OPTIMIZING MAXENT MODEL PERFORMANCE FOR FORECASTING CURRENT AND FUTURE POTENTIAL GEOGRAPHICAL DISTRIBUTION OF QUERCUS GLAUCA IN CHINA

Assessing the influence of occurrence record size on the MaxEnt model is crucial for robust species distribution forecasting. Quercus glauca, a key timber and climate-mitigation species in China, faces challenges in natural regeneration and systematic quality management. In this study, we partitioned 1172 occurrence records into six datasets (10%–100%) to evaluate the impact of sample size on MaxEnt performance. Our findings revealed that increasing the number of records reduced the AUC from 0.958 to 0.921; however, this did not impair the significance of primary environmental variables (Bio14, Bio7, Bio12, Pop, and Bio6). A medium sample size (586–820 records) yielded the most accurate predictions. We proposed an optimization strategy, integrating 50% and 70% datasets, which significantly enhanced predictive precision. Under current conditions, suitable habitats for Quercus glauca are concentrated in central and southern China. Under the SSP245 emission scenario, suitable habitats are projected to remain generally stable, with a moderate expansion towards southwestern China. This study demonstrates that optimizing the quantity of occurrence records is essential for refining the MaxEnt model and provides a systematic framework for accurate species distribution forecasting under changing climates.