Stressor‐induced biodiversity gradients: revisiting biodiversity–ecosystem functioning relationships

Biodiversity–ecosystem functioning experiments typically inspect functioning in randomly composed communities, representing broad gradients of taxonomic richness. We tested if the resulting evenness gradients and evenness–functioning relationships reflect those found in communities facing evenness loss caused by anthropogenic stressors. To this end, we exposed marine benthic diatom communities to a series of treatments with the herbicide atrazine, and analysed the relationship between the resulting gradients of evenness and ecosystem functioning (primary production, energy content and sediment stabilization). Atrazine exposure resulted in narrower evenness gradients and steeper evenness–functioning relations than produced by the design of random community assembly. The disproportionately large decrease in functioning following atrazine treatment was related to selective atrazine effects on the species that contributed most to the ecosystem functions considered. Our findings demonstrate that the sensitivity to stress and the contribution to ecosystem functioning at the species level should be both considered to understand biodiversity and ecosystem functioning under anthropogenic stress. Synthesis Biodiversity loss affects ecosystem functioning, yet biodiversity–ecosystem functioning relations have mainly been investigated using communities with random species loss. In nature however, species are lost according to their sensitivity to environmental stress. In the present study, biodiversity loss and biodiversity–ecosystem functioning relations in randomly composed diatom communities were compared to those induced by the pesticide atrazine. Stress exposure resulted in smaller biodiversity loss but steeper decrease in functioning than in randomly composed communities, due to selective atrazine effects on the best performing species. Therefore, species‐specific sensitivity and contribution to ecosystem functioning need to be considered to predict biodiversity and ecosystem functioning under anthropogenic stress.