Taxonomy and distribution of the Brazilian species of Thylamys (Didelphimorphia: Didelphidae)

Transcription

1 Mammalia (2006): by Walter de Gruyter Berlin New York. DOI /MAMM Taxonomy and distribution of the Brazilian species of Thylamys (Didelphimorphia: Didelphidae) La taxonomie et la distribution des espèces brésiliennes du genre Thylamys (Didelphimorphia: Didelphidae) Ana Paula Carmignotto* and Talitha Monfort Museu de Zoologia da Universidade de São Paulo, Av. Nazaré, 481, Ipiranga, São Paulo, SP, CEP , Brazil, *Corresponding author Abstract The recent increase in mammal inventories and the widespread use of pitfall traps as a major capture method in Brazil have increased the number of non-volant small mammal specimens in scientific collections, providing new information on natural history, geographical range, and taxonomic status of many marsupials and rodents. To date, however, little is known about the Brazilian species of the marsupial genus Thylamys: the number of species in the country is disputed, as are their names and geographic distribution. Thus, the aim of this study was to define the Brazilian species of the genus, delimiting their taxonomic status and distributional range. We provide qualitative and quantitative data on external and skull morphology for specimens housed in scientific collections. A study of the material available allowed us to recognize the occurrence of three species in Brazil: Thylamys karimii (Petter, 1968), T. macrurus (Olfers, 1818) and T. velutinus (Wagner, 1842). For each of these we provide redescriptions, a discussion of local and geographic variations, geographic distribution, taxonomy, age class differentiation, sexual dimorphism, and natural history. Keywords: Brazil; distribution; natural history; taxonomy; Thylamys karimii; Thylamys macrurus; Thylamys velutinus. Résumé L accroissement récent des inventaires de mammifères et l utilisation de plus en plus fréquente des pièges de type pitfall comme principale technique d échantillonnage au Brésil ont contribué à l enrichissement du nombre d espèces de petits mammifères non-volants dans les collections scientifiques. De nouvelles découvertes concernant l histoire naturelle, la distribution géographique, et le statut taxonomique ont été réalisées pour de nombreux marsupiaux et rongeurs du Brésil. Le genre marsupial Thylamys en est un parfait exemple, comme jusqu à récemment trop peu de connaissances existaient à son sujet, et même le nombre d espèces vivant au Brésil était questionné. Les buts de ce travail sont de définir les espèces brésiliennes du genre Thylamys, de déterminer leur statut taxinomique et les synonymies, et d estimer leur distribution géographique. Nous apportons des résultats nouveaux, tant qualitatifs que quantitatifs, concernant les morphologies crânienne et corporelle, par un examen comparatif des spécimens dans les collections scientifiques. Ce travail nous a permis d identifier trois espèces dans le pays: Thylamys karimii (Petter, 1968), Thylamys macrurus (Olfers, 1818), et Thylamys velutinus (Wagner, 1842). Pour chaque espèce reconnue, notre étude présente de nouvelles descriptions, une discussion de la variabilité morphologique au sein des échantillons et entre localités, et divers commentaires ayant trait à la distribution géographique, à la taxonomie, au dimorphisme sexuel, et aux traits d histoire de vie tels que connus à ce jour. Mots cles: Brésil; distribution; histoire naturelle; taxonomie; Thylamys karimii; Thylamys macrurus; Thylamys velutinus. Introduction Thylamys is a genus of small nocturnal, insectivorousomnivorous marsupials restricted to South America, found along the western flanks of the Andes from Peru to Chile, and in the eastern lowlands and middle to high altitudinal areas of Bolivia, Brazil, Paraguay, Uruguay and Argentina (Palma and Yates 1998; Eisenberg and Redford 1999; Flores et al. 2000; González et al. 2000; Solari 2003; Gardner 2005). Thylamys is found mainly in open and semi-arid formations, in contrast to most other mouse opossums, which prefer forested habitats. Species of Thylamys may marginally occur in some forested habitats, such as the subtropical moist forest in Paraguay (Palma 1995a), and in areas of semi-deciduous forest in Brazil (N. Cáceres personal communication). Several morphological characteristics of the genus can be related to life in strongly seasonal or semi-arid environments, such as fat storage in the tail, enlarged auditory bullae, and extensive fenestration of the palate (Creighton 1985). The recent widespread adoption of pitfall traps as a collecting method in Brazil has dramatically increased the number of specimens available for study for several hitherto rare and little known taxa, including Thylamys. We have been able to capture several specimens belonging to this genus in a number of localities across central and northeastern Brazil. Study of these and other specimens housed in collections has demonstrated the need for a 2007/6

2 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. 127 reappraisal of the genus and its species diversity in Brazil. For most of its taxonomic history, Thylamys Gray, 1843 has been considered either a synonym (Tate 1933) or a subgenus of Marmosa Gray, 1821 (Gilmore 1941; Cabrera 1919, 1958; Kirsch and Calaby 1977). Tate (1933), in the most comprehensive taxonomic revision of Marmosa (sensu lato), divided the genus into five informally named units (the cinerea, murina, noctivaga, microtarsus and elegans groups), which closely correspond to the five different genera of marmosines currently recognized: Micoureus, Marmosa, Marmosops, Gracilinanus, and Thylamys, respectively (Creighton 1984; Gardner and Creighton 1989). The elegans group as defined by Tate (1933) corresponds to Thylamys of current usage. According to Tate, this group is characterized by the possession of dorsal fur with a tricolor pattern, relatively small feet and claws (in proportion to body size), tail subject to marked incrassation, nasal bones uniform in width and scarcely expanded at the maxillo-frontal suture, interorbital region much narrowed, and very large auditory bullae. Tate (1933) recognized nine species in the elegans group, namely: Marmosa bruchi, M. elegans, M. formosa, M. janetta, M. marmota, M. pallidior, M. pusilla, M. velutina and M. venusta. Creighton (1984) was the first to associate exclusively the elegans group of Tate (1933) to the name Thylamys, recognizing it as a full genus. Later, Gardner and Creighton (1989) recognized five valid species of Thylamys, including T. elegans, T. macrura (sic), T. pallidior, T. pusillus and T. velutinus. Although the generic rank that Gardner and Creighton (1989) assigned to Thylamys is currently accepted, the species content of the genus was not thoroughly reviewed by those authors, and subsequent changes in the taxonomy have been made. Since Gardner and Creighton (1989), additional species have been recognized as valid, all of them previously treated as junior synonyms (Julien-Laferrière 1994; Palma 1995a; Flores et al. 2000; Solari 2003; Braun et al. 2005). Currently, nine