Integrating multi-model GWAS prior information enhances genomic prediction of cold tolerance traits in Populus simonii.

Abstract

Genomic selection (GS) represents a transformative strategy for accelerating the breeding of cold tolerance in forest trees and other perennial species with long generation intervals. Although integrating genetic loci by genome-wide association studies (GWAS) can enhance prediction accuracy, this potential is frequently constrained by the inconsistency of loci detected across statistical models. Here, we developed a robust GS optimization strategy based on a multi-model GWAS framework using 849 accessions from a half-sib population of Populus simonii. We identified a total of 93 significant loci, among which 29 were co-detected by at least two models, including 8 high-confidence loci consistently detected across all three models. Incorporating these loci as fixed effects in GBLUP improved prediction accuracies ranging from 0.25 to 0.60. Notably, this strategy improved prediction accuracy by up to 60% for complex traits such as superoxide dismutase (SOD) activity, thereby alleviating a key limitation of standard GBLUP in capturing major-effect QTLs. Furthermore, we identified and preliminarily characterized the pleiotropic candidate gene PsiNDHM. Overexpression of PsiNDHM mitigated oxidative damage in poplar under cold stress. Together, these results indicate that leveraging consensus loci from multi-model GWAS offers an effective approach for optimizing GS, providing a methodological framework for precision molecular breeding in species with complex genetic architectures, particularly forest trees.