| Sumario: | This study evaluated the yield stability and environmental responsiveness of 42 pro-vitamin A cassava genotypes across multi-season trials using Finlay-Wilkinson (FW) regression and trait-based clustering approaches. Regression parameters-intercept and slope were used to quantify baseline yield potential and sensitivity to environmental variation, respectively. Hierarchical and k-means clustering grouped genotypes into three biologically distinct clusters with clear agronomic relevance. Cluster 2 genotypes exhibited moderate responsiveness and positive yield baselines, indicating broad adaptability and suitability for regional deployment. Cluster 3 showed high environmental sensitivity but low yield potential, suggesting limited resilience under marginal conditions. Cluster 1 comprised highly responsive genotypes with poor baseline productivity, reflecting unstable performance and strong genotype × environment interaction. One-way ANOVA confirmed significant differences among clusters for both slope (F(2,39) = 40.89, p < 0.001) and intercept (F(2,39) = 102.10, p < 0.001), validating the clustering structure. The dendrogram confirmed the cluster structure and provided a basis for selecting key genotypes. The results inform environment-specific breeding strategies and emphasize the importance of integrating multiple traits into future clustering approaches to improve cultivar selection accuracy.
|
|---|