Extending the results of biodiversity - ecosystem function experiments to regional species collections

Rapid loss of global biodiversity has motivated a number of recent experiments that seek to understand how species richness affects ecosystem functioning. In the first round of studies, researchers have formed local communities differing in richness by selecting species at random from an experimental species pool, which is a subset of the 'regional' collection of species. Often, the effects of local community diversity on ecological processes (e.g., nutrient cycling or primary production) are used to draw more general conclusions about how extinction from the regional pool might influence these and other ecosystem functions. Here, we investigate the extent to which such extrapolations are warranted. We use a simple numerical model to make the intuitive but important point that whenever the experimental species pool is a non-random sample of the regional pool, the results of the typical biodiversity experiment do not lead to valid conclusions about the broader SRF relationship for the regional collection of species. As experimental pools are seldom a random sample of species (most studies focus on taxa that are abundant, commercially available, or that have biological traits of particular interest), more work is needed to assess whether conclusions to date represent generalities. We go on to outline the limited set of circumstances under which current experimental designs can draw broader conclusions safely. Most promisingly, we show that if the experimental pool can be considered a random subset of the regional collection, re-sampling techniques enable inference about the broader SRF relationship on the basis of a single study.