Molecular Techniques:

Genomic DNA was extracted with the DNAeasy kit (Qiagen, Inc., Valencia, CA) from liver preserved in the field in 95% ethanol. The mitochondrial cytochrome b gene (cyt-b) in Oryzomys is 1143 base pairs (bp) long (Smith and Patton, 1999). We amplified the entire gene as two fragments of approximately equal length that overlapped by about 350 bp. Primer pair MVZ05-MVZ16 amplified the initial 800+ bases of the gene, and primer pair MVZ127-MVZ108 amplified the terminal 700+ bases of the gene (primer sequences in Smith and Patton, 1999; Patton et al., 2000; Leite, 2003). Double-stranded DNA was purified using the QIAquick PCR Purification kit (Qiagen, Inc., Valencia, CA) following protocols used in previous studies (Smith and Patton, 1999). The light strand of this template was then cycle-sequenced with primers MVZ05 and MVZ127 and run on an ABI 3730 capillary automated sequencer following manufacturer protocols using Big-Dye chain terminators. Sequences were aligned and edited using the Sequence Navigator software (Applied Biosystems, Inc.).

Samples:

We obtained the entire 1143-bp gene sequence for five individuals of the PNNKM Oryzomys series and added these to a truncated database of 801-bp sequences containing multiple individuals of each of the recognized species of the O. megacephalus group (Musser et al., 1998). Our complete dataset includes 19 sequences of O. perenensis, 41 of O. megacephalus, and 7 of O. laticeps, in addition to those of the new taxon, plus 4 sequences of O. yunganus. We used two sequences each of the five species of the O. nitidus complex (O. emmonsae, O. legatus, O. macconnelli, O. nitidus, and O. russatus; Musser et al., 1998) as the outgroup in phylogenetic analyses, a position supported by other molecular studies (e.g., Weksler, 2003). Sequences not generated for this study have been analyzed previously (Patton et al., 2000; Costa, 2003); provenance and other supporting data for each of these can be found in those publications. GenBank accession numbers for specimens of Oryzomys n. sp. USNM 597565–66 and 597572 are AY940623–5.

Analyses:

We performed two levels of phylogenetic analysis, using the maximum parsimony inference in PAUP*4.0b10 (Swofford, 2002). The first included all 86 sequences of the combined O. megacephalus group and O. nitidus group samples, plus those of O. yunganus. Given the size of the dataset, we restricted this analysis to a single heuristic search using stepwise addition and tree-bisection-reconnection (TBR) branch swapping. This analysis (not presented below) confirmed the previous studies of Patton et al. (2000) and Costa (2003) regarding the monophyly of each of the species in the O. megacephalus-group, with bootstrap support of 96 to 100 at each basal node. We then performed a second analysis using a truncated dataset comprised of 1 sequence from each outgroup species and 3–6 of the most divergent sequences from each species clade identified in the initial analysis. For this smaller dataset, we again used maximum parsimony but with the branch and bound search option, to provide an optimal recovery of minimum-length trees. In both analyses, we treated variable nucleotide characters as unordered and given equal weight, and assessed the robustness of nodes by bootstrapping (Felsenstein, 1985), with 1000 iterations.

Morphometric Analyses:

We obtained external variables (total length, tail length, length of the hind foot, and height of the ear) from specimen labels and recorded 20 cranial measurements with digital calipers to the nearest 0.01 mm. The variables are the same set used by Patton et al. (2000), who provide complete descriptions of each measurement: condyloincisive length [CIL], zygomatic breadth [ZB], mastoid breadth [MB], rostral length [RL], nasal length [NL], rostral width-1 [RW-1], rostral width-2 [RW-2], orbital length [OL], diastemal length [D], molar toothrow length [MTRL], incisive foramen length [IFL], palatal bridge length [PBL], alveolar width [AW], occipital condyle breadth [OCB], basioccipital length [BOL], mesopterygoid fossa length [MPFL], mesopterygoid fossa width [MPFW], zygomatic plate length [ZPL], and cranial depth [CD]. We assigned individuals to an age class based on tooth wear, using the pattern categories described by Myers and Carleton (1981), and included only adult specimens (age categories 3–5) in all statistical summaries and comparisons. For the latter, we used StatView, version 4.5 (SAS Institute, Inc., 1998) to calculate standard descriptive statistics (mean, standard error), compared samples of each species by ANOVA using Fisher's PLSD posterior test of significance for each pair of taxa, and visualized the relationship of each species sample in multivariate space using principal components following log-transformation. For canonical discriminant function analysis, also using log-transformed cranial variables for each a priori defined species, we used the software package Statistica, version 4.1 (StatSoft, Inc., 1994). Our selection of samples is not exhaustive, as prior studies (Musser et al., 1998; Patton et al., 2000) document relative character uniformity within each of the cis-Andean species of the O. megacephalus complex. The comparisons documented below are based on 18 specimens of the Oryzomys, n. sp., from eastern Bolivia, 4 of O. laticeps from coastal Brazil, 21 O. megacephalus from the southwestern Brazilian states of Mato Grosso and Mato Grosso do Sul, and 34 O. perenensis from western Amazonian Brazilian states of Amazonas and Acre, and the eastern Peru departments of Cusco and Madre de Dios. We measured all but the Brazilian samples of O. perenensis specifically for this study; for the latter we used a subsample of the 258 specimens analyzed in Patton et al. (2000). We measured width of the footpads with calipers on eight fluid-preserved specimens of O. perenensis, four of O. laticeps, four of O. megacephalus, and two of O. n.sp.

