081
Genomic Predictions in Aquaculture: Reliabilities in an Admixed Atlantic Salmon Population
Reliability of genomic selection (GS) models was tested in an admixed population of Atlantic salmon, originating from crossing of several wild subpopulations. The GS models included ordinary genomic BLUP models (IBS-GS), using varying marker densities (1 to 220K) and a genomic IBD model (IBD-GS) using genomic relationships estimated through linkage analysis of sparse markers (ignoring LD). The models were compared based on 5-fold cross-validation. The traits studied were log density of salmon lice on skin (logDL) and fillet color (FC), with respective estimated heritabilities of 0.14 and 0.43. IBD-GS and IBS-GS (220K) had similar reliabilities’ for FC, while IBS-GS was superior for logDL. The IBS-GS model was remarkable robust to marker density, especially for logDL, and outperformed pedigree-based models at all densities, which may be explained by admixture of the population, introducing long-range LD. Increasing SNP densities beyond 22K had limited effect for both traits.
Keywords:
Atlantic salmon, genomic selection, admixture
