An Introduction to Marmosops (Marsupialia: Didelphidae), with the Description of a New Species from Bolivia and Notes on the Taxonomy and Distribution of Other Bolivian Forms Author VOSS, ROBERT S. Author TARIFA, TERESA Author YENSEN, ERIC text American Museum Novitates 2004 2004-12-30 3466 1 40 http://www.bioone.org/doi/abs/10.1206/0003-0082%282004%29466%3C0001%3AAITMMD%3E2.0.CO%3B2 journal article 3781 10.1206/0003-0082(2004)466<0001:AITMMD>2.0.CO;2 b3293c54-11ba-43d8-8c3c-243a8c05d6cd 0003-0082 4735089 Marmosops creightoni , new species Figure 9 HOLOTYPE : An adult male specimen in the Colección Boliviana de Fauna ( CBF 6552 ) collected by Eric Yensen (original number 1705) on 5 August 1999 near the Saynani hydroelectric generating station (ca. 16°07'S , 68°05'W ; 2500 m above sea level ) in the valley of the Río Zongo , Departamento La Paz , Bolivia . The holotype is a whole (unskinned) fluid­preserved specimen with an extracted skull and mandibles. Fig. 5. Lateral view of posterior braincase in Marmosops impavidus ( A, AMNH 272709) and Marmosa murina ( B, AMNH 272816) illustrating the presence of a fenestra ( fen ) that exposes the petrosal ( pet ) between the parietal ( par ) and squamosal ( sq ) in the former taxon. The petrosal is not laterally exposed by a fenestra between the parietal and squamosal in Marmosa . PARATYPES : Five additional specimens (listed as unidentified ‘‘ Marmosa ’’ by Anderson, 1997: 151 ) were collected by G.K. Creighton in 1979 at two adjacent localities in the valley of the Río Zongo , Departmento La Paz , Bolivia . Two of these ( UMMZ 155999 , 156000 ) were taken above the type locality near the Cuticucho hydroelectric generating station (ca. 16°08'S , 68°07'W ) at a recorded elevation of 2967 m , whereas three others ( UMMZ 156001–156003 ) were trapped below the type locality at a recorded elevation of 2000 m . All five paratypes are adult animals preserved as whole (unskinned) fluids with extracted skulls and mandibles . DISTRIBUTION : Known only from wet montane forest in the valley of the Río Zongo , Departamento La Paz , Bolivia , between 2000 and 3000 m above sea level (fig. 10) . ETYMOLOGY: For G. Ken Creighton, collector of the first known specimens of this species and author of the first modern analysis of didelphid phylogeny based on morphological characters ( Creighton, 1984 ). DESCRIPTION: Marmosops creightoni is a medium­sized species (see table 3 for measurements and weights), much larger than members of the M. parvidens group (ca. 20– 35 g adult weight; Voss et al., 2001 : tables 9 , 10 ) but substantially smaller than some lowland forms like M. incanus (adults of which can weigh as much as 120 g ; Lorini et al., 1994 ). The body pelage is almost entirely dark, the colors probably somewhat faded in our fluid­preserved specimens, but still a rich chocolate­brown dorsally on the holotype (collected in 1999) and only a little more bleached on the 24­year­old paratypes . The ventral body coloration is slightly paler brown, but there is no sharp distinction between dorsal and ventral color zones, and all of the thoracic and abdominal fur is graybased. The chin is self­white in all of the material at hand, and white fur extends posteriorly onto the throat of one paratype (UMMZ 155601). The blackish mask that surrounds the eye on each side of the head does not extend anteriorly to include the mystacial pads (which are covered with short brownish fur) or posteriorly to the base of the ears. The mystacial and genal vibrissae are blackish, whereas the submental and interramal hairs are white. The pinnae are uniformly dark and appear naked (a sparse auricular pelage is only microscopically visible). Laid back alongside the head, the short mystacial vibrissae do not extend to the tips of the pinnae. A small patch of almost­naked skin on the midline of the throat suggests the presence of a weakly developed gular gland in all of the male specimens examined. The metacarpals and metatarsals are covered with dark (brownish) hairs, but the digits of both manus and pes are abruptly whitish. All of the males we examined have a knob­like glabrous tubercle on the lateral aspect of the wrist that is absent in both females. The scrotal epithelium is pigmented (mottled­grayish in preservative) and sparsely