Genomic selection in a commercial winter wheat population.

Area of Science:

• Plant breeding
• Quantitative genetics
• Genomics

Background:

• Genomic selection (GS) is a powerful tool for accelerating crop improvement.
• Accurate prediction of breeding values is crucial for efficient selection in wheat (Triticum aestivum).
• Optimizing training population composition is key to maximizing GS prediction ability.

Purpose of the Study:

• To evaluate the impact of additive and epistatic genetic effects on GS prediction ability in European winter wheat.
• To assess GS prediction ability using historical or less-intensively phenotyped training populations.
• To explore GS prediction ability in subpopulations selected by reliability criteria.

Main Methods:

• Implementation of GS in a large commercial population of 2325 European winter wheat lines.
• Comparison of prediction abilities between additive-only and additive-plus-epistatic models.
• Analysis of prediction accuracy using varying phenotyping intensities and reliability filtering.
• Evaluation of prediction ability in selected subpopulations.

Main Results:

• A 5% increase in prediction ability was observed when including epistatic effects alongside additive effects.
• GS models demonstrated stable prediction accuracy (0.50) when predicting subsequent year's genotypes from historical data (0.65).
• Excluding single-location evaluated genotypes maximized prediction ability; increased phenotyping intensity beyond two locations reduced accuracy.
• Subpopulations selected based on reliability criteria showed substantially higher genomic prediction ability.

Conclusions:

• Modeling epistasis enhances genomic prediction ability in winter wheat.
• GS models are robust and can be calibrated using historical or less-intensive phenotyping data.
• Optimizing training population composition by excluding single-location data and selecting reliable individuals is critical.
• Genomic selection offers significant potential to advance commercial wheat breeding programs.