Evaluation of Genome-Enabled Selection for Bacterial Cold Water Disease Resistance Using Progeny Performance Data in Rainbow Trout: Insights on Genotyping Methods and Genomic Prediction Models

Abstract

Bacterial cold water disease (BCWD) causes significant economic losses in salmonid aquaculture, and traditional family-based breeding programs aimed at improving BCWD resistance have been limited to exploiting only between-family variation. We used genomic selection (GS) models to predict genomic breeding values (GEBVs) for BCWD resistance in 10 families from the first generation of the NCCCWA BCWD resistance breeding line, compared the predictive ability (PA) of GEBVs to pedigree-based estimated breeding values (EBVs), and compared the impact of two SNP genotyping methods on the accuracy of GEBV predictions. The BCWD phenotypes survival days (DAYS) and survival status (STATUS) had been recorded in training fish (n = 583) subjected to experimental BCWD challenge. Training fish, and their full sibs without phenotypic data that were used as parents of the subsequent generation, were genotyped using two methods: restriction-site associated DNA (RAD) sequencing and the Rainbow Trout Axiom® 57 K SNP array (Chip). Animal-specific GEBVs were estimated using four GS models: BayesB, BayesC, single-step GBLUP (ssGBLUP), and weighted ssGBLUP (wssGBLUP). Family-specific EBVs were estimated using pedigree and phenotype data in the training fish only. The PA of EBVs and GEBVs was assessed by correlating mean progeny phenotype (MPP) with mid-parent EBV (family-specific) or GEBV (animal-specific). The best GEBV predictions were similar to EBV with PA values of 0.49 and 0.46 vs. 0.50 and 0.41 for DAYS and STATUS, respectively. Among the GEBV prediction methods, ssGBLUP consistently had the highest PA. The RAD genotyping platform had GEBVs with similar PA to those of GEBVs from the Chip platform. The PA of ssGBLUP and wssGBLUP methods was higher with the Chip, but for BayesB and BayesC methods it was higher with the RAD platform. The overall GEBV accuracy in this study was low to moderate, likely due to the small training sample used. This study explored the potential of GS for improving resistance to BCWD in rainbow trout using, for the first time, progeny testing data to assess the accuracy of GEBVs, and it provides the basis for further investigation on the implementation of GS in commercial rainbow trout populations.

Applications of genotyping by sequencing in aquaculture breeding and genetics

“…, ; Vallejo et al . ) data, demonstrating the clear advantages over pedigree‐based methods. Studies using varying marker densities for prediction in salmonids have highlighted that as few as a thousand SNPs may be adequate for achieving the gain in selection accuracy versus pedigree approaches (Ødegård et al .…”

“…Indeed, the potential of this approach has already been highlighted for resistance to bacterial cold water disease in rainbow trout (Vallejo et al . ), for growth in Yesso scallop (Dou et al . ), and for resistance to pasteurellosis in gilthead sea bream (Palaiokostas et al .…”

“…IPNV resistance, Houston et al 2008;Moen et al 2009), improvement of polygenic traits using genomic data is more effectively achieved using genomic prediction of breeding values (Meuwissen et al 2001). Studies of genomic selection in aquaculture were first carried out in salmonid fish, with simulated (Sonesson & Meuwissen 2009;Lillehammer et al 2013) and empirical (Ødeg ard et al 2014;Tsai et al 2015Tsai et al , 2016bVallejo et al 2016) data, demonstrating the clear advantages over pedigree-based methods. Studies using varying marker densities for prediction in salmonids have highlighted that as few as a thousand SNPs may be adequate for achieving the gain in selection accuracy versus pedigree approaches (Ødeg ard et al 2014;Tsai et al 2015Tsai et al , 2016b.…”

“…Therefore, it is reasonable to assume that RAD-like techniques may be useful for genomic selection in aquaculture breeding, as typical RAD SNP datasets comprise a few thousand SNPs. Indeed, the potential of this approach has already been highlighted for resistance to bacterial cold water disease in rainbow trout (Vallejo et al 2016), for growth in Yesso scallop (Dou et al 2016), and for resistance to pasteurellosis in gilthead sea bream .…”

“…This application of genomic selection in aquaculture breeding is at a formative stage, and most examples to date have focussed on improved breeding for resistance to infectious diseases (e.g. Ødeg ard et al 2014;Tsai et al 2015Tsai et al , 2016bVallejo et al 2016;Dou et al 2016;Palaiokostas et al 2016). The majority of high-resolution genetic studies in aquaculture species, and applications of genomic selection, have been underpinned by GBS techniques, either by directly providing genotype data or by discovering markers for the design of SNP arrays, which are currently only available for a handful of aquaculture species (e.g.…”