Impact of climate change, soil water, agroforestry system design and management on cropand tree performance: a sensitivity analysis with the Hi-sAFe model

Conference abstract for oral presentation World Congress on Agroforestry (Kigali, Rwanda, 20-24 October 2025) & IALE 2025 (Bratislava, Slovakia, 2-5 September 2025)

Loup Petitjean¹ , Pierre Barbillon² , Christian Dupraz¹ , Marie Gosme^1
1ABSYS, Univ Montpellier, CIHEAM-IAMM, CIRAD, INRAE, Institut Agro, Montpellier, France
²MIA, AgroParisTech, INRAE, Université Paris Saclay, Paris, France

Mechanistic soil-crop models are necessary to predict the performance of agroforestry (AF) systems in a changing climate. However, they are rarely used as a decision support system for cropping system design by farmers or farm advisors, because they are quite difficult to parameterize for a given situation. Our objective was thus to perform a sensitivity analysis of an AF model, as a first step towards building a metamodel to support farmer’s decisions (AF or not AF, choice of plot, AF spatial design).We used the Hi-sAFe model to perform a sensitivity analysis of tree and crop yield in a Mediterranean cereal-walnut alley cropping system. The tested input variables were climate change (past, present, future), soil (water table
depth, stoniness), cropping system (wheat monocrop, maize monocrop, wheat-maize rotation), tree density, tree root pruning.The results show that stone content had a major effect on both tree and crop yield, so much that neither the crop nor the tree grew well in the 70% stone condition. Excluding the data from the highly stony soil, the relative importance of the other inputs depended in the output variable and varied with system age. For tree yield, tree density has the highest effect during the whole simulation (with highest yield for intermediate density), climate change was the second most important determinant of tree yield, while water table had an impact only for trees of intermediate ages. For crop yield, the main determinant was crop species (the yield of maize was higher than wheat), followed by period for wheat (lower yield in the past) and the presence of trees for maize (lower yield in agroforestry). The relative yield
of maize ranged from 0.82 to 0.94, was lowest for the intermediate tree density (100 trees/ha) in particular with 30% of stones. For the lowest and highest tree densities (50, 200 trees/ha), the relative yield of maize is predicted to increase in the future compared to the past and present climate. The relative yield of wheat ranged from 0.85 to 0.98; it was lowest for the past climate and then was lower for 100 trees/ha than for 50 and 200 trees/ha. Contrary to maize, the relative yield of wheat is predicted to decrease in the future compared to the present. In conclusion, some variables had a small effect on yield of both tree and crop, and can be neglected in the future decision support system: rotation (it can be simplified into presence of wheat), tree root pruning, water table depth (it only had a transient effect on tree yield), stone content when it is below a threshold that remains to be determined. Tree density had a non-linear impact and future studies are needed to better untangle the effects of distance between tree rows vs distance
between trees within the rows. Finally, climate change had a complex impact that varied according to the species (maize, wheat, walnut) and conditions, so results obtained in one climate might not be generalizable to other climates.

This work was supported by the DigitAF project (Grant Agreement N° 101059794).