PREDICTING BODY MASS IN RODENTS USING NON-TRADITIONAL 3D DENTAL SHAPE DESCRIPTORS

Body size is closely related to numerous aspects of an organism’s ecology and as such is a useful tool for paleontological inferences. The most common proxies for estimating body mass for fossil mammals, especially rodents, is the rectangular occlusal area (i.e., tooth length multiplied by width) of the lower first molar (m1) or the lower fourth premolar (p4). However, as rodent teeth are highly variable in shape, traditional linear measures often misrepresent tooth dimensions and thus lead to over- and underestimated sizes for many species. To test whether a more accurate body mass could be predicted using non-traditional measures, we microCT-scanned the lower tooth rows of 151 extant rodent species. Using various software packages (Geomagic, Morphotester, molaR, ArcGIS, etc.), we collected 429 variables from the isolated first, second (m2), and third (m3) molars and from entire tooth rows. Variables included basic volume and surface area values, replicated measures including relief index and topographic complexity, novel methods such as platonic shape extraction, and various ratios. When available, we collected body mass in grams from museum tags of the scanned specimen, and compiled average body masses for all species from the literature. All body masses and measures were log-transformed to account for allometry between tooth and body size. Regression analyses between body mass and each of the variables show that the traditional rectangular area of m1 is not the most accurate predictor of body mass for rodents and that the traditional areas of m2 and of m3 were more highly correlated with body mass. Furthermore, the outline of each tooth position was also more highly correlated with body mass than the rectangular area, with m2 and m3 again outperforming m1. Predictions based on the tooth row outline and planimetric area yield the strongest correlations with body mass of the 335 statistically significant regressions. Our results indicate that the entire tooth row is better correlated with rodent body mass than individual tooth positions and therefore should be used to estimate mass when available. However, as isolated teeth are more common in the fossil record, m2 and m3 should be used in place of m1 when estimating rodent body mass, with tooth outline being prefered over traditional tooth area.