Evaluating genomic selection for bacterial cold water disease resistance in rainbow trout using RAD-seq genotyping and a 57K SNP chip.

Background: Bacterial cold water disease (BCWD) causes significant economic losses in salmonid aquaculture. At the National Center for Cool and Cold Water Aquaculture (NCCCWA), we have pursued selective breeding to increase rainbow trout genetic resistance against BCWD and found that BCWD resistance is moderately heritable and responds to selection. Genomic selection (GS) is a recently developed methodology that is revolutionizing animal breeding. In this pilot study, we used GS models to predict genome-enabled breeding values (GEBVs) for BCWD resistance in 10 families from the first generation of the NCCCWA BCWD resistance breeding line, compared the reliability of GEBVs to pedigree-based estimated breeding values (EBVs) and compared the impact of two SNP genotyping methods on the reliability of GEBV predictions.Methods: The BCWD phenotypes survival days (DAYS) and survival status (STATUS) were recorded in training animals (n=583). The animals were genotyped using two methods; restriction-site associated DNA (RAD) sequencing and a 57K SNP chip. The GEBVs were estimated using Bayesian variable selection and single-step GBLUP models. The reliability of GEBVs was assessed using validation animals (n=53) that had progeny testing-based EBV records. The reliability was assessed through predictive ability defined as the correlation between progeny testing EBVs and GEBVs.Results: Overall, the reliability of GEBV estimated with GS models was similar to the pedigree-based EBVs. The RAD genotyping platform (~10K informative SNPs) was as efficient as the SNP Chip (~42K SNPs).Conclusions: This study demonstrated the potential advantages of implementing GS for BCWD resistance in rainbow trout sib-testing selection breeding programs. However, the training sample size in this pilot study was small, and hence we expect higher reliability of GEBV predictions in larger commercial rainbow trout breeding populations. This study provides the basis for further investigation on the use of GS in commercial rainbow trout populations, including the potential for its implementation. Although RAD SNP genotyping is a viable method for predicting GEBVs, we find that the SNP chip is more robust and practical for real time breeding in rainbow trout aquaculture.