Genetic architecture and genomic selection of female reproduction traits in rainbow trout

Abstract

Background: Rainbow trout is a significant fish farming species under temperate climates. Female reproduction traits play an important role in the economy of breeding companies with the sale of fertilized eggs. The objectives of this study are threefold: to estimate the genetic parameters of female reproduction traits, to determine the genetic architecture of these traits by the identification of quantitative trait loci (QTL), and to assess the expected efficiency of a pedigree-based selection (BLUP) or genomic selection for these traits.

Results: A pedigreed population of 1343 trout were genotyped for 57,000 SNP markers and phenotyped for seven traits at 2 years of age: spawning date, female body weight before and after spawning, the spawn weight and the egg number of the spawn, the egg average weight and average diameter. Genetic parameters were estimated in multi-trait linear animal models. Heritability estimates were moderate, varying from 0.27 to 0.44. The female body weight was not genetically correlated to any of the reproduction traits. Spawn weight showed strong and favourable genetic correlation with the number of eggs in the spawn and individual egg size traits, but the egg number was uncorrelated to the egg size traits. The genome-wide association studies showed that all traits were very polygenic since less than 10% of the genetic variance was explained by the cumulative effects of the QTLs: for any trait, only 2 to 4 QTLs were detected that explained in-between 1 and 3% of the genetic variance. Genomic selection based on a reference population of only one thousand individuals related to candidates would improve the efficiency of BLUP selection from 16 to 37% depending on traits.

Conclusions: Our genetic parameter estimates made unlikely the hypothesis that selection for growth could induce any indirect improvement for female reproduction traits. It is thus important to consider direct selection for spawn weight for improving egg production traits in rainbow trout breeding programs. Due to the low proportion of genetic variance explained by the few QTLs detected for each reproduction traits, marker assisted selection cannot be effective. However genomic selection would allow significant gains of accuracy compared to pedigree-based selection.

Keywords: Body weight; Egg size; Egg weight; Fecundity; Fish; GWAS; Heritability; QTL; Spawn weight; Spawning date.

Fig. 1

Manhattan plot of QTL detected under Bayesian GWAS for female reproduction traits. The red line corresponds to the threshold logBF > 6.0 for defining evidence for a QTL; SD: spawning date; SW: spawning weight; EN: egg number; EW: egg average weight; ED: egg average diameter

Fig. 2

Boxplots of accuracy of GBLUP versus BLUP in 40 simulations with a large training population T+ (A) and a small training population T- (B). SD: spawning date; FW: female body weight; SW*: spawning weight (adjusted for FW); EN*: egg number (adjusted for FW); EW: average egg weight; ED: average egg diameter

Fig. 3

Selection accuracy for BLUP (light colors) or GBLUP (dark colors) considering the training scenarios T1 (diagonal hatching bars), T2 (horizontal hatching bars) and T- (full bars). For scenario T-, the sampling standard deviation over 40 replicates is indicated by a black segment above and below the mean. SD: spawning date; FW: female body weight; SW*: spawning weight (adjusted for FW); EN*: egg number (adjusted for FW); EW: average egg weight; ED: average egg diameter

Fig. 4

Pedigree structure of cohorts C1 and C2 produced in 2014 and 2015, respectively. MGP: maternal grandparents of C1 and C2; PGS1 (PGS2): paternal grandsires of C1 (C2); D1 (D2) and S1 (S2): dams and sires of C1 (C2)