Assessing Single-Trait and Multitrait Genomic Prediction Model Abilities Including Significant GWAS Markers for Fusarium Head Blight Disease Resistance in Wheat (Triticum aestivum)

Abstract

Disease resistance traits are complex and quantitative in nature. Breeders regularly evaluate multiple important traits across diverse environments to employ them in genomics-assisted breeding. In this study, we evaluated the prospects of genomic prediction models by incorporating genome-wide association study (GWAS) results into single-trait and multitrait genomic prediction scenarios, using two distinct panels: the NMBU panel and the GRAMINOR panel. A standard genomic prediction model (Base) and the Base model with the addition of significant GWAS markers as fixed covariates (Base + GWAS) were tested on both panels. The predictive ability of models was measured in terms of prediction ability by using Pearson's correlation method. An improvement of 0.05% to as high as a two-fold improvement was observed in both the panels for single-trait and multitrait scenarios. In general, multitrait models outperformed single-trait models regardless of whether the GWAS markers were included. This study further concludes that multitrait-based genomic predictions are superior to single trait–based ones when the associated traits are used and are well correlated.