Adaptive Agents Simulation of Freshwater Ecosystems: Artificial Intelligence Framework to Discover and Forecast Emergent Ecosystem Structures and Behaviours in Response to Environmental Changes

Adaptive Agents Simulation of Freshwater Ecosystems: Artificial Intelligence Framework to Discover and Forecast Emergent Ecosystem Structures and Behaviours in Response to Environmental Changes. The project aims at intelligent adaptive agent models for lakes and rivers in order to improve understanding and proactive management of these highly complex ecosystems. Little is known about species succession in freshwater ecosystems in response to local and global environmental changes. Evolutionary algorithms embodied in differential equations, neural networks and rules allow adaptive agents to simulate emergent structures and behaviours of algae and zooplankton communities interacting by competition and predation. The agents are trained and tested by ecological time-series of twelve lakes and rivers, and validated for the Mediterranean Myponga Reservoir, South Australia, and the temperate Burrinjuck Reservoir, NSW.