Proechimys steerei Goldman, 1911

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 : 248-259

Proechimys steerei Goldman, 1911

TYPE LOCALITY: ‘‘ Rio Purus , a southern tributary of the Amazon, in northwestern Brazil’ ’; recorded as Hyutanahan, upper Rio Purus , Provincia Lábrea, Estado do Amazonas, Brazil by Moojen (1948: 338) .

DESCRIPTION: Proechimys steerei is the largest species of spiny rats found in the Rio Jurua´, with individuals from true várzea sites during the flooding season maximally weighing nearly 1 kg. The ears and hind feet are large, and the tail is proportionately short approximately two thirds that of the body (tables 60 and 64). The color of the tail is dark brown above and white to cream ventrally it is clothed by hair but the scales remain conspicuous to the eye. The color of the dorsal surface of the hind foot is characteristic of this species: a pale to dark brown outer band and whitish inner band along the length of the foot, from the tarsal joint to the end of the toes, in most individuals. Another distinctive feature is the narrow, short, and rath­ er lax aristiform hairs on the dorsum (da Silva, 1998: fig. 3), that contribute to a characteristically softer fur than is found in any other spiny rat, not only along the Rio Juruá but in all of western Amazonia. As is true of most species of Proechimys , there is no lateral stripe and the reddish color of the sides of the body contrasts sharply with the pure white venter. The texture of the ventral fur in P. steerei also seems thicker and distinctly more velvety to both the eye and touch than in other species.

The baculum of our specimens of P. steerei is similar to that described by Patton (1987) for his goeldii ­ group; it is moderately long and narrow, especially when compared to bacula of P. brevicauda and P. cuvieri although it is shorter and wider than that of P. simonsi (fig. 137).

The skull is large, with a long and narrow rostrum (figs. 138 and 139) and well­developed supraorbital ledge (Patton, 1987: fig 21b). The incisive foramen is lyrate to oval in outline, with slightly to well­flanged posterolateral margins in the great majority of

individuals (130 out of 134 specimens) that form grooves extending onto the palate (fig. 140). The premaxillary portion of the septum is short, less than half the length of the opening, the maxillary portion is distinctly narrow, and both are in contact in most specimens (104 out of 135); the vomer is usually not visible (101 out of 133 specimens). A groove is present on the floor of the infraorbital foramen, but development of the lateral flange is weak (of 134 specimens, 79 had a shallow groove but without lateral flanges and only four had a groove present with moderately developed lateral flanges; all 51 remaining specimens had a smooth floor without any groove). The mesopterygoid fossa is relatively broad, but penetrates the palate to a level between the posterior and anterior margins of M3 (fig. 141). Most specimens show three folds in PM4, M1, and M3 (118, 97, and 93 out of 134 specimens, respectively; the remaining individuals have four folds in those teeth) and four folds in M2 (91 out of 134 specimens, with 3 folds in all others). Overall, these characters observed for specimens from the Rio Juruá compare with those described by Patton (1987) for specimens of his goeldii ­ group from other localities outside the Rio Juruá Basin.

SELECTED MEASUREMENTS: Means and ranges of selected external and cranial measurements are given in table 64.

COMPARISONS: Proechimys steerei is perhaps the most easily recognized species of spiny rat in the Rio Juruá basin, being distinguished from all other Proechimys by the combination of very large body size, relatively short and bicolored tail, laterally bicolored dorsal surface of the hind foot, and the distinctly soft adult pelage covering the entire body, especially along the dorsum where the aristiforms are not stiff to the touch. The relatively short and narrow baculum is indicative of a short and thin phallus in the male, a structure that also easily separates this species from all other sympatric spiny rats throughout western Amazonia. Cranially, P. steerei can be distinguished from other sympatric species by the combination of its large size, typically four folds on M2, structure of the incisive foramina, and broad but relatively deep mesopterygoid fossa. Illustrations of these features are given in figures 139 and 140 as well as Patton (1987).

