The Evaluation of Grain Yield Stability in Sunflower Varieties Using Univariate Analysis and AMMI Method

Evaluation of the effects of different environments on grain yield of sunflower hybrids (Helianthus annuus L) and determination of superior and stable genotypes in different places is considered as a model for stability and adaptation to the environment.

In order to investigate the interaction of yield stability on 12 sunflower hybrids, an experiment was conducted in a randomized complete block design (RCBD) with three replications in five regions (Karaj, Birjand, Firooz-Abad, Kashmar and Arak).

The evaluation of variance analysis in 1% probability indicated that the genotypes effect was significant for kernel yield. Also, the data collected from mean value showed that genotype Zaria is the most desirable genotype and genotype Progress the most undesirable. According to AMMI method variance analysis environment effect, genotype effect and genotype × environment interaction in 1% and 5% probability level was significant and the first two factors have covered more than 80% of data variance totally (IPC1= 52.4%, IPC2= 27.9%). AMMI1 model clarified due to having the least first main factor in interaction, genotype Zaria have known as the stable genotype. AMMI2 model shows that genotypes Zargol and Master in Arak, genotypes SHF81-90 and Zaria in Kashmar, genotypes Gabur, Favorite, Record and Progress in Birjand, genotypes Armaverski, Azargol, Lakumka in Karaj and Firooz-Abad have specific adaptation. Based on stability parameters (environmental variances, coefficient of variation, wrick's ecovalance, Shukla variances, regression coefficient, Eberhart and Russell) genotype Zargol recognized as the stable genotype and genotype SHF81-90 as the unstable.

Genotypes Zaria and Zargol identified as the most stable genotypes and genotype SHF81-90 as the unstable genotype generally. The result of stability parameters correlation describes that environmental variance and coefficient values parameters have positive and significant correlation with Eberhart and Russell, Shukla variance and Wricke's ecovalence parameters.