The aim of this paper was to verify the feasibility of using multivariate statistical methods through structural variables for successional mosaic characterization in a semideciduous forest section. Plots measuring 10m x 10m were allocated for structural analysis (phytosociological survey plus the variables Coverage Percentage (CP), Canopy Height (CH) and Liana Cover (LC). The following statistical methods have been used: Principal Components Analysis and Cluster Analysis - more specifically, Ascending Hierarchical Classification. 43.96% of the total variance has been explained by the first principal component and 25.66% by the second one. The variables Basal Area (BA), Average Diameter (AD) and Average Dominance (ADOM) presented positive correlations (higher than 0.75) among themselves. Therefore, AD and ADOM may be considered as variable groups. The variables Number of Individuals (NI) and Number of Species (NE) showed a 0.60 correlation. The variables CH, LC and CP presented lower correlations. These findings show that their inclusion in this analysis was important. The hierarchical classification and the division of the groups in four parts have been performed considering two first factorial axes. Results showed two different types of behavior: 1) low values for CH and BA - Group 1 with low values also for NI, NE and CP (Gap Phase) and Group 2 with high values for NI and LC and low values for ADOM and AD (Building Phase); 2) high values for CH and BA - Group 3 with high values also for NI, NE and CP and low value for LC (Mature Phase) and Group 4 with high values for ADOM and AD, and lower for LC (Degradation Phase). The multivariate statistical methods allowed the forest mosaic developing phase characterization through structural variables. The estimative of CH, LC and CP variables must be improved. Other variables should be included in order to better differentiate the phases.
Multivariate statistical methods; structural analysis; forest mosaic; phytosociology; semi-deciduous forest; principal components analysis; cluster analysis
