Author/Authors :
TOPAL, Mehmet Atatürk Üniversitesi - Ziraat Fakültesi - Zootekni Bölümü, Biyometri ve Genetik Anabilim Dalı, Turkey , YILDIZ, Necati Atatürk Üniversitesi - Ziraat Fakültesi - Zootekni Bölümü, Biyometri ve Genetik Anabilim Dalı, Turkey
Abstract :
Genotype Environment interaction is the most important issue for animal and plant breeding. Parametric and nonparametric methods used in determination of Genotype × Environment interaction are based on the yield values of genotypes and their ranks in each environment. Genotype × Environment interactions are individually established by stability methods. At the beginning, variance analysis was applied to determine the interactions related to data simulated according to normal and discrete uniform distribution, and then environmental variance, Shukla’s stability variance, variation coefficient, ecovalance, Lin and Bins’ Pi method, Finlay and Wilkinson’s regression coefficient, Perkins and Jinks’ regression coefficient and Eberhart and Russell’s deviation parameter from parametric methods were applied to present data. Differences among the methods applied for data displayed normal and discrete uniform distribution were not significant. When the interactions obtained according to variance analysis in both normal and discreate uniform distribution were insignificant, coefficient values from parametric stability estimation methods were also found insignificant. In addition, when interaction was determined to be significant, difference among the coefficient values was significant. While the highest correlation in parametric methods for normal and discrete uniform distributions was established among CV, VK, EV, SSV and ERSKT, the correlation of these methods with FWbi and PJbi methods were found to be very low. Also, the highest correlations were determined between SSV and EV, CV and VK, FWbi and PJbi methods. It was observed that similar results may be obtained when any method showing high correlation with other methods was used to determine the Genotype × Environment interaction. It was concluded that one of EV and SSV methods individually partitioning Genotype × Environment interaction values to genotype or any method showing high rank correlation with these methods is preferable to other methods.
