29962); San Luis, 7 km E San Francisco del Monte de Oro (OMNH 23489), 12 km by road N Varela (OMNH 23490), 15 km E Salinas del Bebedero (OMNH 23488); Tucumán, Tafí del Valle (AMNH 41723–41727; FMNH 41397, 41398). Bolivia—Chuquisaca, 68 km by road N Camargo (AMNH 262405–262407, MSB 57003); Oruro, Challapata (BMNH 2.2.2.116 [holotype of pallidior], USNM 121157), ca. 10 km by road SW Pazña (UMMZ 155830, 155831, 156015); Tarija, 1 km E Iscayachi (AMNH 262408), 2 km SE Cieneguillas (FMNH 162495), Serranía Sama (MSB 87099). Chile—Antofagasta, Muelle de Piedra E of Taltal (AMNH 143240); Atacama, Altamira (USNM 391776), El Transito (USNM 391775); Coquimbo, Paiguano (FMNH 22302 [holotype of coquimbensis]); Tarapacá, 5 km S Belén (USNM 541600), 1 km W Belén (USNM 541595, 541596), Camarones Valley (USNM 391777), Chapiquiña (USNM 541599), Enquelga (MSB 133108), 7 km SE Socorama (USNM 541593, 541594). Peru—Arequipa, 1 km N Chivay (MVZ 173937–173939), 3 mi N Mollendo (MVZ 145531); Tacna, 1.5 mi N Tarata (MVZ 139215), 4 km N Tarata (MVZ 115634), 65 km W Tacna (MVZ 143695, 143696).

Thylamys tatei (Handley, 1957)

Synonyms

None.

Distribution

According to Solari (2003), who examined more material than we have seen, Thylamys tatei is known only from the Peruvian department of Ancash and from the Lomas de Lachay in the northern part of the department of Lima, between 300 and 3000 m above sea level. This species is not known to occur sympatrically with any other congener.

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present); ventral pelage with narrowly continuous median streak of self-white (or whitish) fur from chin to anus, flanked by broad lateral zones of gray-based hairs; plantar pads of manus separate, surrounding a concave central palmar surface, and with well-developed dermatoglyphs; manual claws long, extending well beyond fleshy apical pads of digits; tail longer than combined length of head and body (LT/HBL × 100  =  113%; N  =  9), usually with pale tip (whitish above and below for distal 2–18 mm); prehensile ventral surface of tail tip well developed. Nasal bones usually very long, extending posteriorly well behind lacrimals; lacrimal foramina concealed from lateral view inside orbit; infraorbital foramen above P3 or above P3/M1 commissure; nasolabial fossa shallow; supraorbital margins almost always rounded or squared; maxillary fenestrae almost always absent; unworn crown of second upper incisor (I2) consistently larger than crown of I3; stylar cusp C variably present or absent on M1, usually absent or indistinct on M2; metaconule absent on M3.

Comparisons

Comparisons between Thylamys tatei and other currently recognized species of the Elegans Group are provided in the preceding accounts. Measurement data from sequenced specimens of T. tatei and specimens of related but unnamed forms from the Peruvian departments of Arequipa, Ayacucho, and Lima are provided in table 21.

TABLE 21

Measurements (mm) of Sequenced Specimens of Thylamys sp. and T. tatei from Peru^a

Remarks

Originally described as a valid species, tatei was subsequently listed as a synonym of Thylamys elegans by Gardner (1993). Although our results are obviously consistent with recent inferences from molecular and morphological studies that tatei and elegans are distinct species (Meynard et al., 2002; Palma et al., 2002; Solari, 2003; Braun et al., 2005), other taxonomic problems remain. Foremost among these is the identification of specimens from the Peruvian provinces of Lima (e.g., MVZ 119913, 137585) and Huancavelica (e.g., MVZ 136249) that resemble tatei in having broad lateral zones of gray-based ventral fur but which lack some of the diagnostic craniodental traits of tatei and have divergent mtDNA sequences. Although Solari (2003) identified these as pallidior (as discussed above), they cluster with tatei in our phylogenetic analyses.

Specimens examined (N  =  13)

Peru—Ancash, Chasquitambo (USNM 302915 [holotype], 302916), Huaráz (FMNH 81443), 1 km N and 12 km E Pariacoto (MVZ 155503–135512).

