The overarching aims of most research modelling land use change are to i) understand the spatial and temporal patterns of behaviours and their environmental implications, and ii) to forecast how these patterns may change under hypothetical future scenarios. Two approaches to address such research questions have emerged, agent based simulation and statistically based approaches, which have often been used in isolation. In this paper, we explore the potential for integrating these two approaches by using choice modelling to parameterise agent behaviour in an agent based simulation model. The choice models have the potential to provide a theoretical and empirical foundation for the link between changing environments and responses of land managers.

We test this approach in the upland ecosystems of the United Kingdom. These regions provide important ecosystem services such as water supply for much of the UK’s rural and urban population, carbon storage and sequestration, aesthetic values for tourism and recreation, and habitats for rare plants and animals. The potential for these ecosystems to provide these services are directly and indirectly influenced by land management that is currently dominated by sheep and game bird production. We use a choice experiment performed with land managers in the Peak District National Park to reveal preferences for alternative production strategies and the dependence of such choices on the environmental characteristics of the land. This data forms the basis for a behavioural model, which is integrated with complex environmental process-based models to predict environmental impacts on carbon fluxes, water quality, soil erosion and biodiversity. The integrated model allows us to create different projections of how land use may change in the future under different environmental and policy scenarios and the environmental impacts this might have. We illustrate this by showing future projections of landscape changes related to hypothetical changes to current EU level agricultural management incentives.

This approach allows the simulation of choices in response to the environmental consequences of past actions, and helps to model the interdependence between individual decision making, factors that are usually ignored in studies using choice modelling. Furthermore, the behavioural rules are revealed through empirical data rather than dictated through expert opinions, derived from the literature or simply assumed. This allows the research to test the statistical significance of various determinants of choice. Moreover, the behavioural choice model generates probabilities of alternative behaviours which makes it ideally suited for integration with simulation models.