Molar size and shape in the estimation of biological affinity: a comparison of relative cusp locations using geometric morphometrics and interlandmark distances

Abstract

The study of teeth has been a central tenet in biological anthropology since the inception of the field. Teeth have been previously shown to have a high genetic component. The high heritability of teeth has allowed researchers to use them to answer a myriad of anthropological questions ranging from human origins to modern variation due to microevolution. Traditionally, teeth have been studied either morphologically, through the assignment of nonmetric character states, or metrically, through mesiodistal and buccolingual crown measures. Increasingly, geometric morphometric techniques are being used to answer anthropological questions, especially dentally. However, regardless of analytical technique utilized, the biological affinity of modern U.S. individuals has often been limited to examination under a forensic lens (classification of either American Asian, black, Hispanic, or white) without consideration of parent populations. The current study uses geometric morphometric techniques on human molars for two main goals: 1) to examine biological affinity of each of the four largest population groups in regard to population history; and 2) examine the variation within and among the four modern groups as a means of classification. A total of 1,225 dentitions were digitized. Each of the four modern U.S. groups was compared to possible parental groups via discriminant function analysis (DFA). Additionally affinity was examined using Mahalanobis generalized distances (D²) wherein significance of distances between groups was calculated via permutation tests. Furthermore, the D² values were subjected to principal coordinate analysis, or classical multidimensional scaling, to visualize group similarity and dissimilarity. Each group demonstrated affinity with potential parental groups and geographically similar groups as expected given population histories; however, each was also significantly unique from the comparison groups. The four modern U.S. groups were then compared to one another using the same statistical tests. Total among-group correct classifications ranged from 33.9-55.5%, indicating a greater classification than random chance (25%). These classifications were negatively correlated with the reported intermarriage rates for each group: American whites and blacks have the lowest intermarriage rates, which resulted in the highest correct classifications. Conversely, American Asians and Hispanics have the highest intermarriage rates, which resulted in the lowest total correct classifications. Still, the DFA model created from the modern U.S. sample was able to accurately classify a holdout sample. Lastly, a comparison of the three most abundant groups in the U.S. (black, Hispanic, and white), achieved a total correct classification of 72.3%, which is comparable to other studies focusing on the same populations. Restricted gene flow through sociologically constructed barriers and positive assortative mating are the likely factors in the observed variation.