Continental-scale variation in controls of summer CO2 in United States lakes

Understanding the broad-scale response of lake CO2 dynamics to global change is challenging because the relative importance of different controls of surface water CO2 is not known across broad geographic extents. Using geostatistical analyses of 1080 lakes in the conterminous United States, we found that lake partial pressure of CO2 (pCO(2)) was controlled by different chemical and biological factors related to inputs and losses of CO2 along climate, topography, geomorphology, and land use gradients. Despite weak spatial patterns in pCO(2) across the study extent, there were strong regional patterns in the pCO(2) driver-response relationships, i.e., in pCO(2) regulation. Because relationships between lake CO2 and its predictors varied spatially, global models performed poorly in explaining the variability in CO2 for U.S. lakes. The geographically varying driver-response relationships of lake pCO(2) reflected major landscape gradients across the study extent and pointed to the importance of regional-scale variation in pCO(2) regulation. These results indicate a higher level of organization for these physically disconnected systems than previously thought and suggest that changes in climate and land use could induce shifts in the main pathways that determine the role of lakes as sources and sinks of atmospheric CO2. Plain Language Summary In this study we show that changes in climate and terrestrial landscapes could affect which are the main mechanisms responsible for the widespread emissions of CO2 by lakes. Although mechanisms such as aquatic primary production, respiration by microorganisms, or terrestrial loadings of carbon have been studied extensively, their relative importance across broad geographic extents with different climate or land use remains unknown. Based on an analysis of 1080 lakes distributed across the continental U.S., we show that lake CO2 dynamics depend on the climate and landscape context where these lakes are found, such as precipitation, elevation, percent agriculture, or wetlands in the lakes catchments. We observed a widespread effect of in-lake primary production, while the color of water, which has often been identified as one of the main controls of lake CO2 in northern lakes, was important in only a small fraction of the lakes studied. Our results show that controls on lake CO2 dynamics vary geographically and that considering that variation will be important for creating accurate global carbon models.