Synonyms

janetta Thomas, 1926.

Distribution

Based on specimens that we sequenced, Thylamys sponsorius occurs along the eastern slopes and foothills of the Andes from southern Bolivia (Tarija) to northwestern Argentina (Jujuy, Salta, Tucumán, Catamarca; fig. 9) with recorded elevations ranging from 515 to 3750 m. Although T. sponsorius and T. venustus have broadly overlapping horizontal and vertical distributions, our only sympatric samples are from 3 km SE Cuyambuyo, at 900 m elevation in the Bolivian department of Tarija (fig. 9, gazetteer locality 74). The known geographic range of Thylamys sponsorius also overlaps that of T. pallidior as previously described in the account of the latter species (above).

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler lateral coloration present); ventral pelage gray-based yellowish or whitish, except in some old adults from Carlazo, Bolivia (type series and topotypes of “janetta”), which have mostly self-whitish venters; plantar pads of manus separate, surrounding a concave central palmar surface; manual claws short, not extending much if at all beyond fleshy apical pads of digits; tail substantially longer than the combined length of head and body (LT/HBL × 100  =  136%; N  =  13), without pale tip; prehensile ventral surface of tail tip well developed. Nasal bones usually extending posteriorly as far as or behind lacrimals; lacrimal foramina usually concealed or only partially exposed on orbital rim; infraorbital foramen often over P3/M1 commissure, but sometimes over P3 or M1; nasolabial fossa shallow; supraorbital margins rounded or squared, with weak beads in some old adults but apparently never with distinct processes; maxillary fenestrae consistently present; crown of second upper incisor (I2) consistently smaller than or subequal to crown of I3; stylar cusp C almost always absent on M1 and M2; metaconule indistinct or absent on M3.

Comparisons

Sequenced material of Thylamys sponsorius seems to be indistinguishable from that of T. venustus in external morphological characters. Voucher specimens of both clades have brownish-gray dorsal fur that tends to be darker at lower elevations and paler in the highlands (>2000 m). The ventral fur is predominantly gray based in most specimens, although the fur of the chin and throat is often self-colored, and many specimens of both species have self-colored pectoral markings; the self-colored ventral fur (if any) and the tips of the gray-based ventral hairs are white or off-white (cream) in most highland specimens but yellowish in lowland material. The hind feet of T. sponsorius seem a little darker, on average, than those of T. venustus, but the difference is not sufficiently marked to sort skins, and it is probably useless for field identification.

Sequenced specimens of Thylamys sponsorius and T. venustus are likewise indistinguishable by any qualitative craniodental character that we examined, nor are they consistently diagnosable by any combination of qualitative craniodental traits. Tests for craniodental trait frequency differences (as recorded in tables 7–13) are statistically significant only for the presence/absence of stylar cusp C, which is not present on M2 in any sequenced specimen of sponsorius, but which occurs on that tooth in a substantial number of sequenced specimens of venustus (table 12; p < 0.01 by Fisher's exact test). Among traits that we did not formally score for statistical analysis, small postorbital processes were observed on several sequenced specimens of venustus (e.g., AMNH 186948; MSB 63260, 67005) but not on any sequenced specimen of sponsorius.

Sequenced adult specimens of Thylamys sponsorius and T. venustus differ significantly in most measured dimensions. Two-tailed t-tests for species differences among males, the sex most abundantly represented among our adult tissue vouchers, are significant at the 5% level for all measurements except HBL, Ear, NL, BW, and IBW (table 24). In every case where a significant difference was found, sponsorius is the larger species. Unfortunately, observed ranges of the two species overlap for every measurement, so none is diagnostic. In the absence of other criteria, provisional identifications of unsequenced specimens can be based on discriminant scores (computed as the dot product of the coefficients in table 23 and log_e-transformed craniodental measurement values). If our hypothesis of species status for these clades is correct, then the scores of unsequenced specimens (of which several hundred are preserved in museum collections) should have a distinctly bimodal distribution.

