Specimens

We examined 1481 specimens for this report, most of which are preserved as skins and skulls in the following institutional collections: AMNH, American Museum of Natural History (New York); BMNH, Natural History Museum (London); FMNH, Field Museum of Natural History (Chicago); MCZ, Museum of Comparative Zoology, Harvard University (Cambridge); MSB, Museum of Southwestern Biology, University of New Mexico (Albuquerque); MVZ, Museum of Vertebrate Zoology, University of California (Berkeley); ROM, Royal Ontario Museum (Toronto); and USNM, National Museum of Natural History (Washington, D.C.).

Anatomical Terminology

Our names for external and craniodental structures of Marmosa follow Voss and Jansa (2009), whose anatomical descriptors are largely consistent with standard usage (Brown, 1971; Brown and Yalden, 1973; Bown and Kraus, 1979; Wible, 2003). Principal features of the skull are illustrated in figures 3 and 4. We follow Clemens (1966) in using positional criteria for naming cusps on the stylar shelf of the upper molars (not treated in detail by Voss and Jansa, 2009) as follows:

Fig. 3

Dorsal and ventral cranial views of Marmosa murina showing principal osteological features mentioned in the text. Abbreviations: als, alisphenoid; atp, alisphenoid tympanic process; bo, basioccipital; bs, basisphenoid; cc, carotid canal; ect, ectotympanic; fm, foramen magnum; fpj, frontal process of jugal; fro, frontal; gf, glenoid fossa; gpa, glenoid process of alisphenoid; ip, interparietal; jug, jugal; lac, lacrimal; max, maxillary; mp, mastoid process (of petrosal); nas, nasal; occ, occipital condyle (of exoccipital); of, orbital fossa; pal, palatine; par, parietal; pcf, paracanine fossa; pogp, postglenoid process (of squamosal); pop, postorbital process; pre, premaxillary; prgp, preglenoid process (of jugal); pro, promontorium (of petrosal); ps, presphenoid; pt, pterygoid; rtp, rostral tympanic process (of petrosal); sq, squamosal; tcf, transverse canal foramen; tf, temporal fossa; za, zygomatic arch.

Fig. 4

Left lateral cranial and mandibular views of Marmosa murina showing principal osteological features mentioned in the text. Abbreviations: als, alisphenoid; ap, angular process; atp, alisphenoid tympanic process; conp, condylar process; corp, coronoid process; ect, ectotympanic; exo, exoccipital; fpj, frontal process of jugal; fr, foramen rotundum; fro, frontal; hpp, hamular process of pterygoid; iof, infraorbital foramen; ip, interparietal; jug, jugal; lac, lacrimal; lc, lambdoid crest; lf, lacrimal foramina; maf, masseteric fossa; max, maxillary; mef, mental foramina; nas, nasal; pal, palatine; par, parietal; pc, pars cochlearis (of petrosal); pcf, paracanine fossa; pm, pars mastoideus (of petrosal); pop, postorbital process; pre, premaxillary; rmf, retromolar fossa; sq, squamosal; ssf, subsquamosal foramen; sup, supraoccipital; zps, zygomatic process of squamosal.

Stylar cusp A (styA): A small cusp (often indistinct) at the anterior end of the stylar shelf anterolabial to the paracone, to which it is sometimes connected by the preparacrista ( =  “parastyle” of authors).

Stylar cusp B (styB): A much larger cusp (almost always distinct) posterior to styA (if present) and labial to the paracone, to which it is sometimes connected by the preparacrista ( =  “stylocone” of authors).

Stylar cusp(s) C (styC): One or two small cusps (sometimes indistinct or absent) posterolabial to the paracone and anterolabial to the metacone.

Stylar cusp D (styD): Usually single and sometimes indistinct in Marmosa, this cusp is posterolabial to the metacone, to which it is never directly connected.

Stylar cusp E (styE): A small cusp (often indistinct) at the posterior end of the stylar shelf posterolabial to the metacone, to which it is always connected by the postmetacrista ( =  “metastyle,” “distostyle,” “metastylar spur,” or “metastylar corner” of authors).

Measurements

We recorded external measurements (in millimeters, mm) and weight (in grams, g) from specimen labels. The former include total length (TL), length of tail (LT), length of hind foot (HF), and length of ear (Ear). Length of head-and-body (HBL) was calculated by subtracting LT from TL.

We measured the following 29 craniodental dimensions to the nearest 0.01 mm with digital calipers while specimens were viewed at low magnification under a stereomicroscope. Except as noted, anatomical endpoints are illustrated in figures 5 and 6.

Fig. 5

Anatomical endpoints of 21 craniodental measurements of Marmosa defined in the text.

Fig. 6

Anatomical endpoints of five additional craniodental measurements of Marmosa defined in the text.

Greatest Length of Skull (GLS): From the anteriormost point of the premaxillae to the posteriormost point of the braincase.

Condylobasal Length (CBL): From the occipital condyles to the anteriormost point of the premaxillae.

Rostral Length (RL): From the anteriormost point of the nasals to the ventralmost lacrimal foramen.

Nasal Length (NL): The greatest length of either the right or left nasal bone (whichever is longest).

Palatal Length (PL): From the anteriormost point of the premaxillae to the posterolateral corner of the postpalatine torus.

Length of Maxillary Tooth Row (MTR): From the anterior surface of the base of the upper canine to the posterior margin of M4.

Length of Upper Molar Series (UMS): Crown length of the upper molars, from the anterolabial margin of M1 to the posterior margin of M4.

Length of M4 (LM4): Length (anteroposterior or mesiodistal dimension) of the fourth upper molar crown across the paracone and metacone.

Width of M2 (WM2): Greatest width (transverse dimension) of the second upper molar, from the labial margin of the crown at or near the stylar A position to the lingual apex of the protocone.

Width of M4 (WM4): Greatest width (transverse dimension) of the fourth upper molar, from the labial margin of the crown at or near the stylar A position to the lingual apex of the protocone.

Height of Upper Canine (HC): Height (vertical dimension) of C1, from the exposed labial base to the tip of the tooth.

Palatal Breadth (PB): Measured across the labial margins of the fourth molar (M4) roots.

Postpalatal Breadth (PPB): Least breadth across the anterior processes of the left and right alisphenoids (not illustrated).

Breadth of Basicranium (BB): The least distance between the anteromedial margins of the right and left alisphenoid tympanic processes.

Breadth across Tympanic Bullae (BTB): The greatest distance across the lateral margins of the right and left alisphenoid tympanic processes.

Length of Tympanic Bulla (LTB): From the anterior curvature of the alisphenoid tympanic process to the posteriormost point of the petrosal pars cochlearis.

Tympanic Bulla Opening (TBO): The distance between the alisphenoid tympanic process and the rostral tympanic process of the petrosal, measured across the medial margin of ectotympanic.

Width of Ectotympanic (WET): Greatest width of the ectotympanic (not illustrated).

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

Breadth of Rostrum across Canines (BRC): Measured across the labial bases of the upper canines.

Breadth of Rostrum between Jugals (BRJ): Measured across the triple-point sutures of the jugal, lacrimal, and maxillary bones on each side.

Least Interorbital Breadth (LIB): Measured at the narrowest point across the frontals between the orbits (anterior to the postorbital processes, if any).

Postorbital Constriction (POC): Measured at the narrowest point across the frontals between the temporal fossae (posterior to the postorbital processes, if present).

Breadth of Braincase (BBC): Measured immediately above the zygomatic process of the squamosal on each side.

Zygomatic Breath (ZB): Greatest breadth across the zygomatic arches.

Length of Mandible (LM): Measured from the anteriormost point of the mandible (medial to the alveolus of i1) to the posteriormost point of the angular process.

Length of Lower Molar Series (LMS): Crown length of the lower molars, from the anterolingual margin of m1 to the posterolingual margin of m4.

Length of m4 (Lm4): Length (anteroposterior or mesiodistal dimension) of m4, from the paraconid to the hypoconulid.

Width of m2 (Wm2): Greatest width (transverse dimension) of m2, measured across the hypoconid and entoconid (not illustrated).

Age Criteria

The age classification employed in this report is a refinement of that proposed by Tribe (1990), which was based on the pattern of tooth eruption that he observed in Marmosops incanus and other small didelphids (including Marmosa). In these taxa, the deciduous third upper premolar (dP3) is not replaced until M4 has erupted; therefore, P3 is the last upper tooth to erupt. Animals belonging to Tribe's age class 5 have a completely erupted upper molar dentition (M1–4), but dP3 is either unreplaced or P3 is just starting to erupt. Animals belonging to Tribe's age class 6 have shed dP3, P3 is almost completely erupted, and M4 is slightly worn; specimens in Tribe's age class 7 have P3 fully erupted, and M4 is considerably worn.

In the course of this study, we observed substantial ontogenetic variation among conspecific specimens assignable to Tribe's age classes 6 and 7. In order to take this variation into account for the purpose of taxonomic comparisons, we found it useful to redefine age classes 6 and 7 and to recognize two additional age classes as follows.

Age class 6: Labial cingulum of P3 emergent but slightly dorsal to labial cingulum of P2 (indicating the penultimate stage of P3 eruption); M3 and M4 cristae unworn or with very narrow and discontinuous strips of exposed dentine.

Age class 7: Labial cingulum of P3 aligned with or ventral to labial cingulum of P2 (indicating the complete eruption of P3); dentine narrowly exposed on all or most M3 and M4 cristae.

Age class 8: Labial cingulum of P3 aligned with or ventral to labial cingulum of P2; dentine broadly exposed along preparacrista of M3 but narrowly and discontinuously exposed on at least some of the other cristae of M3 and M4.

Age class 9: Labial cingulum of P3 aligned with or ventral to labial cingulum of P2; dentine broadly and continuously exposed on most M3 and M4 cristae.

Hereafter, specimens belonging to age class 5 are called subadults, and specimens belonging to age classes 6 to 9 are called adults. We additionally distinguish “young adults” (specimens in age class 6) from “mature adults” (in age classes 7 to 9).

Head Pelage

The middorsal fur of the rostrum (between the dark circumocular masks) is conspicuously paler than the fur of the crown of the head (between the ears) in some species of Marmosa. Among the species treated herein, this pigmental contrast is usually sharpest in M. mexicana, M. robinsoni, M. simonsi, and M. rubra. It is weakest and least often developed in M. zeledoni, and it is variably present or absent in M. isthmica and M. xerophila.

A dark median stripe that extends from the rhinarium to between the eyes—bisecting the pale midrostral fur—is consistently present in M. rubra (fig. 7, right). Somewhat resembling the median stripe of dark rostral fur described by Voss and Jansa (2003: character 2) for Caluromys, this trait is usually absent or inconspicuous in other species of Marmosa (e.g., M. zeledoni; fig. 7, left).

Fig. 7

Dorsal views of skins illustrating taxonomic variation in rostral pelage markings. A dark middorsal rostral stripe is absent in Marmosa zeledoni (left, FMNH 128398) but present in M. rubra (right, AMNH 68137). Scale bar  =  1 cm.

The fur surrounding the eye is dark brown or blackish in all species of Marmosa, resulting in a masklike marking that contrasts abruptly with the paler (brownish, grayish, or whitish) coloration of the midrostrum, crown, and cheeks. In most examined specimens, this circumocular mask extends anteriorly among the bases of the mystacial vibrissae, where the dark fur gradually narrows and fades away, becoming less and less distinct toward the rhinarium. In some species, dark fur extends only a few millimeters behind the eye; in such forms (e.g., M. isthmica, M. robinsoni, M. rubra, M. xerophila), several millimeters of paler, brownish fur separate the mask from the base of the ear (fig. 8, bottom). In other species, however, dark fur extends much farther posteriorly, so that the mask reaches the base of the ear; among the species treated herein, this trait is best developed in M. mexicana (fig. 8, top), but it is also seen to a variable extent in M. zeledoni and M. simonsi.

Fig. 8

Lateral views of skins illustrating taxonomic variation in facial markings. The dark circumocular mask extends posteriorly to the base of the ear in Marmosa mexicana (top, MVZ 130251), but not in M. robinsoni (bottom, AMNH 130579). Scale bar  =  1 cm.

Body Pelage

Although species of Marmosa (Marmosa) vary subtly in fur texture, perceived differences in pelage density and softness are somewhat subjective and do not provide a satisfactory basis for identification. Regardless of texture, the coat consists of cover hairs and guard hairs. The former are more abundant, wavy, and range from 7 to 11 mm in length. On the dorsal surface of the body, coat hairs are tricolored, with a basal gray band that makes up some 60% to 90% of the total length of each hair; a short buffy, yellowish, or reddish subterminal band; and a tiny dark-brown terminal band. Guard hairs are sparser and straighter than coat hairs, and range from 9 to 14 mm in length. Dorsal guard hairs are bicolored, grayish basally and paler near the tip.

Dorsal fur color is taxonomically variable in the subgenus Marmosa, but species differences are difficult to characterize by objective standards. Whereas much of the older literature made frequent reference to Ridgway's (1912) color standards, Ridgway's esoteric nomenclature—“Sayal Brown,” “Hair Brown,” “Mars Brown,” “Pecan Brown,” “Mikado Brown,” “Natal Brown,” “Sanford's Brown,” and “Burnt Sienna” are examples of color names that Tate (1933) used to describe the dorsal pelage of Marmosa—is not helpful without a copy of this rare book in hand,1 so more easily comprehensible vernacular descriptors (e.g., “reddish brown”) are preferable in most circumstances. Taxonomic color differences are primarily useful in side-by-side comparisons, where they are sometimes decisive for identification.

Among the species treated in this report, M. simonsi has much grayer fur than the others, whereas most specimens of M. mexicana, M. zeledoni, and M. rubra are distinctly reddish. Marmosa xerophila could appropriately be described as pale grayish brown, M. robinsoni as yellowish brown, and M. isthmica as orangish brown. Typically, the dorsal fur is darker and more richly colored on the middle of the back than along the flanks, which tend to be slightly paler. Juvenile and subadult specimens usually have grayer and/or darker dorsal fur than adults. Young adults often have longer, softer, more brightly colored, and glossier coats than old adults, which often have dull and shabby-looking fur.

The ventral pelage of most species consists of both self-colored (unicolored) and gray-based (bicolored) fur. In most of the species treated in this report, self-colored fur—which may be orange, yellowish, or whitish—extends continuously along the ventral midline from the chin to the anus. Usually, however, the width of this pale self-colored median streak (when present) is narrowed by darker lateral zones of gray-based ventral hairs, the tips of which are either colored like the flanks or like the self-colored median streak. The lateral zones of gray-based ventral hairs may be restricted to the abdomen (as in M. isthmica and M. xerophila), or they may extend from the throat or chest to the inguinal region, including the ventral surfaces of the fore- and hind limbs (as in M. mexicana, M. zeledoni, M. robinsoni, and M. rubra). Indeed, the median streak of self-colored fur is sometimes discontinuous in M. rubra because it is interrupted by gray-based abdominal fur. By contrast, the entire ventral surface (except for the chin and throat) is covered with gray-based hairs in M. simonsi.

In reproductively active adult females, the fur of the inguinal-abdominal region becomes remarkably modified. Ordinary ventral fur in such animals is replaced by a distinctively short, thin, sparse, brownish mammary pelage. Although the color and extent of the mammary pelage varies among conspecific specimens, some taxonomic differences are also apparent (table 3).

TABLE 3

Reproductive Features of Female Marmosa^a

Ears

The external ear or pinna is a prominent, membranous organ in Marmosa. Among the species treated herein, the ears are largest in M. isthmica, M. robinsoni, M. simonsi, and M. xerophila (averaging 25–26 mm in adult males; 24–25 mm in adult females; tables 4, 5) and smallest in M. mexicana, M. zeledoni, and M. rubra (22–23 mm in males, 20–22 mm in females). The internal and external surfaces of the ears are macroscopically naked, but a sparse covering of short hairs is visible under low magnification. The exposed auricular skin ranges in color from pale brown to dark brown, and it is usually more heavily pigmented distally than basally.

TABLE 4

Summary Statistics^a for External Measurements (mm) and Weights (g) of Adult Male Specimens of Seven Species of Marmosa

TABLE 5

Summary Statistics^a for External Measurements (mm) and Weights (g) of Adult Female Specimens of Seven Species of Marmosa

Gular Gland

In most species of Marmosa there is a distinct median patch of glandular (naked or sparsely haired) skin between the chest and the throat (Tate, 1933; Barnes, 1977). Although this gland is usually best developed in adult males, it is also visible in most subadult and female specimens. Among the species treated in this report, only M. rubra consistently lacks any external evidence of glandular activity in the gular region.

