Population variation and macromorphoscopic traits in a sample of South African subadults

Skeletal analyses of subadults are lacking reliable methods to estimate biological parameters, such as population affinity. In adults, macromorphoscopic (MMS) traits capture nonmetric cranial variation and assist in assessing population affinity, but its application on subadults remains underexplored in South Africa. This study evaluated the use of MMS traits to assess population affinity in a sample of South African subadults. The sample comprised of cone-beam computed tomography (CBCT) scanned crania of subadults between 3 and 16 years, including black, white, coloured and indian South Africans. Nine MMS traits were examined across population groups. Substantial variation and overlap were observed, with no trait unique to any population. Kruskal-Wallis tests revealed significant population differences (p < 0.05) in six traits, suggesting some population-level patterning. Random Forest Modelling (RFM) was employed to test the ability of these traits to estimate population affinity. The unweighted multivariate model showed moderate accuracy (61%) but low kappa values, indicating limited discriminatory power. Other RFM iterations produced accuracies between 64% and 96.6%, with age-stratified and class-weighted models performing better. Despite these results, overall findings suggest MMS traits alone are insufficient for reliable classification in South African subadults. The observed overlap between groups indicates that craniofacial variation during development is not distinct enough for accurate population affinity estimation using this method. Nevertheless, the significant differences in certain traits and the variable performance of classification models highlight the potential of MMS traits as part of a broader analytical framework. Future research with larger samples is needed to better capture developmental trajectories and refine methods for forensic application.