evaluation of stability parameters for discrimination of stable, adaptable and high flower yielding landraces of rosa damascena

in order to determinate appropriate stability parameters, six statistics were studied for flower yield stability of 35 rosa damascena landraces in 8 locations over two years (2007-8) in iran, using a randomized complete blocks design with 3 replications. a positive correlation between environmental variance (s2) and flower yield suggested that only low yield landraces develop a similar phenotype over a range of environments. the stable and adaptable landraces using the environmental coefficient of variation (cv) produced a flower yield about average for landraces or higher. although all of the stable landraces by s2 produced very low yield, some of adaptable ones by cv (e.g. yz2) showed high flower yield and stability simultaneously. a negative correlation was observed between cv and flower yield. the regression coefficient of yield over environments (b) was positively correlated with flower yield; the regression coefficients of all studied landraces were statistically different from zero therefore were not stable with static stability concept (b equal to zero). the stable and adaptable landraces according to dynamic stability concept (b equal to unity and sd2 or variance due to deviation from regression equal to zero) produced a flower yield higher than average for landraces or near it. the superiority index (p) determined some of the highest flower yield as adaptable landraces. the stable landraces with the least variance of years within places (msy/p) produced the least flower yield; because of a mixing of effects (year with plant age), msy/p isn’t a favourable parameter for flower yield in perennial plants. some high flower yield landraces were found (e.g. yz2 and is5) showing stability and adaptability with varying statistics such as cv, b, sd2 and p. it could be concluded that a genotype can demonstrate both static and dynamic stability with high flower yield. in addition, the coefficient of variation (cv), dynamic view statistics (b equal to unity and sd2 equal to zero) and superiority index (p) are proposed as desirable parameters for evaluation of flower yield stability with different concepts in damask rose genotypes.