SNP and haplotype-based genomic prediction of fruit quality traits in sweet cherry (Prunus avium)

Abstract

The use of haplotypes in genomic prediction, instead of the widely used individual SNP approach, may improve the accuracy of genomic predictions due to a high probability of complete linkage disequilibrium (LD) between haplotypes and causative quantitative trait loci (QTLs), more accurate estimation of realized relationships between pairs of individuals, and ability to capture local epistatic interactions. We estimated genomic breeding values for three sweet cherry traits (fruit firmness (FF), soluble solids content (SSC) and titratable acidity (TA)) using individual SNPs and genealogy-based haplotypes fitted as pseudo-biallelic loci in a population of 334 sweet cherry individuals obtained from the RosBREED project. 1617 high-quality SNPs obtained from the sweet cherry 6K Array and 2087 haplotypes from 196 genealogy-based haploblocks were used to perform 5-fold cross-validations in a GBLUP framework. We found no differences in predictive ability (PA) between pseudo-biallelic loci and individual SNPs. The highest genomic PA was observed for fruit firmness (0.67), and the lowest PA was observed for titratable acidity (0.45). Predictive abilities (0.67, 0.51 and 0.45 for FF, SSC and TA, respectively) from GBLUP models using both pseudo-biallelic loci and individual SNPs were significantly (p<0.05) higher than pedigree-based predictive abilities (0.64, 0.48 and 0.35 for FF, SSC and TA, respectively). We concluded that using pseudo-biallelic loci instead of individual SNPs did not improve or reduce PA. Further research will be undertaken to compare these models for traits with lower heritability and populations with different structures. © 2023 International Society for Horticultural Science. All rights reserved.