Towards precision limnology: An explainable AI framework decoding spatiotemporal algal dynamics in Chinese major...

Lake ecosystems are increasingly threatened by eutrophication and harmful algal blooms under climate change and anthropogenic stressors, posing significant challenges to sustainable water resource management. To address this, we developed an interpretable machine learning framework to systematically analyse algal cell density (AD) dynamics and identify key drivers across 21 major lakes in China. By integrating high-resolution water quality and meteorological data (2021-2024), we constructed lake-specific predictive models that achieved high accuracy (mean R² > 0.8), with model structures varying substantially among lakes and highlighting the need for regionally tailored approaches. Notably, automated feature selection reduced input variables by 50 % while preserving predictive performance (ΔR² < 0.03). Through explainable AI (XAI) and nonlinear modelling, we uncovered spatially varying driver-response relationships, including critical thresholds linked to regional environmental baselines. These findings provide a XAI framework for optimizing monitoring networks and targeting algal bloom mitigation, offering practical tools for evidence-based lake management in rapidly changing environments.