Carpal Tubercles

Large tubercles, supported internally by carpal ossifications, project from the lateral and medial surfaces of the wrist in mature adult male specimens of all the Marmosa species treated in this report (other species, such as M. murina, lack carpal tubercles). The medial (or “radial”) tubercle, which is supported internally by the prepollex (Lunde and Schutt, 1999), exhibits noteworthy taxonomic variation that can be recorded from suitably preserved dried skins. In M. mexicana (fig. 9A), the medial carpal tubercle is long, extending to the base of the pollex (manual digit I), and uniform in thickness (without distinguishable segments). By contrast, the long medial carpal tubercles of M. zeledoni (fig. 9B) and M. isthmica are divided by a shallow sulcus into similar but consistently recognizable proximal and distal parts. In the remaining species treated here—M. robinsoni, M. simonsi (fig. 9C), M. xerophila, and M. rubra—medial carpal tubercles can be long or short (not reaching the base of pollex), but they are more conspicuously segmented into globular proximal and comma-shaped distal parts.

Fig. 9

Semidiagrammatic plantar views of the right manus and wrist, illustrating taxonomic differences in carpal tubercle morphology. The medial carpal tubercle (mct) is long (extending to the base of the pollex) and undifferentiated in Marmosa mexicana (A), but it is separated by a shallow sulcus into distinct proximal and distal segments in M. zeledoni (B). In M. simonsi (C), the medial carpal tubercle is short (not extending to the base of the pollex) and segmented. Scale bars  =  5 mm.

Mammae

Because species of Marmosa are pouchless, the mammae of lactating females are exposed. Confined to the abdominal/inguinal region, these form a more or less circular abdominal/inguinal pattern that encloses an unpaired median teat (see Tate, 1933 [fig. 3] for an illustration of the mammae in M. mexicana). By convention, didelphid mammary complements are summarized by formulae representing the right-side (R), median (M), left-side (L), and total (T) teat counts in the format R–M–L  =  T. Mammary formulae vary taxonomically, but they also exhibit variation among conspecific individuals (table 3). In M. mexicana, from which we obtained mammary counts of four individuals, the number of teats ranges from 11 to 15, a range of variation that overlaps the teat numbers observed from most of the other species treated herein. Only M. rubra (with 7 to 9 mammae) seems like an outlier, but no definite conclusions are possible with such small sample sizes. Although we were not able to find any stuffed or fluid specimen of M. isthmica with countable mammae, a skin label recording 12 nursing young suggests a mammary count of at least 6–1–6  =  13 for this species. No information is available about mammary formulae or litter sizes for M. simonsi.

Tail

A slender, muscular organ in Marmosa, the tail is always longer than the combined length of the head and body, but the ratio of tail length to head-and-body length exhibits modest taxonomic variation among the species treated in this report (tables 4, 5), tending to be shortest in M. simonsi, M. robinsoni, and M. xerophila than in the other species treated herein (especially M. mexicana and M. rubra). The naked skin of the tail, brownish or grayish in life, is often slightly paler ventrally than dorsally, but the contrast is never abrupt. However, the distal one-third to one-half of the tail is unpigmented (whitish) in M. simonsi, a visually conspicuous and diagnostically useful marking.

The basal part of the tail, corresponding to less than 10% of the length of this organ (≤15 mm on stuffed skins), is covered by soft fur resembling the dorsal body pelage in texture and coloration. The rest of the tail is covered by epidermal scales that exhibit taxonomic variation in shape and arrangement. The caudal scales are distinctly rhomboidal (diamond-shaped) and arranged in consistently spiral series in Marmosa rubra, but the remaining species treated herein have caudal scales that vary in shape and exhibit no clear pattern of arrangement: both spiral and annular patterns can be found on the same individual, the former pattern usually on the dorsal surface of the tail and the latter on sides and ventral surface.

Of the three hairs that typically emerge from the posterior margin of each caudal scale, the central hair is usually slightly longer (by some 20%–50%) and thicker than the lateral hairs (fig. 10). Among the species treated in this report, these hairs are shortest in Marmosa rubra (in which the central hair of each caudal-scale triplet is only about as long as one caudal scale) and longest in M. robinsoni, M. simonsi, and M. xerophila (in which the central hair of each triplet is slightly longer than two caudal scales). The prehensile ventral surface of the distal part of the tail is scale- and hairless; it is marked by transverse pleats, a shallow median groove, and an apical pad bearing dermatoglyphs.

Fig. 10

Semidiagrammatic views of caudal scales and scale-hairs in Marmosa. The central hair in each caudal-scale triplet is longer than two scales in M. robinsoni (A), about two scales long in M. zeledoni (B), and less than one scale long in M. murina (C).

Rostral Process of the Premaxillae

In many small didelphids, including most species of Marmosa, the premaxillae project anteriorly to form a more or less acutely pointed shelf-like process that extends the bony rostrum well beyond I1 and contains a distinct suture between the left and right bones. Among the species treated in this report, a rostral process of the premaxillae is absent only in Marmosa xerophila, which lacks any distinct bony projection anterior to the alveolar rim of I1. By contrast, most of the remaining species (M. isthmica, M. mexicana, M. rubra, and M. zeledoni) have long rostral processes that are typically about as long as I1 is tall (fig. 11A). An intermediate condition is seen in both M. robinsoni and M. simonsi, which have short rostral processes that are about half as long as I1 is tall (fig. 11B).

Fig. 11

Lateral views of rostrum illustrating taxonomic variation in the development of a rostral process of the premaxillae (arrows). A, Marmosa mexicana (ROM 96090), with a long rostral process (about as long as I1 is tall). B, Marmosa robinsoni (AMNH 130584), with a short rostral process (about half as long as I1 is tall).

Orbitosphenoid Suture

The orbitosphenoid is a small bone that is wedged among the palatine, frontal, and alisphenoid bones in the posteromedial wall of the orbit; behind it is a large opening, the sphenorbital fissure. The length of the orbitosphenoid-alisphenoid suture relative to the height of the sphenorbital fissure exhibits noteworthy taxonomic variation in Marmosa. Whereas the suture is about twice as long as the fissure is high in M. mexicana, M. zeledoni, M. isthmica, and M. simonsi, the suture is visibly shorter in M. robinsoni and M. xerophila; in M. rubra, the suture is only about as long as the fissure is high.

Interorbital Region and Dorsal Braincase

The morphology of the dorsal surface of the frontal bones between the orbital fossae is taxonomically variable among small didephids. All species of Marmosa have well-developed supraorbital ridges (or “beads”), from which laterally projecting postorbital processes are sometimes developed. Behind the postorbital processes (when present), the supraorbital ridges are continuous with temporal ridges that mark the dorsalmost extension of the jaw adductor musculature onto the braincase. All of these bony structures are variously developed in relation to sex, age, and species. As a rule, they are better developed in males than in females, and they are better developed in mature adults than in juveniles, subadults, and young adults. Therefore, among large series of conspecific specimens, supraorbital ridges, postorbital processes (if any), and temporal ridges are almost always best developed in old adult males.

In mature adult specimens of Marmosa zeledoni (fig. 12B) and M. rubra (fig. 12G), the supraorbital ridges are typically slender, subparallel, and do not project laterally over the orbital fossae; postorbital processes in these taxa are usually absent or indistinct. By contrast, in mature adult specimens of M. robinsoni (fig. 12D) and M. xerophila (fig. 12E), the supraorbital ridges are thicker, more posteriorly divergent, and project slightly over the rear part of the orbital fossae; postorbital processes in these taxa are sometimes developed, but they are usually indistinct. In mature adult specimens of M. mexicana (fig. 12A) and M. isthmica (fig. 12C), the supraorbital ridges are usually thick, posteriorly divergent, and project laterally over the orbital fossae; postorbital processes are typically indistinct or small in M. mexicana, but they are usually present in large specimens of M. isthmica. In mature adult specimens of M. simonsi, the supraorbital ridges are thick, strongly divergent and project laterally as they widen to form well-developed, acutely pointed postorbital processes (fig. 12F).

Fig. 12

Dorsal cranial views illustrating taxonomic differences in the morphology of supraorbital ridges, postorbital processes, and temporal ridges in mature adult male specimens. A, M. mexicana (ROM 99608); B, Marmosa zeledoni (AMNH 29542); C, M. isthmica (USNM 456815); D, M. robinsoni (AMNH 130583); E, M. xerophila (AMNH 276582); F, M. simonsi (AMNH 66883); G, M. rubra (MVZ 153280).

The left and right temporal ridges are widely separated on the dorsal surface of the braincase in most species of Marmosa. These muscle scars remain widely separated and subparallel throughout adult life in M. rubra and M. zeledoni, but they tend to converge posteriorly in large adult specimens of the other species treated in this report. Uniquely, some old adult specimens of Marmosa robinsoni and M. xerophila develop a weak sagittal crest where the left and right temporal ridges are joined along the dorsal midline of the parietal and interparietal bones (fig. 12E).

Palatal Perforations

The bony palate in all species of Marmosa is perforated by paired incisive foramina (anteromedial to C1), maxillopalatine fenestrae (medial to M1–3), and posterolateral palatal foramina (posteromedial to M4; fig. 13). Although the posterolateral palatal foramina tend to be larger in M. robinsoni, M. simonsi, and M. xerophila than in the other species treated in this report, taxonomic differences are more readily apparent in other palatal features. Among these, the most striking interspecific variation consists in the presence or absence of palatine fenestrae.

Palatine fenestrae, which consist of one or more distinct openings in each palatine bone behind the maxillopalatine fenestrae, are consistently absent in several species of Marmosa, including M. zeledoni (fig. 13B), M. isthmica, and M. rubra. In other congeneric species, palatine fenestrae vary in number, size, and shape. In M. mexicana (fig. 13A), palatine fenestrae sometimes consist of several irregular openings on each side, and in a few individuals these can be quite small. By contrast, In M. robinsoni (fig. 13C), M. simonsi, and M. xerophila, palatine fenestrae usually consist of a single large, rounded opening on each side.

Fig. 13

Palatal morphology of three species of Marmosa, illustrating diagnostic traits mentioned in this report. All species of Marmosa have incisive foramina (if), maxillopalatine fenestrae (mp), and posterolateral palatal foramina (plpf), but palatine fenestrae (p) are present in some species and absent in others. Additionally, posterolateral palatine foramina exhibit noteworthy taxonomic variation in size. In M. mexicana (A, ROM 96090) palatine fenestrae are present and posterolateral palatal foramina are small. In M. zeledoni (B, AMNH 29542) palatine fenestrae are absent and posterolateral palatal foramina are small. In M. robinsoni (C, AMNH 130584) palatine fenestrae are present and posterolateral palatal foramina are large.

Ear Region

As in other didelphids, the anterior part of the middle ear in Marmosa is enclosed by the tympanic process (or “wing”) of the alisphenoid, which exhibits subtle but useful taxonomic shape variation. In M. mexicana (fig. 14A), M. zeledoni, M. isthmica, and M. rubra, the tympanic process of the alisphenoid tends to be small and laterally compressed with a slightly to moderately pointed or bluntly keeled ventral surface. By contrast, the alisphenoid tympanic process is larger and more or less globular (smoothly rounded) in M. robinsoni (fig. 14B), M. simonsi, and M. xerophila.

Fig. 14

Ventral view of the left ear region of Marmosa mexicana (A, AMNH 243700) and M. robinsoni (B, AMNH 266954) illustrating taxonomic differences in the morphology of the alisphenoid tympanic process (atp) and the ectotympanic (ect). The bold dashed line marks the course of the sulcus tympanicus, along which the tympanic membrane is attached on inner circumference of the ectotympanic. A small medial process of the ectotympanic (mp) contributes to the floor of the middle ear cavity in M. robinsoni.

The ectotympanic exhibits correlated taxonomic variation. In species with small, laterally compressed alisphenoid bullae, the sulcus tympanicus (to which the tympanic membrane attaches) closely follows the inner border of the ectotympanic, which does not contribute significantly to the floor of the middle ear cavity. By contrast, in species with larger, more globular alisphenoid bullae, the ectotympanic is wider and contributes a small medial process to the floor of the middle ear.

Dentition

All of the species of Marmosa treated in this report have the normal didelphid dental complement of 50 teeth (I5/4, C1/1, P3/3, M4/4), and they are undifferentiated with respect to both incisor and canine morphology: I1 is hypsodont, the crown of I2 is distinctly wider than its root, and C1 lacks any accessory cusps. By contrast, some congeneric South American forms differ by having a nonhypsodont I1 (M. andersoni), by usually lacking I1 as adults (M. tyleriana), by having a distinctively narrow-crowned I2 (M. quichua), or by having a small posterior accessory cusp on C1 (M. lepida). Distinguishing features of the upper premolars and of the lower dentition also seem wanting. Instead, the only qualitative dental traits that appear to be useful for identifying the species treated herein are those of the upper molar stylar cusps, for which minimally worn teeth must be examined.

As defined by positional criteria (after Clemens, 1966; see above), all five stylar cusps (A, B, C, D, and E) are usually present on M1 and M2 of Marmosa, but styD and styE are sometimes indistinctly separated (see below), and styC is often absent or indistinct on M3. Of the two cusps that sometimes occupy the C position, the one posterior to the ectoflexus is almost always larger. StyB, always the largest stylar cusp, is the labial terminus of the preparacrista in all species except M. rubra, in which the preparacrista usually terminates at styA.

Among the species treated in this report, we found noteworthy variation in the presence or absence of a distinct notch between stylar cusps D and E. This notch is usually present on unworn teeth, clearly separating styD from styE, in Marmosa robinsoni, M. rubra, M. simonsi, and M. xerophila (fig. 15A). However, no notch is present (with the result that styD and styE are indistinctly separated) in M. isthmica, M. mexicana (fig. 15B), and M. zeledoni.

Fig. 15

Labial views of unworn second upper molars (M2) in Marmosa xerophila (A, USNM 443815) and M. mexicana (B, MVZ 159445). A distinct notch separates stylar cusp D (styD) from stylar cusp E (styE) in M. xerophila, but stylar cusps D and E are not separated by a notch in M. mexicana. Other abbreviations: met, metacone; styA, stylar cusp A; styB, stylar cusp B; styC, stylar cusp C.

Type Material

The holotype (by original designation, USNM 71526) is a young adult (age class 6) male specimen consisting of a skin and skull. Although the skin is in good condition, the skull lacks most of the upper incisors, and the posterior palatal region is slightly damaged.

Type Locality

The holotype was collected by E.W. Nelson and E.A. Goldman on 28 February 1895 at “Juquila” ( =  Santa Catarina Juquila, gazetteer entry 11; appendix), a town on the Pacific slopes of the Sierra Madre del Sur in southwestern Oaxaca, Mexico. According to Goldman (1951: 215), Juquila is at 5000 ft (1524 m) “near the frost line and just above the limits of coffee culture. The climate is cool and damp. Much fog and mist prevail, as shown by the moss on the trees on the north slopes of the adjacent hills and by the abundance of vegetation belonging to the temperate mountain flora.”

Morphological Diagnosis

Midrostral fur pale, usually contrasting sharply with darker fur of crown; dark median rostral stripe absent or inconspicuous; dark facial mask usually extending posteriorly to contact base of ear; dorsal body pelage reddish brown; dorsal cover hairs 7–11 mm in length, guard hairs 9–14 mm; gray-based ventral pelage (covering the sides of chest, abdomen, and upper inguinal region, but sometimes also extending to the sides of neck and inner parts of arms and legs) with yellowish or orangish hair tips; self-colored ventral pelage (extending as a continuous median stripe of variable width from chin to anus) yellowish or orangish. Exposed skin of tail dark brown (indistinctly paler ventrally in some individuals), without any consistent pattern of scale arrangement (both spiral and annular patterns coexisting); 14–16 scales/cm on dorsal surface at caudal midlength; caudal scale-hairs reddish brown or light brown, detectable without magnification; central hair of each caudal-scale triplet as long as two scales (rarely 1.5 scales long). Gular gland present. Mammae 5–1–5  =  11 to 7–1–7  =  15. Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle in mature adult males long (reaching the base of the pollex) and unsegmented, without externally distinguishable proximal and distal segments.

