Increased prediction accuracy in wheat breeding trials using a marker × environment interaction genomic selection...

Area of Science:

• Plant breeding
• Quantitative genetics
• Bioinformatics

Background:

• Genomic selection (GS) models predict genetic values using genome-wide markers.
• Traditional GS models often overlook genotype × environment interaction (G×E), limiting prediction accuracy in diverse conditions.
• Existing extensions for G×E in GS are complex or assume constant marker effects across environments.

Purpose of the Study:

• To introduce and evaluate a marker × environment interaction (M×E) genomic selection model.
• To compare the predictive performance of the M×E model against standard and stratified GS approaches.
• To investigate the decomposition of marker effects into environment-specific and stable components.

Main Methods:

• Developed and implemented a marker × environment interaction (M×E) genomic selection model.
• Utilized explicit regression of phenotypes on markers and covariance structures for model implementation.
• Analyzed three wheat datasets (W1, W2, W3) with over 1000 lines genotyped and evaluated under varied simulated environmental conditions.
• Compared M×E model predictions with across-environment (ignoring G×E) and within-environment (stratified) analyses.

Main Results:

• The M×E genomic selection model demonstrated substantially greater prediction accuracy than across-environment models that ignore G×E.
• The M×E model achieved similar or superior prediction accuracy compared to stratified (within-environment) analyses, depending on the specific prediction task.
• The M×E model successfully decomposed marker effects and genomic values into environment-stable (main effects) and environment-specific (interaction) components.

Conclusions:

• The marker × environment interaction (M×E) genomic selection model offers a significant advancement for predicting genetic values in plant breeding.
• This model provides a conceptually simple yet powerful approach to account for G×E, improving prediction accuracy.
• The M×E model facilitates the identification of genetic variants with stable effects versus those contributing to G×E, aiding breeding strategies.