On the predictive power of numerical and Braun-Blanquet classification: an example from beechwoods

Abstract. A numerical classification method and the Braun-Blanquet method, based on external criteria, were compared with the aim of clarifying the differences in predictive power. The numerical analysis leads to a changed perspective on the floristic data and produces, as a whole, an ecologically more differentiated classification. The groups produced by numerical analysis are almost identical with the Braun-Blanquet classification with respect to marginal site conditions but they may differ in cases of intermediate site conditions. Nevertheless, a classification emerges which is on the whole, both floristically and as to site, very similar to the Braun-Blanquet classification. The discriminatory importance of the external variables is, to a large extent, the same in both methods, but discriminant analysis shows that the numerical classification is somewhat more predictive.