Fig 1.
Experimental framework to study the effects of plant and microbial diversity interactions on ecosystem multifunctionality.
(a) Conceptual framework exploring the relative contribution of plant and microbial diversity and their combination, i.e., multitrophic diversity to multiple ecosystem functions; (b) locations of the 101 sites included in the global grassland survey; (c) full-factorial design of the microcosm study comprising four plant diversity levels of plant species and 3 microbial diversity levels obtained from a dilution-to-extinction approach. (Fig 1A and 1C were created with BioRender.com, while Fig 1B was created using open source QGIS software version 3.34).
Fig 2.
Links between ecosystem multifunctionality and plant and soil microbial richness in the grassland field survey.
(a) Significant correlations (Spearman; p ≤ 0.05) between the richness of groups of organisms and weighted multifunctionality and individual ecosystem functions. P-values showing Spearman correlations in Table E/1 in S1 Text. (b) Variation partitioning modeling was used to evaluate the unique and shared portions of variation in ecosystem properties explained by plant richness, microbial richness, environment, and aridity index. Plant richness | microbial richness shared refers to the percent of shared variation in ecosystem properties explained by plant and microbial richness. P-values associated with the unique portions explained by different groups of predictors are available in Table F/1 in S1 Text. The contribution of plant and microbial (bacteria and fungi) richness, environment (distance from the equator, plant cover, soil pH, % clay, soil C, and mean annual temperature) and aridity index to weighted multifunctionality and individual functions (glucose mineralization—soil respiration with glucose addition; soil inorganic pools—ammonium content, available P, nitrate content; plant productivity—net primary production). Microbial richness corresponds to a composite metric of their joint richness (standardized between 0 and 1) and environment properties correspond to a standardized principal component analysis first axis obtained from the multiple properties. R2 values express total variances corresponding to model adj. Plant pathogen only includes fungal pathogens. The data underlying this figure can be found in S1 Data.
Fig 3.
The contribution and relationship of plant and soil biodiversity to weighted multifunctionality in the microcosm experiment.
(a) Significant correlations (Spearman; p ≤ 0.05) between richness groups and weighted multifunctionality, services and individual belowground and aboveground functions. The different background colors are used to highlight which individual functions belong to services categories. Plant and microbes richness corresponds to a composite metric of their joint diversity (standardized between 0 and 1). Fungal phylotypes include mycorrhiza, saprotrophs, and plant pathogens fungi. P-values showing Spearman correlations in Table E/2 in S1 Text. (b) Variation partitioning modeling was used to evaluate the unique and shared portions of variation in ecosystem properties explained by plant richness, microbial richness, plant composition, and drought. Plant richness | microbial richness shared refers to the percent of shared variation in ecosystem properties explained by plant and microbial richness. P-values associated with the unique portions explained by different groups of predictors are available in Table F/2 in S1 Text. Plant pathogen only includes fungal pathogens. The data underlying this figure can be found in S1 Data.
Fig 4.
Direct and indirect drivers of ecosystem functions.
Piecewise structural equation models for (a) weighted multifunctionality derived from the global grassland survey (n = 101) and for (b) above- and belowground functions derived from the microcosm study (n = 157). We aimed to identify the direct relationship between plant and soil biodiversity (bacterial and fungal richness) and multiple ecosystem functions. Soil microbial richness was included as a composite variable, including information of soil bacterial and fungal taxa. Please note, the higher the aridity index the less arid environment it represents. Numbers adjacent to arrows are indicative of the effect size of the relationship. Dashed lines indicate that none of the variables included was significant (p-value >0.05). For each model, the proportion of variance explained (R2) and the various goodness-of-fit statistics are shown below the response variables. Significance levels are as follows: *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001. AIC, Akaike information criterion; EMF, ecosystem multifunctionality; MAT, mean annual temperature. Information about a priori models can be found Tables E/1 and E/2 in S1 Text. The data underlying this figure can be found in S1 Data.