ANALYSING G x E INTERACTION IN SUGAR CANE USING THE ADDITIVE MAIN EFFECTS AND MULTIPLICATIVE INTERACTION (AMMI) MODEL

摘要

Variety trials are conducted with several genotypes in different environments to provide reliable information for selecting high yielding varieties. The Genotype x Environment interaction should be considered while selecting varieties for specific sites. As this interaction must be determined with accuracy, the statistical model utilised must be appropriate. In this paper, the AMMI model is applied using matrix algebra and the statistical software GENSTAT. Cumulative results of plant cane and three ratoons for the characters cane yield (TCH), Industrial Recoverable Sucrose % Cane (IRSC), and sugar yield (TSH) were analysed. The commonly used analysis of variance (ANOVA) failed to detect a significant interaction whereas AMMI analysis revealed a highly significant interaction for TCH and TSH. The mean square (MS) of the principal component analysis (PCA) axis 1 was five to seven times the MS for the residual of the three characters. The AMMI model partitioned the interaction sum of square (SS) and summarised the data quite effectively. Environments were characterised in terms of low and high yielding ones. The Biplot method grouped the genotypes and environments and displayed their response patterns where their potential and weaknesses were clearlyillustrated. Varieties for both wide and specific adaptations were depicted, e.g. variety R570 had a low principal score indicating wide adaptation. As the AMMI analysis gives a more accurate picture of the phenotypic stability of each of the varietiesgrown under different environments, this information can be very useful when recommending varieties for commercial cultivation.