Oligoryzomys microtis (Allen, 1916)

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 : 133-138

Oligoryzomys microtis (Allen, 1916)

TYPE LOCALITY: ‘‘Lower Rio Solimoes (fifty miles above mouth),’’ Estado do Amazonas, Brazil.

DESCRIPTION: This is a small­bodied, longtailed, and short­haired mouse, yellowish brown above and grayish white below. The skull (fig. 92) is small, with a somewhat elongated braincase, broad but long rostrum, and hourglass­shaped interorbital region having squared edges.

NONGEOGRAPHIC VARIATION: Table 33 provides descriptive statistics for four external and 20 cranial characters from all adult individuals pooled across their sampled localities. We examined the effects sex and age within localities on morphometric variation as well as differences among localities through an ANOVA nested by locality, sex, and age. These are also summarized in table 33 as variance components, or the percentage of total variation due to the successive nested effects. Sexual dimorphism is virtually nonexistent, accounting for an average of only 1.54% of the total pool of variation, with only tail length, hind foot length, and interorbital constriction exhibiting significance (at p <0.05). Age accounts for a considerable amount of the variation, however, with an average of 22.0%. Most variables, particularly cranial, exhibit a significant age effect (p <0.01 for all variables except TAL, HF, E, IOC, MTRL, OCB, and BOL). These variables generally increase in size with increasing age, a pattern of variation also found for O. fornesi (listed as a synonym of O. microtis by Musser and Carleton, 1993) and other species of Oligoryzomys from Paraguay by Myers and Carleton (1981). Comparisons between localities, therefore, should take differences in age distributions into account, as geographic patterns can be confounded by the age profiles of different samples.

GEOGRAPHIC VARIATION: Relatively large samples are available from at least three localities within each of the Upper Central and Lower Central sampling regions of the Rio Jurua´; we caught only four individuals in the Headwaters Region and none in the Mouth Region. We examined the extent of variation among localities, either as a general function of isolation by distance or as a function of which the side of the river from which samples came. All but four variables (MTRL, IFL, BOL, and MPFW) exhibit significant (one­way ANOVAs, p <0.01) differences in comparisons between localities, even when comparisons are limited to those six samples of the Upper and Lower Middle regions. However, locality per se exhibits a relatively minor overall effect, averaging only 8.4% of the total variation, less than one half the age effect (table 33).

Because most dimensions increase with in­ dividual age, and since age distributions are different between localities (X 2 = 57.137, df = 16, p <0.0001), we used a multiple groups principal components (MGPCA) approach (Thorpe, 1983) to avoid confusing within­group with among­group size. In this analysis, the first MGPCA axis is a general within­group allometric ‘‘size’’ vector, to which all individual variables are positively and significantly correlated (p <0.001 in all comparisons). Overall size differences between locality samples, due in large measure to differences in their respective age distributions, were then ignored in a canonical discriminant analysis using individual scores for all MGPCA axes except MGPCA­1 (see Thorpe and Baez, 1987, and discussion in Patton and Smith, 1990). Exclusive of any overall size differences, there are no apparent differences between any pair of sampled lo­ calities in multivariate discriminant space. Figure 93 illustrates bivariate plots of the scores on the first three axes, which collectively account for 96.1% of the total variation. The analysis does not support multivariate separation of localities, either between Upper Central versus Lower Central regions or right bank versus left bank. Similarly, clustering of locality­average­squared Mahalanobis distances exhibits no pattern of among­locality geographic relationships (fig. 93, lower left). Finally, there is also no correlation between the matrices of Mahalanobis D 2 and geographic distances (Mantel’s matrix r = 0.173; t = 1.055; p = 0.854), and thus no evidence for a general morphometric pattern of isolation by distance.

MOLECULAR PHYLOGEOGRAPHY: The mor­ phological uniformity, at least along the approximately 200 km stretch of the middle Rio Jurua´, is mirrored by lack of population differentiation, based on hierarchical analyses of haplotypes of the mtDNA cytochromeb gene (Patton et al., 1996a). In these genetic traits, as with the morphological ones, most of the total pool of variation is that contained within populations (81.7% for the mtDNA haplotypes, roughly equivalent to the average error term of 66.1% for external and cranial variables, table 33). Again, we found no genetic differences between localities grouped either by geographic region or river bank. A weak isolation by distance relationship in among­population genetic traits exists in comparisons between localities from all sampled regions but not when analyses were re­ stricted to the Upper and Lower Central localities. Moreover, estimates of gene flow rates (Slatkin’s [1993] M) between adjacent localities, either paired on the same or opposite sides, were all above 10.0. These values are sufficient to prevent local populations from diverging by the simple actions of genetic drift (reviewed by Mills and Allendorf, 1996).

