Predictions of native American population structure using linguistic covariates in a hidden regression framework

Background

The mainland of the Americas is home to a remarkable diversity of languages,and the relationships between genes and languages have attractedconsiderable attention in the past. Here we investigate to which extentgeography and languages can predict the genetic structure of Native Americanpopulations.

Methodology/Principal Findings

Our approach is based on a Bayesian latent cluster regression model in whichcluster membership is explained by geographic and linguistic covariates.After correcting for geographic effects, we find that the inclusion oflinguistic information improves the prediction of individual membership togenetic clusters. We further compare the predictive power ofGreenberg''s and The Ethnologue classifications ofAmerindian languages. We report that The Ethnologueclassification provides a better genetic proxy than Greenberg''sclassification at the stock and at the group levels. Although highpredictive values can be achieved from The Ethnologueclassification, we nevertheless emphasize that Choco, Chibchan and Tupilinguistic families do not exhibit a univocal correspondence with geneticclusters.

Conclusions/Significance

The Bayesian latent class regression model described here is efficient atpredicting population genetic structure using geographic and linguisticinformation in Native American populations.