covered by brownish fur. Both female specimens have 4–1–4 = 9 abdominal­inguinal mammae. Body fur extends onto a short (ca. 1 cm ) segment of the tail base, but the remainder (>90%) of the tail is virtually naked and covered by epidermal scales in predominantly spiral series. Under high magnification (25–50×), each caudal scale is seen to have three hairs projecting from its posterior margin; the central hair is visibly thicker than the lateral hairs, but perhaps not quite twice as wide. On the dorsal surface of the tail, these hairs are no more than one to two scale rows long, but they are longer (three to four scale rows) on the ventral caudal surface. The exposed caudal epithelium is dark (grayish) above and below basally, but it is indistinctly bicolored (dark above, pale below) for at least half the length of that organ, and the tail­tip is entirely white for the last 1–3 cm . Fig. 6. Ventral view of left ear region in Marmosa murina ( A, AMNH 267368) and Marmosops pinheiroi ( B, AMNH 267346) illustrating generic differences in the presence/absence of a secondary foramen ovale. In Marmosa , the mandibular branch of the trigeminal nerve (V 3 , reconstructed course shown by heavy arrow) emerges from the endocranial lumen via the foramen ovale ( fo ), which is bordered by the alisphenoid ( als ) and the petrosal ( pet ); the extracranial course of the nerve is unenclosed in this taxon. In Marmosops , however, the extracranial course of V 3 is partially enclosed by a bony strut ( st ) that extends from the anteromedial surface of the tympanic wing of the alisphenoid ( twa ) across the transverse canal foramen ( tcf ); the nerve then emerges from a so­called secondary foramen ovale. Other abbreviations: bs, basisphenoid; bo, basioccipital; cc, carotid canal; ect, ectotympanic. In dorsal view the skull is distinctively proportioned with small orbits, narrow zygomatic arches, and a very broad interorbital region. The nasal bones are long and conspicuously widened posteriorly near the maxillary/premaxillary suture. The broad interorbital region is due to the considerable inflation of the cranial capsule that encloses the olfactory bulb, behind which there is a shallow postorbital constriction; the supraorbital margins are rounded, with only indistinct beading even in the largest male specimen examined. 6 The braincase is inflated and smooth, without conspicuous temporal scars. In lateral cranial view the premaxillary/maxillary suture is lateral to (rather than behind) the alveolus of I 5 in most examined specimens (the suture is anterior to the alveolus on the left side of UMMZ 155999, and it is behind the alveolus on the right side of UMMZ 156000). The infraorbital foramen is very large (its anterior opening approximately as high as P3 is tall), and the ventral deflection of the zygomatic arch beneath the orbit is shallow. The lacrimal foramina are contained within the anterior orbital margin. The petrosal is laterally exposed through a fenestra between the squamosal and parietal bones. In ventral cranial view the incisive foramina extend between (but not behind) the canine alveoli. The maxillopalatine fenestrae are long and narrow, extending on each side from a point opposite P3 to M3 or to the M3/ M4 commissure. The palatine fenestrae are present as a cluster of larger and smaller openings between the maxillopalatine fenestrae and the postpalatine torus. The alisphenoid tympanic wings (auditory bullae) are large, almost hemispherical capsules without ventral processes or lateral constrictions. The upper canine lacks accessory cusps, and the last upper molar is not much wider (transverse dimension) than M3 and is approximately triangular in outline. Fig. 7. Occlusal views of right dp3 and m1 illustrating the generic difference in trigonid morphology of the deciduous tooth. Top, Marmosops impavidus (MUSM 13286) with a complete (tricuspid) dp3 trigonid. Bottom, Marmosa murina (MUSM 15297) with an incomplete (bicuspid) dp3 trigonid. Note that whereas taxonomic variation in this trait was correctly described by Voss and Jansa (2003 : character 64), the accompanying plate ( Voss and Jansa, 2003 : fig. 15) was mislabeled. 