Proechimys Allen, 1899

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 : 202-213

Proechimys Allen, 1899

Terrestrial spiny rats

The spiny rats of the genus Proechimys are often the most abundant nonvolant mammals

of the lowland Neotropical forests, and perhaps the most easily recognizable genus as well. In contrast to other echimyids of Amazonia, all species of this diverse genus are terrestrial, with elongated heads and long rostra, large and erect ears, and narrow and long hind feet. The tail is always shorter than head­and­body length, the dorsal pelage comprises a mixture of soft hairs (= setiforms; Moojen, 1948) and expanded, varyingly stiffened spines (= aristiforms); the dorsal and lateral color is generally a reddish brown and the venter white, although tinged reddish or grayish, especially in the throat, thoracic region, and inner thighs in some species. The narrow and elongate hind feet have slender toes and small plantar tubercles. The ears are distinctly larger than those of all similar­sized arboreal genera, absolutely so in all comparisons. These animals can be heard scurrying through the leaf litter and readily seen at night, by virtue of their bright red eye shine, freezing when caught in a spotlight, or bounding off with a distinctive loping gait.

The quality of our knowledge of the systematics of Proechimys stands in stark contrast to their ubiquitous presence in all forest types, disturbed and pristine. Three to five species might be present at single localities, may be actually syntopic, and may even occupy the same burrows at different times (e.g., Emmons, 1982; Malcolm, 1992). Although it is sometimes relatively easy to distinguish species when sympatric, identifying living animals to species takes a skilled eye, and defining species boundaries over larger segments of geography has often proven extremely difficult. Only a few studies have succeeded in documenting and adequately describing sympatric entities (e.g., Patton and Gardner, 1972), and only Gardner and Emmons (1984) and Patton (1987) have made much progress in defining geographic character trends within definable taxa over any but the shortest distances, at least for those taxa within Amazonia. As noted by most earlier workers, the diagnosis of species and thus the delineation of their geographic boundaries have been greatly hampered by an extreme level of character variability, both within and among population samples (e.g., Thomas, 1928; Moojen, 1948; Hershkovitz, 1948). Even karyotypes, which have proven useful in differentiating sympatric taxa (Patton and Gardner, 1972; da Silva, 1995, 1998) may be highly variable geographically (Reig et al., 1980; Gardner and Emmons, 1984) Patton and Rogers (1983) showed that individuals essentially grow continuously throughout life, so that mensural variables increase substantially with advancing age. As a consequence, one must be exceedingly careful in morphometric comparisons between taxa and not be confused by age­related variation.

Patton (1987) evaluated several qualitative craniodental and bacular characters for their utility in defining taxa of Proechimys , and allocated the 59 available names to one of nine species groups, five of which occur within Amazonia. For each of these, he provided hypotheses of species units and remarked on their probable geographic limits This work was, of course, available to us both during the field phases of the Juruá research and subsequently while we worked to identify the large series of spiny rats we obtained. Our collections comprise nearly 1200 specimens, obtained at all 16 primary sites Despite our previous experience with the genus throughout eastern Perú and in central Brazil, we had great difficulty in allocating many specimens to the taxa diagnosed by Patton (1987), both in the field and during initial examination of prepared specimens The chromosomal preparations and molecular sequences of the mtDNA cytochrome­b gene that we examined subsequent to returning to the laboratory gave us reason to understand our difficulties in the field. We encountered eight species along the Rio Jurua´ In the sections immediately below, we provide the molecular evidence for these eight lineages, describe general morphological characters of each, and provide multivariate morphometric comparisons. Finally, in the accounts for each species, we further characterize their external and craniodental morphology, describe their karyotypes, provide available molecular data to place each into a broader geographic context within Amazonia, and summarize life history characteristics Although it is clear that both chromosomal and molecular data greatly aid in the discrimination of local species, we emphasize mor­

phological features that will aid workers identify these animals in the field.

Molecular Phylogenetics and Species Limits of Spiny Rats of the Rio Juruá Basin

At the initiation of our survey of the Rio Juruá we were reasonably comfortable with our ability to distinguish species of Proechimys in the field, and with the probability that five taxa would be found within the river basin. These species included P. brevicauda , P. cuvieri , P. simonsi , P. steerei , and a species undescribed at the time but that we’ve known about since 1972 (see Patton and Gardner, 1972; Patton, 1987) and that has recently been described by da Silva (1998) as P. pattoni . We based this prospective species list on our prior experiences in nearby localities in eastern Perú (Balta, on the Río Curanja, Departamento de Ucayali; see Voss and Em­ mons, 1996) and central Brazil (the upper Rio Urucu, southeast of Tefe´, Estado do Amazonas). It quickly became apparent however, that either the characters outlined by Patton (1987) to diagnose these species had grossly underestimated the true nature of morphological variation exhibited by them or that there were more taxa present along the river than could be accounted for by the five taxa identified above. The second alternative proved to be the case, with eight species among our materials from the Rio Juruá now recognized, four of which were undescribed at the time we made our collections (da Silva, 1998). In recognizing these species, we generally follow a species concept based both on demonstrable monophyly of mitochondrial DNA clades and diagnosability by a combination of morphological and chromosomal characteristics. In many cases

this view is supported by sympatry of the taxa in question without evidence for reticulation by hybridization, so that the taxa we recognize also conform to the biological species concept commonly used by many mammalogists. As many as four species were found in terra firme forest at single sites (Igarapé Porongaba [locality 1], Sobral [locality 4], and Barro Vermelho [locality 12]) and five species may be found at localities with both terra firme and várzea forests (Sobral and Barro Vermelho) (fig. 133). At least three species occur at most localities, al­ though true várzea forest contains only a single species ( P. steerei ).

