Oecomys Thomas, 1906

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 : 118-122

Oecomys Thomas, 1906

Arboreal rice rats

Members of this genus are small to medium­sized mice, scansorial in their habits and usually arboreal, commonly trapped in dense vine tangles within a few meters of the ground, and often in relatively disturbed forest. Considered a subgenus of Oryzomys in most early literature, Oecomys was revised by Hershkovitz (1960), who consolidated some 25 scientific names into only two species, the larger O. concolor and the smaller O. bicolor . The inadequacy of this arrangement has been recognized by many fieldworkers, who often can identify three or more distinct forms at a single locality (e.g., Patton et al., 1982; Woodman et al., 1991). The genus is under review by G. G. Musser, M. D. Carleton, and J. L. Patton. In their 1993 compilation of sigmodontine rodents, Musser and Carleton provisionally recognize 13 species, some of which they suggested were composites. Ten of these 13 have ranges that include greater Amazonia.

Oecomys shares a number of generalized characters with other oryzomyine rodents of Amazonia, but is readily separated from all other co­occurring genera. They lack the spinose fur of either Neacomys or Scolomys , and the webbed feet and large size of either Holochilus and Nectomys . Member species span the body size range represented by Oligoryzomys and Oryzomys , but can be distinguished easily from either by their short and broad feet with proportionately long toes and from Oryzomys by a long tail that often terminates in a slight pencil. In foot and tail characteristics, Oecomys is closest to the climbing rats, Rhipidomys , and may be difficult to distinguish in the hand from small species of that genus. However, these two genera typically differ in the coloration of the dorsal surface of the hind feet, in the length of the tail pencil, and in number of mammae as well as in those cranial characters that define the Tribe Oryzomyini relative to the ‘‘Thomasomyini’’ (see Voss, 1993; Voss and Carleton, 1993).

General external and cranial characters of Oecomys include: four pair of mammae in females; tail longer than head and body (averaging 112% to 135% for those species within the Rio Jurua´); tail clothed in sparse short hairs and usually terminating in a short pencil of hairs, 3–4 mm in length or less short and broad hind feet with proportionately long toes; dorsal surface of hind foot without extensive dark patch as in Rhipidomys but which still can be slightly darker relative to sides, particularly over metatarsals. General dorsal coloration ranges from orangish to reddish brown in all species in the Rio Jurua´; ventral color ranges from pure white to gray­based fur tinged with either white or buff. The skull has a short and relatively broad rostrum; the zygomatic plate is relatively narrow, its anterior border straight and lacking a anterodorsal spine resulting in a shallow zygomatic notch when viewed from above; the upper incisors are slightly opisthodont; the interorbital region is broad with well­developed supraorbital ledges diverging strongly posteriorly and extending onto braincase as parietal ridges; the braincase relatively broad and short, distinctly more rounded in comparison to Oryzomys ; all species within the Rio Juruá region have the primitive cephalic arterial pattern, with an enlarged stapedial foramen, squamosal­alisphenoid groove, and sphenofrontal foramen present; an alisphenoid strut is varyingly present in most species; hamular process of

squamosal short and broad, often with subsquamosal foramen totally occluded, or nearly so; the mesopterygoid fossa is rather broad with the anterior border either smoothly arched or squared; sphenopalatine vacuities are small to totally occluded so that the roof of the mesopterygoid fossa is solid; the parapterygoid fossae are typically rather deeply etched; and the cranium is vaulted in lateral view, not flat.

Oecomys of the Rio Juruá

Five species of Oecomys are clearly recognizable from the Rio Juruá and the names we employ were provided by G. G. Musser, who examined all of our specimens. Two or more species were found at 9 of the 16 major localities (fig. 81), with the small­bodied O. bicolor and the larger O. roberti most commonly together. However, three species were recorded at some sites ( O. bicolor , O. roberti , and O. trinitatis at Barro Vermelho [locality 12] and O. roberti , O. superans , and O. trinitatis at Penedo [locality 7] and Altamira [locality 9]), and four species were taken at Vira­Volta, locality 14 ( O. bicolor , O. species, O. roberti , and O. trinitatis ). As detailed below, each of these has a distinct morphology that makes them relatively easy to distinguish in the field, and each is well differentiated in molecular characters.

Three points are of significance in the bootstrap consensus maximum parsimony tree for haplotypes of cytochrome­b (based on 801 base pairs) for representatives of each of the five morphological species from the Rio Juruá illustrated in figure 82. First, the five species are nearly equidistant from one another, with interspecific comparisons ranging between 7.2% and 9.8% (Kimura twoparameter distances, table 28). Interestingly, these values are substantially less (50% or more) than those of other polytypic genera from the Rio Juruá (e.g., Neacomys and Oryzomys ). If the rate of molecular evolution within each of these genera has been more or less equivalent, then diversification within Oecomys has been considerably more recent than in the other two. Second, as well differentiated as these species are, however there are no clear patterns of phyletic relationships among them. All bootstrap values for hierarchical relationships among them are always less than 50%. Hence, the consensus bootstrap tree (fig. 82) has a basal polytomy with hierarchical relationships only between haplotypes of single species. A more expanded analysis, including more taxa and many more localities, will be presented as part of the forthcoming revisionary effort by Musser et al. Finally, there is substantial among­population differentiation for two species, O. bicolor and O. roberti , with average divergences of 5.43 to 4.1%, respectively, for reciprocally monophyletic geographic clades within each (table 28). These will be discussed at greater length in the individual accounts that follow.

There are two distinctly small­bodied species ( O. bicolor and O. species) and three larger ones ( O. roberti , O. trinitatis , and O superans ) (table 29). Oecomys species is significantly smaller in nearly all measurements than O. bicolor (p <0.01 or 0.001 for 20 of 24 measurements; 2­tailed t­tests), and O. bicolor is, in turn, significantly smaller than the other three for all 24 variables (p <0.001 in all comparisons). Oecomys roberti and O trinitatis are both medium­large in size, and differ from each other only in the length of the incisive foramen (p <0.05) with the samples available to us (fig. 83). Finally, O superans is consistently larger than both O roberti or O. trinitatis in all variables, significantly so in most. In general, O. superans has a proportionately longer and wider rostrum but shorter nasals than either O. roberti or O. trinitatis , but the overall proportional

the variance, separates O. superans from the others, with condyloincisive length and diastema contributing most strongly. The two species O. roberti and O. trinitatis remain totally overlapping in discriminant space along all axes. When analyses are restricted to the three larger species, however, O. roberti and O. trinitatis are somewhat more separable, but even here two of the four specimens of O. trinitatis are misclassified as O roberti , with posterior probabilities of 0.528 and 0.696, respectively.

The less clear­cut discrimination between our samples of O. roberti and O. trinitatis is not surprising, since the two differ significantly in only one cranial variable, length of the incisive foramen. These two species differ in qualitative features, however, especially in both dorsal and ventral coloration and pelage quality, as noted in the individual accounts, below.

pattern of character variation is very similar among the three larger species (fig. 83).

We performed two separate multiplegroups discriminant analyses in comparisons between these species, using log­transformed values for the cranial variables. One analysis included all five species; the other was restricted to the large­bodied taxa. Tables 30 and 31 provide the standardized discriminant coefficients for these analyses, and figure 84 illustrates the bivariate relationships between individual scores on the first two axes in comparisons between the different sets of species. In the analysis including all five species, the first axis explains over 92% of the total variation and serves to separate O. species, O. bicolor , and the three larger species, not surprising given the great disparity in size for all measurements (fig. 83). The second axis, while accounting for only 4.5% of