Daily net ecosystem production in lakes predicted from midday dissolved oxygen saturation: analysis of a five-year high frequency dataset from 24 mesocosms with contrasting trophic states and temperatures

Much information can be gained from the net ecosystem production (NEP) of freshwater lakes, and NEP has attracted new interest due to climate change, which may change the importance of freshwaters as a source and sink of CO2. Direct measurement of NEP in freshwater lakes is, however, time-consuming, and high frequency monitoring of diel variations in oxygen levels for metabolism estimation has only recently been commonly employed. However, midday snap-shot oxygen data is available from numerous monitoring programs worldwide, occasionally covering decades. We hypothesize that midday oxygen saturation levels may provide information on NEP in lakes as oxygen super-saturation, indicative of high NEP, is observed in very productive lakes, whereas very low oxygen levels may occur in lakes with great input of organic matter or in lakes experiencing sudden decline in primary producers, indicative of low NEP. By analysis of a high frequency dataset encompassing 24 fully mixed mesocosms with contrasting trophic states and temperatures, we show that midday oxygen saturation provides a reasonable description of daily NEP only marginally affected by trophic state, temperature, and season. Oxygen sampling conducted in the afternoon gave a slightly better prediction than at midday, whereas predictions based on NEP representing an average of the previous 3 days led to a 2-fold increase in R-2. Moreover, an analysis of high frequency data sampled in a shallow Danish lake suggests that the method is transferable to natural shallow lakes. This method may therefore allow estimation of NEP based on oxygen measurements available from monitoring programs.