Modelling Socio-Psychological Drivers of Land Management Intensity

TL;DR

This paper addresses the underrepresentation of socio-psychological factors in land-use models by introducing a Theory of Planned Behaviour–based behavioural extension for agent-based land-use models. The extension integrates environmental attitudes, descriptive social norms, and behavioural inertia into land managers’ decisions on land management intensity and is coupled to the CRAFTY framework to study emergent dynamics under stylised conditions. Key findings show that socio-psychological drivers can reshape intensity shares, landscape configuration, and ecosystem-service provisioning through nonlinear feedbacks, resulting in clustering, path dependence, and multiple stable regimes. The work provides a reusable, modular modelling component that is transferable across ABMs and can be empirically parameterised in future studies, offering insights for policy design and transformative land-management pathways.

Abstract

Land management intensity shapes ecosystem service provision, socio-ecological resilience and is central to sustainable transformation. Yet most land use models emphasise economic and biophysical drivers, while socio-psychological factors influencing land managers' decisions remain underrepresented despite increasing evidence that they shape land management choices. To address this gap, we develop a generic behavioural extension for agent-based land use models, guided by the Theory of Planned Behaviour as an overarching conceptual framework. The extension integrates environmental attitudes, descriptive social norms and behavioural inertia into land managers' decisions on land management intensity. To demonstrate applicability, the extension is coupled to an existing land use modelling framework and explored in stylised settings to isolate behavioural mechanisms. Results show that socio-psychological drivers can significantly alter land management intensity shares, landscape configuration, and ecosystem service provision. Nonlinear feedbacks between these drivers, spatial resource heterogeneity, and ecosystem service demand lead to emergent dynamics that are sometimes counter-intuitive and can diverge from the agent-level decision rules. Increasing the influence of social norms generates spatial clustering and higher landscape connectivity, while feedbacks between behavioural factors can lead to path dependence, lock-in effects, and the emergence of multiple stable regimes with sharp transitions. The proposed framework demonstrates how even low levels of behavioural diversity and social interactions can reshape system-level land use outcomes and provides a reusable modelling component for incorporating socio-psychological processes into land use simulations. The approach can be integrated into other agent-based land use models and parameterised empirically in future work.

Figures (17)

• Figure 1: Visualisation of the different behavioural factors included in the model extension.
• Figure 2: Simplified architecture of the behavioural model extension integrated into CRAFTY. The behavioural weights include the importance of social norms, the inertia coefficient, and the upper limit of the giving-in threshold. Only those components of CRAFTY that are relevant for the behavioural extension are shown. A full sequence diagram of CRAFTY and a detailed flow diagram of the behavioural extension are provided in the Appendix (Figures \ref{['fig:flow_diagramm']} and \ref{['fig:ext_CRAFTY_embedded']}).
• Figure 3: Capital distribution across the modelled landscape. Lighter colours indicate higher capital levels.
• Figure 4: First-order (S1), second-order (S2), and total-effect (ST) sensitivity indices from a global sensitivity analysis of intensity shares, varying behavioural parameters and demand levels (Table \ref{['tab:sobol_ranges']}). According to the SA, intensity shares are independent of initial land use distribution, steepness of the logistic function $k$ and network characteristics (number of teleconnections $N_{tele}$ and neighbourhood radius $S_{nb}$). The upper limit of the giving-in threshold is set to $L=1$ for all cells. Node and edge sizes are proportional to the corresponding Sobol sensitivity indices, with larger circles or wider lines indicating greater influence on the outcome.
• Figure 5: First-order (S1), second-order (S2), and total-effect (ST) sensitivity indices from a global sensitivity analysis of total supply of non-material and material ES, varying behavioural parameters and demand levels (Table \ref{['tab:sobol_ranges']}). According to the sensitivity analysis, total ES supply is independent of the initial land use distribution, the steepness of the logistic function $k$, and network characteristics (number of teleconnections $N_{tele}$ and neighbourhood radius $S_{nb}$). The upper limit of the giving-in threshold is set to $L=1$ for all cells. Node and edge sizes are proportional to the corresponding Sobol sensitivity indices, with larger circles or wider lines indicating greater influence on the outcome.