valid species are commonly recognized: T. cinderella (Thomas, 1902), T. elegans (Waterhouse, 1839), T. karimii (Petter, 1968), T. macrurus (Olfers, 1818), T. pallidior (Thomas, 1902), T. pusillus (Desmarest, 1804), T. tatei (Handley, 1957), T. velutinus (Wagner, 1842) and T. venustus (Thomas, 1902). Despite recent taxonomic research on Peruvian, Chilean, Bolivian, Paraguayan and Argentinean species of Thylamys (Flores et al. 2000; Palma et al. 2002; Solari 2003; Braun et al. 2005), taxonomic problems within the genus persist, especially with regard to the Brazilian species. Whereas some authors have recognized four valid species in Brazil: T. karimii, T. macrura (sic), T. pusilla (sic) and T. velutinus (Eisenberg and Redford 1999), others have recognized only three: T. karimii, T. macrurus and T. velutinus (Gardner 2005), or even just one: T. velutinus (Palma 1995a; Solari 2003; Braun et al. 2005). Published range maps also vary according to different authors, resulting in much taxonomic uncertainty. The main reason behind this state of affairs is probably the limited number of specimens available to previous researchers. The objective of this study is to provide an account of the Brazilian species of Thylamys, with redescriptions and updated distribution maps. Materials and methods We obtained qualitative and quantitative data on external and skull morphology from specimens housed in the following collections: Museu de Zoologia da Universidade de São Paulo, São Paulo (MZUSP); Museu Nacional do Rio de Janeiro, Rio de Janeiro (MN); Departamento de Zoologia da Universidade de Brasília, Brasília (UNB); Departamento de Zoologia da Universidade de Minas Gerais, Belo Horizonte (UFMG); American Museum of Natural History, New York (AMNH); Field Museum of Natural History, Chicago (FMNH); Museum of Vertebrate Zoology, University of California, Berkeley (MVZ); National Museum of Natural History, Washington, DC (USNM); and the Sam Noble Oklahoma Museum of Natural History, Norman (OMNH). In addition, we analyzed two specimens collected by Nilton Cáceres (NC), and examined photographs of types and other important specimens housed in the following institutions: Natural History Museum, London (BMNH); Muséum National d Histoire Naturelle, Paris (MNHN); Naturhistorisches Museum, Wien (NMW); and the Universitetets Zoologiske Museum, Copenhagen (UZM). Geographic data for all Brazilian specimens examined are given in the Gazetteer (Appendix A). For each species, we have used the following subheadings: Type material and etymology, Emended description, Taxonomy, Geographic distribution, Karyotype, Age classes, Sexual dimorphism, Geographic and local variation, Natural history, Comparisons, Sympatric species, and Material examined. The section termed Emended description is based on both the material type (original description, skin and skull photographs) and on the other specimens examined. We follow the terminology and definitions of Voss and Jansa (2003) for external and skull morphology, except for the vibrissae, which follows Hershkovitz (1977), and for the stylar cusps of upper molars and shape of ectoflexus, which is based on Solari (2003). Our quantitative characteristics are based on Tate (1933), Lemos et al. (2000), Patton et al. (2000), and Voss et al. (2001). The following measurements were taken: HBL, head-andbody length; LT, length of tail; ear, length of ear from notch; HF, length of hind foot; Wt, weight; GSL, greatest skull length; CBL, condylobasal length; NAS, nasal length; NB, nasal breadth; LIB, least interorbital breadth; LPB, least postorbital breadth; ZB, zygomatic breadth; ROS, breadth of rostrum; PL, palatal length; PB, palatal breadth; MTR, maxillary toothrow; LM, length of the upper molar series (M1 M4); CB, upper canine breadth; CL, upper canine length; P3L, upper third premolar length; CD, cranial depth; BAB, breadth across bullae; BBB, breadth between bullae; BSB, breadth of single bulla; BBr, least breadth of pterygoid bridge; BB, breadth of braincase; MAD, length of the mandible; and LM1, length of the lower molar series (m1 m4) (Figure 1). External body measurements were taken from museum tags and field notes (where they were usually recorded to the nearest millimeter), whereas the 23 cranial measurements were taken with digital calipers to the nearest 0.01 mm. Measurements of the type specimen of T. karimii were taken from the original description (Petter 1968).

3 128 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. Figure 1 Dorsal, ventral and lateral views of the skull and lateral view of the mandible of Thylamys showing measurements used in the quantitative analysis (Figure modified from Solari 2003). Measurements of the type specimen of T. velutinus and of five specimens (UZM 164, 166, 167, 168 and Halle 360) from Lagoa Santa, MG were taken from Tate (1933). We also used the data provided by Robert S. Voss, who remeasured specimens UZM 164, 165, 166 and 168 from Lagoa Santa, MG. Karyotypic data for the specimens we surveyed in E.E. Urucuí-Una, PI, P.N. Serra das Confusões, PI, P.N. Grande Sertão Veredas, MG, and Fazenda Califórnia, MS, were provided by Ms. Roberta Paresque and Dr. Maria José de J. da Silva. To assess intrapopulation variation, we sorted specimens by sex and age class. Age classes were based on the denomination and tooth-replacement criteria of Tribe (1990), and are similar to those used by Solari (2003), except for age class 4, which showed an intermediate pattern (P3 and M4 erupting) as the typical one. Basic descriptive statistics include the mean, standard deviation, range observed, and sample size for each external and cranial measurement. Two-way ANOVA, performed to test for the existence of significant sexual dimorphism and age variation in cranial variables, was applied to our largest sample (T. karimii from E.E. Urucuí- Una, PI). Evaluation of geographic variation within each taxon was not generally possible owing to small sample sizes. Differences between Thylamys species were evaluated by discriminant function analyses (DFA) based on cranial measurements of the adult specimens analyzed, and are summarized in the Statistical comparisons section. Statistical analyses were performed with STATISTI- CA software (StatSoft 2001). Results and discussion: Brazilian species of Thylamys We recognize three valid species of Thylamys in Brazil: T. karimii (Petter, 1968), T. macrurus (Olfers, 1818) and T. velutinus (Wagner, 1842). The specific names have been modified to agree with the generic name in accordance with the International Code of Zoological Nomenclature (ICZN 1999; see Solari 2003). These three taxa can be distinguished by a combination of skull and external morphology, and size characteristics, as described below. We also provide data on T. pusillus (Desmarest, 1804), an extralimital taxon, to differentiate it from the Brazilian species that were once considered to be its junior synonyms. Thylamys karimii Marmosa karimii Petter, 1968 in Mammalia 32(3):313. Original description. Type locality, Nord-Est du Brésil (région d Exu, Pernambuco), wbrazilx.

