ESTIMATION AND COMPARISON OF STABILITY PARAMETERS TO DETERMINE GRAIN YIELD RESPONSE OF CORN CULTIVATIONS. PANAMA, 1996-98.

Abstract

Obtention of genotypes that less possible interact with environment has been the objective that follow plant breeders. This is with the goal of developing materials that guarantee good productivity through the contrast environments. There are different methods of genotype-environmental interactions (Westcott, 1980; Becker and Leon, 1988; Crossa, 1990; however, the aplication of any of these methodologies can imply diferent concepts of phenotipic stability (Lin y col, 1986; Becker y Leon, 1988). This study was implented with the objective of estimate and compare stability parameters of eight corn cultivations evaluated in 25 environments (three years) using linear regression analysis of Finlay and Wilkinson (1963) (F&W); Eberhart and Russell (1966) model with three variants: 1) all environments (E&RG); 2) only environments with possitive indexer (E&RP), and 3) environments with negative indexer (E&RN); principal component analysis (PCA) and AMMI model. Despite of the analysis of variance with the different models, detected significant genotype-environmental interaction, AMMI is the one that offer the most appropriated statistical analysis to determinate the cultivation response to the environment. In this sense, AMMI plot (PCA1) explains only 42% of the interaction; furthemore, AMMI Biplot (PCA1 vs PCA2), explains 73% of the genotype-environmental interaction in this study. On the other hand, ER model recovers best interpretations of genotype behaviors when evaluate their bi in high and low yield environments. Estimated parameters from the models were correlated among them and it was observed high and significant correlation coefficients between E&R vs F&W (0.98), E&R vs PCA (-0.97) and E&R vs AMMI (0.88).