MOLECULAR PHYLOGEOGRAPHY: We have sampled 25 localities that span a substantial portion of the mapped range of the goeldii ­ group of species, as diagnosed and mapped by Patton (1987: fig. 2). Included are samples of P. steerei from 14 of the primary localities within the Rio Juruá basin (fig. 153; table 77). Up to five individuals were sequenced per population. Three major haplotype clades are recognizable, one of which is further divisible into two groups (fig. 154). Haplotypes from specimens allocated to P. goeldii from two localities in Estado do Pará are so well differentiated from all others (13.2% relative to other members of the group) that it is basal to all other species of Amazonian Proechimys in analyses containing all taxa and geographic representatives (da Silva, 1998: fig 13). The other two clades separate samples from largely south of the Rio Solimões, but including those from the Rio Jaú west and south of the Rio Negro, from those in northern Perú, southern Venezuela, and Brazil east of the Rio Negro (fig. 153). These differ by an average of 10.9%. Although samples from the northern clade are sparse, sequence divergence is low, averaging only 2.6% across the 2000 kilometers between northern Perú and central Brazil north of Manaus. The southern clade, however, does exhibit substantial geographic variation, with the samples from north of the Rio Solimões along the Rio Jaú differing from those to the south by an average of 6.6%. A more extensive analysis of haplotype variation among localities within the Rio Juruá is in preparation for publication elsewhere (M. D. Matocq, J. L. Patton, and M. N. F. da Silva). A synopsis of these data is given below in the section on riverine barriers. However, from the tree in fig. 154 it is apparent that a reasonable degree of haplotype diversity is present along the river, as 801 bp haplotypes from 11 different localities differ by an average of more than 5%. Importantly, those haplotypes from the Headwaters form a monophyletic assemblage distinct from those from the other three sample regions.

We consider the three clades identified in the mtDNA tree (fig. 154) to represent sep­ arate species. Based both on comparisons of specimens of each of these clades to holotypes of the various named forms that Patton (1987) allocated to his goeldii ­ group, the eastern mtDNA clade represents P. goeldii Thomas, 1905 , although the oldest name for the southern clade is P. steerei Goldman, 1911 . The northern clade is P. quadruplicatus Hershkovitz, 1948 , a name published in the same year but somewhat earlier than its junior synonym, P. amphichoricus Moojen, 1948 . Thorough analyses of morphological character variation for this group have yet to be performed, but the data summarized by Patton (1987) suggest that each clade is diagnosable by morphological traits as well as cytochrome­b sequences. For example, most individuals of P. quadruplicatus (more than 60%) possess four folds on all four upper cheekteeth, while this character state is much less frequent in P. steerei , and few individuals of P. goeldii have four folds on any teeth (Patton, 1987: table 5).

MORPHOMETRIC VARIATION: As with P. simonsi , above, we summarize the variation in mensural characters for adults (toothwear classes 8, 9, and 10) of P. steerei that is due to locality, sex, and age effects by a nested ANOVA (table 78). No character exhibits significant interlocality differences, and the average contribution of this factor to character variation is only 5.6%. However, considerable differences due to both sexual dimorphism and age variation are apparent More than two thirds of the variables exhibit significant variant components for sex and age, with an average of 23.3% of the total pool of variation due to sex and 20.6% due to age. Hence, the degree of sexual dimorphism and the extent to which dimensions

continue to increase in older toothwear classes even in fully adult individuals is somewhat greater in this species than it is in P. simonsi , although the differences in patterns between the two species is not great (compare tables 72 and 78). When an analysis nested by sex and age is applied to our largest sample from Nova Empresa (locality 8, n = 41), seven cranial variables exhibit significant sexual dimorphism (p <0.05: CIL, MB, RL, MPFW; p <0.01: NL, D, and PL) and six show significant age differences (p <0.05: MB, RL, NL, MPFW; p <0.01: D, PL). Again, within­locality variance is partitioned somewhat differently than it is in P. simonsi , particularly in the degree of character sexual dimorphism.

The small amount of morphometric variation attributable to interlocality differences contrasts somewhat with the mtDNA haplotype data, which indicates both a reasonable amount of sequence differentiation among haplotypes (over 5%) and some geographic structuring into two reciprocally monophyletic clades along the river. In order to examine geographic trends in morphology more thoroughly, we used both principal components and discriminant function analyses, and compared samples from the four sample regions as well as those from the left and right banks of the river. Despite negligible interlocality variation in univariate di­ mensions, limited regional effects are apparent. In a principal components analysis, samples from each of the four geographic regional sample areas overlap extensively in multivariate space in combinations of the first three axes, which combine to explain 76.1% of the total pool of variation (fig. 155 top). Nevertheless, there are significant differences between the regions in mean PC scores on each axis (PC­1: F 3,161 = 7.086, p = 0.0002; PC­2: F 3,161 = 9.111, p = 0.0001 PC­3: F 3,161 = 15.410, p = 0.0001, respectively). The first PC axis represents a general size axis, as indicated by high and positive factor coefficients (table 79) and by positive correlations of individual scores with their respective mensural variables. The correlation between PC­1 scores and individual values for logCIL, for example, is 0.976 (p <0.001).

