Genomic prediction of 10 metamorphic traits of captive-bred Japanese eels (Anguilla japonica) using the GRAS-Di genotyping method

The Japanese eel (Anguilla japonica) is one of the most important aquaculture fish species in Eastern Asia. A major bottleneck in the mass production of glass eels is the long larval period that they undergo in captivity. In the present study, 1689 glass eels, originating from a partly factorial cross (43 sires x 32 dams), were used to investigate the potential value of genomic selection for 10 traits related to the timing of metamorphosis and the body size at that time. Genotyping by random amplicon sequencing-direct (GRAS-Di (R)) sequencing was used to identify and genotype single nucleotide polymorphism (SNP) markers, which were used to assign parentage and as a basis for genomic selection. The genomic heritabilities of the 10 metamorphic traits obtained by genomic best linear unbiased prediction (GBLUP) models were moderate (range 0.21-0.38), and the standard errors were smaller than those determined by the pedigree-based best linear unbiased prediction (ABLUP) models. The genomic prediction of breeding values outperformed the pedigree-based prediction, resulting in a 3%-19% improvement in the prediction accuracy. The impact of reduced SNP densities on the predictive ability of GBLUP was tested using a linkage disequilibrium-based pruning and random thinning method to reduce 17,017 SNPs to 2117 SNPs. An SNP set of 8000 was considered to be sufficient to achieve predictive ability comparable to that obtained using higher SNP densities for both methods. The selective genotyping of the top individuals in the training population led to a severe reduction in the predictive ability compared with the random genotyping of the same number of samples. These results will be useful for decision-making in establishing best-practice breeding programs when applied to genomic selection for metamorphic traits for shortening the larval period and improving seed quality in Japanese eels.