Details

Autor(en) / Beteiligte

• Duarte, João Batista
• Zimmermann, Maria José de O.

Titel

Correlation among yield stability parameters in common bean

Ist Teil von

• Crop science, 1995-05, Vol.35 (3), p.905-912

Ort / Verlag

Madison, WI: Crop Science Society of America

Erscheinungsjahr

1995

Beschreibungen/Notizen

• Phenotypic yield stability is a trait of special interest for plant breeders. Many statistical procedures are available for stability analysis, each of them allowing for different interpretations. The objective of the present study was to determine the degree of correlation among the 13 statistical parameters that can be used for the analysis of phenotypic stability. Such correlations could be used to assess the extent to which these 13 parameters identify unique genetic effects. Yield data were obtained from 12 yield trials involving 76 common bean (Phaseolus vulgaris L.) genotypes and 12 location-year production environments in Brazil. The stability statistics were divided in four groups according to the structure from which they were derived. On the basis of rank correlation, it was concluded that (i) there were highly significant correlations between many of the stability statistics (among and within groups) indicating that several of the statistics probably measure similar aspects of phenotypic stability; (ii) mean yields were positively correlated with many of the stability statistics; (iii) there was an association between the Group A statistics (variances and ranges) and the Group C statistics (regression and determination coefficients), and a similar association between the Group B (ecovalence) and Group D (variance of deviations from regression) statistics; (iv) the segmented linear regression coefficient (b1i) was overall the most independent parameter, indicating that the other stability statistics do not satisfactorily reflect genotypic responses in poor environments; (v) the strong correlation between the regression coefficients and the coefficients of determination indicates that the latter are not needed to measure the predictability of the estimated genotypic response; and (vi) the variance of the deviations from regression can provide assessment of the relative contribution of the genotype to the genotype x environment interaction.