Genome-wide association analysis of salmon lice (<i>Lepeophtheirus salmonis</i>) resistance in a North American Atlantic salmon population

Abstract

Our objective was to detect single nucleotide polymorphisms (SNPs) associated with resistance to the salmon louse in the Saint John River aquacultural population of North American Atlantic salmon using estimated breeding values (EBVs) and 6K genotypes from the parent-generation and lice count phenotypes from the challenged, but ungenotyped, offspring-generation. In 2011 and 2012, we challenged recent smolts with approximately 100 copepodids each. Fish were euthanized once the lice reached the chalimus stages and lice count, sex, tank and salt water weight were recorded. We used a multiple trait model to estimate breeding values for the parent-generation using their own fresh water weights and the salt water weights and lice counts of the offspring-generation. Salmon lice count heritability for untransformed and transformed data was 0.17 and 0.29 respectively. Two different genome-wide association study methods were compared: (i) forward multiple linear regression and (ii) a mixed linear model using principal components to correct for population stratification as implemented in the egscore function of GenABEL. The two methods detected different SNPs located on different chromosomes. The multiple regression method incorporated 70 SNPs found on chromosomes 2, 7, 9, 12, 14, 15, 21, 22, 1p/23, 24. Many SNPs entered into the forward multiple regression are likely to be false positives from not correcting for the observed population stratification and cryptic relatedness. In contrast, the mixed linear model identified only two SNPs, one on chromosome 1p/23 (6.9%) and one on chromosome 1q (6.1%) consistent with louse-resistance being a quantitative trait.