4 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. 129 Type material and etymology The type specimen is an adult male, preserved as a round skin with skull and hemimandibles, all of which are in good condition. It is housed in the Muséum National d Histoire Naturelle as number The name karimii honors Dr. Y. Karimi, head of the expedition that collected the holotype. The collection site (locality 21, Figure 7) is in the Caatinga, an open and xeric vegetation formation. Although we have not examined the type specimen, photographs of the skin and skull of dorsal, ventral and lateral views were provided to us. Emended description Thylamys karimii is a small species (Table 1) with brownish dorsal coloration, with the generic tricolor pattern inconspicuous, just paling slightly on the sides (Figure 2). The central portion of the dorsum is composed of dark brownish guard hairs and four-color banded hairs. The guard hairs are sparse and long (8 9 mm), gray-based (1/3 of the fur length) with dark distal portions (2/3 of the fur length). The banded hairs are very abundant and shorter (6 7 mm), graybased (4/6), changing to brownish in the middle (1/6), then cream-colored in the distal portion (1/6), and with black tips. On the flanks, the guard hairs are fewer and brighter, and the banded hairs lack the brownish middle portion. The ventral pelage is shorter (3 4 mm), creamywhite, with self-colored hairs occupying most of the underparts from the chin and cheeks to the inguinal region (Figure 2). A gular gland is present on both males and females, and is moderately conspicuous as a patch of almost-naked skin in the midline of the throat. Lateral gray-based incursions onto the venter vary in extent (3 8 mm wide) and are formed by gray-based hairs with creamy-white distal portions. According to the proportion of gray in the fur, these incursions can be faint or conspicuous. Generally, the lateral incursions are narrow and faint. The cranial pelage between the eyes and in front of the ears is brighter, composed of creamy-white-based hairs with dark distal portions, and three-color banded hairs (gray-based, creamy in the middle, and with dark tips). A narrow midrostral streak of brownish fur is present in most individuals. The eye ring is dark brownish and very narrow, extending slightly towards the snout and the ears. Supraorbital, suborbital, ciliary and mystacial vibrissae are mostly black, although some ventrally located hairs have white tips. By contrast, the genal, interramal, mental, digital, antebrachial, and ulnar carpal vibrissae are white. The ears are uniformly brownish and are covered by short hairs that are not visible without magnification. The forelimbs are creamy-white from dorsal and ventral views, with hairs that are self-colored as on the belly. The hindlimbs, however, are brownish on the dorsum and creamy-white ventrally. The dorsal surfaces of the fore- and hindfeet are covered with creamy-white hairs, and all of the toes have ungual tufts that extend to nearly 1/2 to 2/3 from the claw base (or apical pad in the case of plantar digit I). The manus and pes are very delicate and small (Table 1), with short toes, and long but delicate claws that extend beyond the fleshy apical pads. The palmar and plantar surfaces are densely tubercular (composed of small granules), and lack dermatoglyphs. Three plantar pads are present on the manus; these consist of four interdigitals fused together to form one pad, plus the thenar and hypothenar pads, of which the hypothenar is slightly larger. Three plantar pads are present on the pes, apparently consisting of: (1) the thenar plus interdigital 1 fused to form one pad; (2) interdigitals 2, 3, and 4 fused to form one pad; and (3) the hypothenar (see Figure 3 for plantar views of both the manus and pes). The tail length is less than the combined length of the head and body (68 88%), and less than 1 cm is furry at the base. The tail is incrassated, mm in diameter, and is slightly bicolored. The dorsal caudal hairs are cream-based (1/3), with a brownish distal portion (2/3), and some have white tips. The ventral caudal hairs are self-colored. The caudal scales are arranged in annular series, with three hairs per scale; the central hair is slightly thicker and longer than the lateral hairs of each triplet, varying from 3 to 5 scales long. The prehensile tail-tip (sulcus) is inconspicuous, at 1.5 mm long (Figure 4). The scrotal epidermis is darkly pigmented and covered with creamy-white hairs. Females do not have a pouch, and nine mammae are arranged circularly in the abdominal region (4-1-4 in three females: MZUSP 30566, 32247, 32250; and in one female: MZUSP 32246). The skull has a short and broad rostrum (Figure 5). The interorbital region is not greatly narrowed and generally has beaded supraorbital margins or pointed supraorbital ridges, giving rise to a well-marked postorbital constriction. The lambdoidal crests are weakly developed. The nasals are long and wide, slightly expanded at the maxillo-frontal suture, and conspicuously narrowing behind it. The lacrimal foramina are apparent and laterally exposed in the orbital margins. The zygomatic arches are expanded and the external surface of the jugals exhibits a conspicuous concavity along the anterior margin of the zygomata. The petrosals are exposed laterally in a fenestra between the squamosal and parietal bones. The palate is highly fenestrated, with maxillopalatine, palatine, and maxillary openings. The posterolateral palatal foramina are oval and very large, always extending anteriorly well beyond the lingual apices of the protocones of the fourth molars (Figure 6). The incisive foramina exhibit anterior and posterior margins that parallel and extend posteriorly between the canines. The auditory bullae are large and inflated. The upper incisors increase in size from I2 to I5, and P3 is taller than P2. The canines vary in size, being slender and long in some specimens, and short and robust in others (a variation related to age and sex), almost always with small posterior accessory cusps (Table 4). The molars are compressed antero-posteriorly, especially the fourth one (length of postmetacrista/length of postprotocristas ; width M4/width M1s ). Stylar cusp C is developed on M1 M3, but is smaller than stylar cusps B and D; the ectoflexus is serrated (the stylar shelf has a serrated profile in lingual view) (Figure 6). Solari (2003) examined specimens USNM (cited as T. velutinus) and described a lack of ungual tufts on the toes and the absence of stylar cusp C on the upper molars, characterizing a notch-shaped ectoflexus (or V-shaped). We also examined these specimens, and disagree with these observations. Taxonomy Following the description of Marmosa karimii (Petter 1968), this species was recognized by several

