Vincent de Leijster

144 Chapter 7 were higher in agroforestry coffee farms than in monoculture coffee farms. More specifically, above-ground carbon, erosion control and biodiversity were higher in agroforestry farms. Thus, both cases confirm the hypothesis that agroecological management improves ecological performance. Although the results are based on only two case studies, they are in line with multiple larger-scale meta-analyses that have demonstrated that agroecological management improves ecological performance compared to conventional management, in for example Mediterranean fruit crops (Morugán-Coronado et al., 2020), agroforestry systems (De Beenhouwer et al., 2013; Torralba et al., 2016), and diversified farming systems (Rosa-Schleich et al., 2019). However, we also found that most agroecological practices promote a specific set of ecological processes. In the almond farms that we studied, no tillage and green manure were less effective in improving below-ground ecosystem services than compost application, while compost application was less effective in improving plant species richness. In a recent study in the same region in Spain, other researchers found that a combination of compost and vegetation cover can more comprehensively promote the physical, chemical and biological conditions of the soil than each of these separately (Luján Soto et al., 2021). Furthermore, in Chapter 5 we found that the spatial arrangement of trees in coffee agroforestry plantations plays an important role in the effectiveness with which agroforestry improves ecosystem service supply. The local biophysical and social context should therefore play an important role in selecting the agroecological practices, so that the correct ecological processes are targeted. This adaptability to the context has been recognized by the Food and Agriculture Organization of the United Nations (FAO) as one of the most important elements of agroecology (FAO, 2018). RQ 2: How do ecosystem service trajectories develop following agroecological interventions? This thesis looked into the development of ecosystem services following the implementation of agroecological interventions in the short (Chapter 3) and in the long term (Chapter 5). In the almond case study, we measured ecosystem services in greater detail, and there we concluded that nutrient cycling, habitat provisioning and above-ground carbon developed most rapidly (i.e. within one year). We did not find any strong developments in pollination, pest control, almond production or soil organic carbon: there were no differences in these ecosystem services between the agroecological treatment and the reference situation. This lack of effect may either be because land management change is not a driver for change in ecosystem service supply, or because changes in ecosystem service supply follow slow processes. If there is such a slow initial phase, a sigmoid shape (S-curve) may be expected, which represents a trajectory that is at first slow, then fast and then slow again. In Chapter 3