6 The terms ‘‘bead’’ and ‘‘beading’’ as traditionally used by systematic mammalogists to describe the dorsally­grooved supraorbital ridges developed in some marsupials and rodents (fig. 4B) are derived by analogy from the upturned rims used to strengthen sheet metal or other flat surfaces in manufacture (i.e., the last of four definitions listed for the noun ‘‘bead’’ in Webster’s Seventh New Collegiate Dictionary, 1963). MORPHOLOGICAL COMPARISONS: Marmosops creightoni is the only known brownishfurred species of the genus to lack whitish thoracic or abdominal markings; all others have at least some white or cream­colored midventral fur that contrasts abruptly with the darker color of the dorsum and flanks. 7 In many species, whitish midventral fur (which may be either self­colored or graybased) extends continuously from chin to anus, but in a few Amazonian and northern Andean forms (e.g., M. neblina ; see Gardner, 1990 : fig. 2) whitish midventral markings are narrow and sometimes discontinuous. Although narrow whitish midventral markings might occur polymorphically in larger samples of M. creightoni , the six examples at hand suggest that most individuals have completely dark venters. Qualitative and morphometric comparisons with other Bolivian congeners (the identification and taxonomy of which are discussed below) reveal many additional differences ( table 4 ). Marmosops bishopi , for example, is much smaller than M. creightoni , has dull reddish­ or grayish­brown (versus rich chocolate­brown) dorsal fur, has palefurred metapodials that do not contrast in color with the digits (versus dark­furred metapodials contrasting with whitish digits), does not have a pale tail­tip (versus pale tailtip present), lacks palatine fenestrae (versus palatine fenestrae present), and has a small posterior accessory cusp on the upper canine (versus upper canine without accessory cusps). Also, adult males of M. bishopi usually lack a gular gland (versus gular gland weakly developed), have large bladelike (versus knobby) lateral carpal tubercles, and have a pale scrotum with white fur and unpigmented skin (versus scrotum dark with brown fur and mottled­grayish skin). Fig. 8. Palatal morphology in Marmosops noctivagus ( A, AMNH 262403) and Gracilinanus agilis ( B, MVZ 197437) illustrating generic differences in patterns of fenestration. Both Marmosops and Gracilinanus have incisive foramina ( if ), posterolateral palatal foramina ( plpf ), and maxillopalatine fenestrae ( mp ); in addition, some species of Marmosops and all Gracilinanus have palatine fenestrae ( p ). Maxillary fenestrae ( m ), however, are present only in Gracilinanus . Scale bars equal 5 mm. 7 Marmosops invictus , a small Panamanian species with blackish­gray fur, has dark underparts that are only superficially washed with white. Fig. 9. Dorsal, ventral, and lateral views of the skull, and lateral view of the left hemimandible of the holotype of Marmosops creightoni (CBF 6552); however, some missing details (e.g., the nasal tips, broken in the holotype) are reconstructed from UMMZ 155999 (a paratype). Note that the illustrated left hemimandible of CBF 6552 is unusual in possessing only a single mental foramen; the right hemimandible has two foramina, as do both hemimandibles of all five paratypes. Scale bar equals 10 mm. Fig. 10. Collection localities of Marmosops bishopi and M. creightoni in Bolivia. Numbers are keyed to gazetteer entries in appendix 2. Marmosops impavidus , another Bolivian species, overlaps with M. creightoni in many measurements, although it averages smaller in most dimensions. However, M. impavidus has dull grayish­ or reddish­brown (versus rich chocolate­brown) dorsal fur; pale­furred metapodials that do not contrast in color with the digits (versus dark­furred metapodials contrasting with whitish digits); a much narrower interorbital region (LIB = 5.6–6.3 versus 6.6–7.2 mm); and much smaller bullae. In addition, adult males of M. impavidus appear to have no trace of a gular gland (versus gular gland weakly developed), and they have a pale scrotum with white fur and unpigmented skin (versus scrotum dark with brownish fur and mottled­grayish skin). TABLE 3 Measurements (mm) and Weights (g) of Marmosops creightoni Bolivian specimens of Marmosops noctivagus overlap broadly with M. creightoni in most measured dimensions, but they usually have dull reddish­brown (versus