As will be emphasized in the detailed descriptions below, each of these eight species is diagnosable by morphological, chromosomal, and molecular characteristics, and each strongly diverges from all others by their respective cytochrome­b sequences. da Silva (1998) provided a preliminary assessment of phylogenetic relationships among these and other Amazonian species based on the initial 798 bp of cytochrome­b sequence In figure 134 we offer molecular evidence for the distinctness of the eight species we recognize within the Rio Juruá basin. In both our data and those presented by da Silva (1998), geographic samples allocated to most species exhibit close sequence similarity with Kimura two­parameter distances averaging less than 3%, and usually less than 1%. The one exception to this pattern is P. kulinae , which comprises two very divergent haplotype clades that differ by 9.6%, nearly as much as that between P. brevicauda and P. cuvieri (fig. 134). Other than this exception, however, differences between the species we recognize are uniformly large, always above 10% and often above 15%, and haplotypes we have identified for each are always strongly linked in phylogenetic analyses, with bootstrap values consistently at 100%.

Our wish in presenting the data in figure 134 is solely to show that eight separate clades of spiny rats are readily identifiable within the Rio Jurua´. Nevertheless, the length of cytochrome­b sequence examined provides little phylogenetic resolution among any of these species, with the minor exception of the three small­bodied taxa described recently by da Silva (1998). Not surprisingly the terminal nodes (individual haplotypes) of each taxon are highly supported, with bootstrap values of 100% in all cases except that of P. kulinae . However, only P. pattoni and P. gardneri group as species with relatively strong support (72%), with this pair grouping with P. kulinae at a lower bootstrap value of 67. No other taxon exhibits close relationship to any other. This is true even when the data are weighted to accommodate for site saturation (fig. 134; see also Lara et al., 1996).

Characters and the Identification of Species

of Proechimys of the Rio Juruá Basin

We characterize each of the eight species from the Rio Juruá basin in the individual accounts below. In so doing, we provide greater detail than that provided for other mammals in this volume, emphasizing external characteristics which we hope will aid the identification of live animals in the field. The craniodental characters outlined by Moojen (1948), Patton and Gardner (1972), and especially those figured extensively by Patton (1987) provide features upon which museum series can be segregated. Body size, stiffness of the dorsal aristiform hairs, dorsal color pattern and especially ventral coloration, col­ or and color pattern of the hind feet, number and relative size of plantar tubercles, and length and hairiness of the tail can all be used to identify specimens of adults in the field. The assignment of young individuals still in juvenile pelage, however, can be very difficult without examination of the skull or knowledge of the karyotype. Special care must be taken in species assignments of these specimens. Finally, we emphasize that adult males of most, if not all species, can often be distinguished by characteristics of the phallus, including the baculum which is usually visible through the thin dorsal skin of that organ. The phallus of live animals can be easily everted for examination in hand. The soft structures of the phallus have been given scant attention in the taxonomic and morphological literature (but see Patton and Gardner, 1972, and da Silva, 1998), although all eight species from the Rio Juruá can be distinguished by phallic structures.

Table 60 compares each of the eight species from the Rio Juruá by selected external craniodental, phallic/bacular, and karyotypic features. As emphasized by Patton (1987), the best suite of qualitative craniodental characters for species identification are those of the incisive foramina, anterior palate, mesopterygoid fossa, floor of the infraorbital foramen, parietal ridging, and number of flexi in the cheekteeth. There is, admittedly, considerable individual variation in these and the other characters listed in table 60 for any given taxon, and we have tried to indicate the degree of such variation in our descriptions of each species below. Finally, we caution that the species we recognize, and the characters we use to distinguish them, apply to specimens from the Rio Juruá and, certainly adjacent areas in western Amazonia. Given our experiences with these taxa, and in other geographic areas, it is likely that additional species of Proechimys will be uncovered in regional faunas. Also, given the degree of sequence divergence and strong regional monophyly in our limited samples, it is also likely that even well­characterized and reasonably well­known species of spiny rats will be found to be composites of indepen­

dently evolving lineages that may demand subsequent taxonomic recognition.

