Grouping Fusarium section Discolor isolates by statistical analysis of quantitative high performance liquid chromatographic data on secondary metabolite production

The profiles of secondary metabolites of Fusarium culmorum, F. graminearum and F. crookwellense were determined by high performance liquid chromatography (HPLC) of culture extracts. On a qualitative basis there were no differences between the profiles of chromophore families of the three species. The quantitative data (i.e., the heights of chromatographic peaks) were, however, found to contain much more information. The log-transformed heights of 61 peaks were used as the initial raw data material, and analysed by two multivariate statistical methods: (i) SIMCA (soft independent modelling of class analogy) and (ii) CLA (cluster analysis). These methods showed a distinct grouping of the data sets into three separate classes (SIMCA) or three separate clusters (CLA). Irrespective of the method used, the grouping was consistent with identifications based on conidium morphology, demonstrating that statistical evaluation of quantitative data are useful for fungal chemotaxonomy.