South African cranial variation : a combined metric-macromorphoscopic method for ancestry estimation

Abstract

Ancestry is a fundamental parameter of the biological profile. To date South African forensic anthropologists are only able to successfully apply a metric approach to estimate ancestry from skeletal remains. While a non-metric, or macromorphoscopic (MMS) approach exists, limited research has been conducted to explore its use in a South African population. The method has not been sufficiently tested and validated which is required for anthropological methodology to be compliant with standards of best practice. This study aimed to explore the MMS traits and its covariation with cranial measurements to develop improved methodology for the estimation of ancestry from skeletal remains in South Africa. A suite of 17 MMS traits and 25 standard linear measurements were collected from 660 crania of black, white and coloured South Africans. Inter- and intra-observer agreement was closely scrutinized as visual methods have been shown to be prone to error. The intra-observer agreement ranged from moderate to perfect, with three traits (inferior nasal margin, nasal bone shape, and nasal overgrowth) yielding slightly lower repeatability. Inter-observer agreement was assessed among five individuals with varying levels of general experience and familiarity with the traits. Overall, the observers demonstrated poor to substantial agreement. A group discussion on the scoring procedure, followed by subsequent rescoring of the crania showed a slight increase in overall agreement, with kappa values ranging between moderate and substantial. While general experience does not appear to translate to proficiency with the method, familiarity with the traits and scoring procedure contributes to consistent scores. Thus, method-specific training is essential prior to employing the MMS traits in practice. Technical error of measurement was used to assess the repeatability of the measurements, where the intra-observer error was noted to be lower than the inter-observer error. The greatest disparity was observed with the inter-orbital breadth and mastoid height for both the inter- and intra-observer assessments. The MMS trait frequency distributions revealed substantial group variation and overlap. Ultimately, not a single trait can be considered characteristic of any one population group. Kruskal-Wallis and Dunn’s tests demonstrated significant population differences for 13 of the 17 traits. Black and coloured South Africans, and coloured and white South Africans shared similarities for many of the traits, but black and white South Africans did not present with significant overlap for any trait. ANOVA and Tukey’s honestly significant difference (HSD) test revealed that all measurements were significantly different for ancestry, except the foramen magnum length. Substantial variation and overlap were observed for the measurements among all three groups. Random Forest Modelling (RFM) was used to develop classification models to assess the reliability and accuracy of the variables in identifying ancestry. Models were created for the traits and measurements separately to gauge the discriminatory power of each dataset. A combined model including all data was also created to test if mixed data can better capture cranial variation than individual methods. The MMS model outperformed the metric model, with classification accuracies of 79% and 72%, respectively. Ultimately, the best results were obtained with the mixed model, which yielded an accuracy of 81%. The results indicate that the combination of size and shape data (as quantified with the mixed model) can effectively distinguish between black, white and coloured South Africans despite significant group overlap. Thus, this study has shown the MMS traits to be a valid and tested method, and the population-specific data from this study can be used to add MMS analyses to forensic casework and skeletal analyses in South Africa.