Genomic Selection in Citrus Breeding: Improved Accuracy with Additional Population Data

Genomic selection (GS) is an attractive approach for citrus breeding by avoiding time-consuming test cross required for developing DNA markers of the trait of interest. We developed a preliminary GS model using 111 parental citrus varieties for 17 traits, and the GS models showed high prediction accuracies (>0.7) on fruit weight, fruit hardiness, peelability, colors of rind or flesh and firmness of locule membrane. In this study, we incorporated additional trait data for 676 progenies from 35 families that were obtained from last three years, then reconstructed the GS model with 1,841 SNPs marker data. The improved GS models that were constructed from the data with the parental varieties and the additional breeding population data showed the improved accuracies on 15 of 17 traits except bitterness and taste, and it indicated the validity for the use of wide genetic data. These additional data also identified more SNP that showed significant correlations for these traits by genome-wide association study (GWAS). The results suggest that additional family data obtained from various genetic backgrounds are effective to improve the accuracy of prediction, and also valuable for identifying candidate loci on GWAS approach.