Vincent de Leijster

145 Synthesis and conclusion 7 we concluded that the development of arthropod-mediated ecosystem services was too slow for our study period, or that our experiments were too small to affect arthropod population dynamics, because in other, similar studies agroecological interventions were found to affect arthropod populations in almond orchards (Eilers & Klein, 2009; Saunders et al., 2013). In Chapter 5 we measured ecosystem service supply in the long term, which provided us with valuable information that enabled us to analyze the trajectories of the development in specific ecosystem services. We found that the trajectories of some ecosystem services could be explained by non-linear models, which supports the findings of other studies (Bullock et al., 2011; Sutherland et al., 2016). For example, the increases in above-ground carbon, butterfly diversity and epiphyte abundance followed an asymptotic shape, which is characterized by a rapid development in the initial phase followed by a slower development until a saturation level is reached (Figure 4-2). Timber volume increased following a sigmoid shape. Negative effects also follow non-linear pathways, since coffee yield and understory vegetation cover decreased first rapidly then slowly. We found that most changes occurred within the first ten years after intervention since most half-time coefficients were lower than five years, and that most saturation levels were reached after approximately twenty years. However, the speed of change cannot be compared between the Mediterranean case study and the tropical case study as we already noticed. We can postulate that the supply of most ecosystem services related to the development of vegetation complexity, and that plant biomass growth is likely to be faster in tropical regions than in Mediterranean regions, as productivity rates are slower in Mediterranean areas due to water limitations (Foley, 1994). Even though speed may vary between geographical regions, we still expect the shapes of trajectories to respond relatively consistently to land management changes since similar non-linear trajectories of ecosystem services were found in other studies in a wide range of geographical locations (Bullock et al., 2011; Nyafwono et al., 2014; Sutherland et al., 2016). This consistency may be explained by the succession of vegetation, which has been found to followan asymptotic shape in systems inmany geographical locations (Lohbeck et al., 2012; Rawat & Johnsingh, 1998). In addition, the relationship between vegetation and other biophysical elements such as vertebrates (Figure 1-1) may also follow consistent patterns via processes such as herbivore, frugivore, nectarivore, and granivore relationships or because the structural aspects of the vegetation provides habitat or opportunities for foraging or hunting other organisms (Rawat & Johnsingh, 1998). However, to better understand how ecosystem service trajectories behave in terms of shape and speed, more temporal analysis is needed in a wide range of locations that vary in biotic and abiotic conditions.