Functional identity is more important than diversity

Summary 1. Experimental studies have provided significant knowledge of how biodiversity can influence ecosystem processes. However, there is a growing need to relate these findings to natural communities. 2. Here we identify two major hypotheses for how communities may influence ecosystem processes: the 'diversity hypothesis' (the diversity of organisms in a community influences ecosystem processes through mechanisms such as complementary resource use), and the 'mass ratio hypothesis' (ecosystem processes are determined overwhelmingly by the functional traits of the dominant species). We then test which of these two hypotheses best explain variation in ecosystem properties and processes (biomass pools and fluxes, water use, light interception) in a temperate native grassland. We do this by applying various measures of diversity, functional diversity, and functional identity, whose significant relations with ecosystem processes would support either of the competing hypotheses. 3. Mean trait values best explained variation in five of the eight ecosystem processes examined, supporting Grime's mass ratio hypothesis, which proposes that the functional identities of the dominant species largely determine ecosystem processes. 4. Multi-trait functional diversity indices also explained large amounts of variation in ecosystem processes, while only weak relationships were observed between species richness and ecosystem processes. 5. To explore the mechanistic interactions between variables, we developed structural equation models (SEMs), which indicated that many of the community diversity and trait properties signifi cantly influenced ecosystem processes, even after accounting for co-varying biotic/abiotic factors. 6. Synthesis. Our study is one of the first explicit comparisons of the 'diversity' and 'mass ratio' hypotheses, and our results most strongly support the mass ratio hypothesis, that is, the traits of the dominant species most influenced the ecosystem properties and processes examined. Our results suggest that the management of communities for the maintenance of ecosystem processes should focus on species dominance hierarchies.