Similarity indices as an extension of vegetation classification

TICHY, L.

Department of Botany, Masaryk University, Kotlarska 2, CZ-611 37 Brno, Czech Republic
Email: tichy@sci.muni.cz

Calculation of similarity indices is a method suitable for assignment of relevés which are added a posteriori to an existing vegetation classification. This paper discusses properties of two similarity measures, Frequency Index and Fidelity Index, which were newly designed for assignment of relevés to constancy columns of synoptic vegetation tables. The Frequency Index is derived from the Czekanowski Index. The Fidelity Index depends on the diagnostic potential of species occurring in a relevé to be assigned, in comparison with the total amount of diagnostic species in the constancy column. The behaviour of both indices is demonstrated with a simulated data set in which either dominant or diagnostic species prevail in the tested relevés. Generally, both indices perform well but may fail in situations. The Frequency Index poorly discriminates relevés containing diagnostic species from relevés with only constant species. Fidelity Index is unable to distinguish the relevés composed of generalist species from the relevés with totally different composition. Therefore the combined Frequency-Fidelity Index was proposed. This new similarity index retains the advantages and lacks the disadvantages of both the above described indices. Calculation of all the indices is available in the JUICE program.