Agricultural areas represent around 40% of the earth surface and provide a variety of products and services essential to human societies. However, with policy reforms, market liberalisation and climate change issues, continuous land use and cover change (LUCC) brings uncertainty in the quantity and quality of ecosystem services supplied for the future generations. The processes of LUCC have been explored using top-down approaches at global and regional level but more recent methods have focused on agents’ interactions at smaller scale. This approach is better suited to understanding and modelling complex socio-ecological systems, which emerge from individual actions, and therefore for developing tools which improve policy effectiveness. In recent years, there has also been increasing interest in gaining more detailed understanding of the impacts of LUCC on the range of ecosystem services associated with different landscapes and farming practices. The objectives of this thesis are: 1/ to understand and model the internal processes of LUCC at local scale, i.e. farmer behaviour, 2/ to explore heterogeneous farmer decision making and the impacts it has on LUCC and on ecosystem services and 3/ to inform policy makers for improving the effectiveness of land-related policies. This thesis presents an agent-based modelling framework which integrates psycho-social models of heterogeneous farmer decisions and an ecological model of skylark breeding population. The model is applied to the Lunan, a small Scottish arable catchment, and is empirically-grounded using social surveys, i.e. phone interviews and choice-based conjoint experiments. Based on ecological attitudes and farming goals, three main types of farmer agents were generated: profit-oriented, multifunctionalist, traditionalist. The proportion of farmer types found within the survey was used to scale-up respondent results to the agent population, spatially distributed within a GIS-based representation of the catchment. Under three socio-economic scenarios, based on the IPCC-SRES framework, the three types of farmers maximise an utility function, which is disaggregated into economic, environmental and social preferences, and apply the farm strategy (i.e. land uses, management style, agri-environmental measures) that best satisfies them. Each type of agents demonstrates different reactions to market and policy pressures though farmers seem to be constrained by lack of financial opportunities and are therefore unable to fully comply with environmental and social goals. At the landscape level, the impacts on ecosystem services, in particular the skylark local population, depend strongly on policy objectives, which can be antagonist and create trade-offs in the provision of different services, and on farmer socio-environmental values. A set of policy recommendations is offered that encompasses the heterogeneity of farmer decision-making with the aim of meeting sustainable targets. Finally, further improvements of the conceptual and methodological framework are discussed.