Gracilinanus, PERUANUS (TATE, 1931)

Semedo, Thiago Borges Fernandes, Brandão, Marcus Vinicius, Carmignotto, Ana Paula, Nunes, Mario Da Silva, Farias, Izeni Pires, Silva, Maria Nazareth Ferreira Da & Rossi, Rogério Vieira, 2015, Taxonomic status and phylogenetic relationships of Marmosa agilis peruana Tate, 1931 (Didelphimorphia: Didelphidae), with comments on the morphological variation of Gracilinanus from central-western Brazil, Zoological Journal of the Linnean Society 173 (1), pp. 190-216 : 200-207

GRACILINANUS PERUANUS ( TATE, 1931)

Synonyms (first usage of each synonym only)

Marmosa marica: Thomas, 1927: 608 ; part (includes specimens from Tingo Maria, Peru).

Marmosa agilis buenavistae Tate, 1931: 10 ; type locality ‘ Buenavista, Department of Santa Cruz, Bolivia’ .

Marmosa agilis peruana Tate, 1931: 11 ; type locality ‘ Tingo Maria , Rio Huallaga, Peru, 2000 feet’ .

Marmosa agils peruania: Tate, 1933 : legend of plate 11; incorrect subsequent spelling of Marmosa agilis peruana Tate.

Thylamys rondoni Miranda-Ribeiro, 1936: 387 ; part (includes one specimen of the type series from São João da Serra do Norte , Rondônia).

Marmosa rondoni: Vieira, 1955: 352 ; new name combination; part (based on the type series of Miranda-Ribeiro, 1936).

Marmosa [( Thylamys )] agilis buenavistae: Cabrera, 1958: 28; new name combination.

Marmosa [( Thylamys )] agilis peruana: Cabrera, 1958: 28; new name combination.

Gracilinanus agilis: Gardner & Creighton, 1989: 5 ; new name combination; part (buenavistae and peruana in synonymy).

Gracilinanus agilis buenavistae: Anderson, 1993: 18 (general distribution records).

Gracilinanus agilis buenavista: Langguth, Limeira and Franco, 1997: 7 ; incorrect subsequent spelling of Marmosa agilis buenavistae Tate ; part (based on the type series of Miranda-Ribeiro, 1936).

Type information

BMNH 27.11 .1.268, collected on 20 January 1927 by R. W. Hendee. The holotype is an adult male (age class 6) .

PHYLOGENY AND TAXONOMY OF G. PERUANUS 201

Although the type is a skin and skull in good condition, the right zygomatic arch is broken and the palate is partially broken and unclean, making it difficult to visualize the palatine fenestrae.

Type locality

‘ Tingo Maria , Rio Huallaga’, Huánuco, Peru – (09 ° 08′S, 75 ° 57′W – 600 m., Hershkovitz, 1992) .

Geographic distribution

The collecting localities of G. peruanus comprise central Peru, central Bolivia, and western Brazil, in the states of Rondônia and north-western Mato Grosso. By connecting the most external collecting localities of the species we obtained an approximated geographic distribution that extends from central Peru to central Bolivia and western Brazil, in the northern and central portions of Mato Grosso state ( Fig. 1 ). This area pre- sents two records where G. peruanus and G. agilis occur in sympatry (localities 3 and 15; Fig. 1 ).

Emended diagnosis

Gracilinanus peruanus is a small-sized didelphid marsupial that differs from other congeners by the following combination of morphological characters (see Table 10): dorsal pelage dull reddish brown; greybased ventral pelage from chest to anus; usually poorly developed circumocular mask; head and body length 85.0–121.0 mm; tail length 130.0–155.0 mm; greatest length of skull 24.50–28.91 mm ( Table 4); interorbital region usually smooth, sometimes with angular supraorbital margins of the frontals (i.e. margins with sharpened edges, not described by Voss & Jansa, 2003, 2009), never showing incipient postorbital processes; relative position of maxillary fenestra aligned anteriorly with the paracone of P3 and posteriorly with the paracone of M1 or M2; posterolateral palatal foramen usually smaller than palatine fenestra; second foramen ovale present, usually formed by a wide bony strut on the anteromedial region of the alisphenoid tympanic process; paraoccipital process usually very small and rounded; mastoid process usually incipient or absent; and upper canine usually with a posterior accessory cusp, rarely with a small anterior accessory cusp ( Figs 5–10 ).

