Testing wheat in Kansas: Genotype, environment, and interaction effects

Selection of winter wheat (Triticum aestivum L.) genotypes requires testing programs with complementary locations that sample environments of interest with minimal duplication. Exploiting genotype---environment (GxE) interactions to fit genotypes to appropriate environments requires identification of factors that are responsible for those interactions. The goal of the current study was to improve prediction of genotype performance in the highly variable environments of Kansas by estimating the contributions of genotype, location, and year to wheat yield variability; identifying subgroups of test locations that minimize crossover GxE; and identifying genotypic and environmental parameters that underlie subgroups of test locations. Variance components were estimated from 1982 to 2002 Kansas wheat performance data. Balanced GxE data sets were used to generate genotype, genotype-by-environment (GGE) biplots that could objectively separate locations into groups with the same top-yielding genotype. Location and year and their interaction introduced the greatest proportion of the variability in wheat performance test yields. Frequency of common grouping was used to construct six groups of test locations. Evaluation of the six groups using results from two subsequent years revealed that they generally agreed with location groups observed in the previous 21 years. Smaller regional GxE variance component estimates compared to statewide estimates further confirmed the effectiveness of the proposed six regions for reducing GxE. The proposed groups of test locations were used as a framework to identify important genotypic traits and environmental variables in each region. Genotype, genotype-by-trait (GGT) biplots were generated for each region in each year to examine genotype---trait relationships. Environment, environment-by-variable (EEV) biplots were generated for each year to reveal environment---variable relationships. Although no genotypic traits or environmental variables appeared to be related consistently to genotype performance, conclusions regarding relationships that may underlie the wheat testing regions could be made. Based on the observed genotype---environment relationships, plant breeders and producers can select genotypes with a combination of agronomic traits and disease resistance appropriate for different areas of Kansas. Likewise, several environmental variables (minimum and maximum temperatures, precipitation, and topsoil texture, depth, and pH) should be considered when selecting test sites or inferring performance to other locations.