Genomic selection for improvement of Atlantic salmon aquaculture

Since the beginning of aquaculture, breeders have practiced selection of the most suitable specimens for the improvement of their population and at the same time their production. Atlantic salmon (Salmo salar) aquaculture is not an exception and selection efforts have taken place since the beginning of their culture during the early 1970s in Norway. As expected, most of the selection during the early years was done through phenotypic observations of fish and their response to farming conditions. Initial stages of selection in Atlantic salmon targeted traits like growth, timing of sexual maturation and some disease resistance and the results were very positive. Through approximately 40 years of farming (~12 generations), the time to produce a standard market-size 4 kg fish has halved and the rate of food conversion has been amazingly improved. Today, genomic tools promise to be able to develop more efficient selection systems to improve traditional phenotype-based selection systems. Selection could be improved with the identification of genomic regions controlling specific traits of interest, and the currently available 6.5K SNP array that was developed for Atlantic salmon provides an opportunity to do so. In my study I analyzed a Canadian farmed population (Cermaq), which originates from the Norwegian Mowi strain. I aimed to identify genomic regions controlling traits of interest such as growth and sexual development, but I also analyzed the genomic status of the population to learn how selection has affected its genomic diversity. By performing QTL and GWAS analyses I was able to identify multiple genomic regions potentially controlling growth and sexual maturation, one of these regions associated to Maskin, a gene involved in gonad development that shows a high expression in Atlantic salmon ovaries. Additionally, analysis to identify regions under selection allowed me to identify many genomic regions that seem to be selected in the Cermaq population. The polygenic nature of the traits makes it difficult to postulate particular regions to be controlling specific traits; the most likely scenario is that many regions are controlling a particular trait, each of these regions having some effect on the phenotypic outcome.