Microbial functional changes mark irreversible course of Tibetan grassland degradation

The Tibetan Plateau hosts the world’s largest high-altitude grasslands and contributes 2.5% to global soil organic carbon (SOC) stocks while covering only 0.3% of terrestrial area. Kobresia pastures dominate roughly one-fifth of the Plateau, with Kobresia pygmaea forming low grazing lawns underpinned by dense root mats that protect against trampling-induced erosion and enable rapid regrowth. In recent decades, pasture degradation has increased markedly and approximately 30% of Tibetan grasslands are considered degraded, leading to severe declines in SOC and nitrogen (N) storage. Three mechanisms are implicated—erosion, decreased carbon (C) and N input, and increased soil organic matter (SOM) mineralization—yet their relative importance remains unclear. To quantify SOC and N losses across Kobresia pygmaea’s core area, the authors conducted a meta-analysis of 594 observations from 49 studies (2002–2020). They complemented this with a detailed field study categorizing six successive degradation stages, from intact root mats (S0), through increasing surface cracking (S1–S4), to bare soil patches (S5). The study hypothesizes that Kobresia pastures are approaching a critical point where microbial functional changes cause substantial consequences for SOC and N storage. These abrupt shifts are expected to be characterized by coupled changes in SOM quality and quantity, feeding back to microbial community structure and functions that regulate C and N mineralization.

Pasture degradation on the Tibetan Plateau is widely reported, with about 30% of grasslands degraded and major declines in SOC and N storage. The literature identifies three contributing mechanisms: erosion, reduced C and N inputs, and accelerated SOM mineralization. To synthesize existing evidence for Kobresia pygmaea grasslands, the authors compiled 594 observations from 49 publications (2002–2020) that reported SOC, N, and/or bulk density across clearly defined degradation stages including non-degraded references and comparable depth intervals. The meta-analysis indicated average losses of 42% SOC and 33% N at severely degraded sites relative to intact pastures. These literature-derived patterns guided and corroborated the partitioning of SOC losses observed in the field study and provided a broader regional context for the microbial and biogeochemical shifts associated with degradation.

