Vincent de Leijster

106 Chapter 5 . Figure 5-4. Tree arrangement classes and a) above-ground carbon stock F=4.3, df=41, P=0.01, b) potential soil loss F=4.3, df=41, P=0.01, and c) butterfly diversity F=4.6, df=16, P=0.02. Letters give the results of the Tukey test. 5.5 Discussion We analyzed how the transition from a coffee monoculture system to an agroforestry system affects ecosystem service supply and interactions among ecosystem services, and what were the biotic and abiotic factors influencing this supply and interactions. We found that planting agroforestry trees on coffee farms improved the supply of multiple ecosystem services over time, mostly following an asymptotic relationship with a period of steep increase (0-10 y), then a period of little increase (11-20 y), and finally a stabilization period after 20 years (Figure 5-2). The asymptotic shape best described the development of carbon stock, erosion control, habitat provisioning and pest control, while we found some support for a decreasing relationship for understory vegetation cover and coffee yields with time since implementation of agroforestry. We also found that interactions among ecosystem services during the first 10 years were different from those between 10-20 years, in particular only above-ground carbon and erosion control were consistently associated over the years. Canopy cover, and the farm’s slope, altitude and vegetation control management were the factors that best explained ecosystem services bundles, especially the bundle with above-ground carbon, erosion control and epiphyte richness. Finally, tree arrangement also influenced ecosystem service supply, with particularly high supply if trees are dispersed over the farm, followed by farms with trees in alley formation and farms with trees as living fences, and the lowest supply for monoculture farms. Coffee yields did not significantly differ between tree arrangement groups and monoculture farms.