Regime shifts in shallow lakes: palaeoecological and agent-based model approaches

Environmental problems caused by lake eutrophication have become more widespread at a global level, threatening the safety of water, food and other daily needs of people living in the vicinity of lakes. Although lake types, physical conditions and causes of degradation vary from region to region, all threatened lake ecosystems face the same problem of identifying the mechanisms that underlie the deterioration of water quality or cause algal blooms in lakes, and discovering pathways to recovery is becoming an increasingly urgent matter. Theoretical and mathematical models based on Alternative Stable States (ASS) can be used to model and explain the deterioration and restoration of shallow lakes, while multiple patterns of ecosystem state response to external drivers have been validated by a large number of observations in European and North American lakes. In eastern China, eutrophic lakes account for 86.7% and heavily eutrophic lakes for 12.2% of more than 100 lakes in the Middle and Lower Reaches of the Yangtze River plain. Large lakes, such as Chaohu and Taihu, have experienced large cyanobacterial outbreaks in the early twenty-first century, and these disasters will probably be replicated in other lakes under similar conditions. I therefore need to understand the mechanism underlying such catastrophic shifts in lake ecosystems. Natural and anthropogenic influences on the ecological trajectories of lake ecosystems, response mechanisms, and how to avoid subsequent catastrophes are the main questions addressed by this thesis.To discover the external conditions that predominantly affect lake ecosystems in the lower Yangtse River, I used palaeolimnological tools and selected a typical lake in the middle and lower reaches of the Yangtze River, Lake Taibai, for a case study. The study used information relating to the species composition of subfossil diatom genera as indicators of the reconstructed ecosystem state as well as lake area, depth, chlorophyll content, transparency, ion concentration and nitrogen and phosphorus content as environmental factors to analyse the correlation between environmental changes. In terms of anthropogenic impacts, historical data on lake hydrodynamics, fish farming and nutrient loading in the basin over past decades were recovered using historical records, literature research and proxy reconstruction to calculate the magnitude of the correlation between ecosystem state driven by human and natural factors and to analyse possible responses by examining feedback mechanisms.Since it is difficult to use palaeoecological data alone to reveal the dynamic ecosystem changes and emergent mechanisms under the influence of various external factors, I developed an agent-based model (ABM) to simulate the influence of environmental and human activities on lake ecosystems to help analyse how past patterns of shifts in state occurred, the influence of external conditions on these changes and how to avoid the development of catastrophic ecosystem failures. The ABM was constructed on the basis of the predation-prey relationship and other interactions like competition and providing refuge and on known ecological theories to simulate population dynamics in aquatic food webs in response to external drivers. The developed ABM – LAKEOBS_MIX achieves a reasonable balance between generality and realism, providing insights of how ecosystems were affected by various drivers in MLYB-like lakes. The effects of the external environment are implemented as sub-models to the biotic interactions, and the currently available factor models are lake nutrient levels, temperature, water depth, area and changes in the number of fishes in the lake. In-silico experiments were designed with multiple factors to simulate different conditions and measure ecosystem response patterns. The simulations show that the patterns of equilibria developed from the same initial state can be very different due to stochasticity of spatial distribution and decision making of individuals in functional groups. Decades are needed for the modelled ecosystem to form a dynamic equilibrium without external disturbance, which can be altered by any sudden extinction of a functional group. In experiments where nutrient loading constantly impacts on the formed equilibrium, results show regime shifts that occurred in different time and patterns due to changes in the amount, adding pace and timing of nutrient load, the presence of other influencing factors. Besides, recovery pathways are simulated in hyper-eutrophic system state, confirming moderate natural fishing and nutrient removing are efficient approaches to lower the total nutrient level in lake ecosystems.Combining these two approaches provided us with an improved understanding of the development trajectory of lake ecosystems in the MLYB over the last 100 years. I discovered links between changes in conditions and their corresponding response mechanisms, and according to the results of simulations I was able to develop possible regulatory approaches to avoid abrupt system degradatio