Experimental impacts of grazing on grassland biodiversity and function are explained by aridity

The study addresses how long-term domestic herbivore grazing impacts grassland biodiversity and ecosystem multifunctionality across gradients of aridity. The authors identify key uncertainties: most prior experiments are single-site, many analyses rely on meta-analyses and unstandardized methods, effects are often evaluated over short durations despite potential ecological time lags, and the relative roles of above- versus below-ground biodiversity in supporting multifunctionality under grazing remain unclear. Using a coordinated, replicated, regional experiment, they aim to determine whether grazing effects are general or aridity-dependent, quantify impacts on biodiversity and multifunctionality (EMF), and test whether long-term grazing alters the relative contributions of plant (above-ground) and soil microbial (below-ground) diversity to multifunctionality. They hypothesize stronger negative impacts of grazing on biodiversity and EMF at more arid sites due to lower productivity and nutrient uptake, and predict that soil biodiversity will become more important for multifunctionality in long-term grazed ecosystems where plant communities are more directly altered by grazers.

Previous research has shown that grazing is a dominant land use affecting biodiversity and ecosystem functions in grasslands, but robust experimental evidence across climatic gradients is limited. Many studies are single-site or rely on meta-analyses with heterogeneous methods, and frequently examine short-term grazing, leaving potential time-lagged ecosystem responses underexplored. While biodiversity’s role in supporting multiple ecosystem functions is well established, recent work highlights that both above-ground plant diversity and below-ground microbial diversity can drive multifunctionality, with their relative importance varying by environmental context. However, it is unclear how long-term livestock grazing modulates these relationships, especially across aridity gradients. The present study fills these gaps with a multi-site, long-term experimental approach.

Study region and design: Ten experimental sites across northern China (111.23 E to 123.51 E, 41.25 N to 49.52 N) spanning meadow steppe, typical steppe, and desert steppe along an aridity gradient (aridity 0.438–0.746; MAP 232–435 mm; MAT −2 to 5.4 °C). All sites have experienced decades of grazing (mainly sheep and goats, also cattle and horses). Each site contained paired areas: long-term livestock exclosures (>10 years) and adjacent grazed grasslands, enabling comparisons of grazed versus ungrazed conditions. Sampling: At each site, two 50 × 50 m areas (grazed and exclosed) were established. Within each, five 1 × 1 m plots were placed (four corners and center), yielding five replicates per treatment per site. Sampling occurred in late July–August 2020 at peak biomass. Above-ground biomass was clipped, dried (65 °C, 48 h), weighed, and ground for plant community N and P analyses. Belowground biomass (roots to 30 cm depth) was collected via 7 cm cores, washed, dried (65 °C, 48 h), and weighed. Soil sampling: five 2.5 cm diameter cores to 10 cm depth per plot were combined, sieved (2 mm), and split: one air-dried for soil organic C (K2Cr2O7 titration) and one frozen (−18 °C) for microbial analyses and microbial biomass C and N (chloroform fumigation-extraction), and soil NH4+ and NO3− (Alliance Flow Analyzer). Soil available N was the sum of NH4+ and NO3−. Soil biodiversity characterization: DNA was extracted (PowerSoil kit). Bacteria: 16S rRNA V3–V4 (338F/806R). Fungi: ITS1 (ITS-1F/ITS-2R). Protists: 18S rRNA V4 (TAReuk454F/TAReukREV3R). PCR conditions followed standard cycles; amplicons were purified, pooled equimolarly, and sequenced on Illumina MiSeq (paired-end). Bioinformatics: Quality filtering (Trimmomatic), merging (FLASH), OTU clustering at 97% (UPARSE), chimera removal (UCHIME), singleton removal. Taxonomic assignment: SILVA 138 (16S), UNITE v8.0 (ITS), PR2 4.5 (18S). Non-protist 18S OTUs (Fungi, Streptophyta, Metazoa) were filtered out. Functional guilds for fungi (saprotrophs, mutualists, pathogens) were inferred using FungalTraits; pathogen control was represented as inverse abundance of potential fungal plant pathogens. Biodiversity indices: Species richness (counts) and diversity (exp of Shannon’s entropy) were calculated for plants, bacteria, fungi, and protists. Below-ground diversity combined standardized (min–max) microbial diversities (bacteria, fungi, protists). Multidiversity combined standardized above- and below-ground biodiversity. Ecosystem functions and multifunctionality: Eleven functions were measured as proxies for productivity, nutrient cycling, and nutrient pools: above-ground biomass, below-ground biomass, plant community N, plant community P, soil organic C, soil available N, microbial biomass C, microbial biomass N, decomposers, pathogen control, and mycorrhizal mutualism. Each function was min–max standardized to 0–1. Multifunctionality (EMF) was quantified via: (1) average EMF (mean of standardized functions); (2) weighted EMF (down-weighting highly correlated functions, r > 0.7); and (3) multi-threshold EMF (counts of functions exceeding 25%, 50%, and 75% thresholds). Weighted and average EMF were highly correlated (r^2 = 0.937, P < 0.001). Statistical analyses: Linear mixed-effects models (nlme::lme) tested effects of grazing, grassland type, and their interaction on multidiversity, above-ground diversity, below-ground diversity, and EMF, with plots nested within sites as random effects. Grazing effects along aridity were further assessed via regressions using log response ratios (LRR = log[grazed/ungrazed]) for biodiversity and EMF across sites. Additional LMMs tested whether exclosure duration affected grazing responses (it did not). Relative contributions of plant and soil microbial diversity to multifunctionality were assessed separately for grazed and ungrazed conditions using LMMs with above- and below-ground diversity as fixed effects and plots/sites nested within grassland types as random effects; analyses were also done for single functions and combinations. Climate controls (MAT, MAP) were incorporated using piecewise SEM (piecewiseSEM::psem). Community composition differences were evaluated using NMDS (Bray–Curtis) and PERMANOVA with post hoc pairwise tests (Bonferroni correction). All analyses used R 4.1.0.

