ABSTRACT
Experiments that simulate nonrandom species loss from natural communities can offer a fundamentally different understanding of the impacts of species loss on ecosystem function and their underlying mechanisms compared to seeding experiments where species are randomly assembled from a local species pool. We examined the mechanisms underlying changes in primary productivity following experimental species loss scenarios in Mongolian grassland. The range of species loss scenarios was based on natural patterns of species abundance that reflect the species' contributions to ecosystem processes. We found a clear reduction in productivity due to species loss only when species were lost randomly. Grassland productivity was relatively robust following nonrandom species loss scenarios. Even in the context of density compensation, the decrease in dominant trait values for leaf height would explain the reduction in productivity with random species loss. In contrast, the maintenance of dominant trait values of key productivity traits such as leaf dry matter content and leaf height might contribute to the maintenance of productivity in response to nonrandom species loss. Our experiment demonstrated that the responses and mechanisms of primary productivity to species loss differ according to the scenarios of species loss in natural grassland communities. The effects of diversity on productivity might be weak in mature natural systems when species loss is nonrandom. Understanding the consequences of realistic species loss on ecosystem functioning based on field-based removal experiments will give insights into real conservation strategies in the face of global biodiversity change.
