Nicolas Behrens, Christian Brümmer, Kuno Kasak, June Skeeter, Ian B Strachan, Ype van der Velde, Chris D Evans, Ross Morrison, Carole Helfter, Guillaume Bertrand, Sébastien Gogo, Adrien Jacotot, Carsten Schaller, Karen Yeung, Mana Gharun
Peatlands are the largest terrestrial stores of organic carbon, but drainage has turned them into substantial sources of CO2. While water level is widely recognized as the primary control of CO2 emissions from peatlands, effective future management requires understanding its interaction with rising temperatures under a warming climate. Using 276 site-years of annual CO2 flux observations across temperate and boreal peatlands, we apply explainable machine-learning to disentangle the combined effects of water table depths and temperature on ecosystem CO2 exchange at annual scales. Across peatlands spanning diverse land-cover-including natural fens and bogs, croplands, grasslands, and extraction sites-CO2 emissions exhibit a non-linear response to water table depth. Emissions decline when water tables are raised above 60-75 cm depth. Optimal mitigation requires water tables of 20 cm or higher. We find that deep water tables interact with temperatures to increase emissions at temperate sites. On a subset of 113 site-years of daily CO2 flux data, we show that at warm temperatures, higher water tables suppress, whereas deeper water tables enhance temperature-driven CO2 emissions from peatlands. We demonstrate that hydrology regulates the temperature sensitivity of peatland carbon release, revealing a key control on carbon-climate feedbacks under future warming.