Genome wide association and genomic prediction for growth traits in juvenile farmed Atlantic salmon using a high density SNP array

Abstract

BackgroundThe genetic architecture of complex traits in farmed animal populations is of interest from a scientific and practical perspective. The use of genetic markers to predict the genetic merit (breeding values) of individuals is commonplace in modern farm animal breeding schemes. Recently, high density SNP arrays have become available for Atlantic salmon, which facilitates genomic prediction and association studies using genome-wide markers and economically important traits. The aims of this study were (i) to use a high density SNP array to investigate the genetic architecture of weight and length in juvenile Atlantic salmon; (ii) to assess the utility of genomic prediction for these traits, including testing different marker densities; (iii) to identify potential candidate genes underpinning variation in early growth.ResultsA pedigreed population of farmed Atlantic salmon (n = 622) were measured for weight and length traits at one year of age, and genotyped for 111,908 segregating SNP markers using a high density SNP array. The heritability of both traits was estimated using pedigree and genomic relationship matrices, and was comparable at around 0.5 and 0.6 respectively. The results of the GWA analysis pointed to a polygenic genetic architecture, with no SNPs surpassing the genome-wide significance threshold, and one SNP associated with length at the chromosome-wide level. SNPs surpassing an arbitrary threshold of significance (P < 0.005, ~ top 0.5 % of markers) were aligned to an Atlantic salmon reference transcriptome, identifying 109 SNPs in transcribed regions that were annotated by alignment to human, mouse and zebrafish protein databases. Prediction of breeding values was more accurate when applying genomic (GBLUP) than pedigree (PBLUP) relationship matrices (accuracy ~ 0.7 and 0.58 respectively) and 5,000 SNPs were sufficient for obtaining this accuracy increase over PBLUP in this specific population.ConclusionsThe high density SNP array can effectively capture the additive genetic variation in complex traits. However, the traits of weight and length both appear to be very polygenic with only one SNP surpassing the chromosome-wide threshold. Genomic prediction using the array is effective, leading to an improvement in accuracy compared to pedigree methods, and this improvement can be achieved with only a small subset of the markers in this population. The results have practical relevance for genomic selection in salmon and may also provide insight into variation in the identified genes underpinning body growth and development in salmonid species.Electronic supplementary materialThe online version of this article (doi:10.1186/s12864-015-2117-9) contains supplementary material, which is available to authorized users.

Future directions in breeding for disease resistance in aquaculture species

“…While the initial concept of GS was to detect and utilise population-wide linkage disequilibrium between genome-wide markers and QTL (Meuwissen et al, 2001), the benefits of genomic selection also include a more accurate estimate of the genetic relationship between any two individuals than could be given by pedigree records alone, particularly within families . In all studies of aquaculture species to date, the use of GS has resulted in higher prediction accuracy of breeding values than the use of pedigree information alone (Odegård et al, 2014;Tsai et al, 2015;Dou et al, 2016). A prerequisite for genomic selection is a platform to generate high-density SNP marker genotypes across populations of animals and SNP arrays have been developed for several aquaculture species, including Atlantic salmon Yáñez et al, 2016), rainbow trout (Palti et al, 2015), common carp (Xu et al, 2014), and catfish .…”

Mapping of quantitative trait loci associated with size, shape, and parr mark traits using first- and second-generation backcrosses between European and North American Atlantic salmon (Salmo salar)

“…Ssa11 was also found to contain growth QTL in the transAtlantic backcross families of Boulding et al (2008), and Pedersen et al (2013), and in previous studies using European salmon (Houston et al 2009;Gutierrez et al 2012;Besnier et al 2015). Ssa15 (Reid et al 2005;Gutierrez et al 2012), Ssa18 (Gutierrez et al 2012;Tsai et al 2015), and Ssa19 (Gutierrez et al 2012) have also previously been reported to contain growth QTLs.…”

“…All QTL were located relative to a physical map of the Atlantic salmon genome allowing putative candidate genes to be identified and our QTL to be compared with those found in future studies. Finer QTL mapping and genome-wide association study (Gutierrez et al 2015;Tsai et al 2015) will be necessary to see if the candidate genes proposed to underlie these traits continue to be associated with them in future studies.…”

“…Several studies have mapped QTL for growth-related traits in Atlantic salmon due to their economic importance in cultured fish (Reid et al 2005;Boulding et al 2008;Houston et al 2009;Baranski et al 2010;Gutierrez et al 2012;Pedersen et al 2013;Besnier et al 2015;Tsai et al 2015). A recent analysis of growth QTL mapping studies shows that QTL for length, weight, and condition factor often mapped to the same chromosome in different studies ( fig.…”

Current Advances in Functional Genomics in Aquaculture

“…These traits inluence the commercial value of farmed aquatic organisms. Traits for body weight and length have been identiied in several ish species such as Atlantic salmon [149], rainbow trout [150], Big heard carp (H. nobilis) [151], common carp [152,153] and tilapia (Oreochromis niloticus) [154], nine spined stickleback (Pungitius pungitius) [155] and Arctic char (Salvelinus alpinus) [156]. In shrimps and prawns, body weight and length traits have been identiied in kruma shrimp [157,158], Chinese shrimp [159], Giant fresh water prawn [160], Ridge white prawn [161] and Oriental river prawns [162].…”