AUTHOR=Munyengwa Norman , Wilkinson Melanie J. , Ortiz-Barrientos Daniel , Dillon Natalie L. , Webb Matthew , Ali Asjad , Bally Ian S. E. , Myburg Alexander A. , Hardner Craig M. 

TITLE=Increased genomic predictive ability in mango using GWAS-preselected variants and fixed-effect SNPs

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2025.1664012

DOI=10.3389/fpls.2025.1664012

ISSN=1664-462X

ABSTRACT=Genomic selection (GS) using whole-genome sequencing (WGS) data has potential to improve breeding value accuracy in fruit trees, but previous studies have reported limited gains compared to high-density marker sets. Incorporating preselected variants identified through genome-wide association studies (GWAS) is a promising strategy to enhance the predictive power of WGS data. We investigated whether incorporating GWAS-preselected variants and fixed-effect markers into genomic best linear unbiased prediction (GBLUP) models improves predictive ability for fruit blush color (FBC), average fruit weight (AFW), fruit firmness (FF), and trunk circumference (TC) in mango (Mangifera indica L.). The study used 225 gene pool accessions from the Queensland Department of Primary Industries in Australia, with phenotypes collected between 1999 and 2024. Predictive ability was assessed using models that ignored or accounted for population structure using fixed principal components. Accounting for population structure led to substantial reduction in predictive ability across all traits, suggesting that initially high predictive abilities may have been partly driven by genetic differences between subpopulations. GWAS-preselected variants improved predictive abilities compared to using all WGS data, especially when population structure was accounted for in both parental and 5-fold cross-validation. Gains under parental validation reached 0.28 for AFW (from 0.30 to 0.58) and 0.06 for FBC (from 0.44 to 0.50). In 5-fold cross validation, gains were up to 0.16 for AFW (from 0.32 to 0.48) and 0.10 for FBC (from 0.35 to 0.45). This suggests that prioritizing markers that better capture relationships at causal loci can improve predictive ability. Fixed-effect SNPs improved predictive ability of WGS data, particularly for FBC, with increases of up to 0.18 (from 0.44 to 0.62). The combination of GWAS-preselected variants and fixed-effect markers yielded the highest improvements in predictive ability for FBC and TC. GWAS identified 5 trait-associated SNPs for FBC, 11 for AFW, and 8 for TC. These results demonstrate that leveraging GWAS-preselected variants and fixed-effect SNPs improves predictive ability, potentially enhancing breeding efficiency in fruit trees.