Interpreting the 'selection effect' of biodiversity on ecosystem function

Experimental ecosystems often function differently than expected under the null hypothesis that intra- and interspecific interactions are identical. Recent theory attributes this to the 'selection effect' (dominance by species with particular traits), and the 'complementarity effect' (niche differentiation and/or facilitative interactions). Using the Price Equation, I show that the 'selection effect' only partially reflects dominance by species with particular traits at the expense of other species, and therefore is only partially analogous to natural selection. I then derive a new, tripartite partition of the difference between observed and expected ecosystem function. The 'dominance effect' is analogous to natural selection. 'Trait-independent complementarity' occurs when species function better than expected, independent of their traits and not at the expense of other species. 'Trait-dependent complementarity' occurs when species with particular traits function better than expected, but not at the expense of other species. I illustrate the application of this new partition using experimental data.