More diverse plant communities have higher functioning over time due to turnover in complementary dominant species

Abstract

More diverse communities have been shown to have higher and more temporally stable ecosystem functioning than less diverse ones, suggesting they should also have a consistently higher level of functioning over time. Diverse communities could maintain consistently high function because the species driving function change over time (functional turnover) or because they are more likely to contain key species with temporally stable functioning. Across 7 y in a large biodiversity experiment, we show that more diverse plant communities had consistently higher productivity, that is, a higher level of functioning over time. We identify the mechanism for this as turnover in the species driving biomass production; this was substantial, and species that were rare in some years became dominant and drove function in other years. Such high turnover allowed functionally more diverse communities to maintain high biomass over time and was associated with higher levels of complementarity effects in these communities. In contrast, turnover in communities composed of functionally similar species did not promote high biomass production over time. Thus, turnover in species promotes consistently high ecosystem function when it sustains functionally complementary interactions between species. Our results strongly reinforce the argument for conservation of high biodiversity.

Fig. 1.

Effect of biodiversity on maintenance of biomass production over time. (A) Species-rich communities exceeded a threshold level of function in a larger number of years than species-poor communities did (GLM main effect of species richness, P < 0.001). (B) Communities with a larger number of functional groups present had higher functioning for more years than those with only one or two functional groups (GLM main effect of number of functional groups, P < 0.01). Error bars are always ±1 SE and are calculated by back-transforming SEs from GLMs; error bars can therefore exceed 7 (the maximum number of years).

Fig. 2.

Functional turnover in communities of different diversities. (A) FTO across 2 or 7 y in plots of different species richness. FTO was higher in more species-rich communities, when calculated across 2 y (ANOVA main effect of species richness, P = 0.02) and across 7 y (ANOVA main effect of species richness, P < 0.01). Error bars are ±1 SE and are calculated from linear models. We show FTO as 0 for communities with only one or two species because it cannot be calculated for monocultures and is not meaningful for two-species plots. (B) FTO across 7 y in communities of different evenness; communities with different numbers of species are shown with different symbols. FTO across 7 y was higher in more even plots (ANOVA main effect of evenness, P < 0.001); evenness remained significant after correcting for species richness.

Fig. 3.

The relationship of FTO with functioning over time and complementarity effects. (A) In plots with one or two functional groups, there was no significant relation between FTO and functioning over time (main effect from GLM, P = 0.38; from major axis regression, P = 0.15). (B) In plots with three or four functional groups, there was a significant positive relation between FTO and functioning over time (main effect from GLM, P < 0.01; major axis regression, P < 0.001). The fitted line is the prediction from the GLM with Poisson errors, back-transformed, and therefore a curve on the original scale. (C) There was no relationship between mean complementarity effects and FTO in plots with one or two functional groups (main effect from ANOVA, P = 0.27; major axis regression, P = 0.14). (D) In plots with three or four functional groups, there was a significant relationship between FTO and mean complementarity effects (main effect from ANOVA, P = 0.02; major axis regression, P < 0.001). In all plots, different numbers of species are shown with different symbols; we corrected for species-richness effects on FTO by also fitting sown diversity in the models.

Fig. 4.

Average percentage of plots in which each species contributed to function. Points show the average percentage of plots in which each species contributed to 50% of the biomass production out of the total number of plots into which it was sown. These are average percentage numbers of plots per year and are calculated across 7 y. Species are ranked according to the percentage of plots in which they contributed to function. The error bars show the 95% confidence intervals for the percentage number of plots in which each species would be expected to contribute to function by chance (Methods). Points outside this range denote species that contributed to function more or less frequently than expected by chance.