Abstract: Morphological variation of flowers in Sinocalycanthus chinensis were investigated among natural populations and within the population in three different regions. Cluster analysis was used to cluster these three populations. Results indicated that the morphological characteristics of flowers had certain variation both within and among populations of S.chinensis. The average coefficient of variation of each characteristic of flowers ranged from 6.293(Inner-petal length)~46.90(Flower stalk length), while among populations ranged from 13.430(DMS)~15.720 (DLS). Morphological diversities of flowers were high at both species level and population level. In these three populations, DLS population was the highest. One way analysis of variance (ANOVA) showed significant differences existed in parts of flower traits, from which over 82.9% of the variance was attributed to the differences within population, it put forward that genetic diversity was the dominant factor of morphological variation of flowers. 17.1% of the variance existed among populations which meant that certain phenotypic differentiations existed among populations. Cluster analysis revealed DMS population and LXS population between which have the nearest distance were clustered into one group. There was a significant positive correlation between morphologal variation of flowers and geographic distance. After analysis it can be concluded that heterogeneity of micro-habitat, population isolation and genetic drift can lead to phenotypic differentiation of flowers in S.chinensis. Correlation analysis showed significant correlation within parts of flower traits.

Key words: Sinocalycanthus chinensis, flower, morphological variation, cluster analysis