Length of rostral process of premaxillae similar to I1 height. Orbitosphenoid-alisphenoid suture approximately twice as long as height of sphenorbital fissure in lateral view. Supraorbital ridges parallel or slightly divergent posteriorly, slender and dorsally reflected in some specimens but thicker and produced laterally in others; postorbital processes (when present) usually small but occasionally large (e.g., ROM 96090); temporal ridges usually converging posteriorly. Palatine fenestrae present, but often consisting of many irregular holes on each side (rather than a single large, rounded opening); posterolateral palatal foramina small, usually similar in length to M4 (measured across paracone-metacone). Tympanic wing of alisphenoid laterally compressed, with ventral surface slightly pointed or bluntly keeled; medial process of ectotympanic indistinct or absent. I1 hypsodont, normally present in adults; crown of I2 sharply defined in relation to root; C1 without accessory cusps; preparacrista connected to stylar cusp B on M1–M3; stylar cusps D and E confluent, not separated by a distinct notch on M2.

Comparisons

Marmosa mexicana merits close comparisons with three other species of Marmosa (Marmosa) that have overlapping geographic ranges in Central America, namely M. zeledoni, M. isthmica, and M. robinsoni. Specimens of these four taxa are often misidentified in museum collections, but each can be distinguished by a combination of morphometric and qualitative traits. For completeness, we also provide pairwise comparisons with allopatric congeners (M. xerophila, M. simonsi, M. rubra) that, although unlikely to be confused with M. mexicana geographically, may prove to be closely related.

Adult specimens of Marmosa mexicana average smaller than adult M. zeledoni in most external and craniodental measurements (tables 4–TABLE 5TABLE 67), but there is broad overlap in all measurements, and none provides a satisfactory basis for identification. In qualitative features, M. mexicana is most readily distinguished from M. zeledoni by its brighter dorsal coloration (the dorsal pelage of M. zeledoni appears duller and darker in side-by-side comparisons); by the paler coloration of its midrostral fur, which contrasts sharply with the darker coloration of its crown fur (midrostral and crown fur are nearly the same color, seldom contrasting sharply, in M. zeledoni); by the posterior extension of its dark facial mask, which usually contacts the base of the ear (the mask of M. zeledoni seldom extends posteriorly to the ear base); by the more frequent presence of distinct postorbital processes (postorbital processes are usually absent or indistinct in M. zeledoni); and by the presence of palatine fenestrae (absent in M. zeledoni). Other differences include the coloration of the tips of the gray-based ventral hairs (yellowish or orangish in M. mexicana versus reddish or brownish in M. zeledoni), the morphology of the medial carpal tubercles of mature adult males (long and unsegmented in M. mexicana versus long but divided by a shallow sulcus into proximal and distal parts in M. zeledoni), caudal hair length (the central hair of each caudal-scale triplet is about two scales long in M. mexicana versus about one-and-a-half scales long in M. zeledoni), the size of the posterolateral palatal foramina (usually about as long as M4 measured across the paracone-metacone in M. mexicana versus usually shorter than M4 in M. zeledoni). In dorsal view, skulls of M. mexicana also have more laterally projecting and more posteriorly divergent supraorbital ridges, and temporal ridges that converge more strongly posteriorly, than skulls of like-aged M. zeledoni.

TABLE 6

Summary Statistics^a for Craniodental Measurements (mm) of Adult Male Specimens of Seven Species of Marmosa

TABLE 7

Summary Statistics^a for Craniodental Measurements (mm) of Adult Female Specimens of Seven Species of Marmosa

TABLE 8

Morphological and Geographic Comparisons among Seven Species of Marmosa^a

Adult specimens of Marmosa mexicana are distinctly smaller than adult M. isthmica in all measured external and craniodental dimensions (tables 4–TABLE 5TABLE 67). Although no single measurement is sufficient for diagnosing these species, some of the observed overlap in external dimensions may be methodological artifacts; mean values for hind feet and ears are large enough to suggest that the size of these appendages might often be useful for field identification. It is also noteworthy that morphometric overlap is minimal in several ontogenetically invariant molar dimensions (e.g., UMS; tables 6, 7). Among the qualitative external characters we studied, M. mexicana is most easily distinguished from M. isthmica by the posterior extension of its dark facial mask, which usually contacts the base of the ear (the dark facial mask of M. isthmica does not contact the ear base); by its more reddish- (versus orangish-) brown dorsal fur; by the more extensive distribution of gray-based ventral fur, which covers the sides of its chest, abdomen, and upper inguinal region, sometimes also extending to the sides of the neck and inner parts of the arms and legs (gray-based ventral fur is restricted to the sides of the abdomen in M. isthmica); and by the long, unsegmented medial carpal tubercles of mature adult males (the medial carpal tubercles of like-aged M. isthmica are divided by a shallow sulcus into subequal proximal and distal parts). In qualitative cranial characters, M. mexicana is most easily distinguished from M. isthmica by the presence of palatine fenestrae (absent in M. isthmica) and by the usual absence or small size of postorbital processes (postorbital processes are usually present and much larger in fully mature examples of M. isthmica).

Adult specimens of Marmosa mexicana are similar in size to adult M. robinsoni, although they average somewhat smaller in most dimensions (tables 4–TABLE 5TABLE 67). On average, M. mexicana has a relatively longer tail than M. robinsoni, a difference that is noteworthy despite broad overlap in computed ratios (possibly another methodological artifact). In qualitative external characters, M. mexicana is most easily distinguished from M. robinsoni by the posterior extension of its dark facial mask, which usually contacts the base of the ear (the dark facial mask of M. robinsoni does not contact the ear base); by its more reddish- (versus yellowish-) brown dorsal fur; by its more conspicuous and more extensively distributed gray-based ventral fur, which covers the sides of its chest, abdomen, and upper inguinal region, sometimes also extending to the sides of the neck and inner parts of the arms and legs (gray-based ventral fur is often less conspicuous and is usually restricted to the sides of the abdomen in M. robinsoni); and by the long, unsegmented medial carpal tubercles of mature adult males (the medial carpal tubercles of like-aged M. robinsoni are usually divided by a shallow sulcus into a globular proximal segment and a comma-shaped distal segment). In qualitative cranial traits, M. mexicana is most easily distinguished from M. robinsoni by its longer rostral process of the premaxillae, smaller and more irregularly shaped palatine fenestrae, smaller posterolateral palatal foramina, less inflated auditory bullae, and by the absence of a notch between styD and styE (a distinct notch is usually present in M. robinsoni). In addition, a low sagittal crest that is sometimes present on the skulls of old adult specimens of M. robinsoni is never seen in M. mexicana.

Marmosa mexicana differs from M. xerophila in all of the same traits that distinguish M. mexicana from M. robinsoni, except that the rostral process of the premaxillae (long in M. mexicana, short in M. robinsoni) is entirely absent in M. xerophila.

Marmosa mexicana is longer-tailed and smaller-eared, on average, than M. simonsi, but the two species are otherwise similar in univariate comparisons of measured external and craniodental dimensions (tables 4–TABLE 5TABLE 67). In qualitative external traits, these taxa are most easily distinguished by their dorsal fur color (reddish brown in M. mexicana, distinctly grayish in M. simonsi), ventral fur color (a continuous median streak of self-colored fur is present in M. mexicana but absent in M. simonsi), and tail coloration (all-dark in M. mexicana, one-third to one-half white in M. simonsi). In qualitative cranial traits, they are most easily distinguished by the rostral process of the premaxillae (long in M. mexicana, short in M. simonsi), postorbital processes (usually indistinct or small in M. mexicana, consistently well developed in M. simonsi), and by the notch between stD and stE on unworn M2 (absent in M. mexicana, present in M. simonsi).

Marmosa mexicana is substantially smaller, on average, than M. rubra in most measured dimensions (tables 4–TABLE 5TABLE 67), but only measurements of the upper molar series are diagnostic in same-sex univariate comparisons (UMS, tables 6, 7). In qualitative external traits, these species are most easily distinguished by rostral markings (a dark median stripe is absent in M. mexicana but present in M. rubra), by the posterior extension of the dark facial mask (reaching the base of the ear in M. mexicana but not in M. rubra), by the presence or absence of a gular gland (present in M. mexicana, absent in M. rubra), by the arrangement of caudal scales (indeterminate in M. mexicana, distinctly spiral in M. rubra), and by the length of caudal-scale hairs (the central hair of each triplet is about two scales long in M. mexicana, but only about one scale long in M. rubra). In qualitative craniodental traits, these species are most readily distinguished by their palatal morphology (palatine fenestrae are present in M. mexicana but absent in M. rubra), by the stylar terminus of the preparacrista on M1–M3 (at styB in M. mexicana, at styA in M. rubra), and by the presence or absence of a distinct notch between styD and styE on M2 (absent in M. mexicana, present in M. rubra).

Geographic Distribution and Sympatry

Marmosa mexicana occurs in tropical and subtropical forests and shrubby habitats below 1600 m elevation from the Mexican states of Tamaulipas and San Luis Potosí southward to the eastern Panamanian province of Darién (fig. 17). It is also known from two offshore Caribbean islands, Isla de Roatán (Honduras; locality 53) and Isla del Maíz Grande (Nicaragua; locality 57).

Fig. 17

Geographic range of Marmosa mexicana based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

In Mexico, this species occurs in both dry and moist forests that correspond to the Tropical Zone of Shelford (1926). However, Marmosa mexicana is also known from several localities (gazetteer entries 6, 10, and 11; appendix) where the Central American and Sierra Madre del Sur pine-oak forests are juxtaposed with Central American Pacific dry forests (as defined by Olson et al., 2001). In Central America, M. mexicana occurs in a wide variety of habitats that include dry and moist tropical forests, pine-oak forests, mangroves, and scrublands.

The range of M. mexicana overlaps that of M. zeledoni from northwestern Nicaragua to eastern Panama. The two species have been collected sympatrically at Río Coco (locality 60), Nicaragua; Finca Helechales (locality 79) and Monteverde (locality 80), Costa Rica; and Bugaba (locality 105), Panama. The range of M. mexicana also overlaps the known ranges of M. isthmica and M. robinsoni in Panama, although M. mexicana has apparently not been collected sympatrically with either of those species.

Geographic Variation

Available population samples of Marmosa mexicana exhibit variation in pelage color that is apparently related to habitat. Whereas specimens from pine-oak forests and moist forests tended to have rich reddish-brown dorsal fur and orangish ventral fur, specimens from dry forests and scrublands tend to be paler (with faded-reddish dorsal fur and yellowish buff or buff ventral fur). The palest specimens we examined were collected at Puerto El Triunfo (El Salvador; locality 50) in the Northern Dry Pacific Coast Mangroves ecoregion of Olson et al. (2001). By contrast, we found no morphometric variation that could be unequivocally related to the geographic origin or habitat of the specimens analyzed.

Taxonomic History

The first literature reference to a specimen identifiable as Marmosa mexicana is in Waterhouse (1846), who identified a British Museum specimen from Mexico as Didelphys murina. Other 19th-century authors also used murina for specimens of mexicana (e.g., Alston, 1880; Thomas, 1882b, 1888, 1895; Allen 1893) and, indeed, Merriam (1897) originally described mexicana as a subspecies of M. murina. Although Merriam provided only a brief description of mexicana, based primarily on the pelage of two specimens from the Mexican states of Oaxaca and Chiapas, most subsequent authors (except Allen, 1908, 1910; and Gaumer, 1917) seem to have agreed that mexicana was a distinct taxon. Bangs (1902) provided additional details about the pelage color and cranial morphology of mexicana, which he considered to represent a valid species. However, his conclusion was not immediately accepted by other mammalogists (e.g., Elliot, 1904, 1905).

Osgood (1913) described Marmosa mayensis based on a single specimen that he considered to represent the Yucatecan representative of M. mexicana. Later, Goldman (1917) described Marmosa mexicana savannarum based on five specimens from the Panamanian province of Chiriqui. In the same report, Goldman stated that “the continental Middle American forms now known [except for Marmosa canescens ( =  Tlacuatzin canescens)] may confidently be referred to a single species,” thereby establishing a broad concept of M. mexicana that included not only mayensis and savannarum, but also isthmica and zeledoni.

Goldman's (1917) conclusions were adopted by Cabrera (1919), who listed isthmica, mayensis, savannarum, and zeledoni as subspecies of Marmosa mexicana in his influential checklist of marsupials. Tate (1933), however, recognized only M. mexicana mayensis and M. m. zeledoni as valid subspecies, treating savannarum as synonym of zeledoni, and recognizing M. isthmica as a valid species. In Tate's classification, M. mexicana was referred to the monotypic Mexicana Section of the genus, whereas M. isthmica was referred to the Mitis Section. Tate's (1933) opinions about the validity of subspecific taxa in his Mexicana Section were accepted by Hershkovitz (1951), Hall (1981), and most subsequent authors.

Remarks

Osgood (1913) described mayensis as paler than mexicana, and Tate (1933) used this character to maintain these taxa as valid subspecies. However, our examination of the holotypes of mayensis (FMNH 19994) and mexicana (USNM 71526) revealed no conspicuous differences in pelage color that could serve as an unambiguous basis for taxonomic diagnosis. Although the holotype skull of mayensis is damaged (with only the rostral, palatal, and interorbital regions intact), the absence of any noteworthy qualitative or morphometric differences with the holotype skull of mexicana also support our opinion that mayensis and mexicana are conspecific.

Goldman (1917) described savannarum as smaller and paler than other Panamanian mouse opossums that he identified as “M. mexicana zeledoni” ( =  M. zeledoni) and “M. m. isthmica” ( =  M. isthmica). In fact, the differences in size and pelage color that he observed are marked, and they can be partially attributed to the specific status of the specimens he examined (M. zeledoni and M. isthmica are, indeed, larger and darker than M. mexicana). However, the holotype of savannarum (AMNH 18915) is a subadult with an incompletely erupted M4 (as noted by Tate, 1933), so its dimensions are not those of a fully mature individual. Our examination of the holotypes of savannarum and mexicana revealed no conspicuous differences other than the overall darker fur and broader gray-based lateral zones of ventral fur in the former specimen. Neither of these pelage differences exceeds the range of variation among other specimens that we refer to M. mexicana, however, and in the absence of qualitative osteological differences, we conclude that Goldman's and Merriam's taxa are conspecific.

Marmosa ruatanica was described by Goldman (1911) based on a single specimen (USNM 7785/37700 [skin/skull cataloged separately]) from Isla de Roatán, Honduras (locality 53). Tate (1933) also recognized ruatanica as a distinct species that included isthmica and mimetra as valid subspecies; these three nominal taxa were associated by virtue of their large size, well-developed postorbital processes, proportionately small alisphenoid tympanic processes, and minute or absent palatine fenestrae. By contrast, Hershkovitz (1951) considered ruatanica to be a subspecies of M. mitis, and ruatanica is currently regarded as a valid subspecies of M. robinsoni (e.g., by Hall, 1981; Gardner, 2005).

The large size of the holotype skull (fig. 18) is the principal characteristic that impelled several authors to compare ruatanica with the Trinidadian nominal taxon chapmani ( =  M. robinsoni) or to associate it with M. isthmica (including mimetra). However, the type of ruatanica closely resembles M. mexicana in several qualitative traits, including the posterior extension of its dark facial mask (which contacts the base of the ear), by its medial carpal tubercle (which is long and unsegmented), and by the modest development of its temporal ridges (which tend to be much more strongly developed in other species). In fact, craniodental measurements of the holotype, an old adult male, are not much larger than other like-aged specimens of M. mexicana, some of which exhibit comparably developed postorbital processes and have small irregular holes in place of better-defined palatine fenestrae. Although additional material from Roatan Island (from which only the type is currently known) would be welcome, we are not convinced either that ruatanica is a distinct taxon or that its relationships are with geographically distant forms of mouse opossums. Instead, we provisionally regard it as an insular form of the adjacent mainland species.

Fig. 18

Dorsal and ventral cranial views of the holotype of Marmosa ruatanica ( =  M. mexicana, USNM 37700). Scale bar  =  10 mm.