rich chocolate­brown) dorsal fur; pale­furred metapodials that do not contrast in color with the digits (versus dark­furred metapodials contrasting with whitish digits); distinct supraorbital beads that conceal the postorbital constriction from dorsal view (versus supraorbital beads absent or indistinct and postorbital constriction visible dorsally); and smaller, conical or flask­shaped bullae (versus bullae larger and hemispherical). In addition, adult males of M. noctivagus have a pale scrotum with white fur and unpigmented skin (versus scrotum dark with brownish fur and mottled­grayish skin), and adult females have 5–1–5 = 11 mammae, of which the anteriormost pair is pectoral (versus 4–1–4 = 9 abdominal­inguinal teats). Marmosops ocellatus also overlaps broadly with M. creightoni in most measured dimensions, but it has much paler grayishbrown (versus rich chocolate­brown) dorsal fur; pale­furred metapodials that do not contrast in color with the digits (versus darkfurred metapodials contrasting with whitish digits); distinctly bi­ and particolored tails, with the terminal one­third or more completely pale (versus tail indistinctly bicolored and only the tip completely pale); a much narrower interorbital region (LIB = 5.5–6.3 mm versus 6.6–7.2 mm); and proportionately wider zygomatic arches (ZB/CBL × 100 = 53–54% versus 48–50%). In addition, most adult males of M. ocellatus lack a gular gland (versus gular gland weakly developed), and adult females have 6–1–6 = 13 mammae, of which the anteriormost two pairs are pectoral (versus 4–1–4 = 9 abdominal­inguinal mammae). Although we examined representative material of every species of Marmosops listed as valid in table 1 , additional pairwise comparisons are unnecessary because other currently recognized species exhibit no special similarity with M. creightoni . Indeed, most extralimital (nonBolivian) species have strikingly divergent traits. Marmosops cracens , M. fuscatus , and M. incanus , for example, have uniformly narrow nasals with subparallel lateral margins (quite unlike the posteriorly expanded nasals of Bolivian forms); M. pinheiroi has lacrimal foramina that are laterally exposed anterior to the orbit (the lacrimal foramina of Bolivian species are usually concealed from lateral view inside the orbit); and M. paulensis has very long incisive foramina that extend posteriorly behind the canine alveoli (the incisive foramina are substantially shorter in all Bolivian taxa). These and other taxonomically informative characters merit uniform scoring for phylogenetic analysis in any comprehensive attempt to revise the contents of Marmosops , but such exercises are beyond the scope of this preliminary report. MOLECULAR COMPARISONS: The first 680 base­pairs of the mitochondrial cytochromeb gene were sequenced from the holotype of Marmosops creightoni by J.L. Patton, whose generosity in sharing these data allowed us to make preliminary comparisons with homologous didelphid sequences previously obtained by his laboratory ( Patton et al., 1996 ; Mustrangi and Patton, 1997 ; Patton et al., 2000 ). In order to assess the relationships of M. creightoni with other congeneric forms, we assembled a small dataset representing nine species of Marmosops , and we included Gracilinanus microtarsus as an outgroup ( table 5 ). We selected exemplar haplotypes based on sequence length (the longest available sequence for each species was chosen), previous representation in published reports ( Patton et al., 1996 ; Mustrangi and Patton, 1997 ; Patton et al., 2000 ), and accessibility of cataloged morphological voucher material in museums. Most species are represented by single haplotypes in our analysis, but we included three sequences from geographically divergent samples of M. noctivagus to assess molecular variation within a geographically widespread taxon. As quantified by pairwise uncorrected sequence divergence ( table 6 ), Marmosops creightoni is phenetically most similar to M. noctivagus , from which it differs on average by about 9.7%. Interestingly, of the three analyzed haplotypes of the latter species, M. creightoni is most divergent (10.9%) from the geographically proximal Bolivian sample and least divergent (9.2%) from each of the Peruvian and Brazilian specimens. All other pairwise