VEGETATION SUCCESSIONAL STRATA CLASSIFICATION IN THE AMAZON ESTUARY THROUGH QUANTILE REGRESSION

Vegetation successional strata classification is essential for guiding forest management and conservation, particularly in Amazonian floodplain environments. This study sought to classify successional stages in a floodplain forest of the Amazon estuary by adjusting height–diameter models using quantile regression. Data were collected from a permanent forest inventory in Gurupá, Pará. Two models (Curtis and Stoffels) were tested, and their performance was evaluated using root mean square error (RMSE), Pearson’s correlation coefficient, Akaike information criterion (AIC), bias, and mean absolute deviation (MAD). Residual distribution and trend line plots were also analyzed. Quantile regression proved to be an efficient method for fitting the models, with the Stoffels equation showing the best performance. By adjusting its coefficients, this equation enabled the classification of species across different successional strata, demonstrating the potential of quantile regression as a robust tool for analyzing forest structure in heterogeneous environments.