Vincent de Leijster

85 Ecosystem services trajectories in coffee agroforestry in Colombia over 40 years 5 5.1 Introduction Ecosystem services link ecosystem processes directly or indirectly to people’s quality of life. Globally, the majority of ecosystem services are decreasing, however, the magnitude of changes in ecosystem service supply differs between ecosystem services and across locations (IPBES, 2019). Land management that aims to improve single ecosystem services has resulted in an unbalanced provisioning of ecosystem services where the supply of few services increased at the cost of others (Renard et al., 2015). Over the past decades, the worldwide intensification of agricultural practices has caused trade-offs between food provisioning and regulating, supporting and cultural ecosystem services (Foley et al., 2005; Locatelli et al., 2017). Therefore, rehabilitation of degrading ecosystem services to improve the quality of natural systems and of people’s quality of life is urgently needed. By studying the dynamics of multiple ecosystem services across time we can gain a better understanding of how ecosystem services vary over time, and options for their restoration (Bullock et al., 2011). The behavior of ecosystem service supply over time is called a trajectory, and specifically in case of increasing supply; rehabilitation . Ecosystem service trajectories as well as their mutual interactions vary across time as a function of factors such as management regimes, human use and local biotic and abiotic conditions (Fremier et al., 2013). Recent studies suggest that ecosystem services trajectories often follow non-linear paths, and that trade-offs among services may emerge over time (Bullock et al., 2011; Locatelli et al., 2017). Ecological trajectories are pathways of ecological conditions over time and may represent degradation, adaptation or restoration in response to changes in landmanagement or environmental conditions (Gann et al., 2019). The trajectory of ecosystem services may take linear, asymptotic and sigmoid (S-shape) trajectories (Bullock et al., 2011; Locatelli et al., 2017). For example, the development of carbon stock and wood volume in a recovering forest in Canada followed a sigmoid curve, while a provisioning service (berry producing plants) followed a U-shaped curve in these forests (Sutherland et al., 2016). These curves were related to negative feedbacks; for example, carbon stock would likely develop following an asymptotic curve because plant biomass production is limited by water, nutrients and temperature, which at a certain point will have a damping effect on plant biomass increase (Heimann and Reichstein, 2008; Turner, 2010). Generally ecosystem services behave non- linearly across time, although linear shapes can be observed when study periods are short (Rau et al., 2018). For several ecosystem services, e.g. erosion control, habitat provisioning and pest control, the type of curve that can be expected and how this development is affected by biophysical factors is still largely unknown (Bennett et al., 2009; Bullock et al., 2011). As