Application of Genome-Assisted Prediction (GAP) for Apple Fruit Weight

Apple fruit weight is a critical quantitative trait controlled by multiple genes, with numerous quantitative trait loci (QTLs) having been identified. This study evaluated the potential of genome-assisted prediction (GAP) by genotyping 573 hybrids from six populations with 70 previously identified quantitative trait locus (QTL)-based markers. Genetic diversity and marker-trait association (MTA) analyses identified two optimized marker combinations. Results showed that using the screened marker combination with moderate PIC (Polymorphism Information Content, PIC) and HWE-conforming (Hardy-Weinberg Equilibrium, HWE) significantly increased accuracy compared to the full marker set in specific populations, such as 'YM1' × 'Honeycrisp' (Y × H; r increased from 0.115 to 0.204; p < 0.05) and 'Fuping Qiuzi' × 'Ruixue' (Q × RX; r increased from 0.482 to 0.576; p < 0.001). MTA-based (marker-trait association, MTA) marker combinations achieved the highest accuracy in most populations, particularly in Q × RX (r = 0.5766, p < 0.001) and Y × H (r = 0.2475, p < 0.05). In contrast, the full marker set showed population dependence (r ranging from -0.088 to 0.482). These results indicate that the 70 markers were not generally effective across diverse apple hybrids. We propose a procedure for screening effective markers (PIC, HWE, MTA), providing a practical framework and theoretical foundation for implementing GAP in apple breeding.