Estimating population affinity using dental morphoscopic traits in a South African sample

Abstract

Multiple methods exist to estimate population affinity, with country-specific standards available to effectively assess the variation of South Africans. However, analyses are also case-specific, so the overall presence and condition of certain skeletal elements will dictate which methods can possibly be used. Dental characteristics are of importance in biological anthropology, as teeth demonstrate population variation, and are typically well-preserved and intact when recovered in most contexts. However, the application of dentition in South African forensic casework has been fairly limited. The purpose of this study was to explore dental morphological variation in the South African population in order to provide an additional method for the estimation of population affinity for application in forensic cases. The objectives of the study included to test the observer repeatability of identifying and scoring 15 dental morphological traits described in the Arizona State University Dental Anthropology System (ASUDAS) guidelines; to assess the prevalence of each trait among black and white South African males and females; and to test the accuracy with which the traits can be used to estimate population affinity. The overall study sample consisted of 191 individuals, with varying sample sizes per tooth. Any skulls or traits with post-mortem damage, deciduous dentition, pathology and/or excessive tooth loss and dental work that prevents the accurate scoring of the traits were excluded from the study. A series of 15 dental traits were visually assessed on the relevant incisors, premolars and molars of each individual in the sample and scored according to the Arizona State University Dental Anthropology System (ASUDAS). While the traits were considered repeatable based on the intra-observer agreement, the inter-observer agreement was overall poor. This suggests the need for experience with both teeth and the scoring method itself to apply the method accurately. Frequency distributions revealed a substantial amount of overlap in trait prevalence between black and white South Africans. Ultimately, the results of the Kruskal-Wallis tests revealed that only three traits demonstrated statistically significant differences for population affinity (pegged/reduced/missing third molar, four cusped lower second molar, and the seventh cusp on the lower first molar) and only two traits demonstrated differences for sex (the interruption groove on the upper lateral incisors, and the protostylid). Random forest modelling was used to assess the classification accuracy of the traits. Univariate models assessing each trait individually yielded accuracies ranging from 28% to 70%, III with the number of cusps of the lower second molar (LM2-4C) and the presence of a metaconulid (LM1-C7) on the lower first molar performing the best. The multivariate model including all traits yielded a classification accuracy of 78%. Although the classification accuracy is comparable to other frequently employed methods (such as metric and morphoscopic methods applied to the cranium), several considerations must be accounted for when attempting to use this method. Firstly, the sample was influenced by patterns of antemortem and postmortem tooth loss as well as dental wear, which was primarily a problem caused by the advanced age of many individuals in the skeletal collection. Thus, traits were often not available to score or may have been obscured or altered. Secondly, many traits or specific trait states were noted to be exceedingly rare in the sample. Rare traits and variants may not only affect a user’s ability to correctly identify and score a trait (as is evidenced by the poor interobserver repeatability), but also offers limited information in terms of classification models and the estimation of population affinity. The current study failed to observe the promising results in the South African population reported in other global studies. Thus, based on the results the dental traits should be further explored on a more extensive sample. But in its current state, the morphoscopic method should not be used on its own in skeletal analyses in South Africa.