interspecific comparisons in these results are>10% divergent, with the exception of a relatively small value (5.7%) between M. impavidus and M. ocellatus . A branch­and­bound parsimony analysis of the nucleotide sequence data (implemented with PAUP 4.0b10; Swofford, 1998 ) recovered a single minimum­length tree with two main clades (fig. 11), one of which consists of Marmosops creightoni , M. impavidus , M. noctivagus , and M. ocellatus in the sequence ( noctivagus ( creightoni ( impavidus + ocellatus ))). The other clade contains members of the parvidens complex ( M. bishopi and M. parvidens ), two southeastern Brazilian forms ( M. incanus and M. paulensis ), and the Amazonian species M. neblina . Unfortunately, there is only weak Bremer and bootstrap support for several branches in this topology, notably including all of those that resolve the relationships of creightoni with other congeneric forms. NATURAL HISTORY: All of our material of Marmosops creightoni was collected in the valley of the Río Zongo, one of the principal headwaters of the Río Beni on the eastern slopes of the Cordillera Real. The Valle de Zongo descends precipitously from 4800 to < 900 m above sea level, with flanking slopes that are often in excess of 60°. The climate is very humid, with clouds and mist each morning and afternoon for most of the year. Between 2000 and 3000 m above sea level— the known elevational range of M. creightoni —the natural vegetation of the Valle de Zongo is humid montane (‘‘cloud’’) forest. 8 Habitat information associated with specimens of M. creightoni taken at three localities in the Valle de Zongo is summarized below, in order of their collection dates. The five paratypes that G.K. Creighton collected in 1979 were taken at two localities along a road that descends the valley and provides access to a series of hydroelectric generating stations, power lines, and support facilities. His first collecting site was at Cuticucho (= Cuti Khuchu on recent topographic maps), which he recorded as having an elevation of 2967 m above sea level (but see appendix 2: locality 16). The site is described in his fieldnotes as follows: 8 We use the nontechnical term ‘‘cloud forest’’ for montane rainforests generally, including the formations that Grubb (1977) usefully defined as Lower Montane Rain Forest, Upper Montane Rain Forest, and Subalpine Rain Forest. Synonyms used by Bolivian botanists include ‘‘Ceja de Monte de Yungas’’ (corresponding to Subalpine Rainforest; e.g., at Cuticucho) and ‘‘Bosque Húmedo Montañoso de Yungas’’ (corresponding to Upper Montane Rainforest; e.g., at Saynani). For a critical review of ‘‘cloud forest’’ and its technical synonyms from a field zoologist’s point of view, see Myers (1969) ; for another (botanical) perspective, see Webster (1995) . TABLE 4 Diagnostic Morphological Comparisons Among Five Bolivian Species of Marmosops a TABLE 4 (Continued) TABLE 5 Exemplar Specimens Included in Molecular Analysis Since our arrival [six days ago] we have not seen the sun for more than an hour a day. Clouds move up the valley in the morning at around 8:30AM and form just below our camp. Intermittent rain and mist are the order of the day. The temperature here stays relatively constant, between 55 and 65° F [13–18° C]. The forest here is very damp—many lichens, moss and ferns on branches of trees. We have trapped in three habitats here: cloud forest (relatively undisturbed), 2nd growth forest—with much bamboo; and artificial clearings under power lines... We are very near treeline here; as one looks up the valley the trees thin out and disappear at a fairly distinct ‘‘line’’ about 300 m above camp. Looking down the valley the trees get noticeably higher... [ 3 October 1979 ] Both specimens of Marmosops creightoni from Cuticucho were taken in the cloud forest with snap traps: one (UMMZ 155999) at the base of a tree and the second (UMMZ 156000) on a low branch. Creighton also collected five species of sigmodontine rodents at this site, of which Akodon mimus , Oxymycterus nigrifrons , and two undescribed forms of Thomasomys (‘‘sp. 1’’ and ‘‘sp. 3’’ of Pacheco, 2003 ) were trapped in the cloud forest; Oryzomys levipes was trapped in nearby secondary growth. Additional species of small mammals subsequently trapped near Cuticucho by