Conventional Selection versus Methods that Use Genotype × Environment Interaction in Sweet Corn Trials

Abstract

The analysis of variance of a data set made up of 30 sweet corn (Zea mays L.) hybrids evaluated over 5 years for marketable ears (dozens per hectare) indicated a significant genotype (hybrid) × year (GY) interaction. Three selection methods were compared: 1) a conventional method based on mean yield alone (YA), 2) Kang's ranksum (KRS) method, and 3) Kang's modified rank-sum (KMR) method. The number of hybrids selected on the basis of YA, KRS, and KMR was 13. The KRS selected the lowest number of unstable hybrids (three) compared with the YA and KMR, which selected eight and six unstable hybrids, respectively. The mean yields of the selected hybrids were 3034 dozen/ha for YA, 2945 dozen/ha for KRS, and 3019 dozen/ha for KMR. The mean yield of KRS-selected hybrids and KMR-selected hybrids was <2.9% and 0.5%, respectively, than that of YA-based selections. This yield reduction was regarded as insignificant considering the farmer would be able to choose more consistently performing hybrids on the basis of KRS than on the basis of KMR or YA. Heterogeneity due to environmental index is the mean of all genotypes in the jth year and X is the overall mean) was significant and was removed from the GY interaction. The removal of heterogeneity revealed that hybrids 77-2269, 116-Kandy Korn-EH, Golden Queen, 141-Sundance, Merit, and Stowell Evergreen were unstable because of a linear effect of the environmental index, and that hybrids 76-2681 and 806F-Truckers showed stable performance due to a linear effect of the environmental index.