Résumé
Trees in agroforestry parklands significantly contribute to improving and adapting farming systems while providing ecosystem services. However there is limited information on crop productivity and environmental performance. How do crop growth in such heterogeneous agroforestry systems vary according to distance, crown size, and azimuthal direction ? We use a novel approach based on advanced image analysis derived from medical imaging combined with multispectral imagery (known as the "probabilistic atlas") to analyze the crop, tree and soil interactions. We analyzed the influence of 72 Faidherbia albida trees on 13 millet fields in 2021 and 2022. Using "Voronoi" diagrams to separate individual trees. The Normalized Difference Vegetation Index (NDVI), was used as an indicator to assess the tree effect on pearl millet crop. We observed that, at the early growth stages of millet, the effect of Faidherbia albida on NDVI was stronger near the crown and decreased with distance, reflecting the positive growth influence on millet near canopy. We also observed an effect of azimuthal direction on NDVI. Finally, we found that the effect of Faidherbia albida on NDVI was more significant for trees with a large crown size. These results are a step further in the characterization of spatial tree influence on crops in highly heterogeneous systems, with improvements for e.g. the evaluation of ecosystem services or precision agriculture