Multiple-trait genomic prediction of harvest and fillet traits in Asian seabass (Lates calcarifer, Bloch 1790)

Growth performance and fillet yield are considered economically important traits in Asian seabass farming. However, measurement of fillet traits is difficult, expensive and not possible on the selection candidates. The aim of this study was to assess the utilization of multi-trait genomic selection to improve breeding value prediction accuracy for these traits. A total of 1047 fish were genotyped for 29,521 single nucleotide polymorphisms. Pedigree based on eleven microsatellite loci was available. Phenotypic data at harvest, including body weight (BW), standard length (SL), fillet weight (FW) and fillet yield (FY) were used in the analysis. Traditional best linear unbiased prediction (BLUP) and genomic BLUP (GBLUP) were implemented via single-and two-trait models to predict estimated breeding values (EBV) and genomic (G)EBV. Prediction accuracies and bias were evaluated using five-fold cross-validation with five replications. Compared to BLUP, GBLUP increased the prediction accuracies by 25% for BW, SL, FW and 10% for FY. Within GBLUP, multi-trait prediction performed slightly better than single-trait analysis. Based on the standard errors and bias, GEBV predictions were more accurate for BW, SL and FW than for FY. Prediction accuracy for FY was substantially improved with inclusion of other correlated body size traits in the validation set, and ranged from 5 to 76% relative to multi-trait GBLUP. This strategy provided an alternative to improve fillet yield in Asian seabass.

Automated summary

The study evaluated the utilization of multi-trait genomic selection in improving breeding value prediction accuracy for growth performance and fillet yield in Asian seabass farming. Compared to traditional methods, genomic BLUP significantly increased the prediction accuracies for these traits. Additionally, multi-trait prediction under GBLUP showed slightly better performance than single-trait analysis.