Genetic parameters and selection gains in early clonal evaluation trials: implications for cassava breeding

The objective of this study was to estimate and compare genetic parameters in early cassava breeding phases (clonal evaluation trials—CET and preliminary yield trials—PYT) in full-sib (F₁) and self-pollinated (S₁) families, besides estimating the genetic gains. Twenty-three F₁ and six S₁ families were evaluated using the augmented block design in CET and the randomized complete block design in PYT for fresh root yield (FRY), root dry matter content (DMC) and starch yield (STY). In CET, most of the variance was due to environmental (\(\sigma_{e}^{2}\)) followed by variance within F₁ and S₁ (\(\sigma_{Clone/Fam}^{2}\)) families, with the exception of DMC in S₁ families. PYT presented the opposite behavior. In contrast, specifically for S₁ families, the variance between families (\(\sigma_{ Fam}^{2}\)) was more important than \(\sigma_{e}^{2}\) and \(\sigma_{Clone/Fam}^{2}\) in PYT. The heritability of families (\(h_{Fam}^{2}\)) was lower than individual broad-sense heritability (\(h_{g}^{2}\)) in all trials and families. Regardless of the family type and trial, family accuracy (\(r_{ggFam}\)) was lower than the clone accuracies (\(r_{ggCl}\)). Predicted gains using the selection index (SI) applied to best linear unbiased prediction (BLUP) were higher in PYT compared with CET and higher in F₁ families in comparison with S₁. There was also low coincidence in clone selection in both trials (30 and 45% for F₁ and S₁ families, respectively). For cassava breeding, it is recommended to obtain a higher number of clones per family and to use the SI with moderate intensity, particularly in CET.