GWAS-informed genomic selection for cold tolerance in pepper (Capsicum annuum L.)

Background Breeding for cold tolerance in pepper ( Capsicum annuum L.) is critical to mitigate yield losses caused by unpredictable temperature fluctuations associated with climate change. However, genetic improvement of this trait is hindered by challenges in accurate phenotyping, particularly at the adult stage, and by its complex genetic architecture involving numerous minor-effect loci. While genomic selection (GS) offers a promising solution to accelerate genetic gain, its predictive ability is often limited by statistical noise from uninformative markers within whole-genome marker sets. This study aimed to overcome this limitation by developing a robust phenotypic index and implementing a genome-wide association study (GWAS)-informed GS strategy. Results We phenotyped 192 pepper accessions from a core collection for cold tolerance using a visual survival score (Surv) and a newly developed composite cold-tolerance index (CTI). Both CTI ( h ²  = 0.55) and Surv ( h ²  = 0.53) showed moderate heritability, suggesting a substantial contribution from additive genetic variance to the phenotypic variation of cold tolerance in adult plants. GWAS identified 13 candidate genomic regions associated with cold tolerance; these regions included TRM9 , CAP1 , and PP2A-2 , genes previously implicated in abiotic stress responses. A GWAS-informed GS model using a selected subset of 1,024 markers achieved a prediction accuracy of 0.78, representing a substantial improvement over that obtained with the standard model using the full marker set of 73,502 markers (0.203). Notably, a control model using a random marker set of identical size (1,024 markers) yielded an accuracy of only 0.122, confirming that the greater predictive power of the GWAS-informed GS model was driven by the genetic relevance of the selected markers rather than by lower marker density. Conclusions Our study demonstrates that assessing cold tolerance via the CTI is pivotal for overcoming the limitations of small-scale sample collection. By turning ordinal data into a continuous spectrum, the CTI can effectively unmask hidden genetic variation. Integrating this refined phenotype with GWAS-informed marker selection into prediction models significantly enhanced the accuracy of genomic prediction for cold tolerance in adult pepper plants. This integrated framework offers a practical and efficient roadmap for accelerating breeding cycles and improving selection precision for complex abiotic stress traits in pepper breeding.