Multiple-trait, random regression, and compound symmetry models for analyzing multi-environment trials in maize...

Area of Science:

• Plant breeding
• Quantitative genetics
• Statistical genetics

Background:

• Efficient statistical methods are crucial for analyzing genotype-by-environment interaction (GxE) in maize breeding.
• Multi-environment trials (MET) generate complex data requiring robust analytical approaches.

Purpose of the Study:

• To compare the effectiveness of multiple-trait models (MTM), random regression models (RRM), and compound symmetry models (CSM) for analyzing MET in maize.
• To identify the most suitable statistical model for maize breeding programs.

Main Methods:

• Utilized a dataset of 84 maize hybrids across four environments for grain yield (GY).
• Employed restricted maximum likelihood (REML) for variance component estimation and best linear unbiased prediction (BLUP) for genetic value prediction.
• Selected best-fit models using Akaike information criterion (AIC) and tested genetic effects with likelihood ratio test (LRT).

Main Results:

• Random regression models (RRM) with Legendre polynomials of order two and heterogeneous residuals showed good fit.
• MTM and RRM generally yielded slightly higher estimates for heritability, selective accuracy, and predicted selection gains compared to CSM.
• Genetic variability for grain yield among maize hybrids was successfully assessed.

Conclusions:

• Random regression models (RRM) are a preferential approach for analyzing MET in maize breeding due to parsimony and predictive ability for untested environments.
• RRM provides a more effective framework for understanding GxE and improving selection accuracy in maize breeding programs.