Sea Level Rise and Fisheries Production in Southeast Asia: Trends and Explainable AI Insights

Climate change–induced sea-level rise (SLR) is transforming coastal ecosystems and threatening fisheries-dependent livelihoods in Southeast Asia, a region marked by rapid environmental change and high population density in low-elevation coastal zones. While prior studies have examined the physical and ecological dimensions of SLR, few have integrated large-scale fisheries data with interpretable machine learning (ML) frameworks to identify and quantify socio-environmental vulnerabilities. This study applies ensemble learning models, Random Forest (RF) and Extreme Gradient Boosting (XGBoost), to the Food and Agriculture Organization’s (FAO) Global Fisheries Production dataset (2005–2025) across ten Southeast Asian countries. The analysis incorporates tide-gauge-derived SLR trends and production characteristics (species, production source, and water area) to predict fisheries outcomes and assess regional risk. Model explainability is achieved using SHapley Additive exPlanations (SHAP), allowing transparent attribution of feature importance. Results indicate that species type, production source, and country context are the most influential variables shaping fisheries vulnerability. The Random Forest model achieved superior predictive stability (R² = 0.45, Min–Max Accuracy = 0.71) relative to XGBoost. The findings reveal that Cambodia, Thailand, and the Philippines—regions experiencing some of the highest SLR rates—exhibit heightened fisheries vulnerability. The integration of explainable AI with environmental datasets enhances understanding of where climate and socio-economic stressors converge, providing a replicable framework for climate adaptation and sustainable resource governance in coastal Asia.