STUDIES ON NONPARAMETRIC METHOD OF PHENOT YPIC STABILITY .2. SELECTION FOR STABILITY OF AGRONOMIC CONCEPT

The purpose of this study is to establish the nonparametric measures f or stability according to the agronomic concept. The applicability of theoretical results is demonstrated by analysing data from the 1986-19 87 regional trials of maize composites in Taiwan. Under the null hypot hesis of equally stablility among genotypes, the rank orders of genoty pes in each environment are determined by corrected values which depen d on genotype-interaction effect (x*(ij)) rather than on genotypic ef fect (x(ij), Nassar and Huhn 1987). An analysis of phenotypic stabili ty which performed by ranking these corrected values x*(ij) and x*(ij ) is analogous to the stability of agronomic and biological concept, r espectively. However, they result in the same rank orders. Hence, nonp arametric measures for stability based on ranks provide an useful alte rnative to existing parametric measures. They require less statistical assumptions and are easy to use. Using this approach, stable genotype s can be defined as both types of stability concept. Correlation analy sis revealed that the two nonparametric stability measures (mean absol ute rank difference and variance of ranks) are strongly intercorrelate d with each other even they have different statistical interpretation. The results also indicated that the nonparametric measures are signif icantly correlated with the deviation mean square from regression.