In bold, the five larger values for each function.

Description

Gracilinanus peruanus is a small-sized marsupial ( Table 4) with short (approximately 7 mm in length), smooth, and dull reddish-brown dorsal pelage, not exposing the grey-based hairs ( Fig. 5 ); pelage of rostrum short with grey-based and cream-tipped hairs, usually contrasting sharply with darker crown fur that is longer and grey-based with reddish-tipped hairs; dark circumocular mask small and poorly developed anteroposteriorly when compared with other Gracilinanus species ( Fig. 10 ); colour pelage between eyes and ears usually reddish, similar to the colour of head; cheeks self-coloured cream contrasting sharply with facial mask; mystacial vibrissae entirely black or black with white tips; three or four black genal vibrissae dorsally oriented and three genal white vibrissae ventrally oriented; white submental and interamal vibrissae present; the self-coloured cream portion of ventral pelage extends from chin to chest and forearms, and greybased hairs extend from chest to inguinal region, including hind limbs; tail slightly bicoloured, longer than combined length of head and body; fur of the base of the tail no longer than 5 mm in length; caudal scales arranged in annular pattern; each scale possesses three hairs inserted in its posterior margin, with the central hair slightly longer (two caudal scales in length); hairs on the basal portion of the tail shorter and more sparse, giving the ventral and distal portion of the tail a more hairy aspect; prehensile ventral surface of the distal part of the tail approximately 30 mm in length; ears small (14–19 mm), cream to cream–orange on the base and pale brown distally.

Craniodentally, premaxillary rostral process present; nasals long, extending anteriorly beyond I1; rostrum short and wide; two lacrimal foramina present on each side of the skull, both not visible in dorsal or lateral views; interorbital region with parallel margins; supraorbital margins vary from smoothly rounded (young specimens) to slightly angular (adult specimens); postorbital processes absent ( Fig. 7 ); petrosal laterally exposed through a fenestra in the parietal– squamosal suture; incisive foramen small, reaching the posterior edge of canines; maxillo-palatine fenestrae present; maxillary fenestrae present, with anterior margin aligned with the paracone of P3 and the posterior margin aligned with the paracone of M1, metacone of M1, or the paracone of M2; palatine fenestrae present; posterolateral palatal foramen usually smaller than palatine fenestra ( Fig. 8 ); paraoccipital process very small and rounded; mastoid process usually incipient or absent; alisphenoid tympanic process globular; second foramen ovale present, usually formed by a wide bony strut on the anteromedial region of the alisphenoid tympanic process ( Fig. 9 ); I1 longer than I2–I5; I2–I5 similar in size; upper canine short, usually with posterior accessory cusps and rarely with small anterior accessory cusps; P1 smaller than P2 and P3; P2 either taller or subequal in height than P3 ( Fig. 8 ).

Comparisons with congeners

Gracilinanus peruanus differs from other congeneric species by exhibiting dull reddish brown dorsal fur versus the reddish, brownish, or greyish tones in other species (see Table 10); grey-based buffy ventral fur from chest to anus, as in G. agilis and G. marica versus selfwhite in G. emiliae and with grey-based hairs from throat to anus in G. aceramarcae , G. dryas , and G. microtarsus ; circumocular mask restricted to the area close to the eyes, similar to G. agilis but differing from the other species in which the masks are wider ( Fig. 10 ); posterior accessory cusps in upper canines frequently present versus absent in G. aceramarcae , G. agilis , and G. microtarsus ; interorbital region smooth or with angular supraorbital margins in mature adults versus weakly and distinctly beaded in G. marica and G. emiliae , respectively; and well-developed maxillary fenestrae, which are small or absent in G. emiliae . A summary of the diagnostic characters of G. peruanus in comparison with the other species of the genus is provided in Table 10. Closer comparisons between G. agilis and G. peruanus are warranted, as the latter was until now considered a junior synonym of the former, and both species are present in Bolivia and central–western Brazil, including two sympatric records ( Fig. 1 ). In fact, G. peruanus and G. agilis are very similar in external and craniodental traits; however, the former exhibits dull reddish brown dorsal fur whereas the latter exhibits dorsal fur with brownish/ greyish tones ( Fig. 5 ). In addition, G. peruanus is generally smaller in size when compared with G. agilis ( Table 4).

