STABILITY ANALYSIS OF SOME SOYBEAN GENOTYPES USING PARAMETRIC AND NON PARAMETRIC METHODS IN MULTI-ENVIRONMENTS

Abstract

Seed yields of 14 soybean genotypes were evaluated in four locations i.e. Adana, Şanlıurfa, Antalya and İzmir under second crop conditions through summer seasons from 2014 to 2016. The study aims to estimate the stability parameters in terms of seed yield of 14 soybean genotypes by using different stability analysis methods across eleven environmental conditions and to study interrelationships among these stability methods. The analysis of variance for seed yield revealed that the genotypes and the environments as well as the genotype x environment interactions (GEI) were statistically significant at P<0.01. Environmental effects were contributed 51.04% to the total sum of squares whereas GEI and genotype effects were 20.8% and 2.59%, respectively. According to most stability methods, BATEM 223, BATEM 306, BATEM 317 and KASM 02 were determined to be stable genotypes. These genotypes demonstrated superior adaptability with high yield performances in many environments. Results of correlation analysis indicated that seed yield was positively and significantly correlated with Di 2 (P><0.01), Si (6) (P><0.05) and TOP (P><0.01) and showed a negative and significant correlation with Pi (P><0.01) and RS (P><0.01). In addition, the coefficient of regression (bi) was positively significant associated with CVi, αi (P><0.01) and Ri 2 (P><0.01. Environmental effects were contributed 51.04% to the total sum of squares whereas GEI and genotype effects were 20.8% and 2.59%, respectively. According to most stability methods, BATEM 223, BATEM 306, BATEM 317 and KASM 02 were determined to be stable genotypes. These genotypes demonstrated superior adaptability with high yield performances in many environments. Results of correlation analysis indicated that seed yield was positively and significantly correlated with Di 2 (P<0.01), Si (6) (P<0.05) and TOP (P<0.01) and showed a negative and significant correlation with Pi (P<0.01), and RS (P<0.01), In addition, the coefficient of regression (bi) was positively significant associated with CVi, αi (P<0.01), and Ri 2 (P<0.05).