DISCUSSION

The intraspecific variation documented here for Sigmodon curtisi supports the use of some of the criteria commonly used in the identification of extinct cotton rats, but also points out some of the limitations. Hopefully this work will be augmented by similar descriptions of variation through ontogeny in other species of Sigmodon in order to better understand these limitations.

The length and width of the occlusal surface of the m1 in S. curtisi change little through ontogeny. Therefore dimensions of the tooth are useful in delimiting the species of cotton rats, even in the absence of knowing the age of the individual. All other species of cotton rat, except S. minor and hudspethensis, are relatively more hypsodont, and would therefore show even less change in occlusal dimensions through ontogeny. Although there is significant overlap in the size of most species of Sigmodon, the addition of crown height helps separate the species.

Potential Confusion to Similar Species

Qualitative variation is here examined in light of the occlusal morphology of the other species of Sigmodon most similar to S. curtisi. This is not meant to be an exhaustive review of the morphology of those other species. Instead, the intent is to illustrate the potential problems with taxonomic identification based on some previously published characters, especially with regard to particular wear stages of S. curtisi.

The occlusal pattern of m1 in S. curtisi is similar overall to that of the modern Sigmodon leucotis, but the former is distinguished by the presence of a deep LRA1 on the m2 and a shallow LRA1 on the m3 (Martin 1979). Although these folds are lacking in S. leucotis, as described above, the diagnostic LRA1s will disappear from both the m2 and m3 in S. curtisi in advanced stages of wear, resulting in a pattern identical to S. leucotis. The m1 of S. leucotis is longer and wider than the same tooth in S. curtisi, however, the lengths and widths of the m2s and m3s of the two species are nearly identical (Martin 1979, table 1). Isolated m2s and m3s of S. curtisi in advanced wear stages may be confused with teeth of S. leucotis in occlusal morphology and size. Crown height is required to identify such specimens; S. leucotis is more hypsodont.

Sigmodon hudspethensis is an extinct form known from two Pliocene-age localities in west Texas (Akersten 1970). The species is also morphologically similar to the geologically younger S. curtisi in the brachydont nature of the molars and the shared occurrence of LRA1s on the m2 and m3 (Martin 1979; Ruez 2001). The only distinguishing occlusal feature is the wider reentrants in S. hudspethensis. In S. curtisi reentrants narrow with advancing ontogenetic age. Most of the few specimens referable to S. hudspethensis show advanced wear, yet retain the wide reentrants. These two taxa seem to be distinct when examining older wear stages, but ontogenetically young specimens of S. hudspethensis and S. curtisi might not confidently be distinguished. It should be noted that this author is aware of only six m1s of S. hudspethensis. Therefore, the recovery of new material and documentation of the variation in the species is necessary to better elucidate the relationship between the taxon and S. curtisi.

A third morphologically similar species of cotton rat, S. lindsayi, occurs in the Palm Springs Formation of Anza-Borrego Desert State Park in California (Martin and Prince 1989). Martin referred specimens from this area both to S. curtisi (1979:14, 16) and Sigmodon cf. S. curtisi (1979:23-24). However, when Martin and Prince (1989) named S. lindsayi, they did not include any of the specimens cited in the 1979 study, did not mention whether or not they considered S. curtisi or Sigmodon cf. S. curtisi to be present in the Palm Springs Formation, and did not make any comparisons of the holotype of S. lindsayi (an m1) to S. curtisi. (See Murray [2008] for an extensive description of the taxonomic history of these specimens.) This ambiguity is especially problematic because the published description of S. lindsayi is very similar to that of S. curtisi. One striking difference is in a few m1s of S. lindsayi that exhibit slight wear: two have an enamel atoll in the anteroconid, and one has an anteroconid separated completely from the metaconid. These unique specimens suggest either extreme variation in the m1 of S. lindsayi during early ontogeny or the presence of multiple species within the sample identified as S. lindsayi (Martin and Prince 1989). Molars of S. lindsayi are on average larger and more hypsodont than those of S. curtisi, but there is extensive overlap in molar size between these two species (Martin 1979; Martin and Prince 1989). In the absence of the anomalous slightly worn m1s mentioned above, it is difficult to distinguish small individuals of S. lindsayi from large individuals of S. curtisi (which explains the above-mentioned allocation of the Palm Springs Formation specimens to S. curtisi). Detailed description of the ontogenetic variation in the occlusal morphology of molars of S. lindsayi may provide possible characters that can differentiate individuals of these species that overlap in size.

The number of roots on m1s of Sigmodon curtisi from Inglis 1A and Inglis 1C did not vary with either ontogenetic age or body size. This invariance suggests that the use of the number of roots on the m1 may be a reliable character upon which to identify species of cotton rats and establish phylogenetic scenarios by grouping species as done previously (Martin 1979). Similar study of other large samples of cotton rats known to vary in root-count is needed to make more definite statements about this character.

The presence of two-rooted m1s of Sigmodon curtisi is not entirely unexpected; S. minor overlaps in geological distribution with S. curtisi and typically has two-rooted m1s, which could indicate that the origin of S. curtisi is earlier than previously suggested based on currently known stratigraphic placement (Martin 1979; Peláez-Campomanes and Martin 2005). Additionally, the higher occurrence of four-rooted m1s in S. curtisi than previously documented makes the tentatively proposed evolutionary transition into Sigmodon libitinus (Martin 1979) less likely.