5 130 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. Table 1 Descriptive statistics for external and cranial variables of adult males and females (age classes 6 and 7) of T. karimii, T. velutinus, T. macrurus and T. pusillus. T. karimii T. velutinus T. pusillus T. macrurus ns7 (1f, 6m) ns8 (3f, 4m, 1i) ns39 (19f, 18m, 2i) Holotype ns12 (3f, 6m, 3i) Holotype (4ad, 3oad) (2ad, 5oad, 1i) (21ad, 14oad, 4adi) (m, adi) (3ad, 2oad, 7adi) (m, adi) Paraguay MS and Paraguay BA, DF, GO, MG, PE DF, GO, MG, SP SP MT, PI, TO HBL, mm "14.87 (32) "9.86 (7) "9.31 (6) "10.22 (5) ( ) ( ) ( ) ( ) LT, mm 79.94"7.83 (32) "8.38 (7) "9.61 (6) "6.72 (6) ( ) ( ) ( ) ( ) Ear, mm 20.55"2.26 (30) "2.04 (7) 21.42"1.56 (6) 23.50"2.34 (6) ( ) ( ) ( ) ( ) HF, mm 11.36"1.25 (32) "0.83 (8) "0.76 (6) 17.50"0.87 (5) ( ) ( ) ( ) ( ) Wt, g 28.43"7.59 (30) 24.30"11.45 (3) 25.03"6.15 (6) 38.83"9.22 (6) ( ) ( ) ( ) ( ) GSL, mm 28.91"1.41 (34) "1.99 (6) 26.83"0.94 (5) 32.02"1.46 (6) ( ) ( ) ( ) ( ) CBL, mm 28.24"1.44 (34) 27.39"2.29 (6) 26.31"0.83 (5) 31.47"1.41 (6) ( ) ( ) ( ) ( ) NAS, mm 12.16"0.71 (33) 11.08"0.59 (7) 11.01"0.40 (7) 13.60"1.06 (5) ( ) ( ) ( ) ( ) NB, mm 2.88"0.27 (34) 2.55"0.37 (10) "0.21 (7) 2.96"0.32 (7) ( ) ( ) ( ) ( ) LIB, mm 3.98"0.32 (35) "0.18 (8) 4.00"0.29 (6) 5.14"0.56 (7) ( ) ( ) ( ) ( ) LPB, mm 4.91"0.28 (35) "0.24 (10) "0.24 (6) 5.33"0.22 (7) ( ) ( ) ( ) ( ) ZB, mm 16.06"0.97 (34) "1.10 (6) "0.46 (6) 17.63"1.17 (6) ( ) ( ) ( ) ( ) ROS, mm 4.99"0.37 (34) 4.59"0.65 (4) 4.38"0.42 (7) 5.24"0.47 (7) ( ) ( ) ( ) ( ) PL, mm 15.08"0.76 (35) 14.26"0.93 (10) "0.79 (7) 16.61"0.68 (7) ( ) ( ) ( ) ( ) PB, mm 9.11"0.43 (35) 8.86"0.39 (6) 8.24"0.32 (7) 10.05"0.58 (7) ( ) ( ) ( ) ( ) MTR, mm 10.76"0.34 (34) 10.53"0.32 (9) 10.13"0.29 (7) 12.38"0.42 (7) ( ) ( ) ( ) ( ) LM, mm 5.40"0.21 (35) 5.34"0.19 (9) 5.09"0.08 (7) 6.14"0.21 (7) ( ) ( ) ( ) ( ) CB, mm 1.17"0.16 (34) 1.26"0.22 (5) 1.14"0.18 (7) 1.48"0.11 (7) ( ) ( ) ( ) ( ) CL, mm 2.05"0.31 (34) 2.25"0.51 (5) 1.97"0.30 (7) 2.43"0.37 (7) ( ) ( ) ( ) ( ) P3L, mm 1.51"0.16 (34) 1.50"0.28 (5) 1.26"0.05 (7) 1.76"0.33 (7) ( ) ( ) ( ) ( ) CD, mm 9.97"0.36 (33) 9.59"0.52 (4) 9.20"0.28 (5) 10.26"0.44 (6) ( ) ( ) ( ) ( ) BAB, mm 10.33"0.42 (34) 9.68"0.37 (5) "0.34 (5) "0.58 (6) ( ) ( ) ( ) ( ) BBB, mm 4.25"0.29 (33) 3.90"0.42 (3) 4.16"0.26 (5) 5.20"0.36 (6) ( ) ( ) ( ) ( ) BSB, mm 3.19"0.17 (19) 2.81"0.31 (5) "0.08 (5) 3.11"0.08 (3) ( ) ( ) ( ) ( ) BBr, mm 2.22"0.12 (34) 2.33"0.12 (4) 2.32"0.17 (6) 2.71"0.14 (6) ( ) ( ) ( ) ( ) BB, mm 10.87"0.49 (34) 10.33"0.42 (7) "0.26 (5) 12.09"0.38 (6) ( ) ( ) ( ) ( ) MAD, mm 20.71"2.15 (35) 19.88"1.46 (7) "0.80 (7) 23.22"1.56 (7) ( ) ( ) ( ) ( ) LM1, mm 6.16"0.17 (35) 6.01"0.15 (8) "0.11 (7) 6.93"0.17 (7) ( ) ( ) ( ) ( ) Measurements are presented as mean"sd (n) (range). Abbreviations: f, female; m, male; i, sex indeterminate; ad, adult (age class 6); oad, old adult (age class 7); adi, adult age indeterminate. authors (Pine et al. 1970; Kirsch and Calaby 1977; Creighton 1984; Reig et al. 1985, 1987). Kirsch and Calaby (1977), having adopted the subgeneric arrangement of Marmosa proposed by Cabrera (1958), included M. karimii in the subgenus Thylamys. Creighton (1984) and Reig et al. (1985, 1987) recognized Thylamys as a full

6 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. 131 Figure 2 Dorsal and ventral views of skins of three Brazilian species of Thylamys. From left to right: T. karimii (MZUSP 30557, E.E. Urucuí-Una, PI), T. velutinus (UNB 1463, Reserva Ecológica do IBGE, DF) and T. macrurus (MZUSP 32094, Fazenda Califórnia, MS). genus and included karimii as a valid species. Gardner and Creighton (1989), however, did not cite the name karimii when listing their five recognized species under Thylamys. The name karimii reappears in the work of Gardner (1993), where, for the first time, it was considered a synonym under T. pusilla (sic). Although the section named synonyms in Gardner (1993) lists all names ascribed to a particular taxon, including true synonyms as well as subspecific names (de Vivo 1996), in the case of karimii, the name was probably listed as a true synonym, since it was never placed as a subspecies under any Thylamys species. Since T. pusillus is characterized by a tail that is longer than the head-and-body (see Desmarest 1804), which is not the case for T. karimii (Petter 1968), Palma (1995a) disagreed with the decision of Gardner (1993). Instead, based on the agreement of anatomical and cranial characteristics of populations occurring in northeast, central and southeast Brazil, Palma (1995a) concluded that these forms were conspecific and applied the name T. velutinus to them. Palma s synonymization of T. karimii under T. velutinus was then followed by some authors (Bonvicino et al. 1996; Palma and Yates 1998; Carvalho et al. 2002; Palma et al. 2002; Solari 2003; Braun et al. 2005), but Eisenberg and Redford (1999) considered T. karimii as a valid species following the catalog of marsupial types in the Muséum National d Histoire Naturelle (Paris, France) of Julien-Laferrière (1994). Since neither publication provides any additional taxonomic comments, they have not been followed by subsequent authors. In the most recent list of didelphimorphian species (Gardner 2005), T. karimii is attributed the status of a valid species without explanation. Geographic distribution These taxonomic problems have obviously confused the geographic distribution of T. karimii. Publications in which T. karimii was synonymized with T. pusillus (e.g., Gardner 1993) described a broader range for the latter, erroneously extending its presence to Brazil. On the other hand, authors who synonymized T. karimii with T. velutinus (e.g., Palma 1995a; Bonvicino et al. 1996; Palma and Yates 1998; Carvalho et al. 2002; Palma et al. 2002; Solari 2003; Braun et al. 2005), described a wider distribution for the latter in Brazil that, under our concept of two distinct species, is shown to be wrong (below). Owing to a lack of detailed study of specimens in collections worldwide, even authors who have recently considered T. karimii as a valid species (Julien-Laferrière 