TABLE 24

Measurements (mm) of Sequenced Adult Specimens of Thylamys sponsorius and T. venustus^a

Remarks

This taxon was originally described as Marmosa elegans sponsoria, but Tate (1933) treated sponsoria as a subspecies of M. venusta, and Cabrera (1958) synonymized it with M. e. cinderella. Gardner (1993) listed cinderella, sponsorius, and venustus as synonyms of Thylamys elegans, but allozyme data reported by Palma and Yates (1998) suggested that venustus—the name they used for Bolivian populations of “elegans” (sensu Gardner, 1993)—was a distinct species. Subsequently, Flores et al. (2000) recognized three Argentine species of Thylamys with gray-based ventral fur, for which they used T. cinderella, T. sponsorius, and T. venustus as valid binomials. However, cinderella and sponsorius were synonymized again by Braun et al. (2005), who found genetic evidence for only two Argentine species of Thylamys with gray-based ventral fur, one corresponding to the phenotype that Flores et al. called venustus and the other containing specimens that Flores et al. had identified as cinderella and sponsorius. Gardner (2005) and Creighton and Gardner (2008) recognized T. cinderella, T. sponsorius, and T. venustus (including janetta) as valid following Flores et al. (2000), but Voss and Jansa (2009) followed Braun et al. (2005) in recognizing only T. cinderella (including sponsorius) and T. venustus (including janetta) as valid species.

Our use of the name sponsorius is based on the morphometric analysis described above, which unambiguously associates the holotype (BMNH 21.1.1.85, an adult male from Sunchal, in the Sierra de Santa Bárbara of southeastern Jujuy, Argentina; Thomas, 1921) with the cytochrome-b clade that has larger average measurements. Because the same analysis assigned the holotype of cinderella (BMNH 0.7.9.20, an adult female from “Tucuman,” Argentina; Thomas, 1902) to the clade with smaller average measurements, we conclude that these names are not synonymous. Instead, the only synonym of sponsorius that we are currently able to identify as such is a Bolivian form (janetta) whose relationships have long been controversial.

Marmosa janetta was originally described on the basis of 12 specimens collected at elevations ranging from 1700 to 2300 m in Tarija department, Bolivia, by Thomas (1926), who thought it was related to M. verax (a junior synonym of Thylamys pusillus; see above). Tate (1933: 221) also recognized M. janetta as a valid species and said that it was “the mountain representative of the marmota radiation” (marmota being an unavailable synonym of T. macrurus; see Voss et al., 2009). Subsequently, Cabrera (1958) included janetta without comment in the synonymy of M. elegans venusta. Most subsequent authors (e.g., Solari, 2003; Gardner, 2005; Creighton and Gardner, 2008) have listed janetta as a synonym of T. venustus, but the justification for this synonymy has never been explained.

We examined the holotype of janetta (BMNH 26.1.1.167, from Carlazo), several topotypic paratypes (BMNH 26.1.1.166, 26.1.1.168–26.1.1.170; FMNH 29169, 29170) and several recently collected topotypes (UMMZ 155836, 156034–156036). Oddly, all of these specimens are very old adults with heavily worn molars. The advanced age of these specimens plausibly explains the large size attributed to janetta by Thomas (1926) and Tate (1933); this conclusion is supported by the fact that whereas continuously growing dimensions seem unusually large in this series (e.g., CBL: 29.7–31.6 mm), age-invariant measurements do not (e.g., LM: 5.7–6.0 mm). Other distinctive attributes of this nominal species, however, are harder to explain as age-related, including its mostly self-white ventral coloration, white hind feet, and short posterolateral palatal foramina (which, unlike most other Thylamys, do not extend anteriorly to the level of the protocone of M4). In other respects, janetta resembles other forms that we assign to the Venustus Group; for example, the type and all examined topotypes have short claws, and maxillary fenestrae are bilaterally present.

Our primary justification for synonymizing janetta with sponsorius are partial (609–812 bp) cytochrome-b sequences amplified from a paratype (FMNH 29170) and a topotype (UMMZ 155836) of janetta. Despite the unusual features of their morphological vouchers (as noted above), the sequences we obtained from FMNH 29170 and UMMZ 155836 cluster with other Tarija sequences that we obtained from phenotypically sponsorius-like specimens (fig. 8). Although we are not able to confidently explain all of the odd traits that characterize examined specimens from Carlazo, we assume that they are ontogenetic artifacts and/or local peculiarities, perhaps of a small and closely inbred population.

