Landscape heterogeneity strengthens the relationship between β-diversity and ecosystem function.

Consensus has emerged in the literature that increased biodiversity enhances the capacity of ecosystems to perform multiple functions. However, most biodiversity/ecosystem function studies focus on a single ecosystem, or on landscapes of homogenous ecosystems. Here, we investigate how increased landscape-level environmental dissimilarity may affect the relationship between different metrics of diversity (α, β, or γ) and ecosystem function. We produced a suite of simulated landscapes, each of which contained four experimental outdoor aquatic mesocosms. Differences in temperature and nutrient conditions of the mesocosms allowed us to simulate landscapes containing a range of within-landscape environmental heterogeneities. We found that the variation in ecosystem functions was primarily controlled by environmental conditions, with diversity metrics accounting for a smaller (but significant) amount of variation in function. When landscapes were more homogeneous, α, β, and γ diversity was not associated with differences in primary production, and only γ was associated with changes in decomposition. In these homogeneous landscapes, differences in these two ecosystem functions were most strongly related to nutrient and temperature conditions in the ecosystems. However, as landscape-level environmental dissimilarity increased, the relationship between α, β, or γ and ecosystem functions strengthened, with β being a greater predictor of variation in decomposition at the highest levels of environmental dissimilarity than α or γ. We propose that when all ecosystems in a landscape have similar environmental conditions, species sorting is likely to generate a single community composition that is well suited to those environmental conditions, β is low, and the efficiency of diversity-ecosystem function couplings is similar across communities. Under this low β, the effect of abiotic conditions on ecosystem function will be most apparent. However, when environmental conditions vary among ecosystems, species sorting pressures are different among ecosystems, producing different communities among locations in a landscape. These conditions lead to stronger relationships between β and the magnitude of ecosystem functions. Our results illustrate that abiotic conditions and the homogeneity of communities influence ecosystem function expressed at the landscape scale.