Ecological feedbacks modulate lake surface oxygen responses to heatwaves

Climate warming intensifies lake heatwaves (LHWs) and deoxygenation, yet short-term surface dissolved oxygen (DO) responses to LHWs remain unclear. This study integrates 25-year in-situ observations (2000–2024) and 73-year lake surface water temperature (WT) dataset (1950–2022) from Lake Taihu, a eutrophic shallow lake, with explainable machine learning models to decode seasonally divergent surface DO responses to LHWs. We reveal that surface DO during LHWs was generally lower than normal in winter, but higher in summer. Explainable machine learning revealed that DO is mainly influenced by WT in winter, higher temperatures reduced oxygen solubility, and ultimately led to lower DO during heatwaves. However, reduced ammonia nitrogen and turbidity during summer heatwaves inhibited oxygen consumption by nitrification and increased oxygen production from photosynthesis, ultimately resulting in DO increasing. These results highlight the complex impact of LHWs on DO, underscoring the need for tailored management strategies in a warming climate.