An evaluation of sex- and ancestry-specific variation in sacral size and shape using geometric morphometrics.

Abstract

Sex estimation using the human sacrum has primarily been limited to the use of non-statistical, visual observations of sacral size and shape with no quantifiable measure of certainty. Qualitative sex estimates rely primarily on generalizations of two aspects of sacral morphology: dorso-ventral curvature and the S1/alae breadth ratio. Additionally, the use of sacral morphology for ancestry estimation has largely been ignored. The goals of this study are to detail sex- and ancestry-specific variation in sacral form using geometric morphometrics and to evaluate the contributions of dorso-ventral curvature and the S1/alae breadth ratio to such variation. Twenty-five sacral landmarks were digitized from a sample of 201 human individuals of known sex, age, and ancestry at the Hamann-Todd Osteological Collection. Three-dimensional coordinate data were used to calculate independent measures of shape and size using generalized Procrustes analysis. We computed discriminant functions using two configurations of Procrustes coordinates, representing disparate aspects of sacral morphology, centroid size, and interlandmark distances to evaluate the classification potential of each dataset. Cross-validated classification accuracies ranged from 60.2 to 98.0% for sex estimation and 60.0 to 95.8% for ancestry estimation depending on the variables and aspects of sacral morphology represented by each dataset. Our results confirm the important role of the S1/alae breadth ratio for sex estimation, while dorso-ventral curvature shows more utility in estimating ancestry than sex, contrary to traditional use. Sex estimations based on vague, "eyeballed" observations ignore the amount of variation within sexes and may produce incorrect estimates.