1994; Eisenberg and Redford 1999; Gardner 2005) do not mention the record of Pine et al. (1970), thus restricting the range of the species to its type locality. In addition, some records attributed to T. karimii have been based on misidentifications (e.g., Mares et al. 1981; Streilein 1982a,b). The geographic distribution of T. karimii as recognized herein includes the open (non-forest) formations of the Cerrado and Caatinga biomes in northeastern, southeastern, and Central Brazil (Figure 7). It occurs in the states of Rondônia, Mato Grosso, Tocantins, Piauí, Pernambuco, Bahia, Goiás, and Minas Gerais. Among the Brazilian species of Thylamys, T. karimii exhibits the broadest range and shows an overlap in range with T. velutinus in Central Brazil. Karyotype This species is characterized by a chromosome diploid number of 2ns14, but the fundamental number seems to vary geographically. We observed FNs20 in three specimens (MZUSP ) from P.N. Grande Sertão Veredas (locality 14, Figure 7), in six specimens (MZUSP and ) from E.E. Urucuí-Una (locality 22, Figure 7), and in two specimens (MZUSP ) from P.N. Serra das Confusões (locality 23, Figure 7). Carvalho et al. (2002) reported FNs24 for two individuals surveyed at Niquelândia (locality 6, Figure 7). Age classes We observed strong variation in both qualitative and quantitative characteristics between the different age classes in each T. karimii sample analyzed. Two-way ANOVA performed with 20 cranial measurements of one sample (E.E. Urucuí-Una, PI) confirmed these observations, showing significant age-related differences in 13 cranial variables (Table 2). These results

7 132 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. Figure 4 Ventral view of the tail and tail-tip of the same specimens of T. karimii (a), T. velutinus (b) and T. macrurus (c). Figure 3 Ventral view of the right manus and pes of (a) T. karimii (MZUSP 32246, E.E. Serra Geral do Tocantins, TO), (b) T. velutinus (MZUSP 32098, P.N. das Emas, GO) and (c) T. macrurus (MZUSP 32097, Fazenda Santa Terezinha, MS). Note the different scale bars. coincide with those reported by Solari (2003) for Peruvian samples of T. pallidior and Chilean T. elegans. Old adults (age class 7) have the largest skulls, with maximally developed supraorbital ridges and canines (Table 2). The presence of accessory canine cusps is also a variable trait, the most common being the presence of posterior accessory cusps (Table 4). This feature, however, may be age-related, since accessory cusps are often absent in juveniles (age classes 1 4), and less frequent in old adults due to wearing of their teeth. Sexual dimorphism Qualitative analysis of individuals from E.E. Urucuí-Una, PI showed a strong variation between the sexes in ventral coloration. Females presented a yellowish throat and belly, with broader and more conspicuous lateral gray-based incursions. In general, the males possessed whitish venters. These differences seem to be more conspicuous during the rainy season, which is the probable reproductive period of T. karimii in this region. Specimens from other localities seem to agree with these general patterns of sexual dimorphism. Two-way ANOVA revealed no marked sexual dimorphism in craniodental measurements, since only two variables (LIB and CL) exhibited significant differences (Table 3). Interaction effects between age and sex were significant for only three variables (LIB, LPB and LM) (Table 3). When we analyzed just the adults (age classes 6 and 7), we observed significant sexual dimorphism for three variables (LIB, CB and CL), and a significant interaction effect for only one (LIB). Solari (2003) also found weak sexual dimorphism in T. tatei (one variable) and none in T. pallidior. However, T. elegans males were larger than females in 11 variables. In T. karimii, the larger supraorbital ridges and length of canines of the males may also play a significant role in sex differentiation. Geographic and local variation Within samples, the presence of the dark midrostral stripe is highly variable, as is lateral exposure of the petrosals. The supraorbital ridges also vary in size, with some individuals exhibiting more pointed ridges and others with just beaded supra-

8 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. 133 Figure 5 Dorsal, ventral and lateral views of skull and lateral view of the mandible from (a) T. karimii (MZUSP 30581, GSL mm, E.E. Urucuí-Una, PI), (b) T. velutinus (MN 66461, GSL mm, P.N. Chapada dos Veadeiros, GO) and (c) T. macrurus (MZUSP 32094, GSL mm, Fazenda Califórnia, MS). orbital margins. Among geographic samples, the color of the dorsum can be more grayish, with the brownish portion of the fur narrowed, or dark brownish, with the creamy portion of the fur reduced in length. Specimens from sandy-soil areas (localities 14, 22, 23 and 27, Figure 7) are usually paler (brown-creamy dorsum) than specimens from other areas of the Cerrado. The conspicuous tricolor pattern found in specimen BMNH from Serra do Roncador (locality 20, Figure 7) could be due to the way the skin was preserved (not filled with cotton); the other specimens (all stuffed with cotton) are not conspicuously tricolored. In relation to the venter, two specimens (UNB 1153 and 1158) from Fazenda Bandeirantes (locality 7, Figure 7) exhibit the widest and most conspicuous lateral gray-based incursions on the ventral pelage, while the remaining specimens exhibit narrow or faint incursions. Body size and the morphology of the posterolateral palatal foramen also showed geographic variation, with the sample from P.N. Serra das Confusões (locality 23, Figure 7) being larger, with narrower foramina. One individual (MZUSP 32232) from P.N. Grande Sertão Veredas (locality 14, Figure 7) exhibits anterior accessory canine cusps, and also exhibits greater bullae size than other samples. Larger samples from each locality are needed to investigate the taxonomic status of these populations. Natural history All known specimens of T. karimii have been collected in the Caatinga and Cerrado biomes, where they appear to inhabit only open habitats, as Figure 6 Occlusal surface of the left upper molar series (M1 M4) of T. karimii (upper, MZUSP 30557, LM 5.29 mm, E.E. Urucuí-Una, PI), T. velutinus (middle, MN 66461, LM 5.15 mm, P.N. Chapada dos Veadeiros, GO) and T. macrurus (bottom, MZUSP 32095, LM 5.89 mm, Fazenda Califórnia, MS), showing the level of development of stylar cusp C and the posterolateral foramen