Connecting experimental biodiversity research to real-world grasslands

During the last decades, a number of biodiversity experiments have been established to study the effects of plant diversity on ecosystem functioning. This research has been accompanied by a controversial discussion how "diversity effects" in experimental communities are related to the role of biodiversity in "real-world" ecosystems. To explore potential relationships, we compared plant community characteristics of 12 semi-natural managed reference grasslands to selected 10-year old communities of a large grassland biodiversity experiment (Jena Experiment): two communities initially sown with 60 species (JE60), and two communities assembled naturally during succession from bare soil (JESuc). Compared to semi-natural grasslands (17.6 +/- 5.6), JE60 had higher species richness (27.5 +/- 0.8), while species richness in JESuc (15.2 +/- 0.5) was not different on subplots of 0.64m(2) size. Evenness and spatial heterogeneity were similar among grassland types, but biomass proportions of legumes and forbs were higher in JE60, while JESuc and semi-natural grasslands were dominated by grasses. Structural equation modelling applied to identify the drivers of biomass production in mixtures of the Jena Experiment with similar species richness (sown with 8, 16, and 60 species) and reference grasslands, showed no direct relationships between observed species richness and biomass production. In contrast, functional characteristics related to nitrogen acquisition and use were the most important variables explaining community biomass production. These functional characteristics were either driven by management intensity (fertilisation) in the "real world" reference grasslands or established by sowing in the experimental grasslands. Our results imply that species functional characteristics are key for a better understanding of the mechanisms underlying community assembly and ecosystem functioning and that the drivers of ecosystem functioning are not fundamentally different between experimental and "real-world" grasslands. Thus, experimental studies with designed extinction scenarios may help to predict the consequences of species loss for ecosystem functioning in "realworld" ecosystems.