In the context of climate change, the Mediterranean region will be more frequently prone to cli-mate hazards, such as drought and heat waves. In mature alley cropping systems, understory conditions might buffer adverse conditions encountered by cereals during drought events. However, it is still undecided whether facilitation processes will prevail or not on competition interactions.Rainfall manipulation experiments have been largely used in forests, grasslands or mono-crops, with two main types: continuous partial exclusion (e.g. fixed gutters covering a certain percentage of soil) or temporary total exclusion (e.g mobile shelters). In agroforestry systems, rain exclusion experiments are still rare. Experiments published so far only used continuous partial exclusion, with fixed gutters covering the soil below the crop1 or fixed gutters above the crop2. However these devices modify microclimate, and do not allow testing climate scenarios with extreme drought events.Two different prototypes of rainout shelter allowing temporary total exclusion were designed at the Restinclières Farm Estate, in the south of France. This poster will present the metho-dological framework of the experiment, the proposed design of rainout shelters and the first results concerning their relative efficiency (edge-effects, experimental artefacts, impacts on soil water availability). Pros and cons of both solutions will be discussed.