Leveraging historical trials to predict Fusarium head blight resistance in spring wheat breeding programs

Area of Science:

• Plant breeding and genetics
• Agricultural science
• Genomics

Background:

• Fusarium head blight (FHB) significantly impacts wheat production, necessitating genetic resistance improvements.
• The Uniform Regional Scab Nursery (URSN), established in 1995, contains valuable germplasm for FHB resistance but its data remained unanalyzed.
• Genomic prediction offers a powerful tool for accelerating crop improvement through marker-assisted selection.

Purpose of the Study:

• To analyze historical URSN data for genomic prediction of FHB resistance and agronomic traits.
• To evaluate the impact of various factors (statistical method, marker density, training set size, genetic structure) on prediction accuracy.
• To assess the utility of the URSN population as a training set for predicting performance in current breeding programs.

Main Methods:

• Phenotypic and genotypic data from the URSN and two US Midwest breeding programs were collected.
• Genomic prediction models were applied to eight FHB and agronomic traits.
• Reproducing kernel Hilbert space (RKHS) was identified as a top-performing statistical method.
• The influence of marker density, training set size, and population structure was investigated.

Main Results:

• Reproducing kernel Hilbert space achieved an average prediction accuracy of 0.63 using the URSN population.
• Marker density could be reduced to 500 markers without significant loss in prediction accuracy.
• Training set optimization proved beneficial for specific traits.
• Using URSN as a training set for unrelated breeding populations yielded encouraging, though decreased, prediction accuracies.
• Incorporating URSN data into training sets for breeding populations increased prediction accuracy by up to 0.19.

Conclusions:

• The URSN is a valuable resource for genomic prediction of FHB resistance.
• Optimized genomic prediction strategies, including appropriate statistical methods and training set composition, are crucial for effective wheat breeding.
• Leveraging diverse germplasm like the URSN can enhance prediction accuracy for contemporary breeding programs, even across different populations.