Abstract: The Chinese traditional chrysanthemum cultivar is a marvelous group of Chrysanthemum ×grandiflorum Tzvelv., whose floral characters are rich in variation. However, neither the values of floral characters for the classification of chrysanthemums nor the relevance of different characters are yet known. Moreover, the existing classification systems of chrysanthemums are incomplete and unsystematic. In the present study, we selected 16 floral morphological characters on which to carry out diversity and relativity analyses by seven multivariate statistical analysis methods, and we further studied the cultivar classification. The analyses of nested variance, diversity index, and principle components showed that 10 floral morphological characters were of great value for chrysanthemum classification. Clustering analysis classified these 10 characters into three types, in which the variation of the ray floret factor was the highest. The analysis of partial correlations showed that the petal type was significantly affected by the length of the ray florets, the bending state of the outer-layer ray florets, and the tip shape of the ray florets; besides these three characters, the flower head type was also affected by the angle of the outer-layer ray florets and the length and distribution of the disc florets. The multiple linear regression equations for petal type and flower head type were successfully established. All of the cultivars were classified into four groups and 18 forms by cluster and discriminant analyses. These results provide a reference that can be used for cultivar identification and classification of chrysanthemums.

Key words: chrysanthemum, cultivar classification, flower head type, morphological character, multivariate statistical analysis, petal type.