Plant Variety Selection Using Interaction Classes Derived From Factor Analytic Linear Mixed Models: Models With...

Area of Science:

• Agricultural Science
• Genetics
• Statistical Modeling

Background:

• Analyzing multi-environment trial data in plant breeding is complex due to variety by environment interaction (VEI).
• Effective variety selection requires meaningful and concise summaries of performance across diverse environments.
• Existing methods often struggle to capture the nuances of VEI, hindering optimal breeding decisions.

Purpose of the Study:

• To develop a novel statistical framework for analyzing plant breeding multi-environment datasets.
• To address the challenge of variety by environment interaction (VEI) in variety selection.
• To introduce a method for grouping environments into 'interaction classes' (iClasses) to simplify performance interpretation.

Main Methods:

• Fitting a factor analytic linear mixed model (FALMM) to multi-environment trial data.
• Utilizing factor analytic parameters to define interaction classes (iClasses) with minimal crossover VEI within groups.
• Developing an iClass Interaction Plot as a graphical tool for visualizing VEI patterns across iClasses.

Main Results:

• Identification of distinct interaction classes (iClasses) within the environmental dataset.
• Demonstration that environments within an iClass exhibit minimal crossover VEI.
• Validation of predictions for overall variety performance within each iClass.
• Facilitation of variety selection and matching based on performance across iClasses.

Conclusions:

• Interaction classes (iClasses) provide a robust method for simplifying complex multi-environment trial data.
• The FALMM-based approach effectively manages VEI, enabling more accurate variety selection.
• The iClass Interaction Plot offers a valuable visualization tool for breeders to understand variety performance patterns.