Micoureus demerarae (Thomas, 1905)

PATTON, JAMES L., DA SILVA, MARIA NAZARETH F. & MALCOLM, JAY R., 2000, Mammals Of The Rio Juruá And The Evolutionary And Ecological Diversification Of Amazonia, Bulletin of the American Museum of Natural History 2000 (244), pp. 1-306 : 66-74

Micoureus demerarae (Thomas, 1905)

TYPE LOCALITY: ‘‘Comaccka, 80 miles up Demerara River,’’ East­Demerara­West Coast Berbice, Guyana.

DESCRIPTION: A large­bodied murine opossum, this species is equivalent in size to others in the genus and cannot be distinguished in this respect from sympatric, or near­sympatric M. regina from the Rio Juruá (table 13). It has relatively long and distinctly ‘‘woolly’’ dorsal pelage, averaging 12 mm in length. One of the most distinctive features of this taxon is the extensively furred base of the tail (fig. 53). The fur extends onto the tail for an average length of 29 mm in specimens from the Rio Juruá (measurement tak­ en from dried skins), but may be considerably longer in those from eastern Amazonia and especially the Mata Atlântica. The dorsal color is gray­brown with a pale yellowish or buff wash, the black eye­ring is distinct, and the cheeks are pale buff. Ventrally, uniformly buff hairs are confined to the throat and inguinal region in most specimens, with a broad lateral strip of gray­based, buffy tipped fur extending from the lateral margins of the dorsal fur onto the venter, from the upper thorax posteriorly across the entire abdomen (fig. 54). This gray­based fur coalesces along the midventral region in most specimens (35 of 39 adults); in the remainder there is a narrow (<5 mm), purely buff stripe on the midline through the thoracic and abdominal regions. This buff coloration ranges from Light Buff to Warm Buff, or from Cartridge Buff to Cream Buff (Ridgway, 1912). Two specimens from locality 14 (Vira­Volta) near the mouth of the Rio Juruá have a distinctly pinkish cast to the midline ventral region that separates the lateral gray­based incursions other specimens from this and nearby localities are more typical in color and color pattern. These two individuals share the same cranial dimensions and long­furred tail­base of other M. demerarae . The tail is uniformly colored gray­brown along its entire length in all specimens from the Rio Jurua´; none have a terminal mottle of gray­brown and cream patches, as may be found in specimens elsewhere in the species range.

GEOGRAPHIC AND NONGEOGRAPHIC VARIA­ TION: We performed a 2­way ANOVA to examine the effects of geographic variation among samples of adult M. demerarae along the river as well as sexual dimorphism within individual samples. Although the power of this analysis is limited because of the relatively small sample sizes for each locality significant (p <0.05) geographic variation was found for eight mensural variables

(TAL, HF, E, IOC, NL, C­M4, PL, and PW), and within­locality sexual dimorphism was found for two external characters (TAL and HF). However, when all individuals are pooled to increase sample size, significant sexual dimorphism is apparent in all four ex­ ternal measurements and eight of 15 cranial variables (CIL, MB, IOC, RL, C­M4, PL BOL, and CD; p <0.05, one­way ANOVA)

COMPARISONS: Micoureus demerarae is extremely similar to sympatric, or near­sympatric M. regina throughout the Rio Juruá ba­

sin. The two share the same overall size and general body proportions, but differ in color and color pattern, particularly of the venter in the degree to which the base of the tail is furred, in external and cranial dimensions and in some qualitative cranial features. In ventral color pattern, the pure color area is confined to the throat and upper thorax, with gray­based lateral fur coalescing in the midventral region in M. demerarae but widely separated in M. regina (fig. 54). Micoureus demerarae has a significantly longer ear and furred tail­base (fig. 53 and table 13), but averages fewer scale rows per centimeter along the tail (10 versus 13 in regina ). While the two taxa are the same size in cranial length and depth, M. demerarae is consistently smaller in all other cranial dimensions, except BOL, even when the sexes are compared separately (table 13). The narrower interorbital breadth and palate of M. demerarae relative to M. regina are clearly evident in comparisons of representative skulls of the two species (fig. 55). These and other differences are also apparent in bivariate plots where overall size is represented by condyloincisive length (fig. 56). The slopes of the relationship between individual variables and CIL, a univariate size estimate, are not statistically different, although the Y­intercepts are, with that of M. demerarae significantly lower (ANCOVA, Fisher’s Protected LSD, p <0.05 in each comparison). The same patterns are present when the sexes are analyzed separately.

Principal components summaries of multivariate relationships indicate that PC­1 is a general size axis, with all cranial variables loading positive and highly (table 14). This axis accounts for 67.3% of the total variance PC­2 accounts for 11.9%. The dispersion of individuals along PC­2 is influenced mostly by the width of the skull posterior to the supraoccipital processes (IOC­2) and molar toothrow length (M1­M4), which separate specimens with the generally narrower interorbits and shorter toothrows, characteristics of M. demerarae , from the larger M. regina (table 14 and fig. 57, top). PC­3 accounts for 6.5% of the variation in the data, but all variables load relatively evenly on this axis and it does not serve to separate the two species (fig. 57, middle). In a two­group discriminant

analysis with samples allocated to each species identified a priori, the two taxa have completely nonoverlapping scores, with individual probabilities of group membership of 0.909, or higher (table 14; fig. 57, bottom).

