Vincent de Leijster

100 Chapter 5 Figure5-2. Development of woody vegetationcharacteristics over time sinceagroforestry implementation. Green line is model prediction, grey dashed lines indicate 95% confidence intervals. a) Canopy closure, asymptotic nonlinear model (P < 0.0001, R 2 =0.18, n=54 ). b) Average canopy height in the plot, asymptotic nonlinear model (P=0.006, R 2 =0.12, n=54). c) Tree species richness, GLM Poisson (exponential) (P=0.009, R 2 =0.10, n=54) . Table 5-2. Descriptive statistics of farm characteristics. The ‘*’ indicates whether the values only refer to agroforestry farms, in other cases all farms are included. Mean ± SD Minimum Maximum Canopy cover (%)* 63 ± 19 19 95 Tree density (shrubs ha -1 )* 157 ± 130 25 525 Tree species richness (shrubs ha -1 )* 3.0 ± 1.7 1 9 Farm size (ha) 7.5 ± 12 0.9 93 Coffee density (plants ha -1 ) 5640 ± 970 3333 9000 Altitude (m) 1540 ± 160 1212 1905 Slope (%) 29 ± 9.7 6.9 47 Ecosystem service indicators We found a positive relationship between above-ground carbon, biodiversity indicators (butterflies and epiphytes), and a trend for coffee quality (P=0.06, R 2 =0.07, n=39), and time since agroforestry was implemented (Figure 5-3). These relationships followed a positive asymptotic model, with half-time coefficients lower than 5 y. We also found a positive sigmoid relationship between timber volume and time since agroforestry, and a linear increase between coffee quality and time since agroforestry. Herb cover and coffee yield showed a trend towards gradual decreases with time since agroforestry (P=0.08, R 2 =0.03, n=54). We find that the saturation coefficients of canopy characteristics, above-ground carbon, butterfly diversity, epiphyte abundance, coffee quality and timber volume coincide