Widespread deoxygenation in warming rivers

Deoxygenation is commonly observed in oceans and lakes but less expected in shallower, flowing rivers. Here we reconstructed daily water temperature and dissolved oxygen in 580 rivers across the United States and 216 rivers in Central Europe by training a deep learning model using temporal weather and water quality data and static watershed attributes (for example, hydro-climate, topography, land use, soil). Results revealed persistent warming in 87% and deoxygenation in 70% of the rivers. Urban rivers demonstrated the most rapid warming, whereas agricultural rivers experienced the slowest warming but fastest deoxygenation. Mean deoxygenation rates (−0.038 ± 0.026 mg l−1 decade−1) were higher than those in oceans but lower than those in temperate lakes. These rates, however, may be underestimated, as training data are from grab samples collected during the day when photosynthesis peaks. Projected future rates are between 1.6 and 2.5 times higher than historical rates, indicating significant ramifications for water quality and aquatic ecosystems.

This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: http://dx.doi.org/10.1038/s41558-023-01793-3