Relationships Between Yield and Several Yield Components in a Set of Maize Hybrids

The relationship of yield to yield component estimates obtained from additive, additive plus dominance, and additive plus dominance plus epistasis models of inheritance was examined in a set of 15 single cross and 30 three‐way maize ( Zea mays L.) hybrids, derived from a group of 6 commercially adapted, essentially homozygous inbred lines. The yield components considered were: row number, kernels per row, and weight per 300 kernels. Yield component estimates from each genetic model were converted to logarithms and used as independent variables in a multiple linear regression analysis in which log yield was the dependent variable. A model including additive effects of row number and weight per 300 kernels and additive plus dominance effects for kernels per row adequately accounted for the variation in yield due to variability in these components. Maximum yield with regard to additive effects was characterized by high row number and weight per 300 kernels, whereas yield response to dominance effects was due to kernels per row only. These results indicated that yield response to nonadditive effects was maximized by crosses among lines that, on the basis of additive effects, exhibited high row number and weight per kernel. However, the additive effects yield component estimates provided no more information concerning yield than did the additive effects estimate of yield itself. Therefore, estimates of the additive yield component effects of the parental lines would not have been useful in predicting crosses possessing favorable nonadditive effects in this set of hybrids. In addition, because the yield components exhibited at least as much nonadditive variation as yield, yield component estimates obtained from the inbreds themselves would probably not have been of much more use in predicting hybrid performance than inbred yields themselves.