Plant density determines species richness along an experimental fertility gradient

A number of authors have suggested that, within areas a few square meters to many square kilometers in size, species diversity appears to peak at moderate levels of productivity, and this pattern is currently unexplained. Among the best examples of this pattern have been description of vegetation in which species richness declines as soil fertility increases. We tested two hypotheses that have been proposed to explain this pattern. The interspecific competitive exclusion hypothesis proposes that dominant species suppress the growth of competitively subordinate species and exclude subordinate species as fertility rises. In contrast, the assemblage-level thinning hypothesis proposes that individuals of all species tend to become larger as fertility rises, and individuals of all species tend to exclude subordinate individuals of each species. Because total density declines, samples of finite numbers of individuals will result in fewer species by chance alone. To test these hypotheses, we established an experimental productivity gradient in a first-year old field using four levels of slow-release NPK fertilizer (0, 8, 16, and 32 g N/m(2)). At the end of the growing season, we sampled aboveground biomass and numbers of stems for each species in 72 20 x 20 cm subplots (18 reps x 4 levels), with an average sample size of 260 individual stems per plot. We observed an 80% decline in stem density with increasing fertility, and a 50% decline in species richness along this fertility gradient. A simulation of random thinning along a fertility gradient showed a nearly identical decline in species richness, supporting the assemblage-level thinning hypothesis. We also found that responses of individual species to the soil fertility gradient showed virtually no support for interspecific competitive exclusion. The overwhelming influence of density found in this study suggests that plant species richness along many productivity gradients may be strongly influenced by total stem density, and that differences in competitive ability among species, although generally important, are not necessary to create dramatic changes in species richness along fertility gradients.