Population History and Structure of Modern Horse Breeds

Following the domestication of the horse (Equus caballus) several breeds were developed to serve specific purposes such as harness work, warfare, pleasure, and competition. The introgression of older breeds into these developing breeds is not well documented or understood in most cases. We inferred the history of the breeds and modeled migration events (gene flow) between populations using 49,140 single nucleotide polymorphisms genotyped in 795 individuals from 37 populations using TreeMix (Pickrell & Pritchard, 2012). Positive scaled residuals between populations and a significant increase in explained variance in relatedness between populations as migration events were sequentially added demonstrated that the population structure of horses is not well explained by a simple bifurcating tree, such as those reported by previous studies. The maximum likelihood bifurcating tree inferred from this dataset explained only 73.8% of the variance in relatedness between populations. Furthermore, adding only one migration event, from the Thoroughbred to Maremmano, increased the variance explained to 93.1%. Gene flow between these two breeds is known to some extent, as Maremmano horses were often crossed with Thoroughbreds to increase their size. Thoroughbreds have been particularly influential; in the phylogenetic network with 20 migration edges, 9 involved Thoroughbreds and their ancestors. We also assessed admixture amongst breeds using the variational Bayesian framework implemented in fastSTRUCTURE (Raj et al., 2014). Similar to other domesticated species, admixture has been an important force influencing the diversity of modern horses; values of K between 19 and 28 appear to best explain structure in the dataset.