DISTRIBUTION AND HABITAT: Although this species is distributed widely within the Rio Juruá basin, it has a very restricted and unique habitat range. All but one of the 321 specimens we captured were taken on the ground in seasonally available or disturbed habitats along the river margin during the dry season. A single individual was trapped on the terra firme standardized lines at Altamira (locality 9), approximately 800 m inland from the grassy margins of the river (line C, fig. 28). The vast majority of specimens were taken in the dense grass (capim) that grows rapidly on the upper edges of sand bars, which become progressively exposed as the water recedes seasonally. The remainder were taken in other disturbed habitats close to the water’s edge, such as old or active garden plots, pasture, or adjacent to human dwellings. There have been no populational studies of this species elsewhere within its range, but its occupation of what amounts to grass and disturbed shrub patches within otherwise primary lowland evergreen forest matches the habitat range of many other species in the genus (see, for example, summaries in Emmons and Feer, 1997, and Redford and Eisenberg, 1992).

As a result of the seasonal availability of its primary habitat, very few O. microtis were taken at localities in the Headwaters Region (localities 1 through 4), and none were taken at those in the Mouth Region; both areas were sampled during the high water of the rainy season when this habitat is scarce. Thus, within the Mouth Region, it is not clear whether this species is truly absent, or simply at exceedingly low population numbers in refugial habitats that were not sampled.

Two aspects of importance relating to the seasonal rise and fall of the river must be considered when discussing the population biology and evolutionary potential of O. mi­ crotis. First, as a denizen of riverine edge habitats, particularly seasonally dense grasses (fig. 2), individuals are likely to be carried downriver during flood stages when grass mats are swept from their precarious footholds on exposed beaches. Such floating mats of grasses and other edge vegetation were commonly observed following rapid river rises (fig. 94). These could also explain the dendritic distribution along river courses of O. microtis throughout Amazonia. The second issue regarding the biology of this species involves the absolute seasonal availability of the preferred river­edge habitats Where does it seek refuge during the highwater season, and, as a corollary, what kind of annual fluctuation in population numbers age structure, and breeding strategy characterizes such a species? Both aspects suggest that rapid seasonal exploitation once riverine edge habitats become available can be a profitable adaptive ecological strategy.

REPRODUCTION AND LIFE HISTORY: We caught most specimens during the dry season, in the months of August through November. Although longitudinal studies were not possible, so individual growth patterns are unknown, the overall impression is that animals grow exceedingly rapidly, reach reproductive maturity quickly, and breed in successive reproductive bouts, all within the course of the single dry season. Virtually all females trapped (n = 86) were reproductively active, either pregnant, lactating, or parous with relatively fresh placental scars. We recorded reproductive activity by autopsy at the time of specimen preparation. For females, we noted whether individual females were (1) nulliparous and not in present reproductive condition (uteri nonvascularized thin, and threadlike with no evidence of ripe follicles on the surface of the ovary, mammary nipples tiny), (2) undergoing their first estrous (ripe follicles visible, uterus swollen mammary nipples tiny), (3) pregnant, (4) or postpartum (lactating and/or placental scars visible, nipples enlarged). For males, we recorded testis position (abdominal or scrotal and size (length by width), greatest length of the vesicular glands, and visibility of seminiferous tubules in the cauda epididymides Reproductively active males were those with scrotal testes over 9 mm in length, with ep­

ididymal tubules visible to the naked eye, and with vesicular glands greater than 10 mm in total length. Nonreproductive individuals had abdominal and small testes (always <6 mm), short and non­swollen seminal vesicles (<5mm), and nonvisible epididymal tubules. We also recorded the relative age of each specimen, using the toothwear sequence proposed by Myers and Carleton (1981). Although we do not know the correspondence of these toothwear categories to chronological (calendar) age, M3 becomes fully erupted by day 30–35 post­partum in the piñon mouse, Peromyscus truei (Hoffmeister, 1951) .

Individuals of both sexes apparently reach reproductive maturity early (fig. 95). For males, 50% of age class 1 individuals (those without fully erupted third molars; n = 8) were scrotal with visible epididymal tubules and enlarged seminal vesicles, whereas 82% (31 of 39) age class 2 and all of ages 3 or older were similarly reproductively active. As with males, most age class 1 females were reproductively active (9 of 15), seven of which were pregnant. A few nulliparous individuals persisted into higher age classes but by age class 2, 81% (21 of 26) were either pregnant or post­partum. Indeed, the majority of all individuals in each age class were pregnant at the time of their capture Litter sizes based on embryo counts ranged from 2 to 8, with a mode of 4 (n = 47). For both sexes, age class 1 individuals were all still in juvenile pelage, while those of age class 2 had either completed their adult molt or were in the process of doing so. Although we have no notion of the relation between toothwear age classes and chronological ages, reproduction is obviously in full swing by age 2. Clearly, O. microtis exhibits the reproductive features of an r ­selected life history, with fast growth, early reproductive maturity, and large litter sizes. These features are to be expected for a species that lives primarily in seasonally ephemeral habitats.

KARYOTYPE: 2n = 64, FN = 66. We karyotyped 59 individuals from 12 localities (Porongaba [locality 1], n = 2; Sobral [locality 4], n = 1; Condor [locality 6], n = 11; opposite Condor [locality e], n = 1; Penedo [locality 7], n = 6; União [locality g], n = 4; Caioá [locality h], n = 3; Miranda [locality I], n = 9; Eirunepé [locality j], n = 2; Altamira [locality 9], n = 4; Jainu [locality 11], n = 6; and Barro Vermelho [locality 12], n = 10. The karyotype is described and figured by Gardner and Patton (1976), and an up­dated comparison of the chromosome complement of O. microtis to other species in the genus is provided by Silva and Yonenaga­Yassuda (1997).

SPECIMENS EXAMINED (n = 321): (1) 1m, 1f — MNFS 1321, 1360; (4) 1m, 2 unknown — MNFS 1425, 1576, 1619; (6) 28m, 14f — JLP 15520–15522, 15526–15528, 15537– 15538, 15543–15546, 15550–15557, 15579– 15588, 15606–15608, 15615–15616, 15629– 15631, 15672, 15679, 15692–15693; (e) 7m, 5f — JLP 15714–15718, 15733–15739; (7) 31m, 29f — JLP 15233–15237, 15246, 15251–15252, 15406–15412, 15428–15430 15438–15440, 15451–15454, 15471–15475 15481–15494, 15502–15504, 15508–15513 15515–15516, 15518, MNFS 328, 343, 491– 492; (g) 1m, 3f — JLP 15226–15228, MNFS 327; (h) 8m, 3f — DMN 5–6, JLP 15216– 152222, MNFS 324–325; (i) 11m, 5f — DMN 4, JLP 15205–15213, 15223–15225 MNFS 320–322, 326; (j) 5m, 3f — DMN 1– 2, JLP 15201–15204, MNFS 317–318; (k 1m, 3f — JLP 15190–15191, MNFS 313– 314; (9) 51m, 38f — JLP 15925, 15929– 15938, 15945, 15947–15965, 15969, 15973– 15988, 16001–16012, 16022–16025, 16045 16080, MNFS 835–845, 847–850, 888–891 905–908; (10) 1m, 2f — JLP 15970, MNFS 950–951; (11) 11m, 12f — JLP 15760– 15761, 15908–15914, MNFS 788–792, 798– 806; (12) 26m, 18f — JLP 15774–15776 15779–15791, 15795–15796, 15801–15808 15815, 15831–15832, 15854, 15894–15900 15914, 15918–15919, 15921, MNFS 811– 816, 824–826, 829–832.

Figures

Fig. 93. Bivariate plots of individual scores on the first three axes derived from a canonical discriminant analysis of the cranial ‘‘shape’’ variables (excluding the first MGPC axis) for each of the six Upper Central (localities 5, 6, and 7; broken lines) and Lower Central (localities 9, 11, and 12; solid lines) sample sites of Oligoryzomys microtis. Right bank localities are identified by solid symbols; those from the left bank by open symbols. The proportion of the total variation explained by each MGCAN axis is given. The relationship between localities based on a matrix of squared Mahalanobis distances clustered by the unweighted pair group method, is in the lower left.