Craniodentally, the most distinctive character that distinguishes these species is the relative position of the anterior margin of the maxillary fenestrae, which is consistently aligned with the paracone of P 3 in G. peruanus versus consistently aligned with the paracone of M 1 in G. agilis ( Fig. 8 ). It is important to mention, however, that in young specimens of G. peruanus the position of the maxillary fenestrae is similar to the position found in specimens of G. agilis .

Although we found intraspecific variation in the craniodental traits analysed herein, we found remarkable differences in the frequencies in which they appear in each species ( Table 11). In this sense, these species differ in relation to the posterior accessory cusps in the upper canines, which is frequently present in G. peruanus , but usually absent in G. agilis ; in the rela- tive size of posterolateral palatal foramen when compared with the palatine fenestra, which is usually smaller in G. peruanus and usually larger in G. agilis ; and in the morphology of the alisphenoid tympanic process, as it generally exhibits a wide bony strut in G. peruanus versus a narrow bony strut in G. agilis ( Fig. 9 ; Table 11). The interorbital region appears to be somewhat different between these species. Both species generally exhibit a smooth interorbital region in younger specimens, but with slightly angular margins in older specimens. This latter condition appears to be more developed in G. agilis , in which even an incipient postorbital process may be present (versus absent in G. peruanus ; Fig. 7 ). There are also differences related to the projection of the paraoccipital and mastoid processes (of petrosal), which are more developed in G. agilis than in G. peruanus ( Fig. 9 ). The paraoccipital process is often large and squared in appearance in G. agilis , whereas it is smaller and rounded in appearance in G. peruanus . The mastoid process is much more developed in G. agilis , whereas in G. peruanus it is incipient or absent ( Fig. 9 ).

Intraspecific morphological variation

We observed few pelage variations in examined specimens of G. peruanus . The dorsal pelage followed the

aExtracted from Voss et al. (2009). The holotype of Gracilinanus aceramarcae (AMNH 72568) exhibits a ratio of 1.35; the holotype of Gracilinanus marica (BMNH 98.5.15.1) exhibits a ratio of 1.30.

bReddish forms pass through a gradient of red, from a darker coloration (almost red wine) in G. dryas , red in G. aceramarcae , lighter red in G. peruanus , to a bright red (almost orange) in G. emiliae , and a light red mixed with brown in G. marica . Brownish forms includes G. microtarsus , washed with reddish tones, and G. agilis , presenting a lighter brown coloration mixed with greyish hairs.

cTate (1933) refers to a rather short pelage; examination of the holotype (BMNH 9.3.9.10) and other specimens showed longer pelage (around 8 mm) than in G. agilis (around 6 mm) and G. peruanus (around 7 mm), but shorter than in G. aceramarcae (around 12 mm), G. dryas (around 12 mm), G. marica (around 10 mm), and G. microtarsus (around 10 mm).

dThe holotype of Marmosa beatrix exhibits venter entirely self-coloured cream.

eThe topotype of Marmosa agilis peruana (BMNH 27.11.1.269) exhibits wide and divergent supraorbital margins, probably because of the fact that it is a subadult specimen, age class 5.

fWe found variation in this character for this species (see text and Table 11).

gThe relative maximum length of the fenestrae are in parentheses: for example, P3/M2, from about paracone of P3 to about paracone of M2.

hExtracted from Voss et al. (2009).

same pattern in all examined material, except for a few specimens that exhibited a less homogeneous pelage; however, more expressive variation was observed in the ventral pelage, with several specimens (CM 455, 463, 464, 489, 492, locality 3; Fig. 1 ) exhibiting selfcoloured cream pelage in the inguinal region, and other specimens (UFMT 3816, locality 23; AMNH 209157, locality Costa Marques, Rondônia, not shown in Fig. 1 ) exhibiting self-coloured cream pelage over the entire venter. In most specimens, however, self-coloured fur is restricted to the chin, throat, chest, and frontal limbs. Intraspecific variation observed in the presence of C1 accessory cusps, the size of posterolateral palatal foramen, and the morphology of the anteromedial process of the alisphenoid tympanic process are shown in Table 11.