other researchers include Gracilinanus aceramarcae (see Salazar­Bravo et al., 2002 ), Akodon cf. boliviensis , and an undescribed form of Oligoryzomys (‘‘sp. B’’ of the flavescens group; Carleton and Musser, 1989 ). TABLE 6 Uncorrected Percent Sequence Divergence Among Cytochrome­ b Haplotypes from Nine Species of Marmosops Fig. 11. The unique minimum­length tree recovered by a branch­and­bound parsimony analysis of cytochrome­ b nucleotide sequences representing 11 ingroup haplotypes from 9 Marmosops species and 1 outgroup haplotype from Gracilinanus microtarsus (used to root the tree, not shown). Bremer support and bootstrap values (>50%) are provided above and below each branch, respectively. The bootstrap values were obtained from branch­and­bound analyses of 1000 pseudoreplicated datasets. Tree length (including autapomorphies) = 689 steps, consistency index (excluding autapomorphies) = 0.48, retention index = 0.44. Farther down the valley G.K. Creighton described his camp at a locality he called ‘‘Cement Mine’’ (not identifiable on our topographic maps), with a recorded elevation of 2000 m , as being much warmer than at Cuticucho and with taller forest. However, Much of the forest has been cut or burned at one time or another in the past 30 years, hence most... is 2nd growth except in the higher valleys intersecting this one at more or less right angles. Lichens & moss on trees are solid in places but not as ‘‘lush’’ as near Cuticucho. There are more broadleaf trees and ‘‘papaya­like’’ trees here. Also large (~ 10 m ) tree ferns. Bamboo undergrowth is extensive and very dense near artificial clearings. Little undergrowth beneath the bamboo except for a few ferns. [ 6 October 1979 ] The first four nights of trapping (5–8 October) at this site yielded only rodents (the same species as those found in the cloud forest near Cuticucho plus Microryzomys minutus and Thomasomys cf. oreas ), but two specimens of Marmosops creightoni were taken on 9 October and another on 10 October. Creighton’s notes for these dates indicate that the marsupials were trapped in ‘‘humid, dark, mossy forest’’ on horizontal branches near large trees ‘‘covered with moss, ferns, bromeliads & vines’’. Apparently, they were ‘‘not caught coming to the bait but, rather, blundering into the traps as they tiptoe along their accustomed routes.’’ The area where all three Marmosops were trapped included ‘‘about the only places where high, undisturbed natural forest is accessible enough to set a trapline’’, the only other primary forest being restricted to very steep slopes above the cultivated valley floor. The holotype of Marmosops creightoni was collected by EY near the Saynani hydroelectric generating station at 2500 m on the southeastern side of the valley. The obviously disturbed forest at this site consisted of patches with a closed canopy but also open areas with isolated trees. The trees were laden with epiphytes, and the ground cover consisted of vines, small bamboo, grasses, moss, and ferns. The holotype was taken on the ground, in a snap trap set in dense grass under a small tree. All of the other small mammals trapped at Saynani by EY and other researchers are sigmodontine rodents, including Akodon mimus , Chibchanomys sp. , Microryzomys minutus , Oryzomys levipes , Oxymycterus nigrifrons , and the same undescribed forms of Oligoryzomys and Thomasomys that occur at Cuticucho. TABLE 7 Measurements (mm) and Weights (g) of the Brazilian Holotype and Referred Bolivian Material of Marmosops bishopi In summary, Marmosops creightoni has been collected in both primary cloud forest and disturbed vegetation in the Valle de Zongo, where it occurs sympatrically with Gracilinanus aceramarcae and several species of sigmodontine rodents between 2000 and 3000 m above sea level. A different species of Marmosops , however, inhabits lower elevations in the Zongo valley. Two specimens, one from El Vertigo at 1800 m (CBF 4003) and another from Kahua at 1150 m (CBF 3880), are both referable to M. noctivagus as we recognize that species below. Future collecting in the elevational interval between 2000 and 1000 m would be helpful in pinpointing just where these congeneric species occur sympatrically or parapatrically along the forested eastern slopes of the Cordillera Real.