Journal of Mammalian Evolution Экология, эволюция, поведение и систематика
Dental topographic metrics are not inherently taxon- or homology-free: lessons from a new, large dataset of carnivorous mammals (Carnivora, Hyaenodonta, and Metatheria)
Аннотация
Dental topographic metrics (DTMs) are a popular tool for studying diet in extinct vertebrates. These methods are thought to be largely taxon and homology-free, allowing for direct comparisons among species with dramatically different dental morphologies. Here, we test this assumption using a new, large dataset of 735 molar series representing 277 carnivorous mammal taxa, focusing on the three most commonly used DTMs: orientation patch count rotated (OPCr), convex Dirichlet’s normal energy (convex DNE), and relief index (RFI). All three metrics are strongly influenced by homology and phylogeny and do not yield comparable values for carnivorans and non-carnivorans of similar diets. OPCr is strongly influenced by differences in tooth count and yields spurious results. For example, hypercarnivorous metatherians and herbivorous carnivorans have similar OPCr values, but carnivorans of similar diet do not. Convex DNE yields counter-intuitive values because of its sensitivity to the number of cusps and crests; uncalibrated values in the present dataset are low and nearly identical in frugivorous viverrids and felids and systematically higher in metatherians than carnivorans. Moreover, uncorrected OPCr and convex DNE values in these data largely correlate with occlusal complexity and thus are mostly collinear. In some cases, it is possible to compensate for these confounding factors through “homological calibration”. Dividing OPCr by tooth count (here termed OPCrc) and DNE by cusp number (here termed DNEc) produces values more concordant with theoretical expectations and resolves differences between carnivorans and non-carnivorans. RFI shows slightly better correlation with diet than uncalibrated OPCr and convex DNE but has significant phylogenetic bias, making results non-comparable between carnivorans and non-carnivorans (and between mustelids and non-mustelids within Carnivora). Our results suggest that DTMs are not inherently taxon or homology-free and that these metrics must be appropriately calibrated and deployed to avoid producing misleading results.