Simulating Land-Use Change using an Agent-Based Land Transaction Model

In the densely populated cultural landscapes of Europe, the vast majority of all land is owned by private parties, be it farmers (the majority), nature organizations, property developers, or citizens. Therewith, the vast majority of all land-use change arises from land transactions between different owner types: successful farms expand at the expense of less successful farms, and meanwhile property developers, individual citizens, and nature organizations also actively purchase land. These land transactions are driven by specific properties of the land, by governmental policies, and by the (economic) motives of both buyers and sellers. Climate/global change can affect these drivers at various scales: at the local scale changes in hydrology can make certain land less or more desirable; at the global scale the agricultural markets will affect motives of farmers to buy or sell land; while at intermediate (e.g. provincial) scales property developers and nature conservationists may be encouraged or discouraged to purchase land. The cumulative result of all these transactions becomes manifest in changing land-use patterns, and consequent environmental responses. Within the project Climate Adaptation for Rural Areas an agent-based land-use model was developed that explores the future response of individual land users to climate change, within the context of wider global change (i.e. policy and market change). It simulates the exchange of land among farmers and between farmers and nature organizations and property developers, for a specific case study area in the east of the Netherlands. Results show that local impacts of climate change can result in a relative stagnation in the land market in waterlogged areas. Furthermore, the increase in dairying at the expense of arable cultivation - as has been observed in the area in the past - is slowing down as arable produce shows a favourable trend in the agricultural world market. Furthermore, budgets for nature managers are obviously an important driver for nature expansion, but without a strict zoning plan imposed by a government, it is difficult to achieve a continuous, defragmented nature area. Lastly, the model suggests that with time the trend in ever-increasing farm sizes is gradually levelling out. The decision rules that determine the behaviours of the individual agents in the model (selling land, buying land, or none of the two) are calibrated on historical census records, using multi-nominal logistic regression. Because estimating who will sell and who will buy can only be done with a limited certainty, our model reproduces the volatility / uncertainty in who will do what and when. This makes that each specific future scenario can have numerous realizations of reality. Our stakeholders (including, besides policy makers, also local farmers and nature organizations) indicate that this aspect of the model strongly contributes to its credibility. Nevertheless, within different scenarios certain (spatial) trends are distinguishable, so that the model is -besides credible - also useful for exploring future trends.