CONCLUSION

Based on the present study, the vast majority of specimens from Cerdas can be referred to as 'Plesiotypotherium' minus, though two specimens may pertain to a second species. Our phylogenetic analysis indicates that the species 'P.' minus do not belong in the genus Plesiotypotherium, but should be referred to another genus. Given that variation in Microtypotherium choquecotense is unknown – and that 'P' minus could potentially pertain to that genus – we refrain from creating a new genus at this time. We also refrain from naming a second species at Cerdas, given the uncertain dental homologies of at least UATF-V-000840 and the small sample size of this second morph.

The revised diagnosis of 'P.' minus, the recognition of another diminutive form at Cerdas, plus the presence of similar forms at Quebrada Honda, Chucal, Choquecota, and Nazareno indicate that there was a greater diversity of basal mesotheriines from the middle latitudes of South America than previously imagined. It appears that these basal forms were rather conservative in morphology, particularly in size (Figure 20).

UF 133803 is the most complete cranium of a basal mesotheriine presently known (Movie 1, Movie 2, and Movie 3) and provides new insights into the anatomy of phylogenetically important regions such as the suborbital fossa and foramen, shape of the rostrum, and general morphology of the jugal region; other features such as the internal regions of the bony ear and the posterior neurocranium are now known for these early mesotheriines based on the cranial material from Cerdas. The other specimens from Cerdas provide additional insights into dental development and eruption patterns in this group.

A primary challenge in basal mesotheriine taxonomy is that the holotype of 'P.' minus is a lower dentition, which is not usually diagnostic even at the generic level within Mesotheriinae (see also Croft 2007). Given the additional observations from our study indicating that lower cheek tooth dimensions are not stable over the lifetime of an animal, the use of size as a criterion for identifying species is called into question. The lack of distinguishing occlusal surface features such as fossettids make it difficult to assess wear states within a population. Although lower dentitions remain troublesome to classify, we have demonstrated that their dimensions relate closely to wear, and that the lingual talonid sulcus may be useful for assessing wear state, along with the position of the posterior mental foramen. Such factors should permit comparisons of specimens of similar wear stage, lending greater accuracy to taxonomic interpretations. More work on identifying visible patterns of lower molar shape change with wear would certainly be useful (e.g., the posterior end of the m3 talonid appears to become more rounded with wear).

The 'P.' minus sample from Cerdas provides yet another example of the importance of considering each individual specimen as a member of a once dynamic population. The great pitfall of identifying morphological species in the fossil record is an undesirable return to typology. Identifying and attempting to quantify variation within a population can help prevent such an error. We have provided here a sizeable sample of metric data that illustrate variation for another group of typothere notoungulates. We have also shown that an ontogenetic and ecological criterion, dental wear, must be evaluated prior to any taxonomic analysis. This finding supports prior investigations of other typotheres, suggesting it may be a characteristic of the clade. Further work on the sample from Nazareno must include an assessment of wear as a part of any systematic study.

We used CVs to evaluate the samples from Cerdas, Nazareno, and Chucal to assess some degree of metric homogeneity (often interpreted as taxon homogeneity); this was necessary since the visible morphology did not provide obvious differentiation for the Cerdas sample. Simpson et al. (1960) suggested that a CV exceeding 10.0 is evidence for the presence of multiple species in a fossil sample. Since then, some authors have noted that this criterion is meaningless considering the sample sizes afforded by the fossil record and other sources of error that can affect variation statistics (e.g., measurement error, element size, sample mean; Cope and Lacey 1992, 1995; Polly 1998; Plavcan and Cope 2001). One method of using the CV with fossil samples is to evaluate phylogenetically related and geographically restricted extant samples as references for single-species variation to address whether a fossil sample is a single-species or pooled-species sample (Cope and Lacey 1992, 1995). In the case of the mesotheriines (and notoungulates in general), this is not possible due to the lack of extant representatives. It is now evident that dental wear is a powerful generator of interpopulation metric variation in mesotheriines; combined with what has been reported for archaeohyracids and trachytheriines, it is clear that a high CV does not necessarily indicate a multi-taxon population.

Once again, notoungulates have provided insights into evolutionary questions that have previously been addressed with African, Eurasian, or North American fossil mammals: that is, whether or not a high degree of metric variation in a sample of fossils can be too great to be accommodated within a single species (Simpson et al. 1960; Gingerich 1974; Plavcan and Cope 2001). The mesotheriines of Cerdas have highlighted that conventional methods of identifying morphological species (i.e., identifying a holotype and a hypodigm) are problematic for some taxa. Working with these typothere notoungulates suggests that we should look beyond the traditional morphological criteria and add an ecological component, such as wear state, to taxon diagnoses and definitions. This, in turn, may contribute toward developing a multi-dimensional understanding of these ancient species that we might not have thought possible or practical (Hutchinson 1958).