Leveraging Bayesian network to reveal the importance of water level in a shallow lake ecosystem: A study based on Paleo-diatom and fish community.

Lake ecological processes and nutrient patterns are increasingly affected by water level variation around the world. Still, the long-term effects of water level change on lake ecosystems and their implications for suitable lake level management have rarely been studied. Here, we studied the ecosystem dynamics of a mesotrophic lake located in the cold and arid region of northern China based on long-term paleo-diatom and fishery records. Utilizing a novel Copula-Bayesian Network model, possible hydrological-driven ecosystem evolution was discussed. Results show that increased nutrient concentration caused by the first water level drop in the early 1980s incurred a transition of sedimental diatoms towards pollution-resistant species, and the following water level rise in the mid-1980s brought about considerable external loading, which attributed to eutrophication and caused the miniaturization of fishery structure. In the 21st century, a continuous water level plummet further reduced the sediment diatom biomass and the fish biomass by altering nutrient concentration. However, with the implementation of the water diversion project in 2011, oligotrophic species increased, and the ecosystem developed for the better. From the perspective of water quality protection requirements and the ecological well-being of Lake Hulun, the appropriate water level should be around 542.42-544.15 m. In summary, our study highlights the coupling effect of water level and water quality on Lake Hulun ecosystem and gives shed to lake water level operation and management under future climate change and human activities.