Most lakes and reservoirs have surface CO 2 concentrations that are supersaturated relative to the atmosphere 1 . The resulting CO 2 emissions from lakes represent a substantial contribution to the continental carbon balance 2-4 . Thus, the drivers of CO 2 supersaturation in lakes need to be understood to constrain the sensitivity of the land carbon cycle to external perturbations 4-6 . Carbon dioxide supersaturation has generally been attributed to the accumulation of inorganic carbon in lakes where respiration exceeds photosynthesis 7,8 , but this interpretation has faced challenges 9-11 . Here we report analyses of water chemistry data from a survey of Spanish reservoirs that represent a range of lithologies, using simple metabolic models. We find that, above an alkalinity threshold of 1 mequiv. l −1 , CO 2 supersaturation in lakes is directly related to carbonate weathering in the watershed. We then evaluate the global distribution of alkalinity in lakes and find that 57% of the surface area occupied by lakes and reservoirs-particularly in tropical and temperate latitudes-has alkalinity exceeding 1 mequiv. l −1 . We conclude that lake inputs of dissolved inorganic carbon from carbonate weathering should be considered for the CO 2 supersaturation of lakes at both regional and global scales.There are two main mechanisms leading to CO 2 supersaturation in lakes and reservoirs: in situ net ecosystem production (NEP) imbalanced towards net heterotrophy 7 (that is, respiration exceeding photosynthesis); and inputs of groundwater or surface water with high dissolved inorganic carbon (DIC) content coming from both weathering of minerals and soil respiration in the watershed 10-14 . Among these, NEP is usually considered to be the main factor driving CO 2 supersaturation in lakes 7,8 . However, most studies relating CO 2 supersaturation to NEP have focused on relatively dilute, low-alkalinity lakes, despite the fact that CO 2 concentration in water is strongly modulated by the carbonate equilibrium (that is, the chemical reactions relating the different forms of DIC). Therefore, we lack conclusive evidence of the impact of NEP on CO 2 supersaturation in a range of systems showing contrasting DIC content.To test how DIC content can drive CO 2 supersaturation and modulate the relationship between NEP and CO 2 concentration, we use data from a nationwide study 15 including 202 measurements of dissolved oxygen (DO), DIC, alkalinity and CO 2 concentration in the surface layer of a set of Spanish reservoirs covering a wide range of DIC content and trophic states (Supplementary Table 1 and Data). We use the observed DO disequilibrium relative to the atmosphere as a surrogate for surface lake NEP, and investigate its impact on observed DIC and CO 2 concentrations considering three metabolic models. The models combine assumptions concerning the DIC loading from the watershed and the effect of lake NEP (Fig. 1a). All three models assume that the DIC generated during weathering of minerals in the watershed (DIC W ) dominates DIC...