The study of the genomic selection of white gill disease resistance in large yellow croaker (Larimichthys crocea)

White gill disease (WGD) is one of the major threats to the large yellow croaker, which caused severe economic losses to the breeding industry in recent years. Genomic selection (GS) is a promising approach for breeding WGD-resistant varieties. In this study, we collected 590 WGD-susceptible individuals from the sea area during the outbreak of WGD, and also collected 1493 disease-resistant individuals 2 months later when no fish were dying from WGD. These individuals were whole genome sequenced and 10,201,978 SNPs were discovered. With these SNPs, the heritability was estimated as 0.33. Genomic estimated breeding value (GEBV) was obtained using genomic best linear unbiased prediction (GBLUP). We used cross-validation to evaluate the prediction accuracies, with a large population size (2083 individuals) the prediction accuracy reached 0.54 as compared with 0.43 when the population size was 500. When the marker number was decreased to 30 K, the prediction accuracy was quite close to that using whole genomic variants. When the training and validation sets were closely related, the prediction accuracy was increased by at least two folds relative to distant relatives. These suggest that the GS for WGD needs a relatively big sample size, enough markers (>30 K), and a relatively close kinship between training and validation sets.