We compared pelage colors under natural spectrum light; colors when capitalized, are from Ridgeway (1912). We also give values from Smithe (1975) and descriptive names.

Sequence Divergence:

According to Kimura 2-parameter distances within and among each of the five taxa within the megacephalus complex sensu latu (table 1), the sample from eastern Bolivia is markedly different from all other species in the complex, being closest to O. laticeps from coastal Brazil (K2p mean of 11.99). The remaining comparisons were presented earlier by Patton et al. (2000: table 35). The numerical uniqueness of the Bolivian sample relative to other species is consistent with a species definition based on sequence divergence (e.g., Bradley and Baker, 2000), although we prefer not to base a species concept solely on a simple phenetic distance measure.

Members of the O. megacephalus group of taxa appear monophyletic with respect to other such groups in the genus, at least with the limited taxon and gene sequences available (Smith and Patton, 1999; Bonvicino and Moreira, 2001; Weksler, 2003). However, the level of bootstrap support uniting the four taxa of this complex, including the new Bolivian taxon, relative to O. yunganus and the outgroup members of the O nitidus complex, is only 71 (data from the large, 86-sequence analysis not shown). The branch-and-bound analysis of the truncated dataset of these taxa yielded two minimal-length trees of 740 steps (fig. 1). Of the 801 characters in the analysis, 515 are constant across all sequences, 65 are parsimony uninformative, and 221 are parsimony informative. The two trees differ only in the position of two sequences within the O. laticeps clade; all other nodes are shared in common. Bootstrap support for each species clade ranges from a minimum of 93 (the basal node of O. megacephalus) to 99 (basal node of O. yunganus) and 100 (basal nodes for the other three taxa). Oryzomys yunganus is consistently placed at the base of a clade containing the remaining four species, although again with only limited bootstrap support (74). The four taxa of the restricted megacephalus group (sensu Costa, 2003) exhibit a consistent hierarchical structure in both trees of (O. megacephalus (O. laticeps (O. perenensis + O. n.sp.))). However, none of the connecting nodes has bootstrap support greater than 50. As a consequence, we cannot offer a robust hypothesis of relationships among any of the four megacephalus group species with the limited amount of available sequence data. Thus, the apparent closer relationship between the new taxon and O. laticeps suggested by their more similar sequence divergence (table 1) cannot be supported by the hierarchical analysis. What is clear, however, is that both genetic distance and phylogenetic placement indicate equal uniqueness of the new Bolivian specimens in comparison to the other taxa in the group currently recognized as species (Musser et al., 1998; Patton et al., 2000; Costa, 2003).

Morphometric Divergence:

Summaries of external and cranial mensural variables (table 2) show the sample from Bolivia to be generally intermediate in most external and cranial dimensions between O. megacephalus and O. perenensis, with O. laticeps the largest of the four taxa. In pairwise comparisons, the new taxon is significantly different from one or more of these other three species in 3 of the 4 external and in 19 of 20 cranial variables, with the exception of tail length (Tal) and rostral width-1 (RW-1), different significantly from O. laticeps in 9 variables, from O. megacephalus in 18 variables, and from O. perenensis in 8 variables (table 3).

The first three axes of the Principal Components Analysis accounted for a total of 68.6% of the total variation present, and each of these axes had eigenvalues of 1.466 or greater. Not surprisingly, the unrotated factors on PC-1 were all positive and relatively high (table 4), suggesting a general size component. The position of individual specimens along PC-2 generally is weighted most heavily by rostral length (logRL), mesopterygoid fossa width (logMPFW), and interorbital constriction (logIOC) opposing basioccipital length (logBOL) and nasal length (logNL), while position on PC-3 is weighted most heavily by cranial depth (logCD), occipitalcondyle breadth (logOCB), and maxillary toothrow length (logMTRL) versus diastema length (logD) and mesopterygoid fossa length (logMPFL).