Specimens Examined (N  =  131)

Without locality data (FMNH 34898, FMNH 49927; ROM 99608). MEXICO—Campeche, 44 km S Constitución (ROM 95795), Xpujil (ROM 96090). Chiapas, Chicharras (USNM 77680), Huehuetan (USNM 77681, 77682), Rayón (MVZ 159445, 159446), San José (MVZ 113483), Volcán Tacaná (MVZ 155505, 155506). Oaxaca, Juchitán (AMNH 189209), Lachiguirí (AMNH 213754), San Gabriel Mixtepec (AMNH 189483–189485), Santa Catarina Juquila ( =  Juquila: USNM 71526 [holotype of Marmosa murina mexicana Merriam, 1897]), Tehuantepec (MVZ 149102–149107). San Luis Potosí, El Salto Falls (USNM 329396). Tamaulipas, Aserradero del Infernillo (AMNH 166059). Veracruz, Achotal de Moreno (FMNH 13805, 13806), Jalapa Enríquez (BMNH 97.9.9.82, 97.9.9.83; AMNH 12453–12456/10762–10765), Mirador (USNM 62217–62219, 583340, A22019), Paso Nuevo (AMNH 17135, 17136), San Andrés Tuxtla (BMNH 7.1.1.186), Tuxpan de Rodríguez Cano ( =  Tuxpan: USNM 10719). Yucatán, Chichén Itzá (AMNH 91192), Izamal (FMNH 19994 [holotype of Marmosa mayensis Osgood, 1913]). BELIZE—Cayo, Central Farm (BMNH 64.2028, 64.2029, 65.3870; FMNH 106525; USNM 360468, 360469), Georgeville (BMNH 65.3866–69, 65.3873–65.3879, 66.2329–66.2332), Roaring Creek (BMNH 63.626, 65.3871). Orange Walk, Yo Creek (BMNH 65.3872). Stann Creek, Bokowina Hill ( =  Bokowina: BMNH 84.369; FMNH 63887, 63888), Cockscomb Basin Wildlife Sanctuary (USNM 583001), Silk Grass (FMNH 63891). Toledo, Double Fall (BMNH 84.370), Union Camp (USNM 583214, 583215). GUATEMALA—no other locality data (AMNH 14248). Alta Verapaz, Cobán (BMNH 75.2.27.15; FMNH 42040–42042), La Primavera (AMNH 79250). Baja Verapaz, Chilascó (AMNH 14377), San Jerónimo (BMNH 65.5.18.67). Chiquimula, Esquipulas (USNM 564594). Esquintla, San Jose (USNM 275673). Guatemala, Lago de Amatitlán (USNM 275672). Izabal, Bobos (FMNH 41578, 41677), Escobas (FMNH 41577), La Esmeralda (USNM 564588), Río Frío (USNM 564589). Progreso, 5 km E San Cristóbal Acasaguastlán (ROM 99776), Puerta de Golpe (USNM 564591). San Marcos, Nuevo Progreso (USNM 564592, 564593). Zacapa, Cabañas (AMNH 265851), Río Hondo ( =  Finca El Chahuite: USNM 564590). EL SALVADOR—San Miguel, Laguna Olomega (MVZ 98165). Sonsonate, Chilata ( =  Hacienda Chilata: MVZ 98164,130251), Sonsonate (AMNH 243700). Usulatán, Puerto El Triunfo (MVZ 130252, 130253). HONDURAS—Department unknown (received in banana shipment: FMNH 43309). Cortés, Catacombas (AMNH 123290). Francisco Morazán, Monte Vásquez ( =  Cerro Vásquez: USNM 257078). Islas de la Bahía, Isla de Roatán ( =  Ruatán Island: USNM 7785/37700 [holotype of Marmosa ruatanica Goldman, 1911]). La Paz, Humuya (AMNH 126588–126590). NICARAGUA—Atlántico Sur, Isla del Maíz Grande ( =  Great Corn Island: FMNH 34072, 34073). Chinandega, Volcán San Cristóbal ( =  Volcán de Chinandega: AMNH 28314). Jinotega, Río Coco (AMNH 29270). Managua, Managua (USNM 337503). Matagalpa, Matagalpa (AMNH 29269). Nueva Segovia, Ocotal (AMNH 28506). Rivas, San Emilio (BMNH 97.4.7.12). Santa Barbara, San Jose de Santa Barbara (AMNH 123291). COSTA RICA—Puntarenas, Finca Helechales (USNM 547996, 548001, 548002), Monteverde (FMNH 126076). PANAMA—Chiriquí, Bugaba ( =  Bogava: BMNH 0.7.11.90; USNM 248344), Boquerón (AMNH 18914, 18915 [holotype of Marmosa mexicana savannarum Goldman, 1917]). Darién, Yaviza (MVZ 135833).

Other Specimens

In addition to material positively identifiable as Marmosa mexicana, we examined five specimens that we provisionally refer to this species. Three are very young specimens preserved in fluid (USNM 297454–297456; from Diriamba [locality 58], Nicaragua), and two are skins unaccompanied by skulls (BM 94.11.1.7, 94.11.1.8; from Volcán San Cristóbal [locality 59], Nicaragua). Although the latter specimens exhibit some mexicana–like external traits (e.g., a dark circumocular mask extending to the base of the ear; long and unsegmented medial carpal tubercle), they seem to have shorter tails and yellower fur than is usual in the species.

Fig. 19

Dorsal cranial views of mature adult male Marmosa zeledoni (left, USNM 315005) and Marmosa isthmica (right, USNM 575395). Scale bar  =  10 mm. Atypically, postorbital processes are present on this specimen of M. zeledoni; in most other examined conspecific specimens, postorbital processes are indistinct or absent.

Fig. 20

Ventral cranial views of mature adult male Marmosa zeledoni (left, USNM 315005) and Marmosa isthmica (right, USNM 575395). Scale bar  =  10 mm.

Type Material

The holotype (by original designation, USNM 12885) is a young adult (age class 6) male specimen prepared as a skin and skull; the skin is in good condition (except for a naked area on the left side), but the skull is partially damaged (missing the tips of both nasals, lacking most of the basicranium, and with a loose supraoccipital).

Type Locality

The holotype was collected by J. Cooper on 1 March 1878 at Navarro (gazetteer entry 71; appendix), “[a] small village about 5 miles [8 km] south of Cartago on the Caribbean slopes of the Cordillera Central” in Cartago province, Costa Rica (Goodwin, 1946: 456).

Morphological Diagnosis

Midrostral fur usually not contrasting sharply in color with fur of crown; dark median rostral stripe absent or inconspicuous; dark facial mask usually not extending posteriorly to contact base of ear; dorsal body pelage usually dull, dark, reddish brown; dorsal cover hairs 8–10 mm in length, guard hairs 10–12 mm; gray-based ventral pelage (covering the sides of chest, abdomen, and upper inguinal region, sometimes extending to the sides of neck, rarely to the inner parts of arms and legs) with reddish or brownish hair tips; self-colored ventral pelage (extending as a continuous median stripe of variable width from chin to anus) yellowish or orangish. Tail dark brown (indistinctly paler ventrally in some individuals), without any consistent pattern of scale arrangement (both spiral and annular patterns coexisting); 16–18 scales/cm on dorsal surface at caudal midlength; caudal scale hairs dark brown, usually detectable without magnification; central hair of each caudal-scale triplet as long as 1.5 scales (sometimes two scales) long. Gular gland present. Mammae 4–1–4  =  9 or 5–1–5  =  11. Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle long (reaching the base of the pollex), with subequal but recognizable proximal and distal segments divided by a shallow sulcus.

Length of rostral process of premaxillae similar to I1 height. Orbitosphenoid-alisphenoid suture approximately twice as long as height of sphenorbital fissure in lateral view. Supraorbital ridges subparallel, slender, and dorsally reflected; postorbital processes usually indistinct or absent; temporal ridges subparallel, never strongly convergent posteriorly. Palatine fenestrae absent (but tiny asymmetrical palatal perforations are sometimes present); posterolateral palatal foramina usually shorter than M4 (measured across paracone-metacone). Tympanic wing of alisphenoid laterally compressed, with ventral surface moderately pointed or bluntly keeled; medial process of ectotympanic usually indistinct or absent. I1 hypsodont, normally present in adults; crown of I2 sharply defined in relation to root; C1 without accessory cusps; preparacrista connected to stylar cusp B on M1–M3; stylar cusps D and E confluent, not separated by a distinct notch on M2.

Comparisons

Qualitative and morphometric comparisons between Marmosa zeledoni and M. mexicana were provided in the preceding account. In Central America, M. zeledoni overlaps geographically with two other congeners, M. isthmica and M. robinsoni, and it also co-occurs with the former species in western Colombia.

Marmosa zeledoni and M. isthmica overlap broadly in all measured dimensions (tables 4–TABLE 5TABLE 67), and they are also externally similar in qualitative characters. On average, however, M. isthmica is the larger species, a difference that is best appreciated when size comparisons are made between specimens of the same age and sex, and there are subtle but diagnostically useful differences in some skin characters. In side-by-side comparisons, the midrostral fur of zeledoni is usually darker and contrasts less abruptly in coloration with the fur of the crown than in isthmica (which usually has much paler midrostral than coronal fur), and there is sometimes an indistinct posterior extension of the dark circumocular mask to the base of the ear in zeledoni that is not seen in isthmica. Although the body pelage is similarly pigmented in both forms, the dorsal fur of zeledoni is often a darker and duller hue of reddish brown than the somewhat brighter, orangish brown of isthmica. Additionally, the exposed skin of the tail is usually dark grayish brown above and below in zeledoni, whereas the tail is a lighter shade of brown and is often indistinctly paler below in isthmica. In dorsal cranial view, the supraorbital ridges of zeledoni are subparallel, slender, do not project laterally over the orbital fossae, and seldom form distinct postorbital processes; behind the orbits, the temporal ridges maintain this subparallel configuration onto the braincase. By contrast, the supraorbital ridges of isthmica are thicker and diverge more strongly posteriorly as they project laterally over the orbital fossae; postorbital processes are usually well developed in mature adult male specimens, which have correspondingly better-defined postorbital constrictions, and whose temporal ridges often converge posteriorly on the braincase. In ventral cranial view, the posterolateral palatal foramina are usually smaller in zeledoni than they are in isthmica.

Marmosa zeledoni overlaps broadly in measured external dimensions with M. robinsoni, although it has visibly smaller ears. In side-by-side comparisons of skins, the midrostral fur color exhibits consistently less contrast with the color of the coronal fur in zeledoni than in robinsoni, and the dorsal body pelage of zeledoni is consistently redder than that of robinsoni (whose dorsal fur is usually some shade of yellowish or grayish brown). Whereas the medial carpal tubercles of mature adult male specimens of zeledoni are divided by a shallow sulcus into subequal and morphologically similar proximal and distal segments, the medial carpal tubercles of like-aged male robinsoni consist of a globular proximal segment and a comma-shaped distal segment. Under low magnification, the tail appears to be more sparsely haired in zeledoni than in robinsoni, a visual effect of the shorter caudal hairs in the former species. The rostral process of the premaxillae is long (about equal to the height of I1) in zeledoni, but this process is much shorter (about half as long as I1 is tall) in robinsoni. The supraorbital ridges (described above for zeledoni) are thicker and project laterally to a much greater extent in robinsoni, which sometimes develops distinct postorbital processes and has a more pronounced postorbital contriction (POC; tables 6, 7); the temporal ridges (which are subparallel in zeledoni) converge posteriorly in mature adult specimens of robinsoni, sometimes forming a low sagittal crest. The auditory bullae, which are small and laterally compressed in zeledoni, are much larger and smoothly globular in robinsoni. A distinct notch separates stylar cusps D and E on M2 in robinsoni, whereas this notch is absent in zeledoni.

Marmosa zeledoni differs from M. xerophila in the same characters that distinguish it from M. robinsoni, except that the rostral process of the premaxillae (long in zeledoni) is completely absent in xerophila.

Marmosa zeledoni is, on average, larger in most measured external dimensions than M. simonsi, but it has absolutely and relatively smaller ears (tables 4, 5). In external characters, the two species differ conspicuously in dorsal fur color, which is a rich reddish brown in M. zeledoni versus distinctly grayish in M. simonsi. Whereas a continuous streak of pale self-colored fur extends from the chin to the anus in M. zeledoni, almost all of the ventral pelage is gray based in M. simonsi (only the chin and throat are self-colored). Additionally, the tail is entirely dark in M. zeledoni, but the distal half of the tail is whitish in M. simonsi. Cranial differences between these dissimilar forms include the rostral process of the premaxillae (long in zeledoni, short in simonsi), postorbital processes (indistinct or absent in zeledoni, consistently well developed in simonsi), palatine fenestrae (absent in zeledoni, present in simonsi), and a distinct notch between stylar cusps D and E on M2 (absent in zeledoni, present in simonsi).

Marmosa zeledoni is externally similar in size and coloration to the western Amazonian species M. rubra, from which it principally differs by lacking a dark median rostral stripe (present in rubra), by possessing a gular gland (absent in rubra), and in caudal morphology (the caudal scales of rubra are arranged in unambiguously spiral series and the central hair of each triplet is only about one scale long). Marmosa zeledoni is smaller, on average, than M. rubra in most craniodental dimensions (tables 6, 7), but there is broad overlap in all measurements, none of which is sufficient for identification purposes. Instead, M. zeledoni differs from M. rubra by having a preparacrista that connects to stylar cusp B (the preparacrista connects to stylar cusp A in rubra), and by lacking a distinct notch between stylar cusps D and E (a distinct notch is present between these cusps in rubra).

Geographic Distribution and Sympatry

Based on positively identified specimens, Marmosa zeledoni occurs from northern Nicaragua (Río Coco, locality 60; fig. 21) southward through Costa Rica and Panama to southwestern Colombia (Candelilla, locality 176). Incomplete and/or immature material that we tentatively identify as M. zeledoni, however, includes a skin (AMNH 46682) from Gualea (locality 218) in the Pacific foothills of western Ecuador.

Fig. 21

Geographic range of Marmosa zeledoni based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

Marmosa zeledoni has been collected between 100 m and 2200 m above sea level, but it is apparently more abundant in montane or submontane habitats than it is near sea level: 72% of the collecting localities for which there is altitude information are above 700 m, and 61% are above 1100 m. These upland localities are mainly in the Cordillera Central and Cordillera de Talamanca of Costa Rica, the Chiriquí highlands of western Panama, and the Serranía del Darién of eastern Panama. Vegetation maps suggest that most specimens of Marmosa zeledoni are associated with lowland or montane rain forests.

The range of Marmosa zeledoni partially overlaps that of M. mexicana, with which it sometimes occurs sympatrically as described in the preceding account of that species. The range of M. zeledoni also overlaps the range of M. isthmica from western Panama to southwestern Colombia, but the two species have been collected sympatrically only at San José (locality 185) in western Colombia. The range of M. zeledoni additionally overlaps that of M. robinsoni in Panama, where the two species have been sympatrically collected at Finca Santa Clara (locality 109) and near Escopeta Camp (locality 110).

Geographic Variation

Examined specimens exhibit variation in pelage color that appears to be correlated with elevation. Thus, skins from montane habitats in Panama tend to have darker, brownish dorsal fur and darker and more broadly gray-based ventral fur (the latter extending to the inner sides of the arms and legs); in this material, the gray-based ventral fur is similar in color to the lateral body fur. By contrast, specimens from lowland habitats in Nicaragua, Costa Rica, and Colombia tend to have redder dorsal fur and paler and more narrowly distributed gray-based ventral fur (the latter not extending to the inner side of the arms and legs); in this material, the gray-based ventral fur is substantially paler than the lateral body fur.

Taxonomic History

The first published reference to material that we identify as Marmosa zeledoni was by Allen (1891), who misidentified specimens of this mouse opossum as Didelphys (Micoureus) murina. Conspecific Central American material was also misidentified (as M. murina or M. mexicana) by subsequent authors, until Goldman (1911) described M. zeledoni based on five specimens from Navarro (Costa Rica) and Escondido River (Nicaragua). Allen (1912) treated zeledoni as a subspecies of M. murina, but Goldman (1917) allied zeledoni with M. mexicana, an opinion followed by most subsequent authors, including Tate (1933). Recent authors (e.g., Honacki et. al., 1982; Eisenberg, 1989; Alonso-Mejía and Medellín, 1992; Emmons, 1997; Reid, 1997; Nowak, 1999; Brown, 2004; Gardner, 2005) have consistently failed to recognize zeledoni as a valid taxon.

