The 84th Annual Meeting of the American Association of Physical Anthropologists (2015)


Population Structure in the United States: Using Forensic Data Bank Cases to Link Craniometric, Genetic and Social Information

BRIDGET FB. ALGEE-HEWITT.

Department of Biology, Rosenberg Lab & Stanford Center for Computational, Evolutionary and Human Genomics, Stanford University

March 26, 2015 , Gateway Ballroom 2 Add to calendar

This study combines craniometric clustering results with published genetic findings and documented self-identifications to provide a composite picture of population structure, ancestry variation, and personal identity in contemporary America. The unsupervised, model-based clustering methods of finite mixture analysis are used to reveal latent population structure and generate biological estimates of ancestry and proportions of admixture for craniofacial measurements representing known American Black, White and Hispanic individuals from the Forensic Data Bank. The results of these cranial analyses are compared against published data on genetic variation in the U.S. and the self-identifications for each case. Preliminary cluster results for a subset of the FDB craniometric cases (n≈850) reveal strong but complex population structure. Solutions, for which the number of k clusters exceeds the number of true groups, were favored by the model selection criteria and lead to within-population partitioning, suggesting that considerable heterogeneity exists within in each subgroup. Overlap is equally identified between clusters, implying instances of admixture or shared population history. This study shows how these factors, along with expected variability in self-identification, produce a complicated, often inconsistent, relationship between biologically-derived ancestry and sociogeographically-defined membership. In contrast, it reveals similarities in cluster patterns, ancestry estimates and admixture proportions produced from craniometric and molecular data analyses, recapitulating the relationship between neutral phenotypic and genetic variation. This study also explores changes to these patterns with increased sampling and additional skeletometrics. This study’s findings are used to provide recommendations for cranial data analysis in forensic identification and when genetic data is also available.

This study was funded in part by a National Science Foundation Dissertation Improvement Grant, BCS-676917, and support from the Stanford Center for Computational, Evolutionary and Human Genomics.