Integrating GWAS-guided markers preselection with genomic selection enhances prediction of pulpwood-related traits in slash pine (Pinus elliottii Englem.).

This study aimed to enhance the efficiency of genomic prediction for pulpwood-related traits in slash pine (Pinus elliottii Engelm. var. elliottii) by integrating genome-wide association study (GWAS) information with genomic selection (GS). We evaluated 12 traits related to growth, fiber, and wood chemical composition in a population of 340 individuals genotyped with 319,286 high-quality SNPs, comparing the performance of six GS models, including GBLUP and Bayesian methods, under varying training population sizes and marker densities. The results showed that while both GBLUP and Bayesian Lasso performed well, Bayesian Lasso slightly outperformed GBLUP for fiber traits. Predictive ability (PA) plateaued at approximately 100 K SNPs for fiber traits, 60 K for DBH, and 10 K for wood chemical composition traits in all models. Using 100 K random SNPs, PA ranged from 0.05 to 0.23, which expanded to 0.09-0.35 with GWAS-guided SNP preselection (maximum improvement of 16.26%) and further broadened to 0.01-0.38 by incorporating large-effect QTLs (greatest improvement of 23.54%). Overall, integrating GWAS information into GS frameworks significantly improved prediction accuracy as assessed by t-test, offering a cost-effective strategy to accelerate genetic improvement. These findings provide practical guidance for enhancing breeding efficiency in slash pine and other conifer breeding programs.