in each taxon. attested by data extracted from the examined specimens. The Cerrado biome is located in the highlands of Central Brazil and is characterized by a seasonal and tropical climate, with dry winters and wet summers. The average annual rainfall is approximately mm, and the dry period is restricted to 3 months with a mean monthly precipitation of 30 mm; the mean annual temperature is approximately 228C (Ribeiro and Walter 1998). The vegetation is characterized by a complex mosaic of habitats ranging from open grasslands to dense arboreal savannas and woodlands, with gallery forests along watercourses (Oliveira-Filho and Ratter 2002). In the Cerrado, T. karimii was captured in areas of campo limpo (open grassland), campo sujo (grassland with some trees and shrubs), campo cerrado (shrubby vegetation) and cerrado (arboreal dense savanna). The Caatinga in northeastern Brazil is a semi-arid biome, with a hot and dry climate determined by the extremely seasonal rainfall regime. The mean annual rain-

9 134 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. Figure 7 Known collecting localities of Thylamys species in Brazil. The gazetteer is listed in Appendix A. fall is approximately mm and the dry season is longer than it is in the Cerrado, lasting from 5 to 6 months; the mean annual temperature is approximately 288C (Rizzini 1997). The vegetation can be characterized as a predominantly dry, shrubby forest, with a humid forest along watercourses (Hueck 1972). In the Caatinga, we collected T. karimii in areas of arboreal and low caatinga (dry forest). We recorded lactating females during the months of January (one specimen) and April (five specimens) in the rainy season. However, young individuals were found in both the dry (one specimen) and wet seasons (10 specimens), suggesting that this species may breed throughout the year. Since specimens analyzed from both seasons exhibit tail incrassation, we believe that this characteristic is not a seasonal pattern. Indeed, we have not collected any specimens without an incrassate tail. Comparisons The only species of the genus that closely resembles T. karimii is T. velutinus, which also has a tail that is shorter than the combined head-and-body length, and is similar in foot size and morphology (Tables Table 2 Cranial measurements of four different age classes of T. karimii from E.E. Urucuí-Una, PI. Young (ns2) Subadult (ns3) Adult (ns11) Old adult (ns7) p Age class 4 Age class 5 Age class 6 Age class 7 GSL, mm 25.92" " " "0.37 *** CBL, mm 25.11" " " "0.36 *** NB, mm 2.50" " " "0.10 * LIB, mm 3.74" " " "0.08 ** LPB, mm 5.64" " " "0.08 ** ZB, mm 14.42" " " "0.26 *** ROS, mm 4.72" " " "0.14 n.s. PL, mm 13.56" " " "0.20 *** PB, mm 8.46" " " "0.14 ** MTR, mm 10.04" " " "0.11 ** LM, mm 5.07" " " "0.08 n.s. CB, mm 1.02" " " "0.04 ** CL, mm 1.94" " " "0.07 n.s. CD, mm 9.46" " " "0.09 *** BAB, mm 9.87" " " "0.14 * BBB, mm 4.02" " " "0.09 n.s. BBr, mm 2.12" " " "0.06 n.s. BB, mm 10.41" " " "0.17 n.s. MAD, mm 18.46" " " "0.36 ** LM1, mm 6.38" " " "0.06 n.s. Measurements are presented as mean"sd. Probability level (p) for age differences are given as: n.s. (non-significant) p)0.05; *p-0.05; **p-0.01; ***p

10 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. 135 Table 3 Cranial measurements of male and female specimens of T. karimii from E.E. Urucuí-Una, PI. Age classes 4 7 Age classes 6 and 7 Female (13) Male (10) p =p Female (10) Male (8) p GSL, mm 27.17" "0.38 n.s. n.s " "0.35 n.s. CBL, mm 26.52" "0.38 n.s. n.s " "0.35 n.s. NB, mm 2.71" "0.10 n.s. n.s. 2.84" "0.09 n.s. LIB, mm 3.68" "0.08 * * 3.85" "0.08 * LPB, mm 5.00" "0.08 n.s. * 4.90" "0.08 n.s. ZB, mm 15.32" "0.27 n.s. n.s " "0.25 n.s. ROS, mm 4.90" "0.14 n.s. n.s. 5.00" "0.13 n.s. PL, mm 14.33" "0.21 n.s. n.s " "0.19 n.s. PB, mm 8.85" "0.14 n.s. n.s. 9.08" "0.13 n.s. MTR, mm 10.32" "0.12 n.s. n.s " "0.11 n.s. LM, mm 5.21" "0.08 n.s. * 5.32" "0.07 n.s. CB, mm 1.04" "0.04 n.s. n.s. 1.12" "0.03 ** CL, mm 1.83" "0.07 ** n.s. 1.86" "0.06 ** CD, mm 9.73" "0.09 n.s. n.s. 9.91" "0.09 n.s. BAB, mm 9.97" "0.15 n.s. n.s " "0.13 n.s. BBB, mm 4.19" "0.09 n.s. n.s. 4.31" "0.07 n.s. BBr, mm 2.18" "0.06 n.s. n.s. 2.22" "0.06 n.s. BB, mm 10.61" "0.17 n.s. n.s " "0.16 n.s. MAD, mm 19.35" "0.38 n.s. n.s " "0.35 n.s. LM1, mm 6.24" "0.06 n.s. n.s. 6.16" "0.06 n.s. Measurements are presented as mean"sd. Probability level (p) for sex differences and for the interaction between age classes and sex (=p) are given as: n.s. (non-significant), p)0.05; *p-0.05; **p and 5; Figure 3). However, these species differ greatly in dorsal and ventral coloration, and are easily distinguished from each other (Figure 2): T. karimii is brownish dorsally and whitish ventrally, whereas T. velutinus is reddish-brown dorsally and grayish ventrally. In addition, the skull of T. velutinus differs from that of T. karimii by exhibiting a narrowed interorbital region, a lack of supraorbital ridges, and smaller posterolateral palatal foramina (Figures 5 and 6). Sympatric species Other small marsupial species that may occur in the same geographic region as T. karimii belong to the genus Gracilinanus and Cryptonanus (Voss et al. 2005), which can be readily distinguished from Thylamys by tails that are as long as or longer than head-and-body, are not incrassated, and have longer prehensile tips; palmar and plantar surfaces of the feet with fewer and larger granules; longer toes with short claws; and smaller posterolateral palatal foramina. Species of Monodelphis can be recognized by their larger feet and toes; palmar and plantar surfaces with fewer and larger granules, and with dermatoglyphs; absence of eye rings; skull with a broader rostrum; prominent sagittal crest; lack of the strut of the tympanic wing of the alisphenoid; and palate not highly fenestrated (lacking maxillary and palatine openings). Material examined A total of 86 specimens and two photographs were examined: s/s, skin and skull; si, skin; su, skull; f, fluid; p, skin and skull picture. BA. Cocorobó (ns1), MZUSP f; Fazenda Fundão, Cocos (ns5), UNB 1487 s/s, s/s, 1509 su, 1893 si; Fazenda Jatobá, Correntina (ns3), UNB 1843 s/s, 1851 si, 1859 si; DF. Estacão Ecológica Águas Emendadas, Planaltina (ns1), UNB 1228 su; GO. AHE Serra da Mesa, 55 km N de Niquelândia (ns3), MN su, s/s, si; Fazenda Bandeirantes, Baliza (ns2), UNB 1153 si, 1158 s/s; Reserva SAMA, Minacu (ns1), MZUSP s/s; UHE Caldas Novas, Caldas Novas (ns6), MN si, s/s, si, si, si; MG. Fazenda