Flores et al. (2000, 2007) recognized Thylamys cinderella, T. sponsorius, and T. venustus as distinct species that could be identified by diagnostic morphometric and qualitative criteria. However, none of their relevant key couplets (Flores et al., 2000: 334) accurately distinguish voucher specimens representing the two mitochondrial clades recovered in our analysis of cytochrome-b sequence data. Additionally, we sequenced one specimen that they identified as T. sponsorius (AMNH 185323) and another that they identified as T. cinderella (AMNH 186948); according to our phylogenetic analyses, both specimens belong to haplogroup C of the species we identify as T. venustus.

Specimens examined (N  =  29)

Argentina—Catamarca, 5 km S Las Higuerillas on Hwy 9 (OMNH 29965); Jujuy, 9 km NW Barcena (OMNH 29974), 24.8 km E Santa Clara (OMNH 34534), 10 km W Tiraxi on Hwy 29 (OMNH 29970); Salta, 5 km NW Pulares (OMNH 32545), 25 km SE La Viña (OMNH 32548); Tucumán, 5 km N Las Higuerillas on Hwy 308 (OMNH 29967), Km 42 on Hwy 364 S of San Pedro de Colalao (OMNH 32566), 7 km W Ibatín (OMNH 29977), 5 km SW Siambón (OMNH 32553). Bolivia—Tarija, Carlazo (BMNH 26.1.1.166, 26.1.1.167 [type of janetta], 26.1.1.168–26.1.1.170; FMNH 29169, 29170; UMMZ 155836, 156034–156036), Chuquiaca (FMNH 162505), 3 km SE Cuyambuyo (MSB 67014, 67015), 5 km NNW Entre Ríos (AMNH 275437; MSB 67010, 67012), ca. 10 km by road W Narvaez (FMNH 162507), 1 km E Tucumilla (MSB 67009).

Thylamys venustus (Thomas, 1902)

Synonyms

cinderella Thomas, 1902.

Distribution

Based on specimens that we sequenced, Thylamys venustus occurs along the eastern slopes and foothills of the Andes from Bolivia (Cochabamba, Chuquisaca, Santa Cruz, and Tarija) to northwestern Argentina (Jujuy and Tucumán; fig. 9) with recorded elevations ranging from about 350 to 4000 m above sea level. Although we have not investigated every published extralimital record of this species, those from Lima, Peru (Tate, 1933) and Oruro, Bolivia (Anderson, 1997) appear to be based on misidentified material (see Remarks). Gardner's (2005) record from Neuquén, Argentina, was based on an editing error (A.L. Gardner, in litt., 23 February 2010).

The known geographic range of Thylamys venustus overlaps those of T. pallidior, T. pusillus, and T. sponsorius; known cases of sympatry among these species have already been described in preceding accounts.

Morphological diagnosis

Body pelage tricolored (abrupt line of transition from darker middorsal to paler dorsolateral coloration present); ventral pelage gray-based yellowish or whitish; plantar pads of manus separate, surrounding a convex central palmar surface; manual claws short, not extending much if at all beyond fleshy apical pads of digits; tail substantially longer than the combined length of head and body (LT/HBL × 100  =  131%; N  =  21), without pale tip; prehensile ventral surface of tail tip well developed. Nasal bones usually extending posteriorly as far as or behind lacrimals; lacrimal foramina variable in position (no unambiguously modal condition observed among sequenced specimens); infraorbital foramen usually above P3 or above P3/M1 commissure; nasolabial fossa shallow; supraorbital margins usually rounded or squared, sometimes produced as small postorbital processes; maxillary fenestrae almost always present bilaterally; crown of second upper incisor (I2) consistently smaller than or subequal to crown of I3; stylar cusp C variably present or absent on M1 and M2, but seldom a discrete, well-developed structure (usually partially fused with stylar cusp D); metaconule usually absent on M3.

Comparisons

See the preceding species account for morphological comparisons between Thylamys venustus and T. sponsorius.

