Figure legends
Figure 1 . Scheme of the experimental design. (1)Dilution . Fresh field soil was kept undiluted or diluted 1 × 103 and 1 × 106 times using sterilized soil to create high, moderate and low soil biodiversity inocula, respectively. (2) Incubation . Each soil bag was sealed using a sterilized cotton plug to avoid aerial microbial contamination while permitting gas exchange, and then incubated at room temperature in the dark until similar microbial abundance was observed among dilution treatments. (3) Soil inoculation and transplanting . After incubation, 200 g of soil inoculum was homogenized with 6800 g of sterilized soil and sand mixture (1:1) in each microcosm. Each microcosm received 24 seedlings from 1, 4, 8 or 12 plant species depending on plant diversity treatment. (4) Simulating environmental variation in precipitation . Three wet-dry cycles were implemented by maintaining gravimetric soil moisture of 12-18% during the wetting periods and by watering until most plants started to wilt during the drying periods. At the end of each period, plant shoots were cut at 5 cm above the soil surface to determine shoot biomass production.
Figure 2. The effect of the dilution-to-extinction approach on soil microbial diversity, biomass (indicated by soil respiration rate after incubation) and community composition. (A) The dilution-to-extinction approach successfully reduced soil microbial diversity (solid line) but did not alter the soil respiration rate (dash line). The dilution-to-extinction approach dramatically reduced fungal (B) and bacterial (C) diversity and altered the community composition of soil microbes based on amplicon sequence variant (ASV) detection. Error bar represents standard deviation (n = 3).
Figure 3. Soil biodiversity and plant diversity independently influence stability-related indices. (A - E) The effects of plant species richness (PS) and soil biodiversity on the temporal stability, mean and standard deviation of community biomass production, and plant species asynchrony and population variance. (F - J) The effects of plant functional richness and soil biodiversity loss on these stability-related indices. Effects of treatments on temporal stability were partitioned into effects on plant species asynchrony and population variance, or effects on temporal mean and standard deviation of community biomass production. Light bands represent 95% confidence intervals. Note: moderate and low soil biodiversity effects refer to the differences to the high soil biodiversity treatment. ***P< 0.001; **P < 0.01; *P < 0.05; ns P > 0.05.
Figure 4.Relationships between multitrophic biodiversity and (A) the temporal stability of community biomass production, (B) plant species asynchrony, (D) population variance, (C) the temporal mean and (E) standard deviation of community biomass production. Effects of treatments on temporal stability were partitioned into effects on plant species asynchrony and population variance, or effects on temporal mean and standard deviation of community biomass production. Light bands represent 95% confidence intervals.
Figure 5 . Shoot biomass of grasses, herbs and legumes at each harvest when microcosms were inoculated with high, moderate or low soil biodiversity inoculum in the plant communities with 1, 4, 8 or 12 species.