Brejão, Brasilândia de Minas (ns2), UFMG si; Parque Nacional Grande Sertão Veredas, Formoso (ns8), MZUSP s/s, s/s, 32237s/s, su, f, f, f; Serra Cabral, Pirapora (ns1), MZUSP 3085 si; MT. 264 km N de Xavantina, Serra do Roncador (ns4), USNM s/s, BMNH p; PE. Exu (ns1), MNHN p; PI. Estacão Ecológica Urucuí-Una, Baixa Grande do Ribeiro (ns27), MZUSP s/s, su/f, f, su/f, s/s, f, s/s, su/f, s/s, s/s; Parque Nacional Serra das Confusões, Guaribas (ns12), MZUSP s/s, s/s, su/f, si, f; RO. BR-364, km 55 (ns1), MZUSP su/f; TO. Guaraí (ns2), MZUSP f; Estacão Ecológica Serra Geral do Tocantins, Ponte Alta do Tocantins (ns8), MZUSP s/s, s/s, s/s, f, f. Thylamys velutinus Didelphys velutina Wagner, 1842 in Archiv für Naturgeschichte 8:360. Original description. Type locality Ypanema, wsão Paulo, Brazilx. Type material and etymology The type specimen is an adult male, preserved as a round skin with skull and hemimandibles, all of which are in reasonable condition. It is housed in the Naturhistorisches Museum, Wien as number B This locality (locality 26, Figure 7) is situated in southeastern Brazil, in an area of transition between the Atlantic rainforest and the Cerrado biome. There are no habitat data associated with the type mate-

11 136 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp. rial. The name velutinus is derived from the Latin word for velvety (Brown 1956), in reference to the texture of its pelage. We have examined photographs of the type specimen, including dorsal, ventral and lateral views of the skin and skull. Emended description Thylamys velutinus is also a small species (Table 1), but with a dark reddish-brown dorsal pelage that lacks any apparent tricolor pattern (Figure 2). The dorsum is composed of guard and fourbanded cover hairs. Guard hairs are longer (10 11 mm) and fewer, and are gray-based (1/3 the length of the fur) with dark distal portions (2/3). The banded hairs are abundant and shorter (6 8 mm), gray-based (2/3) changing to brown, and creamy with orange to dark-brown tips (1/3). The lateral fur is similar to or slightly brighter than the dorsal fur, and the brown portions of the lateral hairs are absent or shorter than the corresponding portions of the dorsal hairs. The ventral pelage is shorter (5 6 mm) and grayish, paler than the dorsum (Figure 2). The coloration of the venter varies according to the gray-based proportion of the fur: it is paler when the gray portion is short, and grayish when it is long. All hairs are graybased ventrally, except on the chin, cheeks, and throat, where the hairs are creamy-white to the base. A throat gland is present in adult males as a large patch of sparse, short, discolored hairs. The head is paler than the rest of the dorsum, with creamy-white-based hairs with brown tips, and three-banded hairs: gray-based and creamywhite in the middle, with brown tips. There is a dark midrostral streak in some individuals. The eye ring is narrow and dark, and extends to the snout as a stripe of brownish hairs. The vibrissae are as in T. karimii: dark in the dorsal and white in the ventral regions. The ears are uniformly dark brown and appear naked. The fore- and hindlimbs are grayish brown dorsally and ventrally, and the feet are covered with short white hairs. The feet are small (Table 1), with short toes bearing ungual tufts and claws that surpass the fleshy apical pads. Most of the palmar and plantar surfaces are densely covered with small granules, but dermatoglyphs are present on the apex of tiny vestigial pads. There are three distinct pads, as in T. karimii. On the palmar surface, however, the interdigital pads are almost disjunct, with the thenar pad conspicuously smaller than the hypothenar, and not almost similar in size as in T. karimii (Figure 3). The tail is shorter than the head-and-body length, and less than 1 cm is furry at its base; it is extremely incrassated ( mm in diameter) and slightly bicolored: whereas the dorsal hairs are dark-brown-based with paler tips, the ventral hairs are white. The caudal scales are arranged in annular series, and are provided with three hairs each; the middle hair of each caudal scale triplet is slightly wider and longer (5 scales long) than the others (3 scales long). The tip of the tail (approx. 3 mm) has no hair (Figure 4). The scrotal epidermis is darkly pigmented and covered with white hairs. Females lack a pouch. The skull has a short and broad rostrum (Figure 5). The interorbital region is very narrow and lacks supraorbital ridges, exhibiting beaded margins that converge and reach the parietals posteriorly; a well-marked postorbital constriction is also evident. The lambdoidal crests are only weakly developed. The nasals are long and present only slight expansion at the maxillo-frontal suture, then narrowing again slightly behind it. The lacrimal foramina are large and visible laterally in the orbital margins. The zygomatic arches are only slightly expanded, and the external surface of the jugals exhibits a conspicuous concavity along the anterior margin of the zygomata. The petrosals are exposed laterally in a fenestra between the squamosal and parietal bones. The palate is highly fenestrated, with maxillopalatine, palatine, and maxillary openings. The large posterolateral palatal foramina extend lingual to the M4 protocones, but rarely exceed them anteriorly (Figure 6). The incisive foramina exhibit parallel margins, and extend posteriorly to the midline of the canines. The auditory bullae are smaller than those of T. karimii (Table 1). The incisors increase in size from I2 to I5, and P3 is taller than P2. The canines are slender and long, with or without posterior accessory cusps (Table 4). The molars are compressed antero-posteriorly, especially M4 (length of metacrista/length of postprotocristas , width of M4/width of M1s1.0). Stylar cusp C is developed on the upper molars, particularly on M1 M3, but is smaller than stylar cusps B and D; the ectoflexus is serrated (the stylar shelf has a serrated profile in lingual view) (Figure 6). Taxonomy Tate (1933), based on his examination of the type material, considered Marmosa velutina (Wagner Table 4 Frequency of the presence of anterior and posterior accessory cusps on the upper canine in each age class of Brazilian Thylamys specimens examined. Species Age Sample Accessory cusp Frequency with class accessory cusp None Anterior Posterior Both T. karimii Total T. velutinus Total T. macrurus Total