Specimens Examined (N  =  77)

NICARAGUA—Without other locality data (USNM 361203, 361204). Atlántico Sur, Escondido River (USNM 36348, 50882). Jinotega, Río Coco (AMNH 29271), San Rafael del Norte (AMNH 28285). Matagalpa, El Tuma ( =  Río Tuma: AMNH 29542), Vijagua (AMNH 29541). Río San Juan, San Carlos ( =  La Esperanza: USNM 361194–361202). COSTA RICA—Cartago, Aquiares (AMNH 63999), Cerro Carpintera ( =  La Carpintera: AMNH 3652), Navarro (USNM 12885 [holotype of Marmosa zeledoni Goldman, 1911]). Heredia, 11 km S and 4.5 km W Puerto Viejo (FMNH 128390, 128391, 128393, 128397, 128398), 11 km S and 11.5 km E San Miguel Angeles (FMNH 128400). Limón, Jiménez (AMNH 9593/7930). Puntarenas, Cerro Cañas Gordas (AMNH 142488, 142489), Coto Brus ( =  Costa Brus: USNM 516606), Finca Helechales (USNM 547941, 547995, 547997–548000, 548003), Monteverde (FMNH 123992), San Jerónimo ( =  San Geronomo: FMNH 35180). San José, Escazú (BMNH 3.2.1.10). PANAMA—Bocas del Toro, Almirante (USNM 315001, 315003), Changuinola (USNM 315005–3150007), E of Cerro Pando (USNM 516605), Fish Camp (USNM 520691), Río Changena Camp (USNM 319372). Chiriquí, Bugaba ( =  Bogava: BMNH 0.7.11.87–0.7.11.89, 0.7.11.91, 0.7.11.92, 3.3.3.107), Casa Tilley (USNM 314188–314190), Cerro Fortuna ( =  Reserva Florestal Fortuna: AMNH 269997), El Volcán ( =  Volcán: AMNH 147759, 147760), Finca Santa Clara ( =  Osta Clara: USNM 396509, 516603), Near Escopeta Camp ( =  24 km NNE San Felix: USNM 541001, 541005, 541007, 541010, 541012, 541013, 541016, 541017). Darién, Cerro Malí (USNM 337954–337957), Cerro Tacarcuna (USNM 337958). COLOMBIA—Nariño, Candelilla (FMNH 89565). Valle del Cauca, San José (AMNH 31683).

Other Specimens

In addition to material positively identifiable as Marmosa zeledoni, we examined 42 specimens that we provisionally refer to this species. Most of these are juveniles or subadults that have not developed all of the diagnostic traits that distinguish this species from M. isthmica, but a few are skins unaccompanied by skulls (e.g., AMNH 46682), and others are fluids from which skulls have yet to be extracted.

NICARAGUA—Atlántico Sur, Escondido River (USNM 36349). COSTA RICA—Cartago, Navarro (USNM 12884). Heredia, 11 km S and 4.5 km W Puerto Viejo (FMNH 128389, 128395, 128396, 128399). Limón, Pandora (USNM 284465). Puntarenas, Boruca (AMNH 11794/10063), Monteverde (FMNH 123993, 124100). PANAMA—Bocas del Toro, Almirante (USNM 315002), Boca del Drago (USNM 315004), N of El Volcán (USNM 516948). Chiriquí, Near Escopeta Camp (USNM 541003, 541004, 541006, 541008, 541009, 541011, 541014, 541015, 541018, 541325–541330, 541332, 541333), Finca Santa Clara (USNM 539832). COLOMBIA—Cauca, San José (AMNH 31694). ECUADOR—Pichincha, Gualea (AMNH 46682).

Type Material

The holotype (by original designation, USNM 170969) is a mature adult (age class 7) male specimen preserved as a skin and skull, both of which are in good condition.

Type Locality

The holotype was collected by E.A. Goldman on 16 February 1911 on the Río Indio (gazetteer entry 103; appendix), a small tidal tributary of the lower Río Chagres, near the town of Gatun, Provincia Colón, Panama. According to Goldman (1912: 2), the type “was trapped in an old banana plantation only a few feet above sea level.” A photograph of the type locality was subsequently published by Goldman (1920: pl. 3, fig. 1).

Morphological Diagnosis

Midrostral fur usually pale and contrasting sharply with darker fur of crown; dark median rostral stripe absent or inconspicuous; dark facial mask not extending posteriorly to contact base of ear; dorsal body pelage pale to dark orangish brown; dorsal cover hairs 8–12 mm in length, dorsal guard hairs 10–15 mm in length; gray-based ventral pelage conspicuous but usually restricted to the sides of the abdomen, with yellowish or orangish hair tips; self-colored ventral pelage (extending as a continuous median streak from chin to anus) yellowish or orangish. Exposed skin of tail brown, indistinctly bicolored (paler ventrally) in most large specimens, without any clear pattern of scale arrangement (both spiral and annular patterns coexisting); 16–18 scales/cm on dorsal surface at caudal midlength; caudal scale hairs light brown, usually detectable without magnification; central hair of each caudal-scale triplet usually about 1.5 (sometimes two) scales long. Gular gland present. Mammae 6–1–6  =  13 (Enders, 1935: 408). Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle long (reaching the base of the pollex), with subequal but recognizable proximal and distal segments separated by a shallow sulcus.

Length of rostral process of premaxillae similar to I1 height. Orbitosphenoid-alisphenoid suture usually about twice as long as height of sphenorbital fissure in lateral view. Supraorbital ridges slightly to moderately divergent posteriorly, slender and only slightly produced laterally in younger specimens and females (which usually have indistinct postorbital processes) but becoming much thicker and dorsolaterally produced in mature adult males (which usually have distinct postorbital processes); temporal ridges conspicuous and tending to become strongly convergent posteriorly (behind the postorbital constriction) in mature adult males. Palatine fenestrae absent (but tiny asymmetrical perforations are occasionally present); posterolateral foramina usually longer than M4 (measured across paracone-metacone). Tympanic wing of alisphenoid laterally compressed, with ventral surface globular or slightly pointed; medial process of ectotympanic usually indistinct or absent. I1 hypsodont, normally present in adults; crown of I2 sharply defined in relation to root; preparacrista connected to stylar cusp B on M1-M3; stylar cusps D and E confluent, not separated by a distinct notch on M2.

Comparisons

Qualitative and morphometric comparisons among Marmosa isthmica, M. mexicana, and M. zeledoni, all of which have overlapping geographic ranges, are provided in the preceding accounts. The geographic range of M. isthmica also overlaps with that of M. robinsoni in Panama, and it overlaps with that of M. simonsi in western Ecuador. Morphological comparisons with these and other species are provided below.

Marmosa isthmica is larger (on average) and has a relatively longer tail than M. robinsoni, but the two species overlap broadly in all external dimensions (tables 4, 5) and cannot be identified by measurement data alone. In side-by-side comparisons of skins, the two species are most readily distinguished by the more saturated, orangish-brown color of the dorsal pelage in isthmica, by contrast with the usually paler and more yellowish-brown pigmentation of the dorsal fur in robinsoni. Also, the tail appears more sparsely haired in isthmica (in which the central hair of each caudal-scale triplet is only about 1.5 scales long) than in robinsoni (in which the central hair is > 2 scales long). Additionally, the medial carpal tubercle of mature adult male isthmica is long (reaching the base of the pollex) and divided by a shallow sulcus into subequal and morphologically similar proximal and distal segments, whereas the medial carpal tubercle of adult male robinsoni is either short (not reaching the base of the pollex) or (if long) consists of a bulbous proximal knob and a narrower, comma-shaped distal process.

Cranially, Marmosa isthmica and M. robinsoni differ in the length of the rostral process of the premaxillae (about equal in length to the height of I1 in isthmica, about half as long as I1 is tall in robinsoni), in the development of postorbital processes (usually distinct in large specimens of isthmica, usually indistinct in robinsoni), in the occurrence of palatine fenestrae (consistently absent in isthmica, consistently present in robinsoni), and in auditory morphology. Auditory differences between these species consist in the size and shape of the alisphenoid tympanic process, which is small, laterally compressed, and usually somewhat pointed ventrally in isthmica versus larger and globular (smoothly rounded ventrally) in robinsoni. Another auditory character is the medial process of the ectotympanic, which is indistinct or altogether absent in isthmica, but which is usually rather well developed in robinsoni.

Marmosa isthmica differs from M. xerophila in all the same characters by which it differs from M. robinsoni, but the contrast in rostral morphology is even more pronounced: the rostral process of the premaxillae (long in isthmica, short in robinsoni) is completely absent in M. xerophila.

Marmosa isthmica and M. simonsi are easily distinguished by visually obvious morphological differences. Although the two species overlap in all external dimensions (tables 4, 5), isthmica is much the larger species, on average, and it has a relatively longer tail than simonsi. In side-by-side comparisons of skins, the orangish-brown dorsal pelage of isthmica contrasts conspicuously with the grayish-brown fur of simonsi. Ventral pelage differences are also striking: there is always a continuous streak of self-colored fur that extends from chin to anus in isthmica, whereas almost the entire ventral surface of simonsi is covered by gray-based fur (usually only the fur of the chin, throat, and groin is self-colored). Whereas the medial carpal tubercle of adult male isthmica is long (reaching the base of the pollex), the medial carpal tubercle of simonsi is shorter and does not extend distally to the base of the pollex. The tail, which is entirely dark on its dorsal surface in isthmica, is completely unpigmented (whitish) distally in simonsi. Points of craniodental difference include the rostral process of the premaxillae (long in isthmica, short in simonsi), palatine fenestrae (absent in isthmica, present in simonsi), auditory bullae (small and compressed in isthmica, larger and globular in simonsi), and a notch between stylar cusps D and E (absent in isthmica, present in simonsi).

Marmosa isthmica is externally similar to M. rubra, but it usually lacks the dark midrostral streak that is usually present in rubra. Other external differences are apparent on close inspection: (1) the gular gland is consistently present in isthmica but absent in rubra; (2) the medial carpal tubercle of mature adult male specimens is long (extending distally to the base of the pollex) in isthmica, but this tubercle is much shorter and does not extend to the base of the pollex in rubra; (3) whereas the caudal scales of isthmica are arranged in both spiral and annular series, the caudal scales of rubra are consistently arranged in spiral series. Craniodental differences include the postorbital processes, which are often well developed in large specimens of isthmica, but which are absent or indistinct in all examined specimens of rubra; and the preparacrista of M1–M3, which is connected to stylar cusp B in isthmica, but which is usually connected to stylar cusp A in rubra.

Geographic Distribution and Sympatry

The known distribution of Marmosa isthmica extends from Ñuri (locality 89; fig. 22) in the Panamanian province of Bocas del Toro eastward across the isthmus and then southward along the Pacific coast of western Colombia to Santa Rosa (locality 188) in the Ecuadorean province of El Oro; M. isthmica also occurs in the Caribbean lowlands of northwestern Colombia, and it extends southward into the inter-Andean valleys of the Río Cauca and the Río Magdalena. Most collection localities are moist lowland and premontane forests below 1700 m elevation, but a few specimens have been collected in dry forests and mangroves, and one specimen has been recorded in the Guajira/Barranquilla Xeric Scrub of Olson et al. (2001).

Fig. 22

Geographic range of Marmosa isthmica based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

The range of Marmosa isthmica partially overlaps the distributions of M. mexicana and M. zeledoni, with which it is known to occur sympatrically as described above. The range of M. isthmica also overlaps the distribution of M. robinsoni in Panama, where they have been collected sympatrically at Balboa (locality 91). Lastly, the range of M. isthmica overlaps the distribution of M. simonsi in western Ecuador, where the two species have been collected sympatrically at Santa Rosa (locality 188), Vinces (locality 205), and Río Briceño (locality 210).

Geographic Variation

Among the specimens we examined, geographic variation in morphological characters is apparent in pelage color and in the development of supraorbital and temporal ridges. In particular, specimens from Ecuador tend to have less developed supraorbital and temporal ridges by comparison with specimens from Colombia and Panama. Additionally, some specimens from the southern Ecuadorean provinces of Los Ríos and El Oro have intensely yellow-washed pelage in contrast to the darker orangish-brown or yellowish-brown pelage of specimens from northern Ecuador, Colombia, and Panama. As a result, every specimen with yellow-washed pelage has delicate supraorbital and temporal ridges, but not all specimens with delicate supraorbital and temporal ridges have yellowish fur. Neither of the cranial nor coat-color phenotypes we observed appears to be consistently associated with any of the several ecoregions (Olson et al., 2001) in which the species has been collected. Because coat color and cranial ornamentation are only partially correlated, and in the absence of other phenotypic evidence of genetic divergence, we provisionally consider M. isthmica to be a monotypic species.

Taxonomic History

The first literature reference to a specimen of Marmosa isthmica is in Thomas (1880), who identified his material as Didelphys murina. Other specimens were subsequently identified by authors as either murina or zeledoni (e.g., by Alston, 1880; Thomas, 1882b, 1888; Allen, 1912). Shortly after Goldman (1912) described Marmosa isthmica based on a single specimen from the Canal Zone of central Panama, Anthony (1916) reported a large series from the eastern Panamanian province of Darién. Just one year later, however, Goldman (1917) treated isthmica as a subspecies of M. mexicana, as did Cabrera (1919).

In his monographic revision of Marmosa, Tate (1933) treated isthmica as a subspecies of M. ruatanica (a member of his Mitis Section, which also included M. chapmani, M. mitis, and M. simonsi as valid species). Hershkovitz (1951), however, regarded all of the nominal taxa in Tate's Mitis Section as conspecific, listing M. mitis ruatanica and M. mitis isthmica as separate subspecies. Cabrera (1958) accepted Hershkovitz's taxonomic conclusions, but he correctly pointed out that robinsoni was the oldest available name in Tate's Mitis Section, and he listed M. robinsoni isthmica as one of several valid South American forms. Most subsequent authors (e.g., Honacki et al., 1982; O'Connell, 1983; Eisenberg, 1989; Emmons, 1997; Nowak, 1999; Brown, 2004) have effectively treated M. robinsoni as monotypic, but Gardner (2005) listed M. r. isthmica as a valid subspecies.

Remarks

Thomas (1921) described Marmosa mimetra based on four specimens from “Santo Domingo” ( =  Santo Domingo de los Colorados, Ecuador; locality 221). Although his description of mimetra strikingly resembles Goldman's description of isthmica, Thomas considered mimetra to be most similar to Allen's chapmani from Trinidad. Apparently, Tate (1933: 126) was the first to recognize the essential similarity between isthmica and mimetra—both of which he treated as subspecies of M. ruatanica—remarking that they are separable “only in certain cases by the stronger cinnamon color of the ventral pelage and the smoother, more shiny tail scales of the latter” and adding that “[s]pecimens from southwestern Colombia are probably transitional.” Curiously, however, mimetra was recently included in the synonymy of simonsi by Gardner (2005) and Creighton and Gardner (2008).

We examined the holotypes of isthmica and mimetra, both of which are mature adult males. These specimens are remarkably similar in external and craniodental morphology, except for the slightly more yellow-washed fur of mimetra. Because the pelage color difference is not accompanied by other evidence of genetic divergence, we treat mimetra as a junior synonym of isthmica.

