Predicting and managing risk of unsuitable dissolved oxygen in a eutrophic lake

Eutrophication of lakes is often accompanied by wide diel fluctuations in dissolved oxygen (DO) concentrations that can influence fish assemblage composition and periodically result in highly visible fish kills. We conducted a probabilistic risk assessment to estimate the likelihood that a shallow eutrophic oxbow lake would be affected by critically low DO concentration at dawn during mid-summer. Monte-Carlo simulations with a respiration model that considered DO concentration at dusk, water respiration rate, sediment respiration rate, air–water diffusive exchange, and night duration were used to generate a depth-specific probability distribution of DO concentrations for the following dawn. This probability distribution was combined with a bathymetric model of the lake, constructed in a geographic information system using geo-referenced depth data, to predict the probability that a certain area of the lake would be affected by critically low DO concentration. The area affected was evaluated as a function of lake stage elevation. Results of the risk assessment pointed to the importance of shallow water (<1 m depth) in determining the area of unsuitable DO at dawn. Simulation results suggested that water level manipulation can be used to reduce the probability of frequent unsuitable DO events that affect small areas of the lake, and simultaneously reduce the area of infrequent unsuitable DO events that affect large areas of the lake. High-probability unsuitable DO events affecting small areas of the lake are not of major concern because fish are able to locate refugia of suitable DO in adjacent areas. However, water level management in shallow eutrophic lakes can have a major effect on reducing the risk of low-probability unsuitable DO events that affect large areas of the lake.