Statistical genetic comparison of two techniques for assessing molar crown size in pedigreed baboons

Dental anthropologists and paleoanthropologists commonly use an estimated molar crown area (mesiodistal length multiplied by buccolingual width) to describe and compare individuals, populations, and species. Advances in digital imaging now allow researchers to measure the actual crown area of a molar in an occlusal two-dimensional plane. Because error is reduced by this more accurate measurement, actual crown area is thought to be a better representation of the mechanisms that determine tooth crown size, meriting the additional time required to collect it. We tested this assumption by estimating the heritability of both these measurements for the second left mandibular molar from a sample of individuals (n = 332) from a captive breeding colony of baboons. Heritability estimates of both the actual and estimated crown areas of molars are approximately 0.83. Therefore, both measurements are informative as population descriptors, with no significant difference between the accuracy of either to reflect additive genetic contributions to molar crown size. This is fortunate, because genetic studies and inference can be based on estimated areas rather than actual crown area. The heritability estimates for mesiodistal length and buccolingual width are both substantial but lower: ∼0.67 and ∼0.73, respectively. The best fitting models in these analyses show that sex, body size, and subspecific affinity differentially affect molar length and width. We interpret these results to suggest that potentially some of the genetics underlying these covariates also underlie tooth size. As such, measurements designed to describe molar crown size are useful for general descriptive purposes, but do not conform to the assumption of independence inherent in phylogenetic analyses, such as cladistics (Hennig [1966] Phylogenetic Systematics. Urbana: University of Illinois Press). Therefore, if variables like actual crown area and estimated crown area are to be used in phylogenetic parsimony analyses, we suggest that researchers account for the effects of covariates such as sex and body size in their analyses.