Best in the Field: Statistical Strategies for Plant Breeding Using Multi-Environment Trial

Abstract: Evaluating crop varieties across diverse environments is crucial yet statistically challenging due to complex genotype-by-environment (G×E) interactions, high-dimensional data structures, and frequently unbalanced trial designs. Historically, methods such as ANOVA and AMMI provided foundational tools but struggled to accommodate the irregularities of real-world data, resulting in unstable estimates and limited interpretability. Contemporary approaches, particularly Factor Analytic Mixed Models (FAMM), represent significant methodological advancements, offering robust solutions through dimension-reduction techniques. By modelling G×E interactions using latent factors, FAMMs effectively simplify complex patterns into interpretable components, enhancing estimation stability and facilitating practical decision-making in plant breeding programmes. This talk serves as the project proposal presentation for a year-long Honours Project in statistics, aiming to investigate modern statistical strategies for ranking plant varieties under multi-environment conditions.