Predicting richness effects on ecosystem function in natural communities: insights from high-elevation streams.

. Despite the increased complexity of experimental and theoretical studies on the biodiversity-ecosystem functioning (B-EF) relationship, a major challenge is to demonstrate whether the observed importance of biodiversity in controlled experimental systems also persists in nature. Due to their structural simplicity and their low levels of human impacts, extreme species-poor ecosystems may provide new insights into B-EF relationships in natural systems. We address this issue using shredder invertebrate communities and organic matter decomposition rates in 24 high-altitude (3200-3900 m) Neotropical streams as a study model. We first assessed the effects of stream characteristics and shredder diversity and abundance on organic matter decomposition rates in coarse- and fine-mesh bags. We found the interaction term shredder richness x shredder abundance had the most significant impact on decomposition rates in the field, although water discharge may also play a role locally. We also examined the relative contribution of the three most abundant shredders on decomposition rates by manipulating shredder richness and community composition in a field experiment. Transgressive overyielding was detected among the three shredder species, indicating complementary resource use and/or facilitation. By integrating survey and experimental data in surface response analyses we found that observed B-EF patterns fit those predicted by a linear model that described litter decomposition rates as a function of increasing shredder richness and the relative abundance of the most efficient shredders. Finally, the validity of our approach was tested in a broader context by using two independent but comparable data sets from 49 French and Swedish streams showing more complex shredder community structure. Results revealed that richness and identity effects on decomposition rates were lost with increasing shredder community complexity. Our approach of combining experimental and empirical data with modeling in species-poor ecosystems may serve as an impetus for new B-EF studies. If theory can explain B-EF in low-diversity ecosystems, it may also have credibility in more complex ones.