Taxonomic comments

The combination of diagnostic morphological characters mentioned in the Emended diagnosis section, together with the results of the morphometric and molecular analyses, justify the recognition of a distinct species of gracile mouse opossum. To search for available names that could be assigned to this distinct species we have examined the holotypes of nominal taxa related to G. agilis , such as beatrix Thomas, 1910 ; buenavistae Tate, 1931; peruana Tate, 1931; blaseri Miranda-Ribeiro, 1936 ; and rondoni Miranda-Ribeiro, 1936 ( Voss et al., 2005: table 1).

Regarding the holotype of M. beatrix (BMNH 11.4.23.24), it can be consistently associated with the genus Gracilinanus by the presence of maxillary fenestrae, an anteromedial process of the alisphenoid tympanic process, and P2 taller than P3. Nevertheless, the taxonomic status of this nominal taxon requires further evaluation because the holotype exhibits self-coloured cream pelage covering the entire venter, and is smaller in the majority of the craniodental measurements, differing in the ventral coloration and size of typical G. agilis ( Table 4). In addition, the multivariate analyses showed that this specimen does not group with either G. agilis or G. peruanus in morphometric space ( Figs 3 , 4 ).

Miranda-Ribeiro (1936) described two species that are currently in the synonym of G. agilis : M. blaseri , based on a specimen from São Bento, Goiás (MN 1250) and T. rondoni , based on a type series from two distinct localities: Salto do Sepotuba, Mato Grosso state (MN 1270) and São João da Serra do Norte, Rondônia state (MN 1271, 1272, 1275, and 1276). The holotype of M. blaseri can be assigned to G. agilis based on its greyish brown dorsal coloration, absence of cusps in the C1, relative position of maxillary fenestra between M1 and M2, and a posterolateral palatal foramen larger than the palatine fenestra.

In relation to T. rondoni , the type series is a composite. Miranda-Ribeiro (1955) designated the specimen MN 1270 from Salto do Sepotuba , Mato Grosso, as the lectotype , and the other material, consequently, as paralectotypes (see also Langguth et al., 1997). Among the latter specimens, MN 1276 can be consistently assigned to G. peruanus based on its dull reddish brown dorsal coloration, maxillary fenestra between P3 and M2, a posterolateral palatal foramen smaller than the palatine fenestra, and the presence of a second foramen ovale formed by a wide bony strut on the anteromedial region of the alisphenoid tympanic process. The remaining paralectotypes do not belong to the genus Gracilinanus nor the lectotype.

By contrast, the specimens herein identified as a distinct species of gracile mouse opossum are morphometrically and morphologically similar to the holotypes of M. a. buenavistae Tate, 1931 (BMNH 26.12.4.91) and M. a. peruana Tate, 1931 (BMNH 27.11.1.268). All of the specimens exhibit dull reddish brown dorsal pelage; posterolateral palatal foramen smaller than palatine fenestra; second foramen ovale formed by a wide bony strut on the anteromedial region of the alisphenoid tympanic process; paraoccipital process small and rounded; and incipient mastoid process. Several minor differences observed between the holotype of M. a. buenavistae and our specimens, such as the relatively larger size, the faded greyish brown dorsal pelage coloration, the presence of an incipient postorbital process, and the absence of accessory cusps in the upper canines in the former, can be related to age variation because the specimen is a mature adult female (age class 8). Other morphological traits, such as the anterior margin of the maxillary fenestrae that only slightly reaches the posterior border of P 3 in both holotypes, and the absence of accessory cusps in C 1 in the holotype of M. a. peruana, are within the scope of morphological variation observed in the specimens in central– western Brazil and the surrounding areas; however, among all the specimens examined, only the holotype and a topotype of M. a. peruana (BMNH 27.11.1.268– 269) exhibited developed circumocular masks. In the absence of other specimens from Peru and nearby vicinities, we provisionally treat this difference as a geographical variation in this trait.

