Single-step genomic prediction for body weight and maturity age in Finnish rainbow trout (Oncorhynchus mykiss)

Abstract

The use of genomic information has been proven to be a highly effective in predicting genomic breeding values (GEBV) across various species, including aquatic organisms. In the Finnish national rainbow trout breeding programme, the integration of genomic selection holds particular significance for the traits recorded on sibling fish reared in the main commercial sea production environment, given the selection occurs among the breeding candidates reared in the freshwater nucleus. In the programme, family tanks allow to maintain a pedigree for a large number of fish, and genotyping of a portion of the fish accompanied with a single-step genomic evaluation (ssGBLUP) would maintain high selection intensity and simultaneously make use of possibilities of genomic selection. In this study we used three different statistical approaches to quantify the selection accuracy of ssGBLUP evaluation of body weight and maturity age, relative to the evaluation based on the traditional siredam-offspring pedigree (PBLUP). The data included 600,409 fish in the pedigree among which 214,410 and 4573 were phenotyped for the reported traits and genotyped, respectively. Firstly, a phenotypic cross validation study showed that ssGBLUP had a slightly better prediction power for body weight and maturity age recorded at the sea, with an average 2.7% relative increase in accuracy compared to PBLUP. Secondly, a linear regression (LR) of GEBVs computed using either full or reduced dataset demonstrated that the ssGBLUP model had a consistently lower bias and dispersion compared to the PBLUP model, underscoring its efficacy in dealing with complex datasets like ours. When considering the reliability of [G]EBV predictions, the use of ssGBLUP model resulted in a significant improvement. There is, on average, a notable 50% relative increase in the reliability of predictions for the sea-recorded traits. Thirdly, the enhancement in reliability was further evidenced by the individual assessment of [G]EBVs computed using the reverse reliability methodology. Notably, genotyped individuals experienced an average increase of 0.27 units in reliability, while ungenotyped individuals experienced a corresponding increase of 0.03 units. The results show that the ssGBLUP method had higher prediction accuracy for both sea and freshwater traits compared to PBLUP. The developed ssGBLUP model will be instrumental in Finland’s rainbow trout breeding, facilitating precise and efficient selection of new candidates.