Using maximum entropy modeling to predict the potential distributions of genus Copaifera L. in a conservation unit in the Brazilian Amazon.

de Oliveira Sousa, Jonathan Benathar; de Sousa Conceição Benathar, Isamara; Ebling, Angelo Augusto; Kohler, Sintia Valerio; de Oliveira, Ximena Mendes; de Paula Protásio, Thiago; Rodrigues, Marcos; Goulart, Selma Lopes

Species of the genus Copaifera L. are part of the flora present in the Carajás National Forest and have the potential for sustainable extraction of oil resin. In this study, we aimed to provide useful information regarding the areas within the Carajás National Forest through species distribution modeling based on environmental variables that present a higher suitability for the occurrence of the genus Copaifera L. The Maximum Entropy Algorithm (Maxent) was employed to calculate the environmental similarity (distribution modeling) between 129 points of occurrence. Satellite images for the years 2018 and 2021 were employed in two resampling methods for the replicas, based on rasterized environmental variables, applied in two models that differ in the Normalized Difference Vegetation Index (NDVI). The Area Under Curve (AUC) values for model one were 0.9003 and 0.9029 in the bootstrap and subsample resamples, respectively in model one, and 0.9276 and 0.9351 in model two. Considering all models, the variables altitude, sodium saturation (30–100 cm), and soil pH (0–30 cm) had the greatest importance and contribution in predicting occurrences. Furthermore, the variables altitude and soil pH (0–30 cm) provided greater stability to the model. Between 2018 and 2021, there was an increase in areas classified as very high occurrence suitability (p > 80%), whereas those classified as very low (p ≤ 20%) decreased. Based on the AUC values and the literature, it can be inferred that the models have excellent predictive capacity, contributing to the sustainable management and extraction of Copaifera oil resin.

NORMALIZED difference vegetation index; NON-timber forest products; ENVIRONMENTAL sciences; FOREST reserves; REMOTE-sensing images

Plant Ecology, 2025, Vol 226, Issue 2, p185

1385-0237

Academic Journal

10.1007/s11258-024-01484-9