Based on these results, we recognize M. a. buenavistae Tate, 1931 and M. a. peruana Tate, 1931 as synonyms that can be assigned to a valid species distinct from G. agilis ( Burmeister, 1854) . Because both available names were described in the same work by the same author ( Tate, 1931) and in the same hierarchical category (as subspecies of Marmosa agilis ), the priority between these two names cannot be objectively determined. So, as first revisers of this issue, and according to article 24.2 of the International Code of Zoological Nomenclature ( ICZN, 1999), we establish M. a. peruana Tate, 1931 as the name for the species recognized in this report, with M. a. buenavistae Tate, 1931 treated as its junior synonym. We use the spelling of this taxon as G. peruanus following the agreement in gender according to article 31.2 of the ICZN (1999).

Reproductive data

Within our sample, we found five lactating adult females collected in the dry season (August 2006 and October 2007), as demonstrated by the discoloured inguinal region and enlarged nipples, thus giving a completely cream coloration to the inguinal region (UFMT 870, 872, 873, 1333; CM 492); the abdominal– inguinal mammae formula of these specimens is 4-1-4 = 9; however, the female AMNH 209157 has 3-1-3 = 7 (S. Anderson, annotation in the specimen tag).

Regarding sexually dimorphic characters, the adult males belonging to age classes 6 and 7 (sensu Rossi et al., 2010b) present a knob-like, lateral carpal tubercle (UFMT 876, 3816; CM 455, 457, 464, 489; MZUSP 35121, 35112), whereas a young male (UFMT 1370) belonging to age class 3 has a very small, knob-like lateral carpal tubercle. This structure was absent in all females examined, whereas medial carpal tubercles were absent in both sexes.

Field data

Among specimens of G. peruanus with available field information, two were caught in Sherman live traps placed on the ground and 11 were caught in pitfall traps. Although pitfall traps appear to be more effective in collecting the species, it is not possible to confirm this because we have no information about the trapping efforts. Considering the distinct ecoregions (sensu Olson et al., 2001) within the geographic range of G. peruanus , 12 of our records are located in the Chiquitano dry forests (including the holotype of M. a. buenavistae, locality 67; Fig. 1 ), followed by 11 records in the Cerrado, two records in the Madeira-Tapajós moist forests, one record in the Mato Grosso seasonal forests, and one record in the Ucayali moist forests (the holotype of M. a. peruana, locality 71, Fig. 1 ). For most collecting localities in the Cerrado and for a few localities in the Chiquitano dry forests, the specimens are associated with gallery forests such as those shown in Figure 11 . In the Cerrado, 20 individuals were collected in gallery and deciduous forests, and seven individuals were collected in both disturbed forests and Cerrado sensu stricto. This pattern leads us to consider that this species is associated with forested formations, and may use the gallery forests to enter savanna-like formations in the Brazilian Cerrado.

Figures

Figure 1. Map of the distribution localities of Gracilinanus specimens included in phylogenetic and/or morphological analyses in the present study. Numbers correspond to the localities listed in the Appendix; , Gracilinanus peruanus; ○, Gracilinanus agilis; ★, type material associated with Gracilinanus. Records based on sequenced specimens taken from GenBank used in the phylogenetic analyses are as follows: ▲, Gracilinanus microtarsus; +, Gracilinanus emiliae; ◆, Gracilinanus aceramarcae; *, Gracilinanus dryas (see Table 1 and Appendix). The shaded area corresponds to the G. peruanus distribution based on this study.

Figure 3. Scatter plot of the principal component analysis (PCA) of 24 log-transformed craniodental measurements of Gracilinanus agilis, Gracilinanus peruanus, and other nominal taxa related to these species.

