An entire-process MaxEnt framework for habitat suitability modeling on Google Earth Engine: A case study of the...

The MaxEnt model is a probabilistic statistical model based on the principle of maximum entropy. It can predict species' suitable habitats with limited species distribution data while maintaining high accuracy, making it widely used for extracting potentially suitable habitats for species. Although the model has been integrated into the Google Earth Engine (GEE) platform, few studies have explored its entire process simulation of species' suitable habitats at large spatial scales. The main objective of this study is to simulate the potentially suitable habitat of the oriental white stork in eastern mainland China, with the entire experimental process conducted using the MaxEnt model on the GEE platform. First, environmental variables affecting the habitat, reproduction, and foraging of the oriental white stork were collected. Subsequently, pseudo-absence points were generated using environmental analysis methods that account for suitable habitat conditions. Then, the potentially suitable habitats for the oriental white stork in the eastern part of mainland China were simulated and evaluated. The results are as follows: (1) The MaxEnt model based on GEE successfully predicted the oriental white stork's potentially suitable habitat at large spatial scales. Through repeated iterative training, the maximum test AUC reached 0.94 and the average test AUC was 0.93, while all TSS exceeded 0.70 with an average value of 0.75, indicating that the model has high predictive accuracy for species habitat simulation. In addition, all AUC ratios were above 0.96, with an average value of 0.99, suggesting low overfitting and strong model generalization ability. (2) The total area of potentially suitable habitat in eastern mainland China was 401,414.36 km² (predicted probability ≥0.33), with highly suitable habitat covering 83,656.99 km² (predicted probability ≥0.90), accounting for approximately 2.43 % of the study region. (3) The primary environmental variables influencing habitat suitability were the distance to lakes, elevation, mean diurnal range, isothermality, vegetation type, and land use type. (4) The optimal habitats for the oriental white stork are located in the low-elevation plains or near lakes (0-2000 m), with mild and humid climates, adequate food availability, the mean diurnal range is around 10 °C, and isothermality values ranged between 0.19 and 0.20, and the land categories dominated by cultivated land and cropland. This study confirms the feasibility of using the MaxEnt model on GEE for large-scale suitability habitat simulations and provides technical support for the conservation and habitat management of the oriental white stork.