Prediction of Australian wheat genotype by environment interactions and mega-environments

Area of Science:

• Agricultural Science
• Plant Breeding
• Genetics

Background:

• Wheat (Triticum aestivum) breeding in Australia faces challenges due to widespread genotype by environment interactions (GEI).
• Accurate prediction of GEI across diverse environments is crucial for optimizing crop performance.
• Existing models often struggle to link latent GEI effects to observable environmental covariates (ECs).

Purpose of the Study:

• To associate and predict latent environmental effects with observed environmental covariates in wheat.
• To develop a method for predicting genotype by environment interactions in a wider range of Australian environments.
• To improve the accuracy of genomic prediction for wheat breeding programs.

Main Methods:

• Utilized a large multi-environment trial dataset for wheat.
• Employed factor analytic mixed models to capture common GEI using latent environmental effects.
• Associated latent effects with observed environmental covariates derived from weather and soil data.

Main Results:

• Demonstrated that latent environmental effects can be predicted from observed environmental covariates.
• Identified genotype by environment interaction-based environment classes defined by key environmental covariates.
• Revealed significant year-to-year variability in environmental effects and genotype by environment interactions across the Australian grain belt.

Conclusions:

• Observed environmental covariates provide a basis for predicting latent genotype by environment interactions.
• Findings enable better targeting of wheat genotypes to specific mega-environments while accounting for environmental variability.
• Improved accuracy in genomic prediction is achievable with accurate genetic effects and known environmental covariates for new environments.