Figure 1.
Hypothetical Interactions between Biodiversity and Resource Heterogeneity on Rates of an Ecosystem Function
No effect (null hypothesis): neither diversity nor heterogeneity affect functioning. Positive diversity effect: basic diversity-function relationship, with no effect of heterogeneity. Heterogeneity-moderated diversity function: the strength (steepness) of the positive effect of diversity on functioning increases with increasing habitat/resource heterogeneity.
Figure 2.
The Slope of the BDEF Relationship Increases with Increasing Resource Heterogeneity
In all graphs, the slope of the diversity-function relationship increases with increasing levels of resource heterogeneity. As heterogeneity measures were continuous variables, separation into three categories (1, 2, and 3, representing low, medium, and high heterogeneity, respectively) was necessarily arbitrary. There were an equal number of replicates per heterogeneity level, plus or minus 1 when the number of replicates could not be divided evenly by 3. Equations for the BDEF relationship at each level of heterogeneity are presented with each figure. Both diversity and the interaction between diversity and heterogeneity were significant predictors of the ecosystem function.
(A) Relationship between plant diversity (i.e., Shannon-index), soil heterogeneity (Factor 5 from PCA on the CV in soil chemical concentrations), and standing root biomass (dry weight g/m2).
(B) Relationship between parasitoid species richness, heterogeneity in host (Pseudodynerus sp.) abundance, and rates of parasitism. Heterogeneity in host abundance was calculated as the CV between the nine sampling points per plot. Parasitism rates are the proportion of host larvae parasitized (arcsine square root transformed).
(C) Relationship between pollinator species richness, heterogeneity in flower density, and pollination benefit (difference between percent fruit set in open and bagged control treatments). Flower heterogeneity was calculated as the CV in flower abundance between the three experimental coffee shrubs.
Figure 3.
Final Most Parsimonious Structural Equation Models Relating Biodiversity, Resource Heterogeneity, Species Composition, Resource Abundance, and Ecosystem Function
Functions shown are: (A) belowground plant biomass, (B) aboveground plant biomass, (C) parasitism rates, and (D) pollination. All paths in the original models (Figures S1–S4) were treated as optional, and were thus able to be removed during model simplification using maximum likelihood estimation. As a final model we selected the model with a subset of the parameters in the initial model that included at least one predictor of function (plant biomass, parasitism, or pollination, respectively), and had the lowest AIC score. Values adjacent to paths indicate unstandardized direct effects, with significance indicated by *, p < 0.05; **, p < 0.01; ***, p < 0.001. For each original and final model, Cmin, AIC, and BCC scores, along with unstandardized and standardized total effects for final models are presented in Text S4.