Specimens Examined (N  =  408)

PANAMA—Bocas del Toro, Ñuri (USNM 575395, 575397–575400). Canal Zone, Balboa (USNM 456810, 456811, 456816, 456819, 456823, 456824, 456827–456832, 456834–456841), Camp Piña (USNM 301377, 301378, 301546, 302631, 302635, 302636, 303048, 304634, 304636), Chiva Chiva (USNM 302326), Cristóbal (USNM 456842–456847, 456849–456866), Fort Davis (USNM 297874, 303044), Frijoles (USNM 503421), Gatún (AMNH 36724–36730), Madden Road (USNM 300331–300334, 301135–301140), Río Indio (USNM 170969 [holotype of Marmosa isthmica Goldman, 1912]). Coclé, no other locality data (USNM 303407), El Valle (USNM 304710–304714). Colón, Cuipo Point ( =  La Uyama: FMNH 124153), Piña (USNM 318315). Darién, Boca de Río Paya ( =  Mouth of Río Paya: USNM 306406–306422;  =  Camp Mack: USNM 314553), Cana (FMNH 53996, 53997; USNM 178614–178616, 178620, 178703–178707, 178711, 178712, 178971), El Real (AMNH 37576–37579), Jaqué ( =  Junction of Ríos Jaque and Imamado: USNM 362318–362327), Paya Camp (USNM 314554, 314555), Río Chucunaque (USNM 306404, 306405), Río Setegantí (USNM 318120–318123), Tacarcuna Laguna Camp (USNM 309319–309327), Tacarcuna Village (USNM 309268–309318, 309328, 309329;  =  Mt. Tacarcuna: AMNH 37859;  =  Near G.M.L. Camp: USNM 512870, 512871;  =  Tacarcuna: AMNH 37859–37871, 37873, 37875–37888, 37891–37893), Tapalisa (AMNH 37889, 37890). Panamá, Cerro Azul (USNM 302414–302416, 302418–302423, 302425–302428, 302630, 302632–302634, 302637, 302638, 303043, 303045–303047, 303050, 303052–303057, 303232, 303279, 303280, 304637, 306387–306392, 306397–306403, 314552), Cerro Campana (USNM 303406), Cerro Jefe (USNM 306393–306396), Río Pequeni (USNM 318336), Río Trinidad (AMNH 36731, 36732), Tocumen (USNM 304635). San Blas, Armila ( =  Quebrada Venado: USNM 335031–335038), Mandinga (USNM 305147–305153). COLOMBIA—Antioquia, Caucasia (USNM 499263), La Tirana ( =  25 km S and 22 km W Zaragoza: USNM 499256–499262), Purí (FMNH 69858, 69859), Río Currulao (FMNH 69856, 69857), San Jerónimo (FMNH 69829–69832, 69836, 69852), Valdivia ( =  Quebrada Valdivia: FMNH 69828, 69851, 69861, 69862;  =  Quebrada de Oro: FMNH 69860), Villa Arteaga (FMNH 69833–69835, 69839, 69840, 69853, 69854). Caldas, Samaná (FMNH 70978). Chocó, Condoto (BMNH 13.8.10.15, 13.8.10.16, 14.5.28.29), Unguía (FMNH 69841–69848, 69855). Córdoba, Socorré (FMNH 69316–69318). Cundinamarca, W of Bogotá (BMNH 95.8.1.34). Nariño, Guayacana (USNM 309046). Sucre, Colosó ( =  Las Campañas: FMNH 69319). Valle del Cauca, Lomitas (AMNH 32177), San José (AMNH 31682, 31718), Río Raposo (USNM 334679–334685). ECUADOR—No other locality data (BMNH 59.11.28.4). Chimborazo, Puente de Chimbo (AMNH 62124, 62125). El Oro, Santa Rosa (AMNH 61392). Esmeraldas, Corondelet ( =  Carondelet: BMNH 1.6.5.20), San Javier (BMNH 1.3.19.48, 1.3.19.49; USNM 113319). Guayas, Cerros de Colonche ( =  Cerro de Manglaralto: AMNH 64524), Huerta Negra (USNM 534287). Imbabura, Hacienda Paramba (BMNH 1.6.5.19, 99.12.5.11, 99.12.5.12). Los Ríos, Lima Pareja (USNM 534289), Vinces ( =  Near Puerto Nuevo: AMNH 63347–63349; USNM 534288, 534290). Manabí, Cordillera de Balzar ( =  Balzar Mts: BMNH 80.5.6.89), Cuaque ( =  Coaque: AMNH 64534; FMNH 41248), Río Briceño ( =  Bahia de Caraquéz: AMNH 64527–64529), San José (FMNH 53354, 53355). Pichincha, Mindo (BMNH 74.779, 13.10.24.67–13.10.24.69), Nanegal (BMNH 98.5.1.20), Santo Domingo de Los Colorados (BMNH 15.1.1.53, 15.1.1.54 [holotype of Marmosa mimetra Thomas, 1921], 15.1.1.55, 15.1.1.57, 15.1.1.59, 15.1.1.60).

Other Specimens

In addition to specimens positively identified as Marmosa isthmica, we examined 11 juvenile specimens that we provisionally refer to M. isthmica because they have not developed all of the diagnostic traits that distinguish this species from M. zeledoni.

PANAMA—Canal Zone, Camp Piña (USNM 301642, 301643, 304631), Fort Sherman (USNM 296190). Coclé, El Valle (USNM 304695). Panamá, Cerro Azul (USNM 302413, 302417, 302424, 303231, 304632), Tocumen (USNM 304630).

Fig. 23

Dorsal cranial views of mature adult male Marmosa robinsoni (left, USNM 443895) and Marmosa xerophila (right, USNM 443837). Scale bar  =  10 mm.

Fig. 24

Ventral cranial views of mature adult male Marmosa robinsoni (left, USNM 443895) and Marmosa xerophila (right, USNM 443837). Scale bar  =  10 mm.

Type Material

The holotype (by original designation, MCZ B7749) is a mature adult (age class 8) male preserved as a skin and skull; although the skull lacks both ectotympanics, the specimen is otherwise in good condition.

Type Locality

The holotype was collected by U.S. Army Lieutenant Wirt Robinson on 12 July 1895 on Isla Margarita, a large (ca. 1000 km²) Caribbean island comprising the Venezuelan state of Nueva Esparta. Although the published type locality (“Margarita Island”) is imprecise, Robinson's published itinerary (Robinson and Richmond, 1896) states that he spent the week from 8 to14 July 1895 at El Valle de Espíritu Santo (10°59′N, 63°52′W, ca. 200 m above sea level; Paynter, 1982) on the lower slopes of Cerro Copey a few kilometers inland from Porlamar (10°57′N, 63°51′W) on the arid southeastern coast.

Morphological Diagnosis

Midrostral fur pale, usually contrasting sharply with darker fur of crown; dark median rostral stripe usually absent (an indistinct marking is occasionally present); dark facial mask usually not extending posteriorly to contact base of ear; dorsal body pelage variable but usually some shade of yellowish or grayish brown; dorsal cover hairs 8–11 mm in length, guard hairs 10–14 mm; gray-based ventral pelage conspicuous or not, usually restricted to sides of abdomen (but sometimes extending to the sides of lower chest, rarely to the inner parts of arms and legs), with yellowish hair tips; self-colored ventral pelage (extending as a continuous median strip of variable width from chin to anus) dark or light yellowish buff. Exposed skin of tail dark brown to yellowish brown, indistinctly bicolored (paler ventrally), without any consistent pattern of scale arrangement (both spiral and annular patterns coexisting); 14–22 scales/cm on dorsal surface at caudal midlength; caudal-scale hairs pale brown or whitish, detectable without magnification; central hair of each caudal-scale triplet as long as (or slightly longer) than two scales. Gular gland present. Mammae at least 6–1–6  =  13 (but perhaps sometimes as many as 9–1–9  =  19; see table 3). Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle long or short (reaching the base of the pollex or not), with globular proximal part and comma-shaped distal part.

Rostral process of premaxillae short, about half as long as I1 height. Orbitosphenoid-alisphenoid suture somewhat less than twice as long as sphenorbital fissure height in lateral view. Supraorbital ridges straight or slightly divergent posteriorly, tending to become thick and dorsally reflected in mature adults but seldom produced laterally to form distinct postorbital processes; temporal ridges usually conspicuous and converging posteriorly, sometimes forming a low sagittal crest in old adults. Palatine fenestrae consistently present and usually large (typically as a single rounded hole on each side); posterolateral palatal foramina large, usually similar in length to M3 (measured across paracone-metacone). Tympanic wing of alisphenoid usually large and globular; medial process of ectotympanic usually well developed. I1 hypsodont; crown of I2 sharply defined in relation to root; C1 without accessory cusps; preparacrista connected to stylar cusp B on M1–M3; stylar cusps D and E usually separated by a distinct notch.

Comparisons

Morphological comparisons of Marmosa robinsoni with M. mexicana, M. zeledoni, and M. isthmica are provided in the preceding accounts. Comparisons of M. robinsoni with M. xerophila, M. simonsi, and M. rubra are provided below.

Marmosa robinsoni closely resembles M. xerophila in size and qualitative characters. Although robinsoni averages larger than xerophila in most external measurements (tables 4, 5), the two species overlap broadly in postcranial dimensions and cannot be identified on this basis alone. In side-by-side comparisons of skins, the dorsal pelage of robinsoni is usually darker and somewhat less grayish than that of xerophila, and the ventral pelage of robinsoni is usually yellowish (“cream-colored” sensu Handley and Gordon, 1979) whereas the ventral pelage of xerophila is usually whitish. Also, the lateral zones of gray-based ventral fur are more conspicuous in robinsoni than they are in xerophila because the pale hair tips are shorter (the longer pale hair tips of xerophila more effectively conceal the gray hair bases when the ventral fur is unruffled). Marmosa robinsoni averages larger than M. xerophila in almost all measured craniodental dimensions (tables 6, 7), with minimal overlap in observed ranges for Maxillary Tooth Row (MTR), Upper Molar Series (UMS), and Lower Molar Series (LMS).2 Visual comparisons of skulls suggest that the rostrum is proportionately longer and more slender, and that the orbits are proportionately smaller in robinsoni than in xerophila. The qualitative cranial trait that differs most consistently between the species is the rostral process of the premaxillae, which, although small, is almost always distinct in robinsoni; by contrast, the rostral process seems to be consistently absent in xerophila. Additionally, postorbital processes are less often distinct in robinsoni than they are in xerophila, and the temporal ridges are less strongly convergent posteriorly in robinsoni than they are in like-aged specimens of xerophila.

Marmosa robinsoni can be distinguished unambiguously from M. simonsi in numerous qualitative characters despite their similarity in external and craniodental measurements (tables 4–TABLE 5TABLE 67). Whereas the dark facial mask of robinsoni seldom extends posteriorly to the base of the ear, the mask often extends continuously from the ear to the mystacial region in simonsi, some specimens of which closely resemble mexicana in this respect. Dorsal and ventral pelage coloration is also widely divergent in these species: the dorsal pelage is usually some shade of yellowish brown, and a broad streak of self-colored (usually yellowish) ventral fur invariably extends from chin to anus in robinsoni; by contrast, the dorsal pelage is distinctly grayish and the ventrum is almost entirely covered by gray-based fur in simonsi (only the chin and groin of which are self-colored). The tail, which is uniformly dark (brownish) in robinsoni, is boldly marked with white in simonsi. Postorbital processes are usually absent or indistinct in robinsoni, but they are consistently well developed and often large in simonsi. The temporal ridges are better developed and more strongly convergent posteriorly (sometimes forming a low saggital crest) in robinsoni than they are in like-aged specimens of simonsi (which never develops a saggital crest).

Marmosa robinsoni is unlikely ever to be confused with M. rubra, from which it differs externally by its relatively shorter tail and larger ears (tables, 4, 5), lack of a dark midrostral streak (usually present in rubra), yellowish- (versus reddish-) brown dorsal pelage, presence (versus absence) of a gular gland, inconsistent (versus spiral) caudal scale patterning, and much longer caudal-scale hairs. Marmosa robinsoni differs craniodentally from M. rubra by having a short (versus long) rostral process of the premaxillae; by having palatine fenestrae (which are absent in rubra); by having large, globular alisphenoid tympanic processes (the alisphenoid bullae are small and laterally compressed in rubra); by having a small but usually distinct medial process of the ectotympanic (absent in rubra); and by the preparacrista connecting to stB on M1–3 (the preparacrista usually connects to stA in rubra).

Marmosa robinsoni also merits comparison with M. murina—a distantly related species not treated in this report—with which it occurs sympatrically in northern Colombia, northern Venezuela, and on the island of Tobago (see below). Among other characters, these species differ in caudal scale patterning (inconsistent in robinsoni, spiral in murina), carpal tubercles (present in mature adult male robinsoni, absent in murina), caudal-scale hairs (macroscopically visible in robinsoni but not in murina), the rostral process of the premaxillae (shorter in robinsoni than in murina), occurrence of palatine fenestrae (present in robinsoni, absent in mainland forms of murina), shape of the tympanic process of the alisphenoid (smoothly globular in robinsoni, laterally compressed and sometimes pointed ventrally in murina), medial process of the ectotympanic (well-developed in robinsoni, absent or indistinct in murina), and stylar cusps D and E (usually separated by a distinct notch in robinsoni but not in murina).

Geographic Distribution and Sympatry

The known distribution of Marmosa robinsoni extends from Finca Santa Clara (locality 109; fig. 25) in the western Panamanian province of Chiriquí eastward across the isthmus to Colombia and northern Venezuela. Although most Venezuelan specimens are from north of the Orinoco River, we examined one specimen (AMNH 16132) from Ciudad Bolívar (locality 236) on the south (right) bank of the river in Bolívar state. The species is also known from several islands on the continental shelf of Central America (e.g., Isla del Rey, Isla Saboga) and South America (Isla Margarita, Trinidad, and Tobago), and from the Caribbean island of Grenada.

Fig. 25

Geographic range of Marmosa robinsoni based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

Marmosa robinsoni occurs in a wide variety of habitats from sea level to 2600 m elevation including lowland and montane moist forests, lowland dry forests, mangroves, savannas, and xeric shrublands. Among the collecting localities mapped in figure 25, approximately 32% correspond to open habitats (llanos, xeric shrublands, etc) from which about 45% of the specimens examined were collected. Large numbers of museum specimens with the same locality data suggest that this species is often among the most abundant nonvolant small mammals in such environments (Handley and Gordon, 1979).

The range of Marmosa robinsoni partially overlaps the distributions of M. mexicana, M. zeledoni, and M. isthmica in Panama, as described above in the accounts of those species. The range of M. robinsoni also overlaps the distribution of M. xerophila in the arid and semiarid landscapes surrounding the Gulf of Venezuela in the Venezuelan states of Falcón and Zulia, and in the adjacent department of La Guajira, Colombia. However, there appear to be no collecting localities where both species have been taken sympatrically. The only other congeneric species with a distribution that overlaps the range of M. robinsoni is M. murina, which also occurs in northern Colombia, northern Venezuela, and on the island of Tobago (Goodwin, 1961; Handley, 1976; Pérez-Hernández, 1989; Linares, 1998). In Venezuela, M. robinsoni and M. murina have been collected sympatrically near Urama (locality 239), Ciudad Bolívar (locality 236), and San Augustín (locality 261); on the island of Tobago, these species have been collected together at Charlotteville (locality 294) and Speyside (locality 295).

Geographic Variation

Available population samples of Marmosa robinsoni exhibit noteworthy geographic variation in size and pelage color. Insular specimens—such as those from Trinidad (chapmani), Tobago (luridavolta), and Grenada (grenadae)—tend to be larger, on average, than specimens from the adjacent mainland regardless of the habitats in which they were collected. According to López-Fuster et al. (2000), however, mainland Venezuelan specimens from agricultural landscapes and disturbed forests exhibit larger craniodental dimensions than those from cloud forests and gallery forests. In general, specimens of M. robinsoni from humid habitats (including mangroves and high altitude cloud forests) tend to have darker pelage (grayish brown washed with orange or dark yellow dorsally, and orangish buff or dark yellowish buff ventrally) as well as dark-brown ears, feet, and tail. By contrast, specimens from drier, more open habitats (e.g., dry forests, savannas, and xeric shrublands) tend to have paler pelage (light grayish brown washed with yellow dorsally, and yellowish buffy or cream buffy ventrally) and light-brown ears, feet, and tail (Tate, 1933; Handley and Gordon, 1979).

Taxonomic History

The first published reference to material identifiable as Marmosa robinsoni was Allen and Chapman's (1893) report of Didelphis (Micoureus) murina from Trinidad. Subsequently, Thomas (1896) and Allen and Chapman (1897) also misidentified specimens of M. robinsoni as M. murina.

Bangs (1898a) described Marmosa robinsoni based on three specimens from Isla Margarita (Venezuela). Later, and in the same volume of the Proceedings of the Biological Society of Washington, Bangs (1898b) described Marmosa mitis based on 27 specimens from Pueblo Viejo (Colombia). Although the type series of mitis and robinsoni exhibit obvious external and cranial similarities, Bangs (1898b) provided comparisons only between mitis and murina. Because Bangs did not compare mitis with robinsoni, and because he provided only superficial descriptions of the cranial morphology of both taxa, several subsequent authors failed to recognize their close relationship to each other and to other related nominal taxa described in the years that followed.

