Accounting for spatial trends in multi-environment diallel analysis in maize breeding

Area of Science:

• Agricultural Science
• Plant Breeding
• Biometrics

Background:

• Spatial trends in field trials complicate accurate genetic evaluation in maize breeding.
• Spatial analysis techniques can mitigate these trends, potentially increasing selection accuracy.

Purpose of the Study:

• To compare the effectiveness of spatial analysis (SPA) and non-spatial analysis (NSPA) models for diallel multi-environment trial data in maize breeding.
• To assess the impact of SPA models on key genetic parameters like heritability and selection gain.

Main Methods:

• Utilized 78 inter-populational maize hybrids across four environments with three replications each in a randomized complete block design.
• Employed SPA models incorporating autoregressive matrices (AR1 ⊗ AR1) for row and column autocorrelation, alongside NSPA models.
• Model selection for individual environments was based on Bayesian information criteria; joint analysis used compound symmetry for genotypic effects.

Main Results:

• SPA models were selected for trials E3 and E4, while NSPA was used for E1 and E2.
• Likelihood ratio tests indicated changes in the significance of effects between SPA and NSPA models.
• Heritability, selective accuracy, and selection gain were significantly higher when using SPA models.

Conclusions:

• SPA models demonstrate superior performance in addressing spatial trends within maize breeding trials.
• The incorporation of SPA models is recommended for standard genetic evaluation procedures to enhance reliability and accuracy.
• Parents 2, 10, and 12 were identified as promising candidates for breeding in the studied microregion.