Abstract

Natural landscape morphology emerges from continuous, reciprocal interactions between biological agents and their physical environment. Despite its broad application across diverse scientific fields, agent-based modeling remains underexplored in the context of non-human geomorphological change. This paper presents a bio-inspired multi-agent framework to investigate how individual animal behaviors, specifically those of the North American beaver, shape adaptive landscapes. To capture dynamic task specialization, we introduce an architecture that abstracts alternating behavioral preferences into two operational states: Explorers (focused on resource identification) and Builders (focused on localized engineering). Deployed in a dynamic environment characterized by seasonal vegetation mean-reversion and a dynamic hydrological proxy, our targeted parameter sweeps and Monte Carlo replications demonstrate that decentralized stigmergic heuristics drive emergent spatial patterns. Quantitative metric analysis across varying colony sizes shows that while smaller swarms maintain a stable ecological equilibrium, larger populations trigger an apparent non-linear expansion of the hydrological network via active bank erosion. By establishing this foundational framework, this work provides an open-source tool to further explore non-human agency and regenerative strategies in landscape architecture and environmental design.

IPC Classification

Keywords