Tate (1933) was the first author to recognize robinsoni and mitis as conspecific, treating them as valid subspecies along with casta (from mainland Venezuela) and fulviventer (from Isla del Rey, Panama). However, Tate erroneously used mitis as the senior name, and he recognized chapmani (from Trinidad) as a separate valid species. Finally, Tate grouped M. chapmani, M. mitis, M. ruatanica ( =  M. mexicana), and M. simonsi as members of the Mitis Section of his Murina Group of Marmosa (table 1).

As explained in the introduction, Hershkovitz (1951) treated all of the species in Tate's (1933) Mitis Section, together with isthmica and mimetra, as subspecies of Marmosa mitis. This broad concept of M. mitis was subsequently endorsed by Cabrera (1958) who, however, correctly used robinsoni as the oldest available name. Cabrera's taxonomy was then adopted by Hall and Kelson (1959) and effectively persists (with minor modifications) to the present day (Gardner, 2005; Creighton and Gardner, 2008).

Remarks

Despite the marked geographic variation in size and pelage color noted above, specimens that we refer to Marmosa robinsoni resemble one another closely and differ as a group from other valid species of Marmosa as diagnosed herein. Although molecular data would be welcome to assess the genetic distinctness of and phylogenetic relationships among the many named forms here treated as junior synonyms of M. robinsoni (including casta, chapmani, fulviventer, grenadae, luridavolta, mitis, nesaea, and pallidiventris), only morphological inferences are possible at present.

The larger body size of island populations of small nonvolant mammals by comparison with their mainland counterparts is a well-documented phenomenon for which a multitude of explanatory hypotheses have been proposed (Crowell, 1983; Lomolino, 1985; Adler and Levins, 1994; McNab, 1994). Several nominal taxa have been described for insular populations of M. robinsoni, including the nominotypical form (from Isla Margarita), chapmani and nesaea (Trinidad), fulviventer (Isla del Rey), grenadae (Grenada), and luridavolta (Tobago). Our examination of holotypes and other representative material of these taxa suggest that they are nothing more than large island forms with habitat-correlated coat color differences similar to those observed among many other widespread species of small mammals.

Most continental forms average smaller than typical (insular) M. robinsoni, but the young adult female holotype of pallidiventris (FMNH 18692) is exceptionally small by comparison with the mature adult male holotypes of casta (BMNH 11.5.25.184) and mitis (MCZ B-8123), a difference plausibly attributable to age and sexual dimorphism. Continental forms also differ from typical robinsoni in pelage color. The tropical moist forest ecoregions (Olson et al., 2001) where the holotypes of the former were collected plausibly explain their darker pelage when compared with the type material of robinsoni from the predominantly arid environments of Isla Margarita. Although the holotype of mitis has wider and more conspicuously gray-based ventral fur than the holotype of casta, comparable variation in ventral pelage markings is observable among other species of Marmosa and is not, in our opinion, a sufficient basis for taxonomic diagnosis.

Specimens Examined (N  =  541)

Without locality data (BMNH 95.3.9.6). PANAMA—No other locality data (USNM 303051). Canal Zone, no other locality data (USNM 318316), Balboa (USNM 456818, 456822), Fort Kobbe (USNM 298697, 298698, 300329, 300330, 301141, 303049), Miraflores (USNM 396415), Quarry Heights (USNM 303281–303283). Chiriquí, Near Escopeta Camp ( =  23–25 km NNE San Felix: USNM 541000, 541002;  =  Colorado Camp: USNM 541324), Finca Santa Clara (USNM 520772), Tolé (USNM 331071). Coclé, Río Hato (USNM 331069). Panamá, Pacora (USNM 305146), Isla Saboga ( =  Saboga Island: MCZ 10809), Isla del Rey ( =  San Miguel Island: FMNH 34071; MCZ B8435 [holotype of Marmosa fulviventer Bangs, 1901], B8437, B8438). Veraguas, Santa Fé ( =  Río Santa Maria: USNM 304696–304709). COLOMBIA—Atlántico, Barranquilla (MVZ 135234–135243;  =  “Barranquilla”, place of shipment: MVZ 183339;  =  Vicinity Barranquilla: MVZ 183334–183338). Bolívar, San Juan Nepomuceno (FMNH 69315). Cesar, El Salado (USNM 280814–280816), El Orinoco ( =  Río Cesar: USNM 280820, 280886–280888;  =  Río Guaimaral: USNM 280817, 280819), San Sebastián de Rábago ( =  San Sebastián: FMNH 69320, 69321), Valledupar ( =  Aguas Verdes: USNM 280818). Cundinamarca, Bogotá (AMNH 143521). Huila, Naranjal ( =  Valle de Suaza: USNM 541857–541861, 543120), Villavieja (MVZ 113366, 113367, 113833–113835–113840). La Guajira, El Pueblito ( =  Pueblo Viejo: FMNH 18509; BMNH 9.4.15.18–9.4.15.20; MCZ B8117–B8122, B8123 [holotype of Marmosa mitis Bangs, 1898], B8125–B8127, B8132, B8143; USNM 85531, 85532;  =  Santa Marta: FMNH 18508), La Concepción (FMNH 18507), Las Marimondas (USNM 280876–280880, 280882, 280883, 280885), San Miguel (FMNH 18506), Villanueva (USNM 280853–280875;  =  Sierra Negra: USNM 280821–280852). Magdalena, Bonda (AMNH 14610, 14611, 15357–15361, 23273–23276, 23280, 23281, 23292, 23627), Colonia Agrícola de Caracolicito (USNM 280806), Mamatoco (AMNH 15362), Minca (AMNH 23293), Palomino (USNM 85533), Pueblo Bello (USNM 280807–280813), Santa Marta (AMNH 244887), Taganga (AMNH 15363). Norte de Santander, Cúcuta (FMNH 18692 [holotype of Marmosa mitis pallidiventris Osgood, 1912]). Santa Marta, no other locality data (AMNH 244887). Tolima, Honda (AMNH 34602–34604), Mariquita (AMNH 207766). VENEZUELA—Apure, Hato El Frío (USNM 448524). Aragua, Camp Rangel (USNM 314171), Ocumare de La Costa (USNM 517271–517280), Rancho Grande (USNM 517262–517270). Barinas, Altamira (USNM 418540). Bolívar, Ciudad Bolívar (AMNH 16132). Carabobo, El Trompillo (BMNH 14.9.1.86–14.9.1.97), San Esteban (AMNH 31532; BMNH 11.5.25.178–11.5.25.183, 11.5.25.184 [holotype of Marmosa mitis casta Thomas, 1911], 11.5.25.185, 11.5.25.187;  =  San Esteban Valley: BMNH 11.5.25.186), Urama (USNM 372938–372940, 372942–372944, 372947). Distrito Federal, Caracas (AMNH 130586–130589). Falcón, Cerro Santa Ana ( =  15 km SSW Pueblo Nuevo: USNM 442907;  =  49 km N and 32 km W Coro: USNM 443870–443874, 443877, 443880–443888, 443890–443896), Hacienda Socopito ( =  20 km S and 98 km E Maracaibo: USNM 443801;  =  24 km S and 94 km E Maracaibo: USNM 418531, 418532), Mirimire (USNM 406953). Guárico, Estación Biológica de los Llanos (USNM 385052;  =  9 km SE Calabozo: USNM 442908, 443906, 443908, 443910;  =  7 km S and 5 km E Calabozo: USNM 443897, 443901–443905, 443911), Hato Las Palmitas (USNM 385053–385056, 418518, 418519, 443794, 443797, 443798, 443800). Lara, Caserio Boro (USNM 443913), Puerta Vieja (USNM 443914), La Concordia (USNM 443912), Río Tocuyo (AMNH 130577–130585, 130600). Mérida, Mérida (BMNH 2.3.4.8;  =  Cafetos de Chama: AMNH 24320, 24323, 24324, 33166; BMNH 5.1.1.6), Milla ( =  Cafetos de Milla: BMNH 98.7.1.21; USNM 149005), Pedregosa (BMNH 98.7.1.19). Miranda, Curupao (USNM 385057–385060), San Andrés ( =  8 km SSE Caracas: USNM 385047–385049). Monagas, Hato Mata de Bejuco ( =  47 km SE Maturin: USNM 385068–385072;  =  54–55 km SE Maturín: USNM 443915–443917, 442720), Ipuré (BMNH 0.5.1.59), San Agustín (USNM 406951), San Antonio de Maturín ( =  San Antonio: AMNH 69939, 69940). Nueva Esparta, Isla Margarita ( =  Margarita Island: BMNH 99.11.12.1; MCZ B7749 [holotype of Marmosa robinsoni 21Bangs, 1898]; USNM 63209, 63210, 63212), Salamanca (USNM 388381, 388388–388397, 388399, 388400;  =  3 km S La Asunción: USNM 388398). Portuguesa, Guanarito (AMNH 266951–266954). Sucre, Campo Alegre (BMNH 0.5.1.58), Cuchivano (AMNH 69938), Cumaná (USNM 388377–388379, 388385, 388386), Guaraúnos (AMNH 257208–257210). Trujillo, Isnotú ( =  Near Isnoto: USNM 370050), Agua Viva (USNM 371304), Agua Santa (USNM 370048, 370049), El Dividive (USNM 371305, 371315, 371316), La Ceiba ( =  Hacienda Valle Verde: USNM 371317), Valera (FMNH 22175). Zulia, Cerro Azul ( =  17 km N and 55 km W Maracaibo: USNM 443807;  =  18 km N and 56 km W Maracaibo: USNM 443802–443804;  =  39 km NW La Paz: USNM 443805, 443806), Novito ( =  3 km S and 19 km W Machiques: USNM 418529, 418530), Río Chama (BMNH 98.7.1.20). TRINIDAD AND TOBAGO—No other locality data (AMNH 6127; USNM 197042), El Cerro del Oropuche ( =  Orepouche Heights: AMNH 31229–31231), Bush Bush Forest (AMNH 188357, 189314–189316, 204855–204857, 206595–206597, 206761, 206762, 206764–206768). “Trinidad Island” (BMNH 95.3.9.7, 97.6.7.25, 97.6.7.26). Arima, Arima ( =  St. Patricks: AMNH 169671, 188355). Caroni, Caparo River ( =  Caparo Valley: BMNH 97.6.7.24 [holotype of Marmosa nesaea Thomas, 1911]), Caparo (AMNH 7426, 7429, 7660/6046–7664/6050). Saint Andrew, no other locality data (AMNH 186440), Cumaca (AMNH 188354, 208996, 208999–209003, 212128–212130, 214425–214438, 214444, 234963–234970), Cumuto (AMNH 212303–212305), Sangre Grande (AMNH 173984, 173996, 174000, 174007, 174008, 174012, 174162, 188356;  =  El Reposo Rd.: AMNH 173990;  =  Maingot Estate: AMNH 173998), Turure Forest (AMNH 214440, 234972, 234976), Upper Fishing Pond ( =  Fishing Pond: AMNH 173997). Saint George, Brazil Village (AMNH 208997, 208998), Caura (AMNH 7665/6051, 7666/6052 [holotype of Marmosa chapmani Allen, 1900], 7667/6053–7670/6056, 7672/6058, 7674/6060–7676/6062; USNM 85556;  =  Caura Mts.: AMNH 7430). Saint Patrick, Aripo Savanna ( =  Aripo Valley: BMNH 35.2.10.1), Cedros Point ( =  Cedros: AMNH 234960, 234961;  =  Cedros Ward: AMNH 214424). Tobago, Charlotteville (AMNH 259973, 259983; USNM 537898, 537899, 538075–538078), Speyside (AMNH 184845, 184846, 184848 [holotype of Marmosa mitis luridavolta Goodwin, 1961], 184849). Victoria, Princestown (AMNH 4799–4802, 6046, 6049, 6121, 6123, 6045/4767, 6047/4768, 6048/47669, 6050/4770–6053/4773, 6055/4775, 6056/4776, 6058/4778), Savana Grande (BMNH 1939.3270). GRENADA–Saint George, Annandale (BMNH 87.6.30.5 [holotype of Marmosa grenadae Thomas, 1911]).

Type Material

The holotype (by original designation, USNM 443819) is a mature adult (age class 7) male specimen preserved as a skin in good condition and a skull that is missing the left ectotympanic but is otherwise complete.

Type Locality

The holotype was collected on 28 June 1968 at a locality called “La Isla” near Cojoro (gazetteer entry 163) in the Colombian department of La Guajira by a Smithsonian Venezuelan Project field team consisting of Norman E. Peterson, Fred P. Brown, Jr., and John O. Matson, whose field notes were the basis for Handley's (1976: 67) published description of coastal-desert habitats near Cojoro.

Morphological Diagnosis

Midrostral fur pale, usually contrasting sharply with darker fur of crown; dark median rostral stripe absent or inconspicuous; dark facial mask not extending posteriorly to contact base of ear; dorsal body pelage pale, usually some shade of faded yellowish brown or light grayish brown; dorsal cover hairs about 7 mm in length, guard hairs about 10 mm; gray-based ventral pelage restricted to the sides of abdomen, with whitish or yellowish hair tips; self-colored ventral pelage (extending as a wide median stripe from chin to anus) whitish or yellowish. Exposed skin of tail grayish brown, indistinctly bicolored (paler ventrally), without any clear pattern of scale arrangement (both spiral and annular patterns coexisting); 14–16 scales/cm on dorsal surface at caudal midlength; caudal-scale hairs usually whitish and detectable without magnification; central hair of each caudal-scale triplet as long as (or slightly longer than) two scales. Gular gland present. Mammae at least 5–1–5  =  11 (but perhaps as many as 6–1–6  =  13; see table 3). Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle long or short (reaching the base of pollex or not), with subequal but recognizable proximal and distal parts.

Rostral process of premaxillae absent. Orbitosphenoid-alisphenoid suture somewhat less than twice as long as sphenorbital fissure height in lateral view. Supraorbital ridges parallel or slightly divergent posteriorly, usually thick and dorsally reflected in mature adults, sometimes produced laterally as more or less distinct postorbital processes; temporal ridges conspicuous, usually strongly convergent posteriorly and sometimes joining to form a low sagittal crest in old adults. Palatine fenestrae consistently present and well developed (typically as a single rounded hole on each side); posterolateral palatal foramina large, similar in length to M3 (measured across paracone-metacone) or even longer. Tympanic wing of alisphenoid usually large and globular; medial process of ectotympanic well developed. I1 hypsodont; crown of I2 sharply defined in relation to root; C1 without accessory cusps; preparacrista connected to stylar cusp B on M1–M3; stylar cusps D and E usually separated by a distinct notch.

Comparisons

Morphological comparisons of Marmosa xerophila with M. mexicana, M. zeledoni, M. isthmica, and M. robinsoni are provided in the preceding accounts. Marmosa xerophila differs from M. simonsi, M. rubra, and M. murina in all of the same characters by which M. robinsoni differs from those species, except that the rostral process of the premaxillae (short in M. robinsoni) is entirely lacking in M. xerophila.

Geographic Distribution and Sympatry

The known geographic distribution of Marmosa xerophila is restricted to the coastal deserts surrounding the Golfo de Venezuela (including the Península de Paraguaná) in the Venezuelan states of Falcón and Zulia and in the Colombian department of La Guajira (fig. 26). The habitats in which specimens have been collected correspond to the “Paraguaná Xeric Scrub” and the “Guajira/Barranquilla Xeric Scrub” ecoregions of Olson et al. (2001). Although the species is not known from the Netherlands Antilles, it may once have occurred on the continental-shelf island of Aruba, where several other species of small mammals that now inhabit arid or semiarid mainland habitats have been reported as subfossils (Hooijer, 1960, 1967).

Fig. 26

Geographic range of Marmosa xerophila based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

Although the range of Marmosa xerophila is completely overlapped by the macrogeographic distribution of M. robinsoni, these species have yet to be recorded in sympatry. The range of M. xerophila is also contained within the macrogeographic distribution of M. murina, a rainforest species not treated in this report.

Geographic Variation

According to López-Fuster et al. (2002), Marmosa xerophila is a morphometrically homogeneous species that exhibits no noteworthy geographic differences in measured dimensions. Although our observations and data are broadly consistent with their assessment, we agree with Handley and Gordon (1979) that specimens from the west coast of the Golfo de Venezuela (in the Venezuelan state of Zulia and the Colombian department of La Guajira) have visibly smaller bullae than those on the east coast (in the Venezuelan state of Falcón).

