Consistent Effects of Biodiversity on Ecosystem Functioning Under Varying Density and Evenness

Biodiversity experiments typically vary only species richness and composition, yet the generality of their results relies on consistent effects of these factors even under varying starting conditions of density and evenness. We tested this assumption in a factorial species richness x density x evenness experiment using a pool of 60 common grassland species divided into four functional groups (grasses, legumes, tall herbs and short herbs). Richness varied from 1, 2, 4, 8 to 16 species, total planting density was 1,000 or 2,000 seeds/m2, and species were sown in even or uneven proportions, where one functional group was made dominant. Aboveground plant biomass increased linearly with the logarithm of species richness in all density and evenness treatments during all three years of the experiment. This was due to a convergence of realized density and evenness within species richness levels, although functional groups which were initially made dominant retained their dominance. Between species richness levels, realized density increased, and realized evenness decreased with species richness. Thus, more individuals could coexist if they belonged to different species. Within species richness levels, higher biomass values were correlated with lower density, suggesting an underlying thinning process. However, communities with low realized evenness also had low biomass values; thus high biomass did not result from species dominance. So-called complementarity and selection effects were similar across density and evenness treatments, indicating that the mechanisms underpinning the biodiversity effects were not altered. Species richness was the dominant driver of aboveground biomass, irrespective of variations in total densities and species abundance distributions at the start of the experiment; rejecting the hypothesis that initial differences in species abundance distributions might lead to different “stable states” in community structure or biomass. Thus, results from previous biodiversity experiments that only manipulated species richness and composition should be quite robust and broadly generalizable.