Each of the four taxa of the megacephalus complex examined morphometrically is separable along combinations of PC-1 and PC-2 and, to a lesser extent, on PC2 and PC-3 axes (fig. 2). The new taxon from Bolivia differs significantly from O. laticeps on PC-2 and PC-3 (table 3), from O. megacephalus on PC-1 and PC-2, and from O. perenensis on PC2. Oryzomys laticeps differs from the latter two on all three axes, but O. megacephalus and O. perenensis are broadly overlapping, particularly on the second and third axes.

Three canonical axes were extracted from the same set of morphological variables used in the PCA analysis, the first explaining 69.47%, the second 21.57%, and the third 8.96% of the total variance (table 5). The predefined species groups are each highly significantly different from one another, with the Mahalanobis D² for the comparisons between the Bolivian taxon and each of the other three species 42.56 (O. laticeps; F_(20,40) = 3.029, p = 0.0014), 51.14 (O. perenensis; F_(20,40) = 9.957, p < 0.0001), and 35.53 (O. megacephalus; F_(20,40) = 5.971, p < 0.0001). Not surprisingly, therefore, all individual specimens are correctly assigned to their respective species groups with posterior probabilities of group membership < 0.99989.

As is apparent from the bivariate plots of each pair of canonical axes, the new taxon from Bolivia is readily and completely separable from all other species along the second canonical axis (fig. 3, top and bottom panels), but overlaps with the scores of O. megacephalus on CV-1 and CV-3 (fig. 3, middle panel). Oryzomys laticeps is uniquely separable from the other three species on CV-3, and both O. megacephalus and O. perenensis are completely separable on CV-1 but overlap substantially on both CV-2 and CV-3. Based on the standardized discriminant coefficients, the combination of rostral width-2 and diastema length versus palatal length distinguish the Bolivian taxon from all others on the first axis (table 5). Clearly, in morphometric terms, the Bolivian taxon is markedly distinct from other members of the megacephalus complex of species distributed in the lowland tropical forests of cis-Andean South America.

Holotype:

MNK 3628, collected 6 November 1998 by Louise H. Emmons (field no. LHE1540). A young adult male with hindquarters terminating molt from juvenile to adult pelage, prepared as a skin and skull, liver tissue preserved in ethanol. External measurements recorded on label: TL = 261, T = 121, HF = 32, E = 20, weight = 57 g. Testes scrotal, 8 × 5 mm.

Paratypes: USNM 588193, 597563–79; MNK (uncataloged) field nos. IGP 45, IGP 47, IGP 55– 6, LHE 1582, NRS 09, NRS 13, NRS 16, NRS 25, VCC 01–5, VCC 08, VCC 11, VCC 15; all from the type locality; USNM 588192, from Huanchaca II; and MNK uncataloged LHE 1680, from Los Fierros. Only specimens from localities within PNNKM are designated as paratypes.

Type Locality:

Bolivia: Departamento de Santa Cruz, Provincia Velasco, Parque Nacional Noël Kempff Mercado, El Refugio Huanchaca, an outpost with a few buildings and an airstrip on private property, but within the park (14°42.553 ′S, 061°2.034 ′W [WGS 84]; elev. 170 m); on older published maps, the locality now known as El Refugio Huanchaca is shown as Huanchaca, an estancia. In 2004 the present owners added Huanchaca to their former designation of El Refugio, which appears on specimen labels and is the same locality. Captured in a deciduous forest 3.5 km north of the outpost.

Diagnosis:

A medium-sized terrestrial rat. Dorsal color near Olive Brown to Saccado's Umber (olive brown 28 of Smithe, 1975), cheeks and sides with orange tones, Buckthorn Brown to Ochraceous Tawny (clay color 26 to cinnamon 123A of Smithe); rostrum and top of head to behind eyes dusky, close to Deep Neutral Gray, lined with black-tipped hairs, contrasting with brown top of crown. Pelage long, tips of most hairs at mid-rump reaching 9 mm; ventral fur dense and long, 3–4 mm on chest between forelegs; hairs pale gray at base with long white tips, overall ventral appearance whitish, inner limbs white; white of inner hind legs continuous with whitish tops of feet. Hind foot pads large, interdigital and hypothenar pads wider than 1.3 mm. Cranium lacking squamosoalisphenoid groove and sphenofrontal foramen (Musser et al. 1998); bony palatal excrescences strongly developed (fig. 4). Upper second molar with a short paraflexus and a single fossette; lower second molar with a short hypoflexid opposite a lingual fossettid (fig. 5).

