Predicted responses to recurrent selection in maize (Zea mays L.)
Previous studies with computer simulation models have shown that standardised correlated response in one trait, y, to selection for another trait, x, should be the same in one generation regardless of whether selection is for x or for y. In the study reported herein, we tested the hypothesis that direct response to selection for trait x should be the same as standardised correlated response in trait x when selection is based on trait y. Additionally, we compared, observed and predicted correlated responses to seven cycles of reciprocal recurrent selection for grain yield in two maize populations Results obtained indicated that predictions of direct and standardised correlated responses showed poor agreement. Direct response to mass selection, however, agreed more closely with standardised correlated responses than did half-sib selection and S1 testing, whereas direct response to S1 testing agreed more closely with conventional correlated response than did the other two selection methods. Also, observed and predicted correlated response to reciprocal recurrent selection for grain yield showed poor agreement. However, predictions based on parameters obtained in the fourth cycle of selection agreed more closely with observed changes than similar estimates obtained from the original populations. Implications of these results in determining indirect selection criteria are discussed.