Evaluation of selection indices in a recurrent selection program in corn (Zea mays L.)

Abstract

<p>Index selection is a procedure used for selecting more than one trait at a time in a breeding program. Several types of selection indices have been proposed. The first one formally introduced and probably the most widely known is the Smith-Hazel index which requires estimates of phenotypic and genotypic parameters for its calculation. These estimates are subject to sampling errors that affect the performance of the index. In this study, several selection indices were compared, and the effectiveness of pooling parameter estimates over several cycles of selection on the estimation of the Smith-Hazel index was evaluated for the simultaneous improvement of grain yield, grain moisture, root lodging and stalk lodging in corn;The degree of improvement for each of the traits was influenced by the structure of the correlations between traits. Some indices were more affected than others by these changes in correlation structure. The indices least affected were the base using the inverse of the phenotypic standard deviation, the multiplicative and the rank summation resulting in a more even improvement for the four traits considered;The efficiency of Smith-Hazel indices calculated by pooling parameter estimates over several cycles of selection were compared with indices calculated using the parameter estimates obtained from data of the specific cycle of selection. They were evaluated based on the selection differentials and predicted gains obtained for each of the four traits considered, and predicted gain in the aggregate genotype. In general, the Smith-Hazel selection indices calculated by pooling parameter estimates were neither better for improving the four traits simultaneously nor for improving the aggregate genotype than the selection indices calculated with data of the specific cycle of selection.</p>