Almost all biodiversity-ecosystem function studies focus on species richness. However, a complete diversity concept includes species’ relative abundances, since high evenness across species indicates more compositional heterogeneity than does low evenness (high dominance). If we neglect evenness as a component of diversity, we may be missing BEF patterns linked to mechanisms that depend on relative abundance. Hill numbers are a family of diversity metrics vary how they weight species richness and species’ relative abundances. This weighting is accomplished by continuous exponent q, and Hill numbers include species richness (q = 0), exponentiated Shannon (q = 1) and inverse Simpson (q = 2). Here we determine which value of q produces the strongest diversity-ecosystem function relationship in two large-scale, observational datasets, one on wildflower pollination by bees and the second on carbon storage by tropical forests. We measured the predictive power of diversity for function in two ways: (1) the r2 of diversity as the sole predictor; (2) the partial r2 of diversity in a model that also included abundance (“abundance models”). Unless richness is more predictive than other Hill diversities in both model types, incorporating evenness to some degree is valuable for understanding the drivers of function.
Results/Conclusions
Analyses of pollination and tropical forest data produced different results. For the pollination data, the r2 for diversity as sole predictor was highest at q=0, meaning richness predicts seed set better than metrics that also account for relative abundance. However, when aggregate abundance was included, the r2 for diversity in abundance models was much lower and insensitive to q. This happened because diversity per se is not predictive; rather, abundance is strongly correlated with richness and this correlation disappears as q increases. For the forest data, the r2 for diversity by itself was again highest at q=0. However, the r2 for diversity in abundance models peaked at q=1.75, which roughly corresponds with Simpson diversity. One reason we found different results could be high among-site variation in abundance for bees, but not trees. This may be due to fundamental differences between these communities and functions. Stationary organisms store carbon at fairly consistent levels across time. Pollination is provided by mobile organisms whose abundances are highly sensitive to environmental conditions, and whose functional contributions have a high degree of stochasticity. We explore whether diversity is more likely to drive function when organisms are stationary and abundance is less variable across space.