Vincent de Leijster

135 Chronosequence analysis of economic performance of agroforestry coffee farms in Colombia 6 Federation of Coffee Growers of Colombia (mostly timber and fruit tree species). The value of non-marketed timber and carbon increased over time (Table 6-3). This result is in line with studies that showed increases in biomass over time in agroforestry systems (Nijmeijer and Harmand, 2019; De Leijster et al. 2021). We found that the value of timber saturates after around 20 y (Table 6-3), which is similar to the local harvest cycle recommendations to cut C. aliadora after 18 y to optimize financial returns (FNC - Cenficafé, 2010). The non-marketed value of timber and fruit increases much faster and saturates at a higher value than the actual revenues (Table 6-3). This suggests that there are additional potential gains in gross revenues from co-products, which may improve actual net revenues faster if brought to market. We demonstrated that tree spatial arrangement in agroforestry farms affects economic performance. Agroforestry farms with trees dispersed over the farm had lower coffee gross revenues than monoculture farms, while this was not the case if trees were planted in alley formation or living fences (Figure 6-3). On the contrary, dispersed agroforestry systems had higher actual and potential gross revenues of tree fruits and timber, than agroforestry farms with alleys and living fences. Therefore, income diversification in these farms is stronger and they could benefit more when non-marketed products will be marketed. In an earlier study, we found that dispersed agroforestry farms were also related to a higher supply of ecosystem services, such as carbon stock, erosion control and butterfly diversity maintenance (De Leijster et al. 2021). Living fence and alley agroforestry farms had very similar costs and benefits, suggesting, that in this case the most environmentally friendly option should be preferred. In that earlier study, we found that after planting, alley agroforestry farms more rapidly stored carbon, reduced erosion, improved epiphyte richness and coffee quality than living fence farms (De Leijster et al. 2021). To conclude, first dispersed agroforestry farms should be preferred to optimize economic and environmental outcomes then alley agroforestry and then living fence agroforestry. Economic performance was more strongly related to altitude, farms size, pest control and intermediary choice than to the development of canopy characteristics (Table 6.5). These results are in line with previous findings where management intensity, altitude, and farm size were found to be related to economic performance in coffee farms in Costa Rica and Peru (Cerda et al., 2017; Jezeer et al., 2018). Nevertheless, our finding that choice of intermediary was important has been less frequently reported as an important factor influencing farm-level net revenues (Borrella et al., 2015; Higuchi, 2012). ‘Association-selling farms’ had higher coffee gross revenues, produced better quality coffee and had more income from co-products than farm that sold to other intermediaries (appendix Figure A6-6