• Aridity modulates long-term grazing impacts: Grazing reduced biodiversity and ecosystem multifunctionality (EMF) in more arid desert steppes, but had no detectable effects in less arid meadow steppes; effects in typical steppes were intermediate.
• Increasing aridity strengthened negative grazing effects: Relationships between aridity and grazing effects were significant for multifunctionality (Pearson adjusted r^2 = 0.069, p = 0.036) and multidiversity (adjusted r^2 = 0.086, p = 0.022), indicating more negative impacts with higher aridity.
• Component biodiversity responses: Long-term grazing did not significantly influence plant (above-ground) diversity in the higher aridity steppes, whereas below-ground (soil) biodiversity was especially sensitive and greatly decreased in the most arid (desert) steppes.
• Shift in biodiversity drivers of multifunctionality: In ungrazed grasslands, plant diversity was positively associated with multifunctionality; in long-term grazed grasslands, soil biodiversity became the primary positive driver of multifunctionality. This pattern held for average EMF, multi-threshold EMF, and weighted EMF, and for most individual functions and function combinations.
• Mechanistic insights: Grazing increased the relative abundance of annual, acquisitive plant species, which have limited capacity to enhance functions such as biomass growth and storage under dry conditions, potentially weakening plant diversity’s contribution to multifunctionality. In harsher, nutrient-limited conditions, soil biodiversity’s role in regulating ecosystem functions predominated. Soil fungal and protist diversity were particularly important contributors to multifunctionality.
• Robustness: Results were consistent after accounting for MAT and MAP via SEM, and across different multifunctionality metrics (average, weighted, multi-threshold). Differences in exclosure years (>10 years at all sites) did not affect responses.

The findings support the hypothesis that long-term grazing exerts more negative effects on biodiversity and multifunctionality under higher aridity. This aridity dependence explains divergent outcomes reported in prior, context-specific studies and emphasizes climate–management interactions. Importantly, decades of grazing shifted the primary biodiversity drivers of multifunctionality: from plant diversity in ungrazed systems to soil biodiversity in grazed systems. This shift is likely mediated by grazing-induced changes in plant community composition (e.g., more annual, acquisitive species) and by harsher soil conditions under grazing that elevate the role of microbial diversity in sustaining ecosystem functions. The results underscore that conserving soil biodiversity is crucial for maintaining multifunctionality in grazed grasslands, particularly in arid regions. Management implications include the need to limit grazing pressure (e.g., adjust stocking rates or duration) in increasingly arid areas to prevent biodiversity losses and functional declines, especially under projected warming and drying climates.

This large-scale, multi-site field experiment demonstrates that the long-term impacts of livestock grazing on grassland biodiversity and ecosystem multifunctionality are contingent on aridity, with more arid sites experiencing stronger negative effects. It also shows that long-term grazing shifts the key biodiversity drivers of multifunctionality from plants to soil biota. These insights provide an empirical basis for climate-aware grazing management, highlighting the necessity of conserving soil biodiversity to safeguard multifunctionality in widespread grazed grasslands. Future research should extend to longer timescales, broader spatial coverage, and delve into subsoil functional responses to refine generality and mechanisms.

• Geographic scope is limited to northern China grasslands; extrapolation to other regions and biomes should be done cautiously.
• While exclosures exceeded 10 years, even longer-term dynamics remain to be evaluated; authors call for studies over longer timescales and larger spatial scales.
• Subsoil functions were not directly assessed; future work should examine long-term grazing-induced changes in subsoil processes.
• All sites experienced overgrazing historically; responses under lower-intensity or rotational grazing regimes were not directly tested.