Classifying genotypic data from plant breeding trials: a preliminary investigation using repeated checks

Area of Science:

• Agricultural Science
• Plant Breeding
• Biometrics

Background:

• Genotype classification in plant breeding involves subjective choices impacting trial data analysis.
• Standard methods often rely on raw or standardized data, with environmental contributions needing careful weighting.
• The inclusion or exclusion of non-significant environments/blocks presents another analytical decision point.

Purpose of the Study:

• To evaluate the impact of different weighting methods (raw, standardized, discrimination index [DI]) on genotype classification accuracy.
• To compare the effectiveness of using environmental data versus block data for classification.
• To assess the influence of including or excluding non-significant environments/blocks on classification performance.

Main Methods:

• A factorial combination of 12 options was tested, varying weighting methods, data source (environments vs. blocks), and significance filtering.
• A dataset with five check cultivars, replicated six times in three blocks across six environments, was utilized.
• Hierarchical clustering was employed to assess the classification accuracy of genotype repeats.

Main Results:

• The discrimination index (DI) weighting method generally resulted in superior recovery of known genotypic structure.
• Utilizing block data consistently improved structure recovery compared to environmental data across all weighting methods.
• Excluding environments with non-significant genotypic effects decreased structure recovery, while including blocks generally improved it.

Conclusions:

• The DI weighting method, particularly when applied to block data, offers the most effective approach for genotype classification in plant breeding.
• Analyzing data at the block level, rather than the environment level, enhances the reliability of genotype classification.
• The DI method, combined with block-level analysis, provides robust genotype classification, irrespective of genotypic significance in individual blocks.