Mangrove Growth Suitability Based on an Improved MaxEnt Model: Cases from Gaoqiao and Qi’ao Island

Mangrove ecosystems deliver disproportionate ecological and socioeconomic benefits to coastal zones. However, assessments of the habitat suitability of mangroves seldom explicitly account for anthropogenic pressures. In this study, habitat suitability modeling was enhanced by integrating human disturbance indicators into the MaxEnt framework and introducing the Human-Integrated Mangrove Suitability Index (HMSI) as the probabilistic output of the model when anthropogenic variables are included. We applied this approach to two representative protected areas in Guangdong Province in China to evaluate whether incorporating human activity information improves the predictive performance of the model and its ecological interpretability. The areas studied were Gaoqiao (Zhanjiang) and Qi’ao Island (Zhuhai), which differ markedly in their human footprints. Twelve environmental predictors were compiled, covering the topography (elevation, slope, and distance to the coastline), oceanographic conditions (sea surface temperature (SST) and salinity), and anthropogenic pressures (population density, land-use type, and nighttime light intensity). MaxEnt models were trained using 75% of the occurrence records and tested on the remaining 25% and their robustness was assessed using the area under the receiver operating characteristic curve (AUC) and complemented using threshold-based true skill statistic (TSS). Experiments were performed to compare a natural factor-only configuration with an integrated configuration that yields HMSI values. The incorporation of anthropogenic predictors consistently improved the discriminative capability of the model: the AUC for Gaoqiao increased from 0.826 to 0.887 and that for Qi’ao Island from 0.875 to 0.918, and the TSS values increased in parallel, indicating enhanced sensitivity and specificity under realistic thresholding. The model outputs revealed contrasting controls across the study sites. The suitability of Gaoqiao was primarily driven by natural controls, with the slope, land use, and SST collectively explaining the majority of model gain and the response curves indicating peak suitability on gentle slopes (0.3°~5°) and near 27.5℃ SST. By contrast, the suitability of Qi’ao Island was dominated by anthropogenic pressure, with the population density, land use, and salinity exerting the strongest influence and the suitability elevated at lower population densities (approx.