Abstract
Estuarine salinity serves as a link between climate-driven environmental change and ecosystem health, reflecting interactions among hydrological, atmospheric, and anthropogenic processes. This study aimed to assess the potential impact of Global Climate Change (GCC) on salinity levels within a managed shallow estuary (Biscayne Bay in South Florida). Salinity was chosen as the key water quality indicator because it directly influences the distribution and abundance of estuarine organisms. A machine learning model was developed using XGBoost to evaluate the dependency of salinity on features connected to climate change, including water temperature, tidal elevation, and freshwater inputs from inland canals. The SHapley Additive exPlanations (SHAP) framework quantified each feature's contribution to changes in salinity. The results indicate that the primary features driving salinity are geolocation and canal flow, with the latter being the most manageable. Simulation of future climate scenarios defined by the Intergovernmental Panel on Climate Change (IPCC) revealed that, to achieve the historical average salinity (29.4 PSU) for the study estuary (with current average flow of 55.8 m
/s), a freshwater flowrate of 39 m
/s would be required for all IPCC scenarios in the near term (2021-2040). An increase to between 63 and 87 m
/s would be needed in the long term (2081-2100). At these higher flow rates, greater vertical stratification of salinity is observed, which may require further mitigation to maintain benthic communities. These findings demonstrate how integrative modeling approaches, combining machine learning with SHAP-based interpretation, can advance fundamental understanding of estuarine salinity drivers, thereby supporting effective ecosystem management and resilience strategies under GCC.