Abstract Biodiversity loss may result in the decline in important ecosystem processes. The effect of biodiversity loss on ecosystem functioning is determined by the functional contribution of the species lost and the compensatory responses of the remaining species. It is unknown to what extent the strength of the compensatory response of the remaining species depends on resources availability. Here we evaluate how the primary production of an assemblage of salt marsh plants responds to a realistic sequence of species loss in an 7‐year experiment, with and without the addition of fertiliser. We found near‐full compensation of progressive species loss in gross community primary production by the extinction‐resistant species with fertiliser as long as one species (Triglochin maritima) remained. Without fertiliser, at least four species, including the particularly abundant species Plantago maritima, were needed to maintain gross community primary production. These results suggest that the magnitude of the compensation by extirpation‐resistant species for the decline in ecosystem processes associated with progressive biodiversity loss depends on the resource context, and that compensation after the loss of plant species can be accelerated by increasing resource availability. Ultimately, full compensation appears to be limited by the presence and abundance of species in the remaining community that possess traits that allow them to compensate for the species lost. Synthesis and applications: These findings suggests that the conclusions of a large body of biodiversity‐ecosystem experiments cannot be used for informing the management of natural systems because they do not simulate realistic extinction sequences and therefore cannot quantify the potential for compensation of ecosystem services in the real world.