Marine ecosystems worldwide are increasingly degraded by upstream land use activities, compounding climate change impacts. However, empirically quantifying causal land-sea linkages remains challenging. Using remote sensing data (1987-2019) and four causal inference methods, here we developed an empirical and scalable framework to estimate how land use affects coastal turbidity across spatial scales in southern Costa Rica. We found that riparian natural vegetation (15 m buffer) significantly reduced gulf turbidity up to 800 m offshore, which overlaps with coral reefs and seagrass habitats. In contrast, pasture and gravel roads increased coastal turbidity. Effects were greatest for rivers that are short, steep, or have low discharge. Watershed-scale land uses showed no significant effects. We provide a replicable, scalable framework to identify causal pathways from land to sea, particularly valuable in data-limited regions. Riparian conservation and restoration could serve as effective strategies to align human land use needs with terrestrial, freshwater, and marine conservation.
