Construction of importance value index of species using multivariate analysis for phytosociological analysis of dense rain forest

Abstract

An important tool for community ecology is multivariate analysis, which treats all variables simultaneously, summarizing data and revealing its structure with the smallest possible loss of information. This study aimed to develop, by multivariate analysis, two phytosociological indices derived from the same variables of the Importance Value Index (IVI), in order to compare such results between the indices by ranking of tree species sampled in a dense rain forest inventory, according to three previously established ecological categories: high, medium and low prevalence. The suitability of factor analysis was determined by Bartlett and KMO tests. Bartlett's test evaluated the overall significance of the correlation matrix indicating that correlations in general were significant at probability of p < 0.01. The KMO test indicated that the variables were correlated and the factorial model presented a good suitability level to the data. These results reinforce factor analysis using for factor extraction and estimating factor species scores. The best index of the three was obtained with factor analysis, which added a dummy variable to each variable in the model.