Fresh market tomato (Solanum lycopersicum) is one of the most important vegetable crops in the US, but this labor-intensive industry faces severe labor shortages and rising production costs amid heavy competition from lower-cost imports. With labor accounting for over 30% of total production expenses, much of which is due to harvesting, the industry's long-term sustainability depends on developing more labor-efficient systems. Mechanical harvesting presents a promising solution, but tomato fruit are highly susceptible to bruising, a challenge that could be amplified by mechanized handling. Fruit firmness plays a crucial role in resistance to internal bruising, making it a key breeding target for improving harvest efficiency and post-harvest quality. The UF/IFAS tomato breeding program has developed tomato lines with traits beneficial for mechanical harvesting, including compact growth habit (CGH) and increased fruit firmness. To investigate the genetic basis of fruit firmness in CGH lines, bi-parental populations were developed from firm and soft inbred parents. Genome-wide association analysis identified multiple minor-effect QTLs, confirming the quantitative nature of this trait in the population. Variance component analysis revealed that fruit firmness is primarily controlled by additive genetic variance, suggesting a strong potential for improvement through selection with appropriate strategies such as genomic selection (GS), which has been successfully used to improve quantitative traits in many crop species. GS models were successfully trained to predict fruit firmness, demonstrating the feasibility of integrating GS into the UF/IFAS tomato breeding program. Model optimization, including adjustments to training population size, marker density, and the incorporation of significant QTLs as fixed effects, improved prediction accuracy and computational efficiency. This study confirms the presence of significant fruit firmness variability in UF/IFAS germplasm, supporting its use in breeding firmer CGH tomatoes suited for mechanical harvest. Future research will refine GS models by incorporating multi-trait and multi-environment analyses, leveraging variance-covariance relationships to enhance prediction accuracy and accelerate genetic gains.
