According to the stress gradient hypothesis, poor resource availability should favor facilitation interactions while high resource availability should lead to strong competition between two plants relying on the same resources. However, there are still some discrepancies about the shape of the relation between the stress level of the resource availability, and the nature and strength of the interaction between the interacting plants. Here, we present the results of a glasshouse experiment in which a cereal (wheat) and a legume (white lupin) are grown together or separately along a phosphorus gradient. As white lupin, by exuding carboxylates, protons and phosphatases, can increase P availability in soils containing sparingly available P, we make the hypothesis that the cereal could benefit from this increase as long as P availability is low enough. Moreover, by taking up the bioavailable P, wheat may maintain the level of soil P sufficiently low to stimulate the mechanisms responsible for the facilitation effect from white lupin. We thus investigated the relationships between the P supply level and the nature and strength of belowground interactions between wheat and white lupin, in order to determine the conditions leading to transgressive overyealding (i.e. when the amount a species yields when grown with another species, is higher compared to that when grown alone). Designing more performant agroecosystems is of particular interest in the global context of food insecurity, increasing cost/decreasing efficiency of fertilizers, and better using ecological processes that operate in natural ecosystems is worth an option.