Figure 4. Scatter plot of canonical scores (DA) for discriminant functions 1 and 2 based on craniodental measurements of Gracilinanus agilis, Gracilinanus peruanus, and other nominal taxa related to these species.

Figure 5. Dorsal view of skin. Left to right: Gracilinanus peruanus (UFMT 872, locality 1 in Fig. 1, female, age class 7) and Gracilinanus agilis (UFMT 1039, locality 36 in Fig. 1, male, age class 8). Scale bar: 10 mm.

Figure 6. Dorsal, ventral, and lateral views of the skull and lateral view of the mandible of Gracilinanus peruanus (UFMT 1333, locality 2 in Fig. 1, female, age class 7). Scale bar: 5 mm.

Figure 7. Dorsal and ventral views of the skull of Gracilinanus peruanus (MZUSP 35121, locality 2 in Fig. 1) and Gracilinanus agilis (MZUSP 35286, locality 43 in Fig. 1). Both specimens are male individuals of age class 7. Scale bar: 5 mm.

Figure 8. Upper canine morphology with accessory cusps absent in Gracilinanus agilis, upper left (MZUSP 35286, locality 43 in Fig. 1, male, age class 7); usually present in Gracilinanus peruanus, upper right (UFMT 1333, locality 2 in Fig. 1, female, age class 7). Size differences between posterolateral palatal foramen (plpf) in relation to palatine fenestra (p): smaller in G. peruanus (UFMT 3816, locality 23 in Fig. 1, male, age class 7), and larger or similar in size in G. agilis (MZUSP 35311, locality 30 in Fig. 1, male, age class 7). Note the position of the maxillary fenestra (m) in G. agilis (M1/ M2) from the metacone of M1 to the paracone of M2, and in G. peruanus (P3/M1 or M2) from the paracone of P3 to the paracone of M2 (see Table 10). Pictures are not shown to scale to better visualize these characters.

Figure 9. Distinct morphologies of the anteromedial surface of the alisphenoid tympanic process (top) and the paraoccipital and mastoid processes (bottom) in Gracilinanus. Upper left: narrow alisphenoid tympanic process, MZUSP 35311, Gracilinanus agilis (locality 30 in Fig. 1). Upper central: median alisphenoid tympanic process, MZUSP 35121, Gracilinanus peruanus (locality 12 in Fig. 1). Upper right: wide alisphenoid tympanic process, MZUSP 35117, Gracilinanus peruanus (locality 10 in Fig. 1). The arrows show the pathway of the mandibular branch of the trigeminal nerve (V3) through the secondary foramen ovale. Note that the portion of the arrow concealed by the alisphenoid process is notably smaller in MZUSP 35311, whereas it is partially and almost fully enclosed by the alisphenoid tympanic process in MZUSP 35121 and MZUSP 35117, respectively. It is important to note that the condition found in MZUSP 35117 does not correspond to the alisphenoid lamina of Voss & Jansa (2009: fig. 16C), because the latter refers to a posteromedial wide alisphenoid bony strut on the bullar surface that does not span the transverse canal, whereas MZUSP 35117 exhibits a wider alisphenoid bony strut anteromedially positioned on the bullar surface that spans the transverse canal. Lower right: paraoccipital and mastoid processes (of petrosal) often large and squared in appearance in G. agilis (UFMT 1039, locality 36 in Fig. 1). Lower left: paraoccipital and mastoid processes (of petrosal) small and rounded in G. peruanus (MZUSP 35121, locality 12 in Fig. 1). Scale bar: 5 mm.

Figure 10. Distinct patterns of the circumocular mask in Gracilinanus. A, UFMT 872, Gracilinanus peruanus, female, age class 7 (locality 1 in Fig. 1). B, UFMT 692, Gracilinanus agilis, male, age class 7 (locality 29 in Fig. 1). C, UFMT 1039, Gracilinanus agilis, male, age class 8 (locality 36 in Fig. 1). Scale bar: 10 mm.

Figure 11. Gallery forest where Gracilinanus peruanus and Gracilinanus agilis were found in sympatry (locality 3 in Fig. 1). Note the black arrow showing the deforested area near the trapping site. Photo courtesy of Vitor Azarias.