Land-use changes have dramatically transformed many tropical landscapes from forest-dominated to agricultural landscapes. Agricultural land-uses, such as rubber and oil palm plantations, increase economic benefit at the cost of reduced ecological functions. Our study area is located in Jambi, Indonesia, where land-use decisions are mainly driven by smallholder farmers, creating a spatio-temporal landscape mosaic. Our research aims to investigate the spatio-temporal interactions, feedbacks and trade-offs between economic and ecological functions of the landscape mosaic. Based on field data from the EFForTS project (www.uni-goettingen.de/efforts), we developed the agent-based model EFForTS-ABM. EFForTS-ABM serves as an integrated, exploratory tool to analyze how tropical land-use change affects ecosystem functions. The economic module simulates smallholder land-use management decisions based on a profit maximization assumption. Each household determines factor inputs for all household fields and decides about land-use change based on available wealth. Heterogeneous farming efficiencies and learning dynamics incorporate the heterogeneity of households in knowledge and experience. To understand the importance of heterogeneity and learning we contrast realistic assumptions with model scenarios where all households perform optimally. The ecological module currently includes a simple account of carbon sequestration in above- and below-ground vegetation, which changes due to land-use decisions and growth and ageing of plantations.
For different output price and farming efficiency scenarios, we demonstrate model capabilities with results on household consumption and carbon sequestration. Land-use change towards the more profitable crop was mainly driven by the output prices for rubber and palm oil. However, heterogeneity of farming efficiency created inertia of land-use change and asynchronous behavior of smallholder households. In contrast, optimal farming scenarios with all households fixed to 100 % efficiency, showed less inertia and households reacted more or less immediately to changing prices. In general, model scenarios with higher inertia resulted in more stable landscape mosaics and carbon stock dynamics. The overall model results proved that the interactions between economic and ecological functions were not trivially straightforward, underlining the need of exploratory tools like EFForTS-ABM, which help to gain an understanding of such complex dynamics. Future research will focus more deeply on the effects of the spatial configuration of the landscape mosaic. Additionally, further ecosystem functions such as plant and bird biodiversity will be incorporated into the EFForTS-ABM model.