Multi-location yield trials in sugarcane were conducted in the central and western regions of Thailand. Twelve genotypes grown at 15 locations were used. Thirteen traits including sugar yield, cane yield, number of stalks, stalk height at 6 months after planting (MAP) and at harvest, stalk diameter, single stalk weight, number of tillers at 6 MAP, number of died-tiller, Brix value, Pol (sucrose content), fiber content and CCS (commercial cane sugar) value, were used to analyze the genotype x trait (GT) biplot. All the traits except for the sugar yield and CCS value were used to analyze the trait covariate-effect biplot. In all the crop-cycles, several traits showed large variations among genotypes, whereas some other traits showed small variations. The GT-biplot clearly revealed these relationships between genotypes and traits. The GT-biplot enabled to analyze not only the relationship between each genotype and each trait, but also to compare these relationships among genotypes. In addition, the GT-biplot also displayed the relationships among traits. In all the crop-cycles, the trait covariate x environment biplots clearly showed 4 positively correlated traits with cane yield (number of stalks, stalk height at 6 MAP, that at harvest and stalk weight) at most locations, and there were 4 negatively correlated traits (Brix value, Pol, fiber content and number of died-tiller). Although yearly variations were found, some associations between the contribution of traits to cane yield and locations were confirmed. Therefore, it was suggested that the GT biplot is an efficient and powerful technique to analyze the relationship among traits and their association with genotypes, and that the trait covariate biplot is a useful indicator of the traits correlated with cane yield in association with tested locations.
抄録全体を表示