Evaluation of stability analysis models for the selection of high flower yield and stability in Rosa damascene Mill. genotypes

In order to determination of appropriate stability parameters, the flower yields of 35 Rosa damascena genotypes in eight regions of Iran (Kurdistan, Hamedan, Kerman, Fars, Khuzestan, Isfahan, Markazi, and Khorasan) were evaluated for two years (2007-2008) using randomized complete block design with three replications. Significant differences (P≤0.01) were observed for flower yield among genotypes (G), locations (L), environments (E), and also for genotype× location (GL) and genotype× environment (GE) interactions. The stable and adaptable genotypes with some static concepts of stability parameters such as environmental variance (S2) and regression coefficient (b ≈ 0) had low yield, but with environmental coefficient of variation (CVE) had high performance. The stable genotypes according to Eberhart and Russell model and the concept of stability dynamics (bi ≈ 1, ∑ Sdi2 ≈ 0, and high Ῡi) had moderate to high flower yields. According to the superiority index (P), high yielding genotypes were determined as compatible (stable in location). The stable genotypes with the lowest year by location (YL) variance had the lowest flowers yield. The flower yield was positively correlated with S2, b and MSY/L (p<0.01) and negatively correlated with superiority index (P) and CVE (p<0.01). Based on both flower yield and stability, it was found that genotypes of Yazd 2, Isfahan 5, Isfahan 8, Isfahan 4, and Khuzestan 1 had high flower yield, stability, and general adaptation over locations. The genotypes Isfahan 9 and Isfahan 7 had high flower yield, stability, and specific adaptation with semi-arid and arid areas and also the genotype of Isfahan7 had high flower yield, stability and specific adaptation with semi temperate and cold areas in Iran. It was concluded that a genotype can demonstrate both static and dynamic stability with a high flower yield at the same time. In addition, coefficient of variation (CV), statistics regression (b ≈ 1, Sd2≈ 0), and superiority index (P) suggested as desirable parameters to evaluate different aspects (static and dynamic) of flower yield in R. damascena.