Freshwater ecosystems are increasingly recognized as important components of the global carbon cycle. Research has shown that lakes and streams are highly active sites of carbon processing and that they release large quantities of greenhouse gases including carbon dioxide (CO2). However, the large-scale freshwater carbon emission estimates are still poorly constrained. Improving these estimates require both more observations and improved methods for upscaling existing observations to larger geographical regions. The release of greenhouse gases in freshwater ecosystems is both sustained by processes taking place within and outside the systems, where the intimate connection between freshwater habitats and the terrestrial environment is influential. This is especially the case for small lakes that are highly abundant and make up a significant proportion of the total lake surface area. Nevertheless, they have not received the same attention as the medium- and largesized lakes despite their dominance in the landscape. This thesis explores carbon cycling in small lakes and stream networks and contains four chapters addressing methodology, environmental drivers, and upscaling.
Chapter I describes a novel floating chamber designed to perform automatic measurements of lake CO2 fluxes. The chapter describes how a timer, and a miniature air pump can be used to extend a traditional floating chamber design with a cost-efficient CO2 sensor. This modification enables periodic replacement of the floating chamber headspace with atmospheric air and, thus, enables continuous CO2 flux measurements, alleviating the need for manual intervention. The suggested solution is easy to assemble and very cost-efficient compared to commercial alternatives. These characteristics allow researchers to obtain many measurements of lake CO2 flux covering the entire diel cycle, something which is otherwise very laborious.
Chapter II examines the influence of drought on carbon cycling and CO2 fluxes from small lakes and their air-exposed sediments. The chapter shows the high CO2 fluxes from lake sediments that have recently been exposed to air and their strong dependence on water content. Within the lakes, respiration exceeded gross primary production resulting in poor oxygen conditions and CO2 emissions. The high CO2 fluxes from air-exposed sediments may significantly offset the ecosystem carbon balance even during shorter periods of drought. The chapter highlights the impact of drought on system-scale carbon processes in small lakes and large-scale emission estimates, which is further reinforced by the expected increasing frequency of drought events in the future due to climate
changes.
Chapter III investigates the contribution of lake carbon cycling processes to the CO2 flux in small forest lakes. This is done using the equipment developed in chapter I, accompanied by other high-frequency sensors covering an annual cycle. The chapter shows that aerobic lake metabolism alone cannot account for the observed CO2 flux. The discrepancy can be explained by potential contributions from anaerobic metabolism or groundwater input. A national data set comprising several hundred lakes, was used to further investigate the discrepancy between CO2 flux and aerobic metabolism in relation to lake size. This analysis shows that the discrepancy is most pronounced in small lakes, highlighting the influence of size on the contributions of different sources to lake CO2 flux.
Chapter IV uses national stream monitoring data from Denmark, Sweden, and Finland along with catchment characteristics to predict CO2 partial pressure in the entire stream network. The chapter highlights the applicability of machine learning algorithms to improve predictive performance using remote sensing data products. The most influential drivers were catchment elevation, permanent water cover, and slope, revealing the importance of wetlands and catchment geomorphometry on stream CO2 partial pressure. The chapter shows how data from different sources can be used to improve our ability to upscale existing observations to larger geographical regions and reduce the uncertainty of large-scale freshwater carbon emission estimates. The suggested framework is general and can be applied to other stream chemical compounds as well.