12 A.P. Carmignotto and T. Monfort: Taxonomy and distribution of Brazilian Thylamys spp ) to be a valid species. He also examined six specimens from Lagoa Santa, Minas Gerais, including the holotype of Didelphis pimelura Reinhardt, 1849, which he included as a junior synonym of M. velutina. He provided a detailed description of the specimens he examined, accompanied by external and skull measurements, clearly showing the diagnosable characteristics of this species, such as a tail length shorter than the head-andbody length, gray-based ventral coloration, and a skull with a very narrow interorbital region and lack of supraorbital ridges. We follow the decision of Tate (1933) based on his taxonomic considerations, and on photographs examined of the Lagoa Santa material (UZM 164 and 166). Robert S. Voss has also examined these specimens (UZM 164, 165, 166 and 168) and confirmed their identity as members of Thylamys (R.S. Voss, personal communication). Cabrera (1958) considered two subspecies under Marmosa velutina: M. v. formosa from northern Argentina, and M. v. velutina from southeastern Brazil. The taxonomic status of M. formosa Shamel, 1930, once considered a junior synonym of Gracilinanus agilis (Hershkovitz 1992; Gardner 1993), was resolved by Voss et al. (2004a), who considered it a valid species of a new genus, Chacodelphys. In relation to T. velutinus, however, the taxonomic situation is not as clear, as shown in the T. karimii section. Although T. velutinus was never treated as a synonym or subspecies under other taxa, other species were included under its name, such as formosa (Cabrera 1958; Streilein 1982c) and karimii (Palma 1995a; Bonvicino et al. 1996; Palma and Yates 1998; Carvalho et al. 2002; Palma et al. 2002; Solari 2003; Braun et al. 2005), leading authors to recognize a broader and erroneous geographical range. Geographic distribution Gardner (2005) restricted this species to southeastern Brazil. However, ecological studies conducted in Central Brazil point to the presence of T. velutinus in that region, indicating a major extension of its geographic distribution (Vieira and Baumgarten 1995; Vieira and Palma 1996; Vieira 1999; Rodrigues et al. 2002; Bonvicino et al. 2003). Misidentification of some specimens (Vieira 1949; Lyra-Jorge and Pivello 2001) also led to erroneous records in the geographical range of T. velutinus. As well as the range limits, data on habitat occurrence were also confused. Fonseca et al. (1996) cite the presence of T. velutinus in the Atlantic rainforest biome. However, the two localities previously known for T. velutinus (Ipanema, SP and Lagoa Santa, MG) are located in a transitional zone between the Cerrado and the Atlantic rainforest biome, with open and forested habitats intermingled in a mosaic ecological landscape. Since this species also occurs in areas of the Cerrado in Central Brazil, it is more likely that the specimens from Ipanema and Lagoa Santa were collected in open habitats that are really part of the Cerrado biome. In addition, recent inventories in areas of Atlantic rainforest using live and pitfall traps failed to capture any specimens of this genus (Patton and Costa 2003; R. Pardini and R.V. Rossi, personal communication). Based on specimens examined and apparently reliable literature records, T. velutinus has been collected in areas of the Cerrado in the interior of the states of São Paulo and Minas Gerais (Tate 1933) in southeastern Brazil, and also in the Central plateau in the Distrito Federal (Vieira and Palma 1996) and the state of Goiás (Rodrigues et al. 2002; Bonvicino et al. 2003) (Figure 7). This species appears to be rare and restricted to the southern and central portions of the Cerrado biome. In Central Brazil, its distributional range overlaps with that of T. karimii. Karyotype No chromosomal data have been published for this species. Age classes and sexual dimorphism Even considering the small sample size (ns6) examined for this species, we observed great age-related differences in general size, cranial morphology, and development of various processes, crests, and dental features. In the subadult of age class 5, the general size is smaller, the skull is also smaller and delicate, without marked processes; the lambdoidal crests are slightly developed and small, and the canines are triangular. In the old adult of age class 7, the skull is larger and more robust, with prominent beaded margins convergent and reaching the parietals; the lambdoidal crests are slightly developed, and the canines are slender and longer, with a more robust mandible. Since we do not know the sex of these two specimens, it is not possible to evaluate whether or not sexual dimorphism has also influenced these results. However, when we compared the only specimens analyzed with skin and skull preserved, from the same age classes and different sexes, the holotype male (Vienna 81 photograph) and the female UNB 1463, we noted that they are similar in general size, structure and size of skull, processes and canines. Thus, the pattern found for T. karimii probably applies to this species too. Geographic variation Despite the small sample size, we observed some variation related to skin coloration and skull morphology. The skins from Ipanema (locality 26, Figure 7) and Lagoa Santa (locality 13, Figure 7) are too old and not in a sufficiently good state of preservation to permit detailed comparisons. Skins from the remaining samples exhibit dark dorsal coloration, with grayish venters. The ventral coloration varies from dark gray to creamy gray, according to the length of the gray basal portion of the fur. The skull of MN from P.N. Chapada dos Veadeiros (locality 8, Figure 7) exhibits a nearly flat dorsal profile, greatly differing from the other specimens analyzed. In addition, one specimen (UNB 1463) from E.E. Jardim Botânico (locality 5, Figure 7) does not show a narrow interorbital region. The taxonomic status of these differences can only be investigated with larger samples. Natural history The specimens from Ipanema and Lagoa Santa from southeastern Brazil (localities 13 and 26, Figure 7) were obtained in 1821 (holotype) and 1849 (series analyzed by Tate 1933). The specimen from Cachoeira de Emas, Pirassununga (MZUSP 17079; locality 25, Figure 7) is a poorly preserved skull from None are accompanied by habitat data, but Ipanema and Lagoa Santa are located in transitional areas between the Atlantic rainforest and Cerrado biomes, and Pirassunun-