<![CDATA[Coastal fisheries in tropical archipelagic regions are increasingly affected by climate-induced environmental changes that alter fish distribution and productivity. This study develops a machine learning-based framework to assess the impact of climate variability on fisheries productivity using multiple environmental predictors, including sea surface temperature, chlorophyll-a concentration, dissolved oxygen, and climate oscillation indices. Several machine learning algorithms were evaluated, including Random Forest, SVR, ANN, XGBoost, and LightGBM. Model performance was assessed using R², RMSE, and MAE, while SHapley Additive exPlanations (SHAP) were applied to identify key environmental drivers. Results show that LightGBM achieved the highest predictive accuracy (R² = 0.92), outperforming other models in capturing nonlinear relationships between environmental variables and fisheries productivity. Sea surface temperature emerged as the most influential predictor, followed by chlorophyll-a concentration and dissolved oxygen. Scenario analysis indicates that continued ocean warming may reduce fisheries productivity by 10–25% in vulnerable coastal zones. The study highlights strong nonlinear interactions between climate variables and fisheries dynamics, emphasizing the importance of integrating machine learning into ecosystem-based fisheries management. The proposed framework provides a robust tool for predicting climate-induced changes in fisheries productivity and supporting adaptive management strategies in tropical marine ecosystems.]]>
