An agent-based approach to model farmers' land use cover change intentions

Land Use and Cover Change (LUCC) occurs as a consequence of both natural and human activities, causing impacts on biophysical and agricultural resources. In enlarged urban regions, the major changes are those that occur from agriculture to urban uses. Urban uses compete with rural ones due among others, to population growth and housing demand. This competition and the rapid nature of change can lead to fragmented and scattered land use development generating new challenges, for example, concerning food security, soil and biodiversity preservation, among others. Landowners play a key role in LUCC. In peri-urban contexts, three interrelated key actors are pre-eminent in LUCC complex process: 1) investors or developers, who are waiting to take advantage of urban development to obtain the highest profit margin. They rely on population growth, housing demand and spatial planning strategies; 2) farmers, who are affected by urban development and intend to capitalise on their investment, or farmers who own property for amenity and lifestyle values; 3) and at a broader scale, land use planners/ decision-makers. Farmers’ participation in the real estate market as buyers, sellers or developers and in the land renting market has major implications for LUCC because they have the capacity for financial investment and to control future agricultural land use. Several studies have analysed farmer decision-making processes in peri-urban regions. These studies identified agricultural areas as the most vulnerable to changes, and where farmers are presented with the choice of maintaining their agricultural activities and maximising the production potential of their crops or selling their farmland to land investors. Also, some evaluate the behavioural response of peri-urban farmers to urban development, and income from agricultural production, agritourism, and off-farm employment. Uncertainty about future land profits is a major motivator for decisions to transform farmland into urban development. Thus, LUCC occurs when the value of expected urban development rents exceeds the value of agricultural ones. Some studies have considered two main approaches in analysing farmer decisions: how drivers influence farmer’s decisions; and how their decisions influence LUCC. To analyse farmers’ decisions is to acknowledge the present and future trends and their potential spatial impacts. Simulation models, using cellular automata (CA), artificial neural networks (ANN) or agent-based systems (ABM) are commonly used. This PhD research aims to propose a model to understand the agricultural land-use change in a peri-urban context. We seek to understand how human drivers (e.g., demographic, economic, planning) and biophysical drivers can affect farmer’s intentions regarding the future agricultural land and model those intentions. This study presents an exploratory analysis aimed at understanding the complex dynamics of LUCC based on farmers’ intentions when they are faced with four scenarios with the time horizon of 2025: the A0 scenario – based on current demographic, social and economic trends and investigating what happens if conditions are maintained (BAU); the A1 scenario – based on a regional food security; the A2 scenario – based on climate change; and the B0 scenario – based on farming under urban pressure, and investigating what happens if people start to move to rural areas. These scenarios were selected because of the early urbanisation of the study area, as a consequence of economic, social and demographic development; and because of the interest in preserving and maintaining agriculture as an essential resource. Also, Torres Vedras represents one of the leading suppliers of agricultural goods (mainly fresh fruits, vegetables, and wine) in Portugal. To model LUCC a CA-Markov, an ANN-multilayer perceptron, and an ABM approach were applied. Our results suggest that significant LUCC will occur depending on farmers’ intentions in different scenarios. The highlights are: (1) the highest growth in permanently irrigated land in the A1 scenario; (2) the most significant drop in non-irrigated arable land, and the highest growth in the forest and semi-natural areas in the A2 scenario; and (3) the greatest urban growth was recognised in the B0 scenario. To verify if the fitting simulations performed well, statistical analysis to measure agreement and quantity-allocation disagreements and a participatory workshop with local stakeholders to validate the achieved results were applied. These outcomes could provide decision-makers with the capacity to observe different possible futures in ‘what if’ scenarios, allowing them to anticipate future uncertainties, and consequently allowing them the possibility to choose the more desirable future