Non-parametric approach to the study of phenotypic stability

Area of Science:

• Agricultural Science
• Biometry
• Genetics

Background:

• Phenotypic stability is crucial for consistent crop performance across environments.
• Existing parametric methods for stability analysis can be sensitive to assumptions about data distribution.
• Non-parametric methods offer an alternative approach, particularly when data heterogeneity is present.

Purpose of the Study:

• To theoretically derive and evaluate non-parametric methods for plant genotypic stability analysis.
• To extend existing parametric stability methods using rank order linear regressions.
• To compare the performance of novel non-parametric methods against a standard non-parametric method.

Main Methods:

• Theoretical derivations of non-parametric stability methods based on rank order linear regressions.
• Application of intensive computational techniques, including bootstrap and permutation methods.
• Utilizing data from a plant-breeding program for empirical validation.

Main Results:

• The developed non-parametric methods proved effective in assessing phenotypic stability.
• These methods demonstrated superior discriminatory power compared to standard approaches, particularly under conditions of variance heterogeneity.
• The study confirmed the utility of rank-based regressions for stability evaluation.

Conclusions:

• Non-parametric stability analysis using rank order linear regressions is a robust approach for plant breeding.
• These methods provide a valuable tool for genotype selection, especially in environments with unpredictable conditions.
• The findings support the use of non-parametric techniques for more reliable phenotypic stability assessments.