Pervasive selection pressure in wild and domestic pigs

Abstract

Animal domestication typically affected numerous polygenic quantitative traits, such as behaviour, development and reproduction. However, uncovering the genetic basis of quantitative trait variation is challenging, since it is probably caused by small allele-frequency changes. To date, only a few causative mutations related to domestication processes have been reported, strengthening the hypothesis that small effect variants have a prominent role. So far, the studies on domestication have been limited to the detection of the global effect of domestication on deleterious mutations and on strong beneficial variants, ignoring the importance of variants with small selective effects. In addition, very often, the study of the effects of selection are conducted on genome sequences that contain a non-negligible fraction of missing data, especially in non-model organisms. Hence, appropriate methods to account for these positions are needed. To overcome these difficulties, here we propose to estimate the proportion of beneficial variants using the asymptotic MacDonald-Kreitman (MK) method based on estimates of variability that summarizes the site frequency spectrum (SFS) while accounting for missing data and use them to perform an Approximate Bayesian Computation (ABC) analysis to infer the Distribution of Fitness Effects (DFE) of each population. We applied this approach to 46 genome sequences of pigs from three different populations, one wild and two domestics, with very different demographic histories and selective pressures. The obtained results showed that domestic and wild pig populations do not differ in nonsynonymous fixed mutations. Therefore, differences in estimation among breeds are determined by their polymorphisms. The comparison of for total and exclusive mutations suggests that the different domestic populations have suffered recent divergent changes in their functional versus neutral polymorphisms ratio, while the wild population is compatible with =0. Besides, the DFE inferred with ABC indicates that both wild and domestic pigs display a large number of deleterious mutations at low frequency and a high number of neutral and/or nearly-neutral mutations that may have a significant effect on the evolution of domestic and wild populations. In addition, models not considering beneficial mutations have higher posterior probabilities, suggesting that beneficial mutations are difficult to detect or are scarce. Indeed, for all three populations, the median proportion of the strong favourable mutations are very low ([≤] 0.1%) in those models that includes positive selection, with the average values of weak beneficial mutations around 0.6% for wild boar and 0.8-1.0% for the domestic pigs. Lastly, the analysis based on exclusive mutations showed that recent demographic changes may have severely affected the fitness of populations, especially that of the local Iberian breed.