AbstractClonal evaluation trials of cassava (Manihot esculenta Crantz), where the main selection of this crop takes place, are usually carried out in multiple environments. This study investigated the influence of genotype–environment (GE) interaction on selection and how to explore genetic information across environments in a mixed model approach by modeling different genetic covariance structures. Approximately 240 cassava clones were assessed in an augmented block design during the 2020/2021 growing season in Brazil. The unstructured model was the best suited and used to investigate several strategies of selection. The predicted genetic gains based on individual analyses varied greatly among environments (5.52%–12.62% for root yield; 1.00%–6.09% for dry matter content; and 4.01%–9.42% for dry matter yield), although the clones mean was similar. Moreover, most of the selected clones in each environment outperformed the best check (>80%), except for root yield and dry matter yield in one environment. By multi‐environment analysis, greater local gains were detected in each environment (means of 16.87% for root yield, 5.56% for dry matter content, and 17.27% for dry matter yield) and for mean heritability (0.52 for root yield, 0.76 for dry matter content, and 0.55 for dry matter yield). The coincidence of clones selected by individual and multi‐environment analyses was 64% for root yield, 73% for dry matter content, and 66% for dry matter yield. The best scenario for selection is when all environments are considered simultaneously, for which regional genetic gains of 16.71% were predicted for root yield, 5.40% for dry matter content, and 17.06% for dry matter yield.
Read full abstract