Vincent de Leijster

87 Ecosystem services trajectories in coffee agroforestry in Colombia over 40 years 5 neutral ways. Synergistic interactions are those where ecosystem services respond similarly to drivers that change their supply; all ecosystem services that show positive correlations form a bundle (Spake et al., 2017). Trade-off interactions are those where ecosystem services respond in opposite directions to a driver. Finally, no-interactions occur when ecosystem services do not respond in the same way to a given driver. In coffee agroforestry, there is empirical evidence of trade-offs (Campanha et al., 2005; Farfán V., 2014) or no-interactions (Cerda et al., 2017; Clough et al., 2011; Jezeer et al., 2019) between biodiversity and coffee yield. However, most of these studies have not yet addressed how these interactions behave over time. Temporal pathways of changes in ecosystem service supply can be investigated by using either trajectory analysis, i.e., repeated measures in time, or space-for-time substitution, i.e., spatial locations replace time by selecting locations with different age- since-intervention (Falk et al., 2006). Space-for-time substitution is less time consuming than trajectory analysis, which is an advantage. The space-for-time approach can be used to quantify general developments over longer time periods, however, for investigating subtle changes in time trajectory analysis is preferred (Pickett, 1989; Rácz et al., 2013), as different environmental conditions across locations may cause a higher level of noise in the dataset. Space-for-time substitution approach has been successfully applied to investigate ecosystem service recovery in forests (Sutherland et al., 2016), river restoration projects (Gilvear et al., 2013), and carbon storage develops in cocoa plantations (Nijmeijer and Harmand, 2019). Here we investigate the development of ecosystem services over time in coffee agroforestry systems in Colombia. More specifically we examine the trajectories and interactions of multiple ecosystem services in coffee farms that transitioned from monoculture to agroforestry coffee using a chronosequence analysis. We use a space-for- time substitution to reconstruct the chronosequence of agroforestry coffee farms, with 1-40 years since the planting of shade trees. We analyze six ecosystem services: coffee production (coffee yield, coffee quality), timber stock (standing volume), carbon stock (above-ground carbon and below-ground carbon), habitat provisioning (butterfly diversity and epiphyte richness), erosion control (potential soil loss, soil stability, litter cover, and herb cover), and pest control (coffee berry borer and leaf cutter ant incidence). First, we examine how these ecosystem services develop over time, in terms of trend (positive or negative response), shape (asymptotic or other shape) and speed of the response. Second, we investigate which ecosystem service bundles and trade-offs develop over time. Finally, we assess which biotic and abiotic factors (biotic: vegetation structure characteristics; abiotic: input management, farm characteristics and soil characteristics) explain the development of ecosystem service