Agent-based modelling of social processes in land use change: Which influence have socio-psychological factors on shifts in land use intensity?

Understanding land use change dynamics is crucial for sustainable transformation, as land-use intensity affects ecosystem services and socio-ecological resilience. While much modelling effort has focused on economic and biophysical drivers, the role of psychological and social factors in shaping land use trajectories remains underexplored. However, empirical evidence suggests that socio-psychological factors significantly influence land managers' decision-making alongside economic considerations. To address this gap, we present a novel, generic model for social processes that we incorporate into a large-scale agent-based modelling framework for land use change. Our approach combines agent-based modelling with social network analysis, using the Theory of Planned Behaviour to simulate land managers' decisions on land use intensification or extensification. We examine how attitudes, social influences, network characteristics, and demand-driven competition impact land use outcomes and ecosystem service provision. Using a global sensitivity analysis, we identify key drivers shaping land manager distribution across intensities. Our findings reveal that the demand for ecosystem services is the most influential factor for the abundance of high- and low-intensity land managers. However, once psychological and structural barriers - contributing to an overall inertia to adopting new behaviour - are removed, attitudes toward sustainable practices become the primary driver for low-intensity land use. Social influence significantly increases the prevalence of medium-intensity land use, particularly at the spatial border between high- and low-intensity managers. As adoption surpasses a critical mass, medium-intensity practices rapidly expand, while high-intensity practices decline. Social influence also drives spatial clustering of similar land-use intensities, reflecting homophily within land use communities where neighbouring managers adopt comparable strategies. These local clustering effects reinforce dominant practices, creating path-dependent transitions that are difficult to reverse. In contrast, distant social ties have minimal impact, emphasizing the importance of local network effects. We conclude that incorporating social processes into land use models leads to distinct behaviours, revealing threshold and lock-in dynamics. Our approach offers a generic method for enhancing land use models with social dynamics, providing a more holistic understanding of future trajectories and potential sustainability transitions in the land system.