Genomic prediction and association analyses for breeding parthenocarpic blueberries

Abstract

Parthenocarpy is a desirable trait that enables fruit set in the absence of fertilization. While blueberries typically depend on pollination for optimal yield, certain genotypes can produce seedless fruits through facultative parthenocarpy, eliminating the need for pollination. However, the development of parthenocarpic cultivars has remained limited by the challenge of evaluating large breeding populations. Thus, establishing molecular breeding tools can greatly accelerate genetic gain for this trait. In the present study, we evaluated two blueberry breeding populations for parthenocarpic fruit set and performed genome-wide association studies (GWAS) to identify markers and candidate genes associated with parthenocarpy. We also compared the predictive ability (PA) of three molecular breeding approaches, including i) genomic selection (GS); ii) GS de novo GWAS (GSdnGWAS), which incorporates significant GWAS markers into the GS model as prior information; and iii) in-silico marker-assisted selection (MAS), where markers from GWAS were fitted as fixed effects with no addition marker information. GWAS analyses identified 55 marker-trait associations, revealing candidate genes related to phytohormones, cell cycle regulation, and seed development. Predictive analysis showed that GSdnGWAS consistently outperformed GS and MAS, with PAs ranging from 0.21 to 0.36 depending on the population of study and the specific markers utilized. MAS showed PAs comparable to GS in some cases, suggesting it could be a cost-effective alternative to genome-wide sequencing. Together, these findings demonstrate that molecular breeding techniques can be used to improve facultative parthenocarpy, offering new avenues to develop high-yielding blueberry varieties that are less reliant on pollination.