Competition for light in heterogeneous canopies Application of MAESTRA to a coffee (Coffea arabica L.) agroforestry system

Abstract

In agroforestry systems (AFS), quantifying the competition for light is a prerequisite toward understanding the impact of shade trees on the productivity of the under-crop. Models for homogeneous canopies and shade/full-sun approaches do not address the intra-plot heterogeneity, typical of AFS. For the first time, MAESTRA, a 3D light absorption model, was fully parameterized in a heterogeneous 2-canopy layers AFS. We quantified competition for photosynthetic photon flux density (Q) between shade trees (Erythrina poepiggiana) and coffee (Coffea arabica), with a spatial resolution from the plant to the plot (2.7 ha) and a temporal resolution from half-hour to one full year. The predicted transmittance through the 2- canopy layers was verified against field measurements. The goodness of fit (R2 > 0.75, RRMSE < 26%) was comparable to the predictions from 10 other studies using 3D light models and mostly verified in onelayered systems (mean R2 = 0.89 and mean RRMSE = 17%). Maps of absorbed Q showed that despite their low density in the plot (5.2 trees ha−1), the tall Erythrina trees reduced Q available for the coffee layer by 14% annually. Annual pruning of the oldest unproductive coffee resprouts maintained a large horizontal heterogeneity in coffee LAI, with direct impact on the Q absorption map. This management practice had a strong impact on seasonal variations of absorbed Q by the coffee canopy. We proposed also a simple approach to estimate Q absorbed yearly by the coffee plants in AFS of variable tree density, requiring only few measurements in the field. An extrapolation indicated that the amount of Q absorbed by the coffee canopy would display a negative exponential relationship (k = −0.34) when increasing shade tree density (from nil to 29 trees ha−1 ). The estimated k was similar to the shade tree extinction coefficient of diffuseradiation measured with a plant canopy analyzer. We showed that the presence of shade trees tends to reverse the diurnal time course of the fraction of Qa when compared to a plantation in the open. Overall, MAESTRA proved to successfully unlock the question of intra-plot heterogeneity for light absorption and to provide defensible light budgets as a continuous and mapped covariable, a crucial step for many field experimentations.We found that although irradiance was reduced by 60% below crowns of shade trees, coffee light-use efficiency increased by 50%, leaving net primary productivity fairly stable across all shade levels. Variability of aboveground net primary productivity of coffee plants was caused primarily by the age of the plants and by intraspecific competition among them (drivers usually overlooked in the agroforestry literature) rather than by the presence of shade trees.