CO2 supersaturation along the aquatic conduit in Swedish watersheds as constrained by terrestrial respiration, aquatic respiration and weathering


Christoph Humborg, e-mail:


We tested the hypothesis that CO2 supersaturation along the aquatic conduit over Sweden can be explained by processes other than aquatic respiration. A first generalized-additive model (GAM) analysis evaluating the relationships between single water chemistry variables and pCO2 in lakes and streams revealed that water chemistry variables typical for groundwater input, e.g., dissolved silicate (DSi) and Mg2+ had explanatory power similar to total organic carbon (TOC). Further GAM analyses on various lake size classes and stream orders corroborated the slightly higher explanatory power for DSi in lakes and Mg2+ for streams compared with TOC. Both DSi and TOC explained 22–46% of the pCO2 variability in various lake classes (0.01–>100 km2) and Mg2+ and TOC explained 11–41% of the pCO2 variability in the various stream orders. This suggests that aquatic pCO2 has a strong groundwater signature. Terrestrial respiration is a significant source of the observed supersaturation and we may assume that both terrestrial respiration and aquatic respiration contributed equally to pCO2 efflux. pCO2 and TOC concentrations decreased with lake size suggesting that the longer water residence time allow greater equilibration of CO2 with the atmosphere and in-lake mineralization of TOC. For streams, we observed a decreasing trend in pCO2 with stream orders between 3 and 6. We calculated the total CO2 efflux from all Swedish lakes and streams to be 2.58 Tg C yr−1. Our analyses also demonstrated that 0.70 Tg C yr−1 are exported to the ocean by Swedish watersheds as HCO3 and CO32− of which about 0.56 Tg C yr−1 is also a residual from terrestrial respiration and constitute a long-term sink for atmospheric CO2. Taking all dissolved inorganic carbon (DIC) fluxes along the aquatic conduit into account will lower the estimated net ecosystem C exchange (NEE) by 2.02 Tg C yr−1, which corresponds to 10% of the NEE in Sweden.