A humped-back relationship between species richness and community biomass has frequently been observed in plant communities, at both local and regional scales, although often improperly called a productivity–diversity relationship. Explanations for this relationship have emphasized the role of competitive exclusion, probably because at the time when the relationship was first examined, competition was considered to be the significant biotic filter structuring plant communities. However, over the last 15 years there has been a renewed interest in facilitation and this research has shown a clear link between the role of facilitation in structuring communities and both community biomass and the severity of the environment. Although facilitation may enlarge the realized niche of species and increase community richness in stressful environments, there has only been one previous attempt to revisit the humped-back model of species richness and to include facilitative processes. However, to date, no model has explored whether biotic interactions can potentially shape both sides of the humped-back model for species richness commonly detected in plant communities. Here, we propose a revision of Grime's original model that incorporates a new understanding of the role of facilitative interactions in plant communities. In this revised model, facilitation promotes diversity at medium to high environmental severity levels, by expanding the realized niche of stress-intolerant competitive species into harsh physical conditions. However, when environmental conditions become extremely severe the positive effects of the benefactors wane (as supported by recent research on facilitative interactions in extremely severe environments) and diversity is reduced. Conversely, with decreasing stress along the biomass gradient, facilitation decreases because stress-intolerant species become able to exist away from the canopy of the stress-tolerant species (as proposed by facilitation theory). At the same time competition increases for stress-tolerant species, reducing diversity in the most benign conditions (as proposed by models of competition theory). In this way our inclusion of facilitation into the classic model of plant species diversity and community biomass generates a more powerful and richer predictive framework for understanding the role of plant interactions in changing diversity. We then use our revised model to explain both the observed discrepancies between natural patterns of species richness and community biomass and the results of experimental studies of the impact of biodiversity on the productivity of herbaceous communities. It is clear that explicit consideration of concurrent changes in stress-tolerant and competitive species enhances our capacity to explain and interpret patterns in plant community diversity with respect to environmental severity< Réduire