Agroforestry is of growing interest to winegrowers as a way of coping with climate change, thanks to the buffering effect of trees on the microclimate. Conversely, olive growers in France are interested in diversifying their production, for example by growing vines for wine or table grapes. However, trees and vines compete for light, water and nitrogen resources, and these competitions change across time, as the trees and vines grow and colonize the soil. This raises questions about the dynamics of planting: is it better to plant the trees first, and wait for the tree to be large enough to create a beneficial microclimate to protect the young vines? Or should the two species be planted at the same time, so that their root plasticity enables them to create a spatial complementarity between their root systems? Or can the trees be planted in an existing vineyard (as many winegrowers wish to do, taking advantage of gaps in the vineyard to plant the trees) without the vine exerting too much competition on the young tree? In order to answer these questions, we ran a virtual experiment using the Hi-sAFe model (Dupraz et al. 2019), which has recently been adapted to simulate olive tree (Barbault, submitted) and includes the STICS model for soil-crop simulation, including grapevine (Valdés-Gómez et al. 2009). We compared three planting chronology modalities: tree first (TF), vine first (VF), or both at the same time (BST), and compared the results with vine-only (VC) and tree-only (TC) controls. The second species (olive tree in VF, grapevine in TF) was planted six years after the start of the simulation, as well as both species in the BST system. Simulations were run for 23 years. The olive trees were planted at a density of 222 trees per hectare (15 m between tree rows, 3 m between trees along the row), and 4 rows of grapevine were simulated between two tree lines. The soil and climate were those of a grapevine plot located in Restinclières Agroforestry Platform. The simulated plants were not fertilized nor irrigated. The results (e.g. figure 1) were surprising, as some of our hypotheses were not verified: we expected the yield of vines associated with olive trees to be lower than that of vines alone, due to competition for water, nitrogen and light. But the comparison of grapevine yield in VF vs VC showed that the competition by trees had a minimal impact on grapevine yield. On the contrary, and in accordance with our hypotheses, the grapevine benefited from the microclimate created by olive trees: in years with low grapevine yield, yield was higher in systems with olive trees (e.g. years 14 and 17 in VF, compared to VC), and all the more so when trees were old (compare year 3 of BST, with young trees, and of TF, with older trees). The simulated yield of olive trees was very low and showed an unexpected production pattern (early peak of production) even in the tree-only control. Despite these inaccuracies, the simulation results highlighted the fact that tree-crop interactions are indeed complex and make it difficult to get an intuitive grasp of the functioning of the system. The presence of grapevine seems to accentuate the irregularity of production of olive trees, as there are more no-yield years in TF than in TC. Regarding the chronology of planting, the results confirm our hypothesis that olive trees perform better when planted in existing vineyard (the cumulated yield in VF is higher than in BST between years 6 and 14), although this trend seems to reverse as trees get older due to several no-yield years in VF. This seems to be confirmed by the comparison between VF-BST vs TF (delayed peak of production but with a much higher value in VF and BST compared with TF), but these results must be taken with caution because they do not correspond to the same climatic years. Simulations could be remade on a set of generated climate scenarios to mask this climate effect. Further analysis of the simulation results will focus on understanding the mechanisms responsible for the complex tree-vine interactions patterns, in particular in terms of i) dynamics of soil colonization by roots (explaining the presence or absence of root profile complementarity), and ii) reduction in heat and water stress resulting from the modified microclimate. Even if some aspects of the simulations still require improvements, the model gives us access to variables that would be very difficult (or even impossible) to measure in the field, and enables us to test and compare several systems over the long term. The results of the simulation showed the good potential of olive-grapevine associations, which could help stabilize grapevine yield, under the condition that tree management is adapted to reduce the phenomenon of irregularity of production of olive trees.