The trend in overall size in P. steerei along the Rio Juruá from its headwaters to its mouth is more complex then in P. simonsi with parallel increases in size from the Headwaters Region to the Upper Central Region and from the Lower Central Region to the Mouth Region (fig. 156). The differences between samples from the Upper and Lower Central regions are significant, as are those between the Lower Central and Mouth (Duncan’s multiple range critical differences = 0.479 and 0.495, respectively, both p <0.05). This complex pattern parallels interlocality variation in karyotype, as discussed below.

As was the case for P. simonsi , although regional samples of P. steerei exhibit a pattern of size increase along the length of the Rio Jurua´, there is no differentiation between opposite­bank samples when characters are examined either in a univariate fashion or by multivariate principal components analysis For example, there is broad overlap in the bivariate plot of PC­1 and PC­2 scores for samples pooled by right and left bank localities (fig. 155, bottom), and one­way ANO­ VAs for PC scores were nonsignificant for each of the three axes of table 78: for PC­1 F 1,163 = 1.943, p = 0.1652; for PC­2, F 1,163 = 0.133, p = 0.7161; and for PC­3, F 1,163 = 0.648, p = 0.4221. However, also as was true for P. simonsi , if samples are pooled by river bank and subjected to a discriminant function

analysis (table 74), which maximizes between­group variance while minimizing that within groups, some segregation of oppositebank populations is apparent. A one­way ANOVA of individual scores on the first discriminant axis yields a significant river bank effect (F 1,163 = 20.660, p <0.0001), and histograms of these scores (fig. 151, right) illustrate the slight differences among the samples.

We also examined the relationship between the morphometric and genetic distances among our samples of P. steerei , as well as that between each of these variables and geographic distance. We used the Mahalanobis D 2 matrix generated from the discriminant function analysis that specified localities as the a priori groups as a measure of morphometric distance, and a matrix of genetic similarities (Slatkin’s [1993] M­statistic) generated from the population cytochrome­b haplotypes by the AMOVA program of Excoffier et al. (1992). These were compared to the log 10 of the straight­line geographic distances among localities given in table 1. Morphometric distances increase significantly with an increase in geographic distance among locality pairs (fig. 157; Mantel’s matrix correlation coefficient r = 0.460, p = 0.0015), and genetic similarity decreases sharply with geography (fig. 157; r = ‾0.804, p <0.0001). Not surprisingly, therefore, there is a significant, if weak, correlation between genetic similarity and morphometric distance (r = ‾0.275, p = 0.0192) Populations of P. steerei along the Rio Jurua´ as is true of P. simonsi , exhibit a clinal, isolation­by­distance pattern in both morphometric and genetic traits.

REPRODUCTION: We obtained specimens of P. steerei along the entire river and at all seasons during the year of our sampling. Of the 183 males we autopsied, 88 were reproductively active. These ranged in age from toothwear class 3 to 10, although 82% (68 of 83) were adults (age classes 8 to 10) and 17% (14 of 83) were subadults (age classes 6 and 7). Reproductively inactive males include both young and adult individuals (age classes 1 to 9), with 68% (59 or 87) individuals of age classes 3, 5, and 6, and 9% (8 of 87) full adults (age classes 8 and 9). We have reproductive data for 179 females (table 75). Off these, 65% showed signs of current or previous pregnancy whereas 35% apparently had not yet reproduced. We caught pregnant, lactating, or postpartum females at all sites indicating that at least some females are breeding in all seasons of the year. Of the parous females, 83% (87 of 105) were adults, 16% subadults, and 1% young individuals. Seventy­five females were pregnant. The age class of these individuals ranged from 6 to 10; 87% (65 of 75) were fully adult (age classes 8 to 10) and the remaining subadults Seventeen percent (14 of 84) were both pregnant and lactating. Modal litter size is 3 range, 1–7. The great majority of nulliparous females are young (47 of 58) or subadults (10 of 58). Relative to other species of spiny rats, particularly P. simonsi , which is also distributed along the entire river and for which we also have large samples, P. steerei exhibits reproductive characteristics tending towards a more r ­selected life history, with somewhat earlier reproductive maturity, larg­ er litter sizes, larger percentage of young animals breeding, a larger percentage of postpartum estrous females, and a smaller proportion of nonbreeding adults (table 75) These features might be expected for a species that lives primarily in strongly seasonal habitats, such as the várzea forests of the Rio Jurua´.

