Craniometric evaluation of population affinity among South African subadults

Although standard craniometrics is typically regarded as the method of choice for estimating population affinity in adults, its application to subadults has received limited attention, despite potential population-specific signals demonstrated in cranial growth patterns. This study evaluated the potential efficacy of interlandmark distances (ILDs) derived from three-dimensional (3D) cranial models obtained from Cone Beam Computed Tomography (CBCT) scans of South African black, coloured, indian and white subadults, and linear discriminant analysis (LDA) to estimate population affinity. Results indicated that the combined midface and cranial base model contained the most population-specific variation, with black South Africans achieving the highest classification accuracy. The best-performing multivariate model achieved a 79.17% accuracy and a kappa of 0.67. Age-stratified models produced slightly lower classification rates, suggesting that early ontogenetic trajectories during childhood may obscure population differences. Indian South Africans were consistently misclassified across all models, likely reflecting both challenges posed by limited sample size and substantial heterogeneity and overlap with other groups. While age had a limited influence on the predictive performance, ILDs show considerable promise for subadult classification, particularly when combining measurements from the facial skeleton and cranial base. These findings indicate that population-specific craniofacial variation is detectable across subadult development and that incorporating the full developmental sample may yield more accurate classifications than age-stratified approaches.