Abstract
Game species are often manipulated by human beings, whose activities can deeply affect their genetic make-up and population structure. We focused on a geographically isolated wild boar population (Sardinia, Italy), which is classified, together with the Corsican population, as a separate subspecies (Sus scrofa meridionalis). Two hundred and ten wild boars collected across Sardinia were analysed with a set of 10 microsatellites and compared with 296 reference genotypes from continental wild populations and to a sample of domestic pigs. The Sardinian population showed remarkable diversity and a high proportion of private alleles, and strongly deviated from the equilibrium. A Bayesian cluster analysis of only the Sardinian sample revealed a partition into five subpopulations. However, two different Bayesian approaches to the assignment of individuals, accounting for different possible source populations, produced consistent results and proved the admixed nature of the Sardinian population. Indeed, introgressive hybridization with boars from multiple sources (Italian peninsula, central Europe, domestic stocks) was detected, although poor evidence of crossbreeding with free-ranging domestic pigs was unexpectedly found. After excluding individuals who carried exotic genes, the population re-entered Hardy-Weinberg proportions and a clear population structure with three subpopulations emerged. Therefore, the inclusion of introgressed animals in the Bayesian analysis implied an overestimation of the number of clusters. Nonetheless, two of them were consistent between analyses and corresponded to highly pure stocks, located, respectively, in north-west and south-west Sardinia. This work shows the critical importance of including adequate reference samples when studying the genetic structure of managed wild populations.