We examined the morphometric distinctness of each of the eight species we recognize by a series of discriminant function analyses, using log 10 transformations of 21 cranial measurements for adult specimens (age classes 8, 9, and 10 combined). Four separate groups are evident in plots of the first two discriminant axes, which combine to explain 86.9% of the total variation present (fig. 135, top). Proechimys simonsi and P. steerei are individually separable from each other on the first axis and from the other six species on the second axis. The other six species divide into two triads of species ( P. brevicauda , P. cuvieri , and P. echinothrix versus P. gardneri , P. kulinae , and P. pattoni , respectively) that are separable from each other on the first axis and from P. simonsi and P. steerei on the second. The uniqueness of each of the four groups is ev­ ident in histograms of mean discriminant scores for the first two axes (fig. 135, bottom). Misclassification of specimens to species based on a posteriori probabilities of group membership is relatively minimal (always less than 5%), except for those species in each of the two triads. The length of the rostrum contrasts with cranial depth at M1 diastema length, and length of the maxillary toothrow as those variables contributing most strongly to separation on the first axis; mesopterygoid fossa width contrasts with nasal length on the second axis (table 61).

Separate discriminant analyses were performed for each of the two triads of species to further examine their morphometric relationships. The triad consisting of P. brevicauda , P. cuvieri , and P. echinothrix are each completely separable in the bivariate plot of the first and second discriminant axes (fig 136, top), with greater than 95.8% correct assignments of individuals based on their a

posteriori probabilities. For the second triad, P. kulinae overlaps somewhat with both P. gardneri and P. pattoni , which themselves separate completely on DF­1 (fig. 136, bottom). All individuals of P. pattoni are correctly allocated while only 84% of either P. gardneri or P. kulinae group correctly. These are obviously morphologically closely similar species, as noted by da Silva (1998) in her description of these taxa. Proechimys cuvieri differs from both P. brevicauda and P. echinothrix primarily by rostral length, postpalatal length, and width of the mesopterygoid fossa, while condyloincisive length contrasts with rostral length in separating the latter two on the second axis (table 62). For the triad of small­bodied species, mastoid breadth, diastemal length, and width of the mesopterygoid fossa combine to separate species on the first axis, and rostral depth contrasts with maxillary toothrow length as primary variables on the second (table 62; the number of variables included in this anal­ ysis was necessarily reduced because of the small sample size of P. pattoni ).

Habitat Distribution of Proechimys

within the Rio Juruá Basin

As many as five species of Proechimys can be found at single localities within the Rio Juruá basin (fig. 133). Four of these were typically present in terra firme habitats, although the assemblage of species is different across the sampled regions of the river. In contrast, only one species was found in the seasonally flooded várzea at all localities. Table 63 compares the number of captures of each species by habitat (the standardized terra firme and várzea lines, with all other habitats pooled) for each of the four regional sampling areas. Proechimys steerei is the várzea specialist, especially evident in the central and lower reaches of the river where seasonal flooding occurs every year. Only one specimen of another species was captured in this habitat ( P. simonsi , at locality 9 in the Lower Central Region), and only five specimens of P. steerei (of a total of 461 for which data are available) were taken on all of the eight terra firme standardized trap plots combined. Even in the Headwaters Region, where the várzea is both narrow in extent and only floods in exceptional years, no P. steerei were trapped on our standardized lines in mature terra firme forest, although individuals were obtained in mixtures of disturbed habitats on the elevated terra firme landscape. This difference in habitat between P. steerei and the other seven species is highly significant by contingency table analysis for each of the four sampling regions: Headwaters — X 2 = 60.278, df = 2, p <0.0001; Upper Central — X 2 = 182.081, df = 2, p <0.0001; Lower Central — X 2 = 1878.988, df = 2, p <0.0001; and Mouth — X 2 = 130.579, df = 1, p <0.001 (only two habi­ tats [terra firme and ‘‘other’’] were compared in the Mouth Region since true várzea could not trapped during the high water season). As we will discuss below, this sharp dichotomy in habitat preference between P. steerei and the other species of Proechimys within the Rio Juruá basin is also reflected in differences in life history characteristics of these species.

In contrast to P. steerei , all other seven species were always found on our terra firme plots, even though each was also trapped in other terra firme habitats that ranged from second growth forests to garden plots. There are, however, differences in habitat distribution among those species co­occurring in the terra firme. For example, in the Headwaters Region P. pattoni is nearly restricted to undisturbed terra firme forests whereas the three other co­occurring species ( P. brevicauda , P. cuvieri , and P. simonsi ) are each found commonly in disturbed second growth forests, active and abandoned cultivated plots, and various edge habitats. Interesting­ ly, P. brevicauda and P. cuvieri exhibit nonsignificant habitat distributions (X 2 = 3.855, df = 1, p = 0.1455), and were found in quite different relative abundances at those Headwaters localities where they are sympatric. Since P. cuvieri maintained the same habitat distribution at other sites downriver (X 2 = 5.926, df = 2, p = 0.0517), and P. brevicauda was only taken in the Headwaters, perhaps these morphologically similar species compete with one another, thus affecting each others distributional limits and relative abundances.

Clearly much remains to be learned about habitat requirements of these species, as well as their ecological roles as seed predators and dispersal agents within the habitats in which they occur (e.g., Forget, 1991; Adler, 1995).