Impact of ecosystem metabolism on CO emissions: Insights from high-resolution time series of pH measured in situ.

Lakes and reservoirs are important sources/sinks of atmospheric CO. Primary production and respiration transforming inorganic to organic carbon and vice versa alter CO concentrations in the surface waters and thus affect CO emissions. Here we investigate the link between net-production (NEP) and CO concentrations and emissions at high temporal resolution over more than two months in a German pump storage reservoir. Continuous in-situ pH measurements in combination with few alkalinity measurements provided concentrations of CO and dissolved inorganic carbon (DIC) at high temporal resolution over more than 75 days. Time series of metabolic rates of carbon were determined with an open-water diel pH technique, which utilizes the diel changes in DIC obtained from the observed diel changes in pH and data on alkalinity. During the measuring period, average NEP was positive and CO concentrations were typically substantially under-saturated. On average, the reservoir acted as a sink for CO, whereby CO uptake was 39% larger in the evening than in the morning. Only few consecutive days with negative NEP were sufficient to turn the reservoir temporally into a source of CO. Therefore, the average CO uptake determined from continuous data can be 80% larger to 30% smaller than estimates of average uptake based on bi-weekly data. Daily mean NEP explained only 9% and 4% of the variance of daily mean DIC and CO. Note that NEP is proportional to the time derivative of DIC and therefore not expected to correlate well with DIC in general. Because CO changes nonlinearly with DIC, NEP explains less variance of CO than of DIC. Numerical experiments confirmed the arguments above and revealed that at positive average NEP the total CO uptake over several weeks is not well predicted by average NEP but depends on the detailed temporal pattern of NEP. However, if average NEP is negative, average NEP may be a good predictor of total CO emissions. Similar conclusions apply for high and low alkalinity waters, but uptake rates and temporal variability of CO emissions are smaller in high than in low alkalinity waters. Assessment of the link between NEP and CO emissions requires differentiation between lakes with different alkalinity and, because of the non-linear relationship between NEP and CO, strongly benefits from data with high temporal resolution especially during time-periods with positive net-production.