Vincent de Leijster

125 Chronosequence analysis of economic performance of agroforestry coffee farms in Colombia 6 standardized ‘processing labor time’ of 185 kg fresh coffee per labor day (SD=64, n=13). We calculated ‘processing labor costs’ by multiplying the standardized processing time by ‘fresh coffee yield’. Finally, we asked the farmers their ‘transportation costs’. For the 23 farmers that did not record these transportation costs, we used the average ‘transportation cost’ divided by ‘dry coffee yield’ of all other farms, this average value was then applied to their ‘dry coffee yield’. For ‘plantain, fruit and timber management’, we recorded information on (i) Musa (either bananas or plantains), fruit tree and timber tree density, (ii) cumulative labor time spent to harvest, prune, control pests and fertilize these crops and trees, and (iii) quantity of fertilizer products applied. No chemicals were used to control pests in plantains, and some farmers used plastic bags to protect Musa fruits, but costs of the bags were negligible according to these farmers. To calculate ‘plantain management costs’ and ‘(fruit) tree management costs’ we summed the costs of labor and of the agrochemicals applied. 6.2.4.3 Coffee net revenue, actual net revenue, and potential net revenue ‘Coffee net revenue’ was calculated by subtracting the costs of ‘pest control, ‘herb control’, ‘fertilization’, and ‘harvesting, processing and transporting’ from the ‘coffee gross revenue’ (Table 6-1). The ‘actual net revenue’ was the sum of ‘coffee net revenue’ and ‘plantain, tree fruit and timber gross revenue’ minus ‘costs of plantain and tree management. Finally, the ‘potential net revenue’ was calculated as the sum of ‘actual net revenues’ and ‘potential additional benefits from Musa fruits, tree fruits, timber and carbon’. 6.2.5 Data analysis 6.2.5.1 Development of economic performance and costs and (potential) benefits We examined how economic performance indicators and individual costs and benefits varied over ‘time since transition to agroforestry’ (Table 6-1). To do this, we excluded monoculture farms. For economic performance, we first tested whether the best fit was achieved with a polynomial regression, since we hypothesized that the development of economic performance would at first show a decrease followed by an increase, and then stabilization. We used a 2 and 3 folded polynomial regressionmodel with time since transition as explanatory variable and economic performance indicators as the response variable. The residuals of the models were evaluated for their ability to meet the assumptions of