a study of morphological variations and their relationship with flower yield and yield components in rosa damascena

abstract rosa damascena has attracted considerable attention in horticulture, biochemistry and in pharmacology because of the fragrance of the flowers and the high content of its bio-logically active substances. there is limited information available on factors controlling flower yield and its components. the present study reports an investigation on flower yield and the various factors affecting it, in iranian rosa damascena using sixteen geno-types collected from different regions in iran. these genotypes were planted at isfahan agriculture research center, using a randomized complete block design with three repli-cations in 2005. twelve yield determining characters including flower yield per plant, number of flowers per plant, fresh flower weight, plant height, canopy diameter, length and width of bud, number of petal per flower, length and width of receptacle, fresh weight of petal as well as flowering period were recorded. rosa damascena genotypes sig-nificantly varied for all the traits except for the fresh weight of petals per flower. pheno-typic and genotypic coefficients of variation for flower yield per plant (48.03%, 36.49%), number of flowers per plant (40.65%, 26.99%), number of petals per flower (37.56%, 32.31%) were higher than the coefficients for other tested traits. cluster analysis revealed that khuzestan and shiraz 2 genotypes were the most related ones, while the most inde-pendent ones were the western and eastern azerbaijan genotypes. results of stepwise re-gression analysis showed that 90 percent of total variation of flower yield per plant could be explained by the number of flowers per plant. this suggested that number of flowers per plant is the most important component determining flower yield per plant. according to principle component analysis, parents with a high standing of the following traits: fresh weight of flower, number of petals per flower and bud width, may well be employed for hybridization in a breeding program. regression and cluster analyses proved to be the appropriate multivariate analyses for an identification of rosa damascena genotypes pos-sessing the desirable characters for hybridization to develop improved cultivars.