DISTRIBUTION AND HABITAT: This species occurs throughout the Rio Juruá basin, as we took specimens in all four regional sampling areas. Most of these came from undisturbed terra firme forest or second­growth forest in upland, nonflooded areas. Only three of the 47 specimens collected in the Upper Central, Lower Central, or Mouth areas, where true várzea is present, were taken from seasonally flooded várzea forest. Both M. demerarae and M. regina were taken together at all four localities in the Headwaters region, including the quasi­várzea forest localities (opposite Porongaba, locality 2, and Nova Vida, locality 3). We also took both species in várzea at Boa Esperança (locality 9a) in the Lower Central section of the river. Of the 55 total specimens obtained, 50 were taken in canopy traps placed at heights between 6 and 17 m, two were shot at heights above 2 m, and three were taken in terrestrial traps.

REPRODUCTION: We found parous or reproductive females (those with orange inguinal mammary areas, attached young, or lactating) in the months of February, March, April September, October, and November. These months encompass both the rainy and dry seasons, suggesting that reproduction can occur throughout the year. All reproductive females were at least age class 5, with a completely erupted toothrow (following Gardner 1982). A single female from Barro Vermelho (locality 12) had 7 attached young. All males of age class 5 or older had testes at least 12 mm in length, with the average length 13.9 mm; all individuals of younger age had testes less than 9 mm.

KARYOTYPE: 2n = 14, FN = 20. The au­ tosomal complement consists of four pairs of large biarmed and two pairs of medium uniarmed chromosomes; the X­chromosome is a small uniarmed element, and the Y an even smaller uniarmed chromosome. This karyotype is identical in gross chromosomal characteristics to that of M. regina , described below (fig. 47E). It has been described by Reig et al. (1977) and Palma and Yates (1996), under the name M. cinerea , and by Svartman (1998).

COMMENTS: As currently recognized, this species is distributed throughout Amazonia, from central Perú in the upper Río Urubamba drainage (Departamento de Cusco) east to the Guianan coast and eastern Estado do Para´, around the horn of northeastern Brazil and south through gallery forests along the clear water tributaries of the Brazilian shield to the Mata Atlântica of coastal Brazil. The limited sequence data from the cytochromeb gene show that this range is divisible into at least five reciprocally monophyletic regional lineages, with the Mata Atlântica representatives as nearly distinct from those of Amazonia as are M. demerarae and M. regina , and certainly as distinct as is the Central American M. alstoni (fig. 52). Amazonian and Mata Atlântica representatives differ by an average of 10.7% in sequence, and the four Amazonian clades differ by a maximum of 7.4% among themselves (see da Silva and Patton, 1998). While additional fieldwork is critically needed in geographically intermediate areas between the Mata Atlântica and Amazonia , the suggestion that these represent separate species is supported by the sequence data. If so, the name demerarae would be restricted to Amazonian forms. Alfred L. Gardner (personal commun.) believes that limae Thomas, 1920, with type locality of ‘‘Ceara´,’’ Brazil, is the appropriate name to apply to the coastal Brazilian clade, although this needs to be confirmed by appropriate comparisons and analyses of variation within coastal Brazil. If limae proves to be associated with Amazonian demerarae instead, then travassosi (Miranda Ribeiro) with type locality in Estado do Rio de Janeiro would be the most likely name.

The four haplotype clades of M. demerarae within Amazonia (fig. 52) form regional units, with one in southwestern Amazonia (including our Rio Juruá samples), one north of the Rio Solimões and west of the Rio Negro, one in northeastern Amazonia (east of the Rio negro, Venezuela, and Guyana), and one in southeastern Amazonia (Estado do Para´). The southwestern Amazonian clade contains the haplotype from locality ‘‘k’’ in eastern Bolivia (fig. 51) that represents the taxon M. constantiae based on geographic position (Anderson, 1997). This suggests that constantiae might best be considered a junior synonym of demerarae , at least until a thorough revision of the genus is accomplished. Although there is strong geographic structure in identifying these four clades, there is also a reasonable level of se­ quence divergence within each of those clades with several sampled localities, with haplotype differences ranging from a mean of 2.9% to 5.3% (fig. 52). Clearly, the woolly mouse opossums remain a rich source for much needed revisionary studies. Fortunately, relatively large samples are available in many museum collections, which will facilitate future studies.

SPECIMENS EXAMINED (n = 55): (1) 2m — MNFS 1365, 1672; (2) 1f — MNFS 1256; (3) 1m, 3f — MNFS 1490, 1513, 1616, JUR 209; (4) 1f — MNFS 1672; (6) 4m — JLP 15632, 15652, 15741; MNFS 727; (7) 5m, 3f — JLP 15357, 15368, 15423, 15477, MNFS 382, 397, 434, 475; (9) 4f — JLP 16050– 16051, 16058, 16077; (9a) 2f — MNFS 904–905; (12) 2m, 2f — JLP 15833; MNFS 721, 727, 748; (13) 1m, 1f — JUR 267, 305; (14) 8m, 3f — JUR 411, 439, 452, 460, 469, 479, 507, 518, 535, 542, 565; (15) 1m — JUR 353; (16) 7m, 4f — JUR 412–413, 448– 449, 451, 458, 484, 506, 517, 519–520.