Prioritization of key indicators for the classification of successional stages in regenerating subtropical Atlantic forest, Southern Brazil: a proposal based on multivariate order statistics

Abstract

Statement of problem. Brazilian guidelines establish a set of parameters to classify forest regeneration stages, but there are no criteria or evidence for prioritizing key indicators. Purpose. This study performed a comparative analysis between phytosociological parameters to verify the hypothesis that there is a difference in their explanatory power of the forest regeneration stages. Consequently, it shows the need to prioritize key indicators for classifying forest regeneration stages. Material and methods. The analyzes were carried out using a database of the Floristic Forest Inventory of Santa Catarina, Southern Brazil, composed of 177 sampling units of subtropical Atlantic forest (Mixed Ombrophilous Forest) characterized according to 12 quantitative and qualitative phytosociological parameters. Several statistical tests were performed to analyze the explanatory ability of the phytosociological parameters. Such process includes tests for multivariate ordering, principal component analysis (PCA), discriminant analysis (FDA and QDA), exploratory factor analysis (EFA), and multiple correspondence analysis (MCA). Afterward, is provided a ranking regarding the analyzed indicators based on their usefulness to separate forest regeneration stages, measured in terms of classification accuracy according to the QUEST (Quick Unbiased Efficient Statistical) algorithm. The hypothesis was verified with two-way tests at a significance level (α) equal to 0.05, for a test power (1-β) of 0.8 and a minimum detectable effect of medium size (ρ = 0.3). Results. Statistical significance tests confirmed the research hypothesis. Regarding both the qualitative and quantitative variables, the ranking resulting from the discriminant analysis provided the best accuracy (85.3%). In decreasing priority order, was defined the following parameter order: basal area, number of individuals and species, Shannon diversity index, diameter at breast height, total height, stem height, leaf litter, canopy structure, canopy cover, the density of lianas and epiphytes. Conclusions. The phytosociological parameters have statistically different explanatory power (p < 0.0001), which should be considered when classifying forest regeneration stages. Practical implications. The presented results and conclusions should substantially impact the guidelines and decision-making process for cutting down the forest/native vegetation and applicable compensation measures, which are currently only based on the forest regeneration stages.