Skeletal phenotypic variation and modern human evolution in Asian populations
Loading...
Authors
Skipper, Cassie Elisabeth
Issue Date
2022
Type
Dissertation
Language
Keywords
Asia , Forensic anthropology , Machine learning , Microevolution , Population affinity estimation , Sex estimation
Alternative Title
Abstract
The skills of a forensic anthropologist are often requested by medicolegal entities to assist in the process of identifying unknown decedents in medicolegal cases. Obtaining positive identifications requires efficiency and accuracy in the methods used in estimating the biological profile, including the age, sex, population affinity, and stature of the decedent. Current methods used in the United States are predominately based on individuals of European and/or African descent, and as such, these methods do not produce accurate estimates of the biological profile of individuals who do not belong to these groups as evidenced by this research. The lack of variation in current methods is problematic in the United States as there has been increased migration over the last several decades. According to the Migration Policy Institute, the number of Asian immigrants in the US increased by 2,597% from 1960 to 2014 thus representing 30% of the US foreign-born population. In fact, the ‘Asian’ race category is the fastest growing of any race category in the US. As Asian migration into the US continues to rise so does the need to create and utilize population-specific methods for estimating the biological profile of unknown decedents. Concurrent with the need to accurately estimate the biological profile of unknown Asian decedents, is the need to appropriately situate Asian skeletal and dental variation within a microevolutionary framework. The current study presents a multi-faceted approach to answering these needs through an analysis of various datasets of the dentition and cranium. First, current methods for estimating ancestry and sex were tested using Japanese and Asian samples. Based on the results, it is clear that these methods do not accurately represent the variation present in Japanese and Asian American individuals, and Japanese and Asian American females were misclassified approximately one-half of the time. Second, the effects of genetic drift and gene flow were tested in the study samples, and the study samples were found to exhibit greater degrees of genetic drift. Third, various machine learning (ML) techniques were tested to assess their accuracies in estimating population affinity and sex. The cranial nonmetric and macromorphoscopic ML analyses produced the worst results. Overall, the population affinity ML analyses produced better results than the combined population and sex analyses. Finally, the cranial and dental, metric and nonmetric datasets were combined, and the populations were restructured into three broad geographic groups (African, Asian, and European) and tested using ML techniques. The results of this final analysis generally produced better results than when the datasets were tested alone.The results of this study indicate that larger sample sizes of various Asian populations are needed to better capture the range of variation in these groups. Additionally, combining datasets and employing ML techniques can provide better estimates of population affinity and sex than current methods, and as such, may be useful in forensic casework. Finally, the high levels of genetic drift observed in the results are consistent with the relative degrees of cultural and geographic isolation experienced by the Japanese and Asian American study samples.