Evaluating the effects of age on pubic morphoscopic traits for sex estimation in a South African sample

The pelvis is one of the most reliable skeletal elements for sex estimation, with three pubic morphoscopic features regarded as highly sexually dimorphic. However, the extent to which age-related skeletal remodeling influences the expression of these traits and their utility for sex estimation remains unclear. This study investigated the effect of age on pubic morphoscopic trait expression and sex estimation accuracy in a South African skeletal sample. The ventral arc, subpubic concavity, and medial aspect of the ischiopubic ramus of a total of 429 ossa coxae from documented South African individuals aged 16-108 years were scored. Age-related variation in trait expression was assessed using cumulative link ordinal logistic regression, while sex estimation performance was evaluated using logistic regression (LR) and random forest models (RFM) in pooled and age-stratified samples. Age demonstrated little influence on trait expression, with only the female subpubic concavity showing a significant age effect. Overall classification accuracies remained high across all models, and age stratification produced minor changes in performance while offering no consistent improvement over pooled models. Although LR and RFMs achieved comparable overall accuracies, LR consistently demonstrated lower sex bias and more stable performance across age groups. These findings indicate that adult age has minimal influence on pubic morphoscopic trait expression and sex estimation accuracy, suggesting age-specific correction factors or age-stratified classification models are unnecessary for these traits in this South African sample. The results further support the importance of population-specific validation when applying methods in forensic anthropological practice.