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


A Multi‐Component Analysis of Mexican Variation with Forensic Implications

CRIS E. HUGHES1 and BRIDGET F.B. ALGEE-HEWITT2.

1Department of Anthropology, University of Illinois Urbana - Champaign, 2Stanford Center for Computational, Evolutionary and Human Genomics & Rosenberg Lab, Department of Biology, Stanford University

March 26, 2015 , Gateway Ballroom 2 Add to calendar

A remarkable amount of genomic diversity among indigenous populations has been found in present-day Mexico. This diversity, coupled with regional variation in European admixture, provide a geographically structured genetic landscape. In the forensic context, understanding this variation can improve identification methods for US-Mexico border fatalities. In this study, we aim to reveal this geographic structure using the biological, CODIS (STR) and craniometric (ILD), markers relevant to forensic applications. Our previous work identified K=2 cluster-model for both data types, supporting moderate differences between northern and southern regions of Mexico. Revealing stronger fine-grained, microregional structure was potentially limited by the number of cases with known region of origin. Here, we greatly increase sampling depth with ≈300 unidentified border-death cases from PCOME and reference data for indigenous Maya and Mexican mestizos. Model-based cluster analyses of STRs and ILDs were performed with and without the reference samples to identify clusters capturing indigenous and mestizo variation.

We find, overall, that models with larger K values are preferred for the genetic data, suggesting that, when predictor numbers are similar, CODIS-STRs encode more structure than ILDs. In fact, ILDs fail to distinguish between Mexicans and Guatemalan Mayans. For STRs, our optimal K=5 model allocates mestizos across two clusters and the unidentified PCOME cases into three clusters, of which at least one contains individuals with high indigenous ancestry (> 0.90). Our clustering trends for the known Mexican cases recapitulate previous work, showing how Northern and Central cases cluster more frequently with mestizos, while the Southern cases do not.