Facilitation, an ecological interaction assembling plant communities worldwide, has been shown to modulate both species richness and ecosystem functions. Such a Biodiversity-Ecosystem Functioning (BEF) relationship can be decomposed into different components not only related to species losses and gains but also to the identity of the species and the context in which they live. Using an extension of the classical BEF approach named CAFE (Community Assembly and the Functioning of Ecosystems), we quantified the contribution of these components to the BEF relationship in a Spanish semiarid plant community shaped by facilitation. We used species richness as a measure for biodiversity and plant cover as a proxy of multiple ecosystem functions including plant productivity, soil protection, soil fertility and microbial productivity. Nurse plants doubled the number of species that live beneath them relative to open ground, but caused a five-fold increase in plant cover. The disproportionate increase of plant cover was a consequence of the identity of the species enhanced by nurse plants, which were more productive than the average. We discuss these results in terms of sampling effects (i.e., the higher probability of richer communities to harbour hyperproductive species) and complementary effects (i.e., richer communities enhancing productivity through resource partitioning, abiotic facilitation, or biotic feedbacks). The enhancement of ecosystem functions that plant facilitation produces by incorporating species with high functional values to the community may reverberate among other trophic levels and propagate beyond the local scale where the ecological interaction is produced.

We selected 25 plots under the nurse plant (Ononis tridentata) canopy and 25 adjacent gaps of the same size to estimate the cover of each plant. The size of the plots varied with the size of the nurse plant, ranging from 11435 to 21980 cm². The cover of facilitated plants was estimated by adding the number of cm that each plant species occupied along a variable number of parallel linear transects established under the canopy of the nurse plant. To estimate the contribution of different processes to the BEF relationship, we followed the CAFE approach described in Bannar-Martin et al (2017; Ecol. Lett. 21, 167-180).