Taxonomic History

Remarkably, no specimens of this locally abundant mouse opossum appear to have been collected prior to the Smithsonian Venezuelan Project expeditions to the northwestern coastal lowlands of Venezuela in 1968. Therefore, the brief taxonomic history of Marmosa xerophila effectively began with Handley and Gordon's (1979) original description, which was based on the Colombian holotype and 245 paratypes from the localities listed below. Despite a close resemblance to M. robinsoni, the distinctness of M. xerophila has not been questioned by subsequent researchers.

Specimens Examined (N  =  105)

COLOMBIA—La Guajira, Cojoro ( =  37–55 km NNE Paraguaipoa: USNM 443812–443818, 443819 [holotype of Marmosa xerophila Handley and Gordon, 1979], 443820–443831). VENEZUELA—Falcón, Capatárida (USNM 442721–442727, 442729–442731, 442733–442735, 442744, 443918–443925, 443927–443929, 443931, 443936–443938, 443940–443942, 443946, 443947, 443951, 443952, 443955–443957, 443959, 443960, 443963–443972, 443974–443978;  =  18 km WSW Capatárida: USNM 442728), Moruy ( =  49 km N and 33 km W Coro: USNM 443834–443848, 443851), Yabuquiva ( =  25 km SW Pueblo Nuevo: USNM 442906;  =  48 km N and 46 km W Coro: USNM 443852, 443854–443856, 443862, 443863, 443868–443869). Zulia, Paraguaipoa ( =  114 km N and 32 km W Maracaibo: USNM 443810, 443811, 443832).

Fig. 27

Dorsal and ventral cranial views of Marmosa simonsi (USNM 461643, a mature adult male). Scale bar  =  10 mm.

Type Material

The holotype (by original designation, BMNH 99.8.1.20) is a young adult (age class 6) male, preserved as a skin and skull in good condition.

Type Locality

The holotype was collected by P.O. Simons on 3 November 1898 at Puná (gazetteer entry 197; appendix), a village on the island of Puná in the Ecuadorean province of Guayas. The island of Puná (855 km²) is covered with Ecuadorean dry forests and Gulf of Guayaquil/Tumbes mangroves (Olson et al., 2001).

Morphological Diagnosis

Midrostral fur pale, contrasting sharply with darker fur of crown; dark median rostral stripe absent; dark facial mask often extending posteriorly to contact base of ear; dorsal body pelage distinctly grayish; dorsal cover hairs 8–10 mm in length, guard hairs 10–11 mm in length; most of ventral surface (including chest, abdomen, and the inner parts of the fore- and hind limbs) covered by gray-based fur with yellowish hair tips; self-colored ventral pelage (present only on chin or as a narrow median stripe extending from chin to upper chest) yellowish buff. Exposed caudal skin brownish basally but abruptly whitish on distal ⅓ to ½ of tail; caudal scales without any clear pattern of arrangement (both spiral and annular patterns coexisting); about 14 scales/cm on dorsal surface at caudal midlength; caudal-scale hairs usually whitish, detectable without magnification; central hair of each caudal-scale triplet slightly longer than two scales. Gular gland present. Mammary formula unknown. Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle short and comma-shaped, not reaching base of pollex, and not segmented into distinguishable proximal and distal parts.

Rostral process of premaxillae short, usually a little less than half of I1 height. Orbitosphenoid-alisphenoid suture about twice as long as sphenorbital fissure height in lateral view. Supraorbital ridges moderately to strongly divergent posteriorly, tending to become thick and produced laterally over orbital fossae in adults; postorbital processes consistently present and distinct; temporal ridges weakly developed and moderately to strongly convergent posteriorly. Palatine fenestrae present (usually as a single well-developed rounded or irregular hole on each side); posterolateral palatal foramina large, typically about as long or longer than M3 (measured across paracone-metacone). Tympanic wing of alisphenoid globular; medial process of ectotympanic usually well developed. I1 hypsodont; crown of I2 sharply defined in relation to root; C1 without accessory cusps; preparacrista connected to stylar cusp B on M1–M2; stylar cusps D and E usually separated by a distinct notch.

Comparisons

Comparisons of Marmosa simonsi with M. mexicana, M. zeledoni, M. isthmica, M. robinsoni, and M. xerophila are provided in the preceding accounts. Marmosa simonsi and M. rubra are compared below.

Marmosa simonsi is smaller than M. rubra in most measured external and craniodental dimensions (tables 4–TABLE 5TABLE 67), with the notable exception of its relatively and absolutely larger ears, longer upper canines, and larger bullar dimensions. In qualitative external traits, M. simonsi is easily distinguished from M. rubra by its long facial mask (often extending posteriorly to the ear), distinctly grayish dorsal pelage, almost entirely gray-based ventral pelage, particolored (white-tipped) tail, annularly and spirally arranged caudal scales, and long caudal-scale hairs. By contrast, M. rubra has a short facial mask, reddish dorsal fur, more extensively self-colored ventral fur, an all-dark tail, spirally arranged caudal scales, and much shorter caudal-scale hairs. In addition, whereas M. simonsi lacks a dark midrostral stripe and possesses a gular gland, M. rubra possesses a dark midrostral stripe and lacks a gular gland.

In qualitative craniodental traits, these species are most readily distinguished by the rostral process of the premaxillae (shorter in simonsi than in rubra), postorbital processes (consistently well developed in simonsi, indistinct or absent in rubra), palatine fenestrae (present in simonsi, absent in rubra), tympanic wing of the alisphenoid (relatively more inflated and globular in simonsi than in rubra), medial process of the ectotympanic (better developed in simonsi than in rubra), and by the attachment of the preparacrista (to stylar cusp B in simonsi, to stylar cusp A in rubra).

Geographic Distribution and Sympatry

The known distribution of Marmosa simonsi extends along the Pacific littoral zone and adjacent hilly country of western Ecuador (between sea level and 1600 m elevation) from Jama (locality 208; fig. 28) in Manabí province southward to Naranjito (locality 228) in the Peruvian department of Piura. Although a wide range of terrestrial habitats occur in this region (e.g., mangroves, dry forests, lowland moist forests, and montane moist forests), most collection localities appear to be from mangroves or dry (deciduous) forests.

Fig. 28

Geographic range of Marmosa simonsi based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

The geographic range of Marmosa simonsi partially overlaps that of M. isthmica. Geographic details about the extent of this overlap and the localities where these species occur sympatrically are described in the preceding account for M. isthmica.

Geographic Variation

We did not observe any noteworthy variation among geographic samples of this species.

Taxonomic History

Marmosa simonsi was briefly assigned to the genus Caluromys by Matschie (1916), an obvious mistake that was soon rectified by Cabrera (1919). Tate (1933) recognized simonsi as a valid species that he associated with chapmani and mitis (both treated as synonyms of M. robinsoni in this report) and with ruatanica ( =  M. mexicana) in his Mitis Section of Marmosa. Hershkovitz (1951), however, considered simonsi to be a subspecies of mitis, alleging that it grades into the darker mimetra (a taxon that he also regarded as a subspecies of M. mitis) in northern Ecuador. Hershkovitz's broad concept of M. mitis ( =  M. robinsoni) has been accepted by all subsequent authors (e.g., Cabrera, 1958; Honacki et al., 1982; Gardner, 2005; Creighton and Gardner, 2008).

Specimens Examined (N  =  117)

ECUADOR—El Oro, Río Puyango (MSB 87083–87087), Santa Rosa (AMNH 61195; USNM 513423). Guayas, Cerro Bajo Verde (AMNH 63416), Chongoncito (AMNH 63012, 63401–63414), Guayaquil (BMNH 99.8.1.50–99.8.1.57, 99.9.9.140; USNM 121155, 121156, 121158), Isla Puná (AMNH 140232, 140233), Puná (BMNH 99.8.1.20 [holotype of Marmosa simonsi Thomas, 1899], 99.8.1.21–99.8.1.26), San Ramón (AMNH 66851–66868, 66870–66881, 66883–66885;  =  Hacienda San Ramon: FMNH 41402–41404). Loja, Hacienda Casanga ( =  Casanga: AMNH 47181;  =  Río Casanga: AMNH 93797), Malacatos (FMNH 53356–53366), Zozoranga (USNM 461643). Los Ríos, Vinces ( =  Hacienda Pijigual: AMNH 63343–63346, 63350). Manabí, Jama (FMNH 53353), Los Pozos (AMNH 67282, 67283), Río Briceño (AMNH 64525, 64526, 64530–64533, 64536). PERU—Piura, Huásimo (FMNH 81448–81451), Naranjito (USNM 551640). Tumbes, Matapalo (FMNH 81452–81455), Papayal (USNM 302980).

Fig. 29

Dorsal and ventral cranial views of Marmosa rubra (USNM 274577, mature adult male). Scale bar  =  10 mm.

Type Material

The holotype (by original designation, AMNH 71973) is a mature adult (age class 8) female preserved as a skin and skull; although the skin is in good condition, the auditory region is damaged on the left side of the skull.

Type Locality

The holotype of Marmosa rubra was collected by Carlos Olalla and Sons (an Ecuadorean family of professional collectors) on 7 December 1925 at a locality they called “Boca río Curaray” (gazetteer entry 226; appendix) in territory historically claimed by Ecuador. Tate's (1931: 6) verbatim type locality “mouth of Rio Curaray, Province of Oriente, Ecuador” reflects the Olallas' patriotism, but it is geographically inaccurate because the mouth of the Curaray—a right-bank tributary of the lower Napo—is (and was, de facto, in 1925) in the Peruvian department of Loreto. According to Wiley (in press), the Olallas' collections from “Boca río Curaray” probably include material collected on both banks of the Curaray and the Napo within a radius of several kilometers around the confluence of those rivers. It is not known exactly where the holotype of M. rubra was taken in this general area.

Morphological Diagnosis

Midrostral fur pale, usually contrasting sharply with darker fur of crown; dark median rostral stripe present and usually distinct; dark facial mask not extending posteriorly to contact base of ear. Dorsal body pelage usually dull reddish brown (occasionally dark grayish brown washed with dark orange or red); dorsal cover hairs 8–10 mm in length, guard hairs 10–12 mm in length; gray-based ventral pelage conspicuous, usually covering sides of neck, chest, and abdomen (sometimes also extending to the inner parts of arms and legs), with orangish hair tips; self-colored ventral pelage (usually extending as a continuous median strip from chin to anus) yellowish buff or orangish buff. Exposed skin of tail dark brown dorsally from base to tip, indistinctly paler ventrally; caudal scales arranged in spiral pattern; about 16 scales/cm on dorsal surface at caudal midlength; caudal-scale hairs dark brown and not apparent without magnification; central hair of each caudal-scale triplet about as long as one scale. Gular gland absent. Mammae 3–1–3  =  7 or 4–1–4  =  9. Lateral and medial carpal tubercles present in mature adult males; medial carpal tubercle short, robust, comma shaped, close to the base of pollex, and not bipartite.

Rostral process of premaxillae long, slightly greater than or equal to I1 height. Orbitosphenoid-alisphenoid suture variable, somewhat longer than or similar to sphenorbital fissure height in lateral view. Supraorbital ridges subparallel or slightly divergent posteriorly, slender, and dorsally reflected in mature adults, not projecting laterally over orbital fossae; postorbital process absent or indistinct; temporal ridges conspicuous, parallel or slightly convergent posteriorly. Palatine fenestrae absent (but tiny asymmetrical palatine perforations occasionally present); posterolateral palatal foramina small, similar in length to M4 (measured at paracone-metacone). Tympanic wing of alisphenoid small and laterally compressed, often somewhat pointed ventrally; medial process of ectotympanic absent or indistinct. I1 hypsodont; crown of I2 usually sharply defined in relation to root; C1 without accessory cusps; preparacrista usually connected to stylar cusp A on M1–M3; stylar cusps D and E usually separated by a distinct notch on M2.

Comparisons

Morphological comparisons of Marmosa rubra with other congeneric species treated in this report are provided in the preceding accounts. Brief comparisons with other members of the subgenus Marmosa with which it overlaps geographically in western Amazonia are provided below.

Marmosa rubra is most readily distinguished from M. andersoni by having a dark median rostral streak (absent or indistinct in andersoni), by usually having a continuous median streak of self-colored ventral fur (absent in andersoni), by having spirally arranged caudal scales (the caudal scales are both spirally and annularly arranged in andersoni), by lacking distinct postorbital processes and palatine fenestrae (both present in andersoni), and by the preparacrista of M1–2 connecting to stylar cusp A (the preparacrista of these teeth connects to styB in andersoni).

Marmosa rubra is much larger than M. lepida in all measured dimensions (the length of the upper molar series, for example, is 7.1–7.6 mm in rubra but only 5.6–6.7 mm in lepida). Qualitative character differences that distinguish these species include the dark median rostral streak (present in rubra, absent in lepida), distinct postorbital processes (absent in rubra, present in lepida), a posterior accessory cusp on C1 (absent in rubra, present in lepida), and the connection of the preparacrista on M1–M2 (to styA in rubra, to styB in lepida).

Marmosa rubra can be distinguished from M. murina and M. quichua by having a dark median rostral streak (absent in murina and quichua), by having reddish-brown dorsal fur (the dorsal pelage is grayish brown in murina and quichua), by lacking a gular gland (present in murina), by the presence of carpal tubercles in mature adult males (absent in murina and quichua), by lacking postorbital processes (present in murina and quichua), by the preparacrista of M1–2 connecting to stylar cusp A (the preparacrista of these teeth connects to styB in murina and quichua), and by the usual presence of a distinct notch between stylar cusps D and E on M2 (usually absent in murina and quichua).

Geographic Distribution and Sympatry

The known distribution of Marmosa rubra, the only Amazonian species treated in this report, extends from the Río Mecaya (locality 179; fig. 30) in Departmento Putumayo, Colombia, southward along the eastern Andean piedmont and adjacent lowlands (between 180 m and 730 m elevation) to Hacienda Villa Carmen (locality 225) in Departmento Cusco, Peru. The natural vegetation throughout this region is either lowland or premontane rain forest.

Fig. 30

Geographic range of Marmosa rubra based on specimens examined. Numbered points correspond to collection localities listed in the gazetteer (appendix).

Although the geographic range of Marmosa rubra does not overlap that of any other species reviewed herein, it does overlap the distributions of several distantly related congeners, including M. andersoni, M. lepida, M. murina, and M. quichua. The details of this overlap and lists of localities where M. rubra occurs sympatrically with other species of Marmosa (Marmosa) will be described in subsequent reports.

Geographic Variation

We did not observe any noteworthy variation among geographic samples of this species.

Taxonomic History

Marmosa rubra was originally described by Tate (1931) based on the holotype and 11 paratypes. Tate considered M. rubra to be closely related to M. murina, but provided no arguments to support that hypothesis. Later, based on perceived similarities in mammary formulae, inguinal pelage color, caudal features, and supraorbital morphology, Tate (1933) grouped M. rubra, M. murina, M. quichua, and M. tyleriana to form his Murina Section of the Murina Group of Marmosa. A close relationship between M. rubra and M. murina, however, is not supported by recent analyses of molecular sequence data (e.g., Voss and Jansa, 2003, 2009; see fig. 2). To our knowledge, the validity of Marmosa rubra as a full species of mouse opossum has never been questioned.

Specimens Examined (N  =  44)

COLOMBIA—Putumayo, Río Mecaya (FMNH 70969–70977). ECUADOR—Province unknown, Huachi (BMNH 54.268). Napo, Puerto Napo ( =  Near the River Napo: BMNH 34.9.10.230–34.910.233). Orellana, San José Nuevo ( =  San José: AMNH 68128;  =  San José Abajo: AMNH 68129, 68137–68139), San José de Payamino (FMNH 124612). Pastaza, Montalvo (FMNH 41452–41454, USNM 274577), Río Copotaza ( =  Río Copataza: FMNH 53348, 53350), Río Pindo ( =  Rio Pindo Yaco: FMNH 43181–43183, USNM 274578). Sucumbíos, Marian (FMNH 124611). PERU—Amazonas, Huampami ( =  Headwaters of Río Huampami: MVZ 153283;  =  [Vicinity of] Huampami (MVZ 153280, 153282, 154759, 154765), Puesto Vigilancia Comaina (USNM 581941). Cusco, Hacienda Villa Carmen (FMNH 84253). Loreto, Boca del Río Curaray (AMNH 71950, 71952, 71953, 71972, 71973 [holotype of Marmosa rubra Tate, 1931], 71976).