Remarks

Diligent visual and morphometric comparisons of voucher material has not allowed us to discover any phenotypic attribute that distinguishes the three allopatric haplogroups of Thylamys venustus designated A, B, and C in figures 8 and 10. The skins, skulls, and dentitions assignable to those groups are qualitatively similar and overlap broadly in all recorded measurements. In effect, they appear to represent a single morphological species. Despite the apparent absence of diagnostic morphological traits, however, available names can tentatively be assigned to two of them based on geographic criteria: the type locality of venustus (in the Bolivian department of Cochabamba; table 14) falls within the known range of the haplogroup A, whereas the type locality of cinderella (in the Argentinian province of Tucumán) falls within the known range of haplogroup C. Apparently, no name is currently available for haplogroup B.

We examined one of the two specimens that Tate (1933: 225) identified as Marmosa venusta from the Peruvian department of Lima, on the wrong (western) side of the Andes and over 1000 km from the nearest collection locality of any material conforming to T. venustus as recognized herein. The specimen in question (FMNH 24141), a juvenile male skin and skull collected in 1922 by J.T. Zimmer at Matucana (11°51′S, 76°24′W, 2378 m; Stephens and Traylor, 1983), has a relatively short tail (108% of HBL), long claws, and a narrow but continuous median streak of self-white ventral fur flanked by broad lateral zones of gray-based hairs, traits that better match the characteristics we associate with the Elegans Group than those we associate with the Venustus Group (table 16). Solari (2003) examined this specimen and identified it as T. pallidior, but we believe it to represent one of the unnamed phenotypes related to T. tatei, as discussed above in the account for the latter species.

Three specimens from the Bolivian department of Oruro that Anderson (1997) identified as Thylamys venustus are the only examples reported from the altiplano, where only T. pallidior is known to occur. Although Anderson did not publish the museum catalog numbers of the material he examined, his card files (in the AMNH Department of Mammalogy archives) identify the Oruro specimens as CM 5227, MVZ 119912, and USNM 121157. We have not been able to examine all of these, but CM 5227 (currently identified in the Carnegie Museum database as “T. elegans”) was collected by José Steinbach in Cochabamba (not Oruro; S. McClaren in litt., 26 January 2010), and USNM 121157 (which we did examine) is unambiguously identifiable as T. pallidior. The current identification of MVZ 119912 in the database of the Museum of Vertebrate Zoology (Berkeley) is T. pallidior, in agreement with what we would have supposed based on its geographic origin.

Specimens examined (N  =  43)

Argentina—Jujuy, Highway 9 at border with Salta (OMNH 29952), Santa Bárbara (AMNH 185323), Yuto (AMNH 186948); Tucumán, 12 km WNW Burruyacú along Río Cajón (OMNH 29966), Los Chorillos (OMNH 29976). Bolivia—Chuquisaca, 3.8 km by road E Carandaytí (AMNH 261245), 2 km SW Monteagudo (MSB 63261, 63262), 9 km by road N Padilla (MSB 63267, 63268), 11 km N and 16 km W Padilla (AMNH 263558, MSB 63269), Porvenir (AMNH 261254, 261255, 261260; MSB 55837), 1.3 km SW Porvenir (AMNH 261264), Río Limón (MSB 63264), 2 km N Tarabuco (AMNH 263556), 4 km N Tarabuco (AMNH 263555), 12 km N and 11 km E Tarabuco (MSB 63265, 63272, 63273); Cochabamba, 1.3 km W Jamachuma (AMNH 275427), Parotani (BMNH 2.1.1.118–2.1.1.121 [type series of venustus]), 7.5 km SE Rodeo (AMNH 275429, MSB 87100, 87109), Tholapujru (AMNH [uncataloged]), 17 km E Totora (AMNH 275428, MSB 67001); La Paz, Caracato (AMNH 248704, 248705); Santa Cruz, Cerro Itahuaticua (MSB 63260), 5 km by road SE Comarapa (AMNH 260030), 15 km NE Quiñe (MSB 87106), 6 km NNE Quiñe (MSB 87107), 17 km S Quiñe (MSB 67003), Río Ariruma (AMNH 275433), 5.5 km by road NE Vallegrande (MSB 67005), 5.5 km by road NNW Vallegrande (MSB 67004); Tarija, 1 km S Camatindy (MSB 67007), 3 km WNW Carapari (MSB 67392), Tapecua (MSB 67008).