Evaluation of Genomic Selection for Sea Louse Resistance in Atlantic Salmon (Salmo salar L)

In animal breeding, selecting for improved disease resistance has become a common target. To achieve this goal, selection candidates are commonly evaluated using information of relatives that are challenged with the pathogen of interest and relying on pedigree information. The availability of molecular markers panels has led to the search of the best strategies to accelerate the selection process and to improve the accuracy of the evaluations.

2600 Atlantic salmon smolts were experimentally challenged against Caligus rogercresseyi, sea louse species that causes severe losses in the salmon industry.  Parasite load was counted on each fish, and tank and final weight was recorded after the challenge.  DNA was extracted from fin samples and all were genotyped using a 50K SNP array.

Genomic prediction was evaluated using three different models: GBLUP, BayesA and BayesB. The models differ with respect to assumptions regarding distribution of marker effects.  The predictive ability of genomic predictions was evaluated as the correlation between the predicted phenotype and the true phenotype, using five-fold cross validation.  Preliminary results indicate that the predictive ability of genomic selection approaches show a mean of 0.63 for this trait, outperforming the prediction accuracy obtained with classical pedigree-based approach in up to 4%.

These results indicate that there is potential to include the molecular information to assess the genetic merit of breeding candidates through genomic selection. The use of a reduced-density marker array should be assessed in order to determine the minimum number of markers needed to obtain the highest predictive ability.