DISTRIBUTION AND HABITAT: Although distributional limits of P. steerei remain poorly understood, we have identified this species from localities from central Perú south of the

Río Marañón to northern Bolivia and east through western Brazil in Acre and Amazonas states as far as the west bank of the Rio Negro, north of the Rio Solimões (fig. 152; da Silva and Patton, 1998: fig. 1). We found the species throughout the Rio Jurua´, although most specimens (74.6%) were collected in the extensive várzea forests of the Upper and Lower Central regions. The lower numbers of specimens collected at localities in the Mouth Region (only 15.5% of the total sample) likely reflect our sampling of that area during the rainy season when the habitat of P. steerei was reduced by flooding, probably resulting in a seasonal decrease in population density. However, the relatively low percentage of animals found in the Headwaters Region (9.9%) might be due to other habitat features, or to competitive interactions with other species, since strict segregation by habitat is not evident in this region (table 63) and true, seasonally flooded várzea is not present.

While most common in seasonally flooded várzea, P. steerei was also found occasionally in secondary and disturbed terra firme forests, active and abandoned gardens, and margins of flooded grasslands (table 63). Elsewhere in Amazonia, we have taken this species in igapó (seasonally flooded black water) forests and along small creeks and riv­ er edge habitats within terra firme forest. The totality of our observations strongly suggest that P. steerei prefers flooded forests and riparian habitats within the lowland Amazon.

We captured the species in the Tomahawk and Sherman traps in our standardized lines; in addition we also took specimens by hunting and with Victor and metal snap traps. Although the total number of each kind of traps varied, in our sample of specimens (n = 313 the majority (72%) was caught in Tomahawk traps and only 16% in Sherman traps, 4% were shot, 6% were caught with Victor snap traps, and 2% with the metal snap traps. Of the animals captured in live traps, most young and subadults (129 out of 178) were caught in Tomahawk traps and the remaining 49 individuals in Sherman traps; the same is true for adults with 129 (out of 135) caught in Tomahawk traps, and only six in the Sherman traps. Considering the large size (maximum weight 800 g) of individual P. steerei these results are not surprising.

KARYOTYPE: 2n = 24; FN = 40–42. We prepared karyotypes from 148 individuals, at least one from each locality from which P. steerei was collected. All specimens had an autosomal complement consisting of two pairs of large and seven pairs of medium to small meta and submetacentrics, one pair of large subtelocentrics, and one pair of small acrocentrics. Specimens from the Headwaters Region had a second, smaller pair of subtelocentrics (fig. 158D); those from the Mouth Region typically had this pair as a medium­sized acrocentric (fig. 158B). Finally, specimens from localities in the Upper and Lower Central Regions had one or the other of these two homozygous conditions, or were heterozygous (fig. 158C). The Xchromosome was a small acrocentric and the Y­chromosome an even smaller one in all individuals, regardless of autosomal comple­ ment. We caught heterozygous individuals at localities 9, 10, and 14. At localities 9 and 14, the parental homozygotes with mediumsized acrocentrics were more abundant (9 of 11 and 8 of 10 karyotyped individuals in each locality, respectively); the two remaining individuals from each locality were of both types, the parental homozygote with medium­sized subtelocentric and the heterozygote. At locality 10, three of six karyotyped individuals were homozygotes with medium­sized acrocentrics, and three were heterozygotes. The downriver karyotype with the single pair of medium­sized acrocentrics also characterizes samples from the Rio Jau´, to the north of the Rio Solimões and west of the Rio Negro (fig. 158A). The upriver karyotype with the two pairs of subtel­ ocentrics is also found in samples from central, eastern, and southern Perú (table 80). Despite the slight variation in the karyotype of P. steerei , it is readily distinguishable from that of P. goeldii (2n = 24, FN = 42) from the Rio Xingu in Estado do Pará to the east, or to those of P. quadruplicatus (2n = 26–28, FN = 42–44) in northwestern and north­central Estado do Amazonia (table 75; fig. 159).