The study combined a meta-analysis with an intensive field investigation along a degradation sequence. Literature study: The authors searched Web of Science, ScienceDirect, Google Scholar, and CNKI using combinations of terms related to degradation gradients and Tibetan Plateau alpine meadows. Inclusion criteria required clear degradation stage classification, presence of a non-degraded reference, reported SOC, N and/or bulk density (BD), explicit sampling depths and locations, and 10 cm sampling intervals. Reported SOM was converted to SOC using a factor of 2.0. SOC and N stocks were computed as: stock = 100 × content (g kg−1) × BD (g cm−3) × depth (cm). Effect sizes were calculated as ES = (D − R)/R × 100%, where D is the degraded stage value and R the reference. Field study design and site: Six degradation stages (S0–S5) were selected within a 4 ha area near Nagqu, Tibet (∼4484 m a.s.l.; gentle 2–5% slopes), ensuring comparable environmental conditions, with four field replicates per stage. S0 denotes intact Kobresia root mats; S1–S4 denote increasing polygonal surface cracking; S5 is bare soil without root mats. The site’s intact topsoil (0–25 cm) is a Stagnic Eutric Cambisol (Humic) developed on loess over glacial sediments (50% sand, 33% silt, 17% clay), carbonate-free and near neutral pH (6.8). Total soil depth averages 35 cm. The area is used as winter pasture (Jan–Apr) and hosts plateau pikas. Sampling and measurements: Crack extent and vegetation cover were measured; penetration resistance was recorded at 1 cm increments. Soil pits (30 × 30 × 40 cm) were opened at each plot; horizons were classified and sampled down to 30 cm. Bulk density and root biomass were measured on undisturbed cores (10 cm height, 10 cm diameter). Roots were separated into live and dead; root biomass was quantified up to 25 cm (where >95% of roots occur). Additional soil samples were collected at fixed depths (0–5, 5–15, 15–35 cm) for microbial analyses. Chemical and isotopic analyses: SOC, total N, δ13C and δ15N were analyzed via elemental analyzer–isotope ratio mass spectrometry. Soil pH was measured in water (1:2.5 v/v). Lignin phenols (vanillyl, syringyl, cinnamyl; V+S+C) were quantified after CuO oxidation by GC–MS. Microbial community analyses: Total DNA was extracted (PowerSoil kit). Bacterial 16S rRNA (V3–V4) and fungal ITS regions were amplified using primers suitable for t-RFLP and Illumina MiSeq. t-RFLP used restriction enzymes MspI, BstUI, HaeIII. Illumina MiSeq (2×300 bp) reads were processed with fastp, PEAR, cutadapt, and VSEARCH; denoising with UNOISE3; chimera removal against SILVA and UNITE; amplicon sequence variants (ASVs) classified via BLAST (≥90% identity). Non-fungal ITS reads were filtered via blastn against nt. Rarefaction used the minimum sequencing depth (15,800 bacterial; 20,500 fungal reads). Fungal functional guilds were assigned with FUNGuild. Enzyme activities: Activities of β-glucosidase, xylanase, phenol oxidase, urease, and alkaline phosphatase were measured in situ following Schinner et al. protocols at specified incubation temperatures and durations; reaction products were quantified photometrically. SOC stocks and partitioning of losses: SOC stocks (kg C m−2, 0–30 cm) were calculated from SOC content, BD, and horizon thickness. SOC losses per stage were referenced to S0. Erosion-induced SOC losses were estimated from measured topsoil removal (crack depth/extent) using S0 SOC content and BD. Mineralization/decreased input losses were estimated by applying each degraded stage’s SOC content and BD to S0 horizon thicknesses, attributing declines in erosion-unaffected horizons to mineralization and reduced root C input. This partitioning relied on assumptions that (i) erosion losses originate mainly from topsoil removal, and (ii) SOC decreases in horizons not affected by erosion were driven by mineralization and reduced root C input. Statistics: Means ± SE are reported. One-way ANOVA tested effects of stage and depth (p<0.05), with Shapiro–Wilk and Levene tests for normality/homoscedasticity; LSD/Tukey HSD post hoc or Kruskal–Wallis with Bonferroni for non-normal data. Relationships were assessed by linear/nonlinear regressions (p<0.05). Outliers in microbial datasets were identified by Grubbs’ test and excluded. Community differences (t-RFLP, MiSeq) were tested by MANOVA (Bray–Curtis), with pairwiseAdonis for multilevel comparisons. NMDS visualized community patterns; environmental correlates were assessed by CCA.

• SOC and N losses: Across Tibetan Plateau Kobresia grasslands, the literature meta-analysis indicated average losses of 42% SOC and 33% N at severely degraded sites relative to intact pastures. At the field site (0–30 cm), SOC stocks declined by 7.5 kg C m−2 (≈45%) from S0 to S5, consistent with the meta-analysis.
• Partitioning of SOC loss: Of the total SOC decline, approximately two-thirds were due to topsoil erosion (S5: ~5.0 kg C m−2), and one-third due to decreased root C input combined with accelerated mineralization (≈2.5 kg C m−2).
• Topsoil and texture changes: At S5, about 81 kg m−2 of SOC- and N-rich topsoil had been eroded from the site. Clay content decreased by about 60% at S5 compared to S0, indicating preferential removal of fine particles; bulk density and texture profiles shifted accordingly. Penetration resistance and root density decreased markedly with degradation from S1 to S5.
• SOM quality and isotopes: SOC content decreased with degradation, especially in the upper 20 cm. δ13C of SOC decreased from S0 to S4, coinciding with relative enrichment of 13C-depleted complex compounds (lignin; V+S+C increased from S0 to S4), then δ13C increased from S4 to S5 as lignin content declined, indicating enhanced decomposition of recalcitrant SOM at the most degraded stage.
• Microbial community shifts: Bacterial and fungal communities changed significantly along the degradation sequence (p<0.05, MANOVA). Decomposers of low-molecular-weight compounds (e.g., Actinobacteria) declined with degradation, while lignin-degrading taxa (e.g., Rhizobiales; Agaricomycetes including brown- and white-rot fungi) increased, especially at later stages. Ascomycota decreased. Nitrifying (Nitrospirales, Nitrosomonadaceae) and denitrifying (Pseudomonadales) groups increased from S0 to S3, then declined towards S5, indicