Vincent de Leijster

104 Chapter 5 Table 5-4. Multiple regression model outcomes on the principal components (PC1, PC2, PC3; appendix Table A5-7 and Figure. A5-5). The first column presents which of the ecosystem services (ES) the PC resembles (VegCov= understory vegetation cover, Cqual= coffee quality, Soilst=soil stability, CBBc= coffee berry borer control). The model’s Akaike information criterion is AIC, the Δ AIC is the increase in AIC when the related dependent variable is excluded from the model, R 2 is the adjusted R 2 (· is P < 0.10, * is P < 0.05 , ** is P < 0.01). ES Factors coefficient ± SD Δ AIC P-value R 2 PC1: AGC+Erosion+Epiphyte Canopy cover 0.74 ± 0.10*** 42 < 0.001 0.68 Slope -0.57 ± 0.12*** 18 Vegetation control -0.16 ± 0.10 1 PC2: VegCov–Litter Altitude 0.25 ± 0.09* 3 0.003 0.18 Vegetation control 0.40 ± 0.14** 3 Farm area -0.40 ± 0.21• 1 PC3: CBBc–Soilst Nitrogen content soil -0.07 ± 0.25 3 0.16 0.03 Fertilization management 0.47 ± 0.24• 7 Coffee productivity Soil K content 2250 ± 700 ** 8 0.003 0.32 Last pruning coffee 55 ± 21 * 6 Pest control index 2450 ± 1180 * 3 Altitude 2.7 ± 1.5• 2 Coffee yield Altitude 0.32 ± 0.12* 5 0.04 0.16 Weed control 135 ± 87• 2 Last pruning coffee 3.4 ± 2.0 1 5.4.1 Factors explaining ecosystem service supply We identified three main principal components that represent groups of ecosystem services of farms that vary in time since agroforestry was implemented (60% of the variance explained; appendix Table A5-7 and Figure. A5-5). The first principal component (PC1) included information of above-ground carbon, litter cover, erosion control and epiphyte richness, we refer to this axis as ‘PC1:AGC+Erosion+Epiphyte’. Independent of PC1, the second PC axis included information on understory vegetation cover and was inversely related to litter cover, we refer to this axis as ‘PC2:VegC–Litter’. The third PC axis included information on coffee berry borer control and inversely related to soil stability, we refer to this axis as ‘PC3:CBBc+Soilst’. We found that canopy cover was most strongly positively related to PC1:AGC+Erosion+Epiphytes, while the slope of the farm was negatively related to PC1 (Table 5-4). Both altitude of the farm and understory vegetation control management were positively related to PC2:VegC–Litter, while the area of the farm was negatively related to PC2. The third PC axis could not be explained by the abiotic and biotic factors that were measured (P=0.16, R 2 =0.03). Coffee productivity, as measured in the plot, was best explained