Improving fruit size in sweet cherry via association mapping and genomic prediction.

Large fruit size is one of the most important breeding objectives for sweet cherry. In the Washington State University (WSU) breeding program, large-fruited germplasm exists, and heritability is reasonably high. An important genetic locus for cherry fruit size has been identified but appears to primarily distinguish between wild and improved phenotypes and may have limited utility in the breeding program. We employed association mapping and genomic prediction approaches to identify markers and models for fruit size in germplasm relevant to the breeding program. The germplasm consisted of 247 individuals from two sub-populations. The "RosBREED" sub-population (n = 106) was genotyped with the cherry 6 K Illumina SNP array. The "program" sub-population (n = 141) was genotyped with the cherry 6 + 9 K SNP array. Each sub-population was phenotyped for fruit diameter for two seasons. Sub-populations were analyzed individually as well as combined. SNPs for the combined dataset included those common to both arrays. Significantly associated SNP markers for fruit diameter were identified on all 8 chromosomes using BLINK. Individual markers accounted for up to 33% of phenotypic variance. In all but two cases, the minor allele was associated with smaller fruit. However, minor allele frequencies were 0.25–0.44 for four significant SNPs, indicating the opportunity for continued selection. Seven markers were converted to locus-specific assays for validation and use in marker-assisted selection and showed a concordance rate ≥ 97.6%. Genomic prediction models were developed using BGLR. Selection accuracy ranged from 0.38 to 0.53 depending on the population and year. These results indicate that either association mapping or genomic prediction can be used to select for larger fruit in sweet cherry, although genomic prediction might be more efficient in improved germplasm if a cost-effective genotyping platform is available. [ABSTRACT FROM AUTHOR]