Distribution and Habitat:

Oryzomys acritus were captured by LHE at four localities in the PNNKM and one just outside its borders (fig. 4, localities 1–5). The park is on the Brazilian Shield; its vegetation is a mosaic of humid, semi deciduous and deciduous forests and liana forests, and wet and dry savanna and cerrado formations. The biota and geology are described in detail in Killeen and Schulenberg (1998) and the vegetation also in Panfil (2001). These rats were captured on the ground with Sherman, Tomahawk, and break-back traps and a standardized bait of oatmeal and raisins flavored with fish oil and essences of vanilla, banana, and coconut, in forests that ranged from evergreen riverine or gallery forest to deciduous and semideciduous terra firme and seasonally flooded forest bordering seasonally flooded pampa. One locality (Huanchaca II) is at 700 m on top of the Huanchaca meseta, but its forest is similar in species composition of both flora and fauna to that of the lower elevation sites where the species was found within the park (Killeen and Schulenberg, 1998). These forests either have a seasonally deciduous emergent canopy, sparse understory, and midstory varying in deciduousness depending both on the location and on the dryness of the particular year, or are liana forests, with low canopies and dense lianas at all levels. Additional specimens were identified in collections from near San Joaquín, Departmento Beni, and Guarayos, Departmento Santa Cruz. All localities identified thus far are in the area drained by western tributaries of the basin of the Río Itenez (named Rio Guaporé in Brazil: fig. 6).

Etymology:

From the Greek akritos, “confused, mixed, doubtful” (Brown, 1956), with reference to the mixed characters that can be confused with those of O. megacephalus or even O. nitidus.

Description:

A medium-sized rat (table 2) of similar aspect and color to Oryzomys megacephalus sensu Patton et al. (2000). Above brown to yellow-brown with hairs gray-based, younger individuals darker and less yellowish than old animals; occasional individuals bleached slightly reddish overall. Sides paler than dorsum, yellowish to distinctly orange, brightest on cheeks and below ear; lateral color can extend from thigh forward to cheeks below eye and to whitish spot at base of mystacial vibrissae. Rostrum dusky above to behind eyes, with dusky-tipped cream or whitish hairs. Anteroventral base of ear with inconspicuous preauricular tuft of entirely whitish or pale gray hairs sometimes tipped with pale rufous, probably associated with a cutaneous gland. Pinnae brown, thinly clothed on the interior surface with whitish or, rarely, brown hairs. Skin surrounding eye blackish. Ventral pelage whitish; hairs quite long, pale gray at bases with long white tips; white of inner legs extending to join pure white of fore and hind feet; hair at wrists white-based; chin spot self-white. Pelage at mid-rump with majority of hairs measuring 8–10 mm (a few longer hairs may be present). Ventral pelage on chest between forelegs 3–4 mm. Tail dusky, paler below at base, but not prominently bicolored, clothed with inconspicuous short, fine, dusky hairs. Hind feet long and narrow (terrestrial type), with digits II–IV noticeably longer than I and V; soles brown but not strongly pigmented. Six footpads always present, hypothenar and fourth interdigital pads broad (fig. 7); thenar pad with strongly raised part of pad short and nearly round; squamae between and around pads inconspicuous.

Cranium similar to those of other members of the megacephalus group, with most diagnostic features as outlined by Musser et al. (1998), including no squamosoalisphenoid groove or sphenofrontal foramina; incisive foramen short and broad, teardrop-shaped (figs. 8, 9, 10); and bony palate long. Supraorbital ridge prolonged into strongly developed temporal ridge that extends across parietal to occipital suture. Parietal usually with a small but distinct contribution to lateral braincase wall exhibited by a ventrad extension of the posterolateral border of the parietal forward of the parieto-occipital suture (Musser et al., 1998: fig. 62); area of ventral contribution appears to decrease with age. Foramen ovale large and rounded, such that the opening is visible below the alisphenoid when the skull is viewed from the side (fig. 11). Anterior rim of auditory meatus usually notched, notch depth decreasing with age (fig. 11). Second upper molar with a long paraflexus, a short hypoflexus, and a labial fossette, but no medial fossette; second lower molar with hypoflexid reaching halfway or less across tooth, opposite a lingual fossettid (figs. 5, 12).