How demography shapes linkage disequilibrium with or without recombination

Abstract

Linkage Disequilibrium (LD) is a measure of the statistical association, within a population, between alleles present at different loci. Here, we produce detailed characterizations of LD statistics in different demographic scenarios, with simulations and with analytic results, focusing on several widely used measures of LD (D, |D'|, and r2), and introducing a fourth class of statistics: 2-site configuration probabilities, which correspond to the frequencies of different allelic association patterns. By studying these measures under no recombination, a single recombination event, and between fully independent loci, we are able to gain new insights to how these measures vary across recombination scales. We also classify recombination events into four types and rank them according to their respective frequencies and effects on LD measures. Using previously published human data, we show how tree-sequence information can be used to partition the genome and to analyze LD in relation to theoretical expectations. Finally, we provide recommendations for the use of the different statistics and suggest applications of LD measures in the absence of recombination.