Stability analysis of yield of potato genotypes in Rio Grande do Sul
DOI:
https://doi.org/10.1590/S1678-3921.pab1998.v33.4867Keywords:
Solanum tuberosum, genotype-environment interaction, AMMI model, linear regression, principal component analysisAbstract
The objective of this work was to compare the amount of genotype-environment (GE) interaction described by using the additive main effects and multiplicative interaction (AMMI) model and the analysis of linear regression (LR), and to compare the yield stability of potato (Solanum tuberosum L.) genotypes. Ten genotypes were evaluated in 34 environments (local, cultivation period and year combinations) in Rio Grande do Sul, Brazil, in 1994 and 1995. The sum of squares (SS) for the regressions accounted for only 19.5% of the interaction SS, whereas the first principal component (PC1) of the analysis of the principal components accounted for 44.6% of the interaction SS. The SS of PC1 was more than twice the combined SS of all three regressions (jointed, genotypic and environmental). Therefore, the AMMI analysis was more efficient in describing the GE interaction than the LR. The cultivar Monte Bonito is unstable, but has the highest yield; 'Baronesa' is moderately stable and has high yield; 'Santo Amor' and 'Trapeira' are the most stable, with yield above average; 'Macaca' and 'Cristal' are the most unstable, with yield below average; 'Atlantic' is unstable and has low yield.