Abstract:

 A numerical taxonomic study of Caprifoliaceae is presented. For the sake of ana-

lyzing 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 numeri-

cal analyses.  Raw data for each character were given equal weighting by condensa-

tion 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 mor-

phological.  The estimation of the coefficient resemblance between each pair or OUT’s

was established using the association coefficient method.  The resulting values com-

prise 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 de-

markation 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. Trioste-

um 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 check

on ordinary taxonomic procedures.