Predicting the suitable habitat distribution of greater amberjack using the MaxEnt model with fishery and remote sensing data in the surrounding waters of Taiwan.

The increasing availability of satellite-derived remotely sensed oceanographic data offers significant potential for evaluating changes in habitat suitability driven by oceanographic phenomena and for informing scientific management strategies, particularly when coupled with maximum entropy (MaxEnt) methods. Oceanographic variables, including sea surface temperature (SST), sea surface salinity (SSS), sea surface height (SSH), sea surface chlorophyll-a concentration (CHL), mixed layer depth (MLD), and eddy kinetic energy (EKE), alongside fishery data from Taiwanese fishing vessels, were collected for the period 2014-2019. The MaxEnt model is a widely employed method for predicting species' geographical distributions by analyzing species occurrence data in relation to environmental variables. Our results indicated that the annual response curves of habitat suitability were significantly influenced by environmental factors. Specifically, optimal environmental conditions for high habitat suitability were identified as SST > 24 °C, SSS < 35 PSU, SSH between 0.41 and 0.64 m, CHL > 0.22 mg/m3, MLD is between 10 and 17 m, and EKE < 0.007 m2/s2. The MaxEnt models exhibited strong predictive performance across all seasons and overall, as evidenced by receiver operating characteristic curve (AUC) values exceeding 0.8 and true skill statistics (TSS) values above 0.7, thereby confirming their accuracy in predicting greater amberjack presence, a reference for future conservation and management priorities.