Modeling genetic correlation in microsatellite frequencies associated with covariates and population substructure

Abstract

The survival of an endangered species can depend on how accurately the population structure of that species is identified. By determining the substructuring of a species, wise management can be facilitated. A popular tool for the detection and estimation of population structure is information extracted from genotypic data from individuals within populations. In this dissertation I develop a new method that models genetic correlation structure and relates it to a covariate. Two sources of structure are isolated: (1) substructuring of a population into genetically distinct substocks, and (2) genetic correlation within a substock corresponding to a measurable covariate. My modeling approach is based on match probabilities for different types of allele pairs, and on marginalization of a beta-binomial probability model. My statistical model can be fit by adapting GAM fitting methodologies. Hypotheses can be tested using permutation methods. In order to evaluate the performance of my method I examine diverse simulations. I consider both one-population and two-population cases. In the one-population case, within-stock correlation attributable to a covariate is the influential factor, while for the two-population case both within-substock correlation and population substructuring can be detected. In these studies, I analyze the influence of various simulation parameters and compare the performance of my method with other related methods. Generally, my method is shown to have good power to detect all but the tiniest effect sizes in datasets limited to a small number of loci and samples. I also examine the performance of my method by applying it to real data. The two examples I consider pertain to the Bering-Chukchi-Beaufort Seas stock of bowhead whales and to black-tailed prairie dogs living in northern Colorado. In both cases application of my method is found to corroborate results from previous research. The dissertation concludes with a discussion of some of the strengths and weaknesses of my approach, and some consideration of potential future research.