Assessing of similarity indices for the assignment of the plots to the plant communities of an existing phytosociological classification

In this paper, we evaluate quality of several similarity indices for assignment of new releve to vegetation units, using geometric and non-geometric based evaluators. For this purpose, Buxus hyrcana dataset (437 releves) that belong to Mazandaran and Golestan provinces in Hyrcanian forests was used. 11 vegetation units as a primary groups produced by combination results of TWINSPAN, Modified TWINSPAN and cluster analysis. All of 437 releve were re-assigned to 11 primary groups by six similarity indices indices that consider Phi fidelity index. With the exception of Calinski and Uniqueness evaluators, classification resulting from TPFI similarity index had the highest values of evaluators. Results of PCA analysis showed that ISA evaluator as a non-geometric evaluator with two geometric evaluators consist of PBC and PARTANA are the most important evaluators. Totally, the results of this study revealed that among six similarity indices, TPFI to be an effective one for assignment of new releve to vegetation units.