Bayesian analysis and prediction of hybrid performance.

Area of Science:

• Plant breeding
• Genomics
• Statistical genetics

Background:

• Maize hybrid selection is crucial but field trials are expensive.
• Genomic models predict performance of untested maize hybrids.
• Bayesian models offer flexible frameworks for hybrid prediction.

Purpose of the Study:

• Overview of Bayesian genomic models for maize hybrid prediction.
• Decomposition of genotypic variance into general and specific combining abilities.
• Application to tropical maize hybrid data.

Main Methods:

• Bayesian parametric and semi-parametric genomic models.
• Analysis of 906 single cross tropical maize hybrids.
• Estimation of variance components for combining abilities.

Main Results:

• Non-additive effects significantly impact grain yield genetic variance.
• Non-additive effects are less important for ear and plant height.
• Genomic prediction achieved high accuracy for untested hybrids and pre-screening.

Conclusions:

• Genomic prediction is valuable for pre-screening maize hybrids.
• Bayesian framework provides flexibility in modeling hybrid performance.
• Methodology aids in estimating genetic parameters and predicting hybrid performance with uncertainty.