Can modelling enable us to understand the rôle of humans in landscape evolution?

Landform-evolution models have typically failed to include human actions, or have done so only in a static, scenario-based way. This failure is despite the extensive empirical data that suggest rates of soil erosion are most sensitive to anthropic pressure. The CybErosion modelling framework overcomes this limitation by using an agent-based approach to simulating the dynamic interactions of people and their landscapes. The interactions simulated relate to basic processes of food acquisition (hunting, gathering and basic agriculture) in prehistoric communities. Simulations demonstrate the value of this approach in supporting the vulnerability of landform evolution to anthropic pressures, and demonstrate the limitations of existing models that ignore human and animal agency, which are likely to produce both quantitatively and qualitatively different results. The model is also a useful heuristic tool for understanding human-landscape interactions and for suggesting directions for future research. Despite the acknowledged limitations of agent-based approaches in simulating human populations, it is suggested that further research will be fruitful, especially if combined with a range of field evidence.