Evaluation of maize hybrids stability using parametric and non-parametric methods

The purpose of the study was to compare four parametric and three nonparametric methods for the assessment of agronomic stability of eight maize cultivars chosen from postregistration trials carried out at 16 locations in Poland for a period of three years. Significant correlations were found between mean yield and two parametric measures: bi and Di. Of nonparametric statistics only the new proposed measure (RD) was significantly associated with yield (r=0.86), whereas Kang’s rank sum just missed significance (r=0.68). Correlations between parametric and non¬parametric measures were in general not significant. Full correlation was found between Wricke’s ecovalence and Shukla’s stability variance. Hühn’s stability measures S1i and S2i were also highly positively associated (r=0.92). As expected high correlation was observed between ecovalance and S2di (r=0.98). The results showed that high yield¬ing cultivars can also be stable. The new proposed method based on homogeneous groups ranks can be a useful alternative to Kang’s rank sum parameter in studies of agronomic stability of maize hybrids