J Syst Evol ›› 1983, Vol. 21 ›› Issue (1): 26-33.

• Research Articles • Previous Articles     Next Articles

A Preliminary Numerical Taxonomy of the Family Caprifoliaceae

Hsu Ping‐Sheng   

  1. (Department of Biology, Fudan University)
  • Published:1983-02-18

Abstract: A numerical taxonomic study of Caprifoliaceae is presented. For the sake of analyzing the resemblances between the 33 species or OUT’s selected at random from the total 13 genera of the family, a summation of 32 characters was employed in the numerical analyses. Raw data for each character were given equal weighting by condensation in order to have adequate comparisons, and the characters were converted to 51 states, each with a new range of zero to one. Owing to the lack of sufficient data from other lines for numerical analyses, the characters used in this study were largely morphological. The estimation of the coefficient resemblance between each pair or OUT’s was established using the association coefficient method. The resulting values comprise the 33×33 OUT’s basic similarity matrix. The clustering technique used was unweighted pair-group method using arithmetic averages (UPGMA). It can be stated that the scheme of phenetic relationships shown in the resultant dendrogram (Fig. 1) is on the whole in accord with the concepts hold by most current taxonomists, but with some noteworthy exceptions. If the phenon line of tribal demarkation is drawn at the level of 0.6820, the OUT’s could be roughly divided into five groups or tribes. The fact that the highest degree of correlation between Group I Sambuceae and Group II Viburneae on the one hand, and the great distance between them and the rest genera of the family on the other hand agrees well with the data obtained from morphological (Troll and Weberling, 1966), anatomical (Wilkinson, 1949, Metcalfe and Chalk, 1950), embryological (Moissl, 1941), sereological (Hillebrand and Fairbrother, 1970), and phytochemical (Bohm and Glennie, 1971) researches. These two tribes are most probably members of different phylogenetic origin. Triosteum and Symphoricarpos both show their affinities with Leycesteria of Group V Lonicereae instead of Group III Linnaeea or Group II Viburneae as suggested by some taxonomists, and thus supports the opinion of Troll and Weberling (1966), who suggested that these two genera are members of the tribe Lonicereae. The location of the phylogenetically uncertain genus Heptacodium in the dendrogram shows its close morphological similarity to the tribe Linnaeeae. Because of the relatively small number of characters considered in this work, and “taxonomic judgement” was used in selecting these characters which appeared to be most “basic” to the classification of genera in the family, as well as the limitation of numerical taxonomy in itself, the resultant scheme of tribal relationships presented in this paper is by no means phylogenetic, but one that provides an excellent checkon ordinary taxonomic procedures.