COMMENTS: Patton (1987) included P. steerei within his goeldii ­ group, which he noted varied more over its geographic range than any other species group that he examined, and suggested the existence of an eastern ( P. goeldii ) and a western ( P. steerei ) species. The cytochrome­b analyses (fig. 154), however, subdivide western populations even further and, along with karyotypic data, suggest at least three species within the goeldii ­ group. The senior names that would apply to each clade are: (1) P. quadruplicatus Hershkovitz, 1948 (with amphichoricus Moojen, 1948 a junior synonym), which occurs from Perú north of the Rio Marañón through eastern Ecuador to southeastern Colombia and east across southern Venezuela and adjacent Brazil north and east of the Rio Negro to at least the vicinity of Manaus Available geographic samples of this taxon are 2n = 26 ( Venezuela [Reig and Useche 1976]) or 2n = 28 ( Perú and Ecuador [Gardner and Emmons, 1984] and Brazil [this report]; see table 80 and fig. 159B–C). We referred to this clade as P. amphichoricus in earlier papers (e.g., da Silva and Patton 1998). (2) P. steerei (with pachita Thomas 1923, and rattinus Thomas, 1926, clearly junior synonyms, and probably kermiti Allen 1915, hilda Thomas, 1924, and liminalis Moojen, 1948, as well) with 2n = 24, FN = 40 or 42 karyotype (table 80 and figs. 158) The range of this species is delimited above and in figure 153. And, (3) P. goeldii (with hyleae Moojen, 1948, nesiotes Moojen 1948, and leioprimna Moojen, 1948 synonyms), from Estado do Pará in the eastern Amazon Basin of Brazil. Our samples of this taxon from the lower Rio Xingu and the Serra Carajás are inadequate to do more than simply document the extensive degree of sequence divergence between P. goeldii and either P. quadruplicatus or P. steerei (average from 13.2 to 14.4%). These samples also

have a different karyotype, with 2n = 24 and FN = 42 (fig. 159A; table 80). Additional sampling throughout the range of the goeldii ­ group is needed to further refine the geographic boundaries of the three species we identify here, as well as to determine if other taxa deserve recognition.

SPECIMENS EXAMINED (n = 453): (2) 1f — MNFS 1254; (a) 7m, 4f — MNFS 995–997, 1032, 1046, 1056–057, 1060–1063; (c) 1f — MNFS 1037; (3) 4m, 9f — JUR 206, MNFS 1521, 1543, 1545, 1547–1548, 1589–1592, 1633, 1655, 1682; (4) 8m, 13f — JUR 241, 245, MNFS 1430, 1445–1447, 1459, 1462, 1474–1477, 1506–1507, 1569, 1617, 1623– 1625, 1643, 1662; (d) 1m — JLP 15628; (5) 39m, 31f, 1 unknown — JUR 118–129, 133– 134, 136–143, 170, 172, MNFS 570–576, 585, 594–607, 614–620, 626–628, 641, 647–650, 658–667; (6) 12m, 7f — JLP 15558, 15614, 15690–15692, 15698–15700, 15705, 15709–15712, 15722, 15730–15732, JUR 176, MNFS 526; (7) 10m, 8f — JLP 15244–15245, 15254–15255, 15268–15269, 15458–15459, MNFS 333–334, 337–338, 342, 346, 354, 472, 496–497; (8) 68m, 61f — JLP 15375–15382, 15386–115391, 15396–15401, JUR 15–32, 44, 49, 50–71, 80–109, 111–113, 115–116, MNFS 443– 463, 467–469, 477–483, 501; (9) 3m, 2f — JLP 15926–15927, 16019, 16053, MNFS 851; (9a) 5m, 5f — DMN 16–17, MNFS 926–928, 939, 949, 958–959, 963; (10) 15m, 9f — DMN, 12–13, MNFS 869–877, 895, 898–899, 915–916, 919, 932–933, 940, 945– 947, 956; (11) 12m, 17f — JLP 15749– 15750, 15753–15757, MNFS 679–680, 688– 691, 700–701, 704, 709–711, 714–715, 720, 752–753, 770–771, 784–785, 794; (12) 17m, 13f — JLP 15789, 15799–15800, 15835– 15838, 15856–15858, 15860–15864, 15878– 15880, 15889, 15915, MNFS 687, 730–731, 733–735, 780, 781, 782, 783; (13) 19m, 27f — JUR 252–262, 264–266, 274–282, 293, 307–314, 326–330, 338–341, 345, 347, 349– 351; (14) 1f — MNFS 1777; (o) 3m, 2f — MNFS 1737–1738, 1770–1772; (16) 12m, 6f — JUR 477, 504, MNFS 1749–1750, 1752– 1753, 1755–1756, 1763–1769, 1783, 1792– 1793.