Results

Average changes of ecosystem functions in plant mixtures

In field experiments, we observed significant increases in mixtures relative to monocultures for carbon processes including aboveground biomass (AGB, 64.6%), belowground biomass (BGB, 45.0%), total biomass (TB, 123.1%), soil carbon pool (SCP, 13.2%), soil respiration (Rs, 10.6%), heterotrophic respiration (Rh, 18.9%), microbial biomass (MB, 16.7%), fungal biomass (FB, 16.4%), and bacterial biomass (30.0%) (BB, Fig. 2). We observed similar increases for nitrogen processes including aboveground nitrogen pool (ANP, 26.5%), soil nitrogen pool (SNP, 7.8%), and soil ammonium nitrogen (SAN, 27.0%). Decreases occurred in soil nitrate nitrogen (SNN, -40.4%), soil nitrogen mineralization (SNM, -52.3%), and soil nitrogen leaching (SNL, -76.0%) (Fig. 2).
In greenhouse experiments, we similarly observed significant increases in mixtures relative to monocultures for measures including AGB (60.1%), BGB (53.1%), TB (49.6%), Rs (16.4%), and SNP (5.6%) (Fig. 2). The remaining attributes showed no significant differences.

The effects of plant diversity, experimental age, and climate

Carbon and nitrogen attributes increased linearly with experimental age in the field experiments, with the exceptions of FB and SNN (Fig. 3, Table S2), and increased logarithmically with species richness except for FB, SAN, SNN, and SNM (Fig. 4, Table S3). Importantly, significant interactions between plant diversity and experimental age were found for AGB, BGB, TB, SCP, Rh, MB, BB, ANP, SNP, SAN, and SNM (Table S4), indicating stronger species diversity effects on those variables in longer-term experiments (Fig. 4, Table S5). In greenhouse experiments, AGB, BGB, FB, BB, and ANP increased logarithmically with species richness (Fig. 5, Table S3). Few significant interactions between plant diversity and climate (mean annual temperature and precipitation) were observed, with only BGB, FB, and BB impacted (Fig. 6).

Predicted responses of soil carbon and nitrogen pool

We utilized the differences in species richness and experimental age effects in the field experiments to predict the impacts of long-term diversity declines (Fig. 7). A 10% decrease in species richness (from 100 to 90%) over one year reduced SCP and SNP by 0.5 and 1.1% respectively (Fig. 7). An 80% decrease in species richness (from 100 to 20%) over one year led to 2.3 and 4.8% reductions in SCP and SNP respectively (Fig. 7). The declines in SCP and SNP in response to the decrease in species richness became amplified with time. For example, a 10% decrease in species richness (from 100 to 90%) over five years led to a 2.4 and 5.2% reduction in SCP and SNP, respectively (Fig. 7).