Sex and ancestry variation in ulna morphology in an adult South African cadaveric sample

Abstract

BackgroundSex and/or ancestry estimations based on skeletal elements are vital in forensics, as these variables are key to identification of unknown skeletal remains. Unfortunately, patterns of skeletal variation are often shared between sex and ancestry groups, making independent estimation of such variables less accurate, especially when substantial size differences exist both within and between groups. Geometric morphometric analysis allows isolation of the size component of variation, enabling independent and more sensitive detection of shape variation between groups. This creates the potential for more accurate estimations of sex and ancestry either independently or simultaneously, thus reducing the chances of compounding errors of estimation. This would be especially beneficial in heterogeneous populations, such as that of South Africa, where group separation may be affected by complicated genetic and environmental influences. MethodsThis study assessed sex and ancestry variation in morphology of 1894 ulnae of South African males and females of the country's three largest ancestry groups. Three-dimensional data was submitted to Generalized Procrustes Analysis for superimposition and scaling to a common centroid size. Mean centroid sizes and shapes were compared, and accuracy of sex, ancestry, or sex-ancestry estimation was assessed using Discriminant Function Analysis and leave-one-out cross-validation. Covariation with size, age and year-of-birth were assessed through regression analysis. ResultsMale ulnae were absolutely and proportionally larger than female ulnae, while Black individuals were similarly larger than Colored and White individuals. Based on this variation, sex could be estimated with 68.8 % accuracy, and ancestry with 73.6 % accuracy. Simultaneous sex-ancestry assessment showed similar morphological patterning and yielded a mean classification accuracy of 73.6 %. ConclusionsThese results have practical value for forensic application, where relatively poorly elements such as the ulna are often all that is available for analysis. Additionally, simultaneous estimation of sex and ancestry reduces compounding errors that may arise from first estimating one variable and basing the rest of the biological profile estimations thereon. Such improved estimations are of great potential value, especially in heterogeneous populations such as that of South Africa.