The Impact of Tooth Wear on Occlusal Shape and the Identification of Fossils of New World Porcupines (Rodentia: Erethizontidae)

Although isolated dental remains are a central source of data in mammalian paleontology, the teeth of many taxa remain unidentifiable at low taxonomic levels because of intraspecific variation. Extant New World porcupines, Erethizontinae, illustrate this problem. Since being declared unusually variable over 50 years ago, molar occlusal morphology has been neglected as a potential source of diagnostic traits. Here, we use geometric morphometrics to re-evaluate the occlusal morphology of the third lower molar of erethizontines as the basis for identification of genera while accounting for variation, particularly changes in shape caused by progressive tooth wear. We model the impact and predictability of this source of variation by using CT-based digital sectioning. We use modelled occlusal shapes to analytically search for traits that can be used to improve identification of isolated dental remains, as well as training machine-learning models to assign occlusal surfaces to genus. Nearly half of all shape variation can be explained by variables such as wear and size. Resulting models of shape predict six univariate traits that may diagnose the two extant genera, especially when specimens are compared at the same wear stage. However, high levels of variation previously noted qualitatively also influence quantitative analyses, rendering machine learning models unreliable. Overall, although the identity of some specimens that are highly worn or contain certain combinations of traits will remain ambiguous, large segments of the gnathic erethizontine fossil record have renewed utility when analyzed using these methods.