Methods to detect selection history in a population under ongoing directional selection

The aim of animal breeding is to select the genetically best animals to improve the performance of future generations for a specific breeding goal. Knowledge of the (indirect) selection history of new traits is valuable before adding it to the breeding goal. Two methods have been developed to assess the selection history: BayesS estimates a parameter (s) that reflects the relationship between estimated additive effects and minor allele frequency of markers, while G^ calculates the expected genetic change of a trait based on allele frequency changes and estimated additive effects of markers. We evaluated the performance of both methods in an animal breeding context, focusing on their ability to detect selection for a trait with low heritability under direct and indirect selection. We simulated direct selection in a commercial pig breeding program under phenotypic selection, with varying heritabilities (0.05, 0.1, and 0.3) across 30 generations. In addition, indirect selection was simulated using a correlated trait with heritability 0.05 and a genetic correlation of 0.4 or 0.7 to a trait with heritability 0.1 under direct selection. Both methods were able to detect selection, where higher heritabilities and a larger sample size (for s-value estimation) or a longer selection interval (for G^) increased detectivity. The detectivity of indirect selection was limited; only G^ identified selection in some scenarios, where estimating marker effects in the starting generation increased detectivity. Overall, we observed that both methods have potential to identify selection but that the preferred method depended on the available data of that population.