Genomic selection with GWAS-identified QTL markers enhances prediction accuracy for quantitative traits in poplar (Populus deltoides)

Poplar (Populus deltoides) serves as a model tree species with economic importance for wood and biomass production. Genomic genetic improvement of traits is crucial for accelerating tree breeding programs. In this study, we systematically characterized phenotypic variation across ten traits related to growth, wood properties, disease resistance, and leaf morphology in 237 poplar accessions. Phenotypic variation analysis revealed substantial variability among individuals, with coefficients of variation ranging from 4.86% to 73.49%. Narrow-sense heritability estimates indicated genetic contributions ranging from 6.23% to 66.84% for ten traits. A genome-wide association study identified 69 significant quantitative trait loci (QTL) distributed across various chromosomes, strongly associated with traits and implicating 130 annotated genes such as late embryogenesis abundant protein, uridine nucleosidase, and MYB transcription factor. Furthermore, the effects of QTL alleles were significantly correlated with phenotypic values. The integration of multi-trait QTL as random effects into genomic selection (GS) models significantly enhanced prediction accuracy, with an increase ranging from 0.06 to 0.48. Specially, the Bayesian Ridge Regression (BRR) model exhibited superior prediction accuracy for multiple traits. This study provides critical insights into the genetic basis of important traits in poplar, facilitating accelerated breeding efforts and enhancing genetic gains in forestry.