STATISTICAL APPROACHES FOR PARTITIONING GENETIC VARIANCE IN COMPLEX AGRONOMIC TRAITS

Abstract

Background: Grain yield, drought tolerance and disease resistance are complex agronomic traits controlled by multiple genetic and environmental factors, so accurate estimation of genetic variance components is crucial for crop improvement and precision breeding programs.

Objective: The aim of this study was to assess statistical methods to partition genetic variance of complex agronomic traits using quantitative genetics and genomic prediction methods.

Methods: Methods Phenotypic and genomic data sets from crop breeding populations were analyzed using analysis of variance (ANOVA), mixed linear models (MLM), genomic best linear unbiased prediction (GBLUP) and Bayesian regression approaches. Within computational statistical frameworks, we estimated the variance components, heritability and predictive performance.

Results: The total phenotypic variation explained by the mixed linear models was 72% and the prediction accuracy maximized at 81% with Bayesian regression. Additive genetic variance explained 58% of total trait variance and environmental variance explained 29%. Genomic prediction models considerably improved the efficiency of heritability estimation and trait prediction compared to traditional statistical methods.

Conclusion: Improved statistical and genomic prediction methods provide powerful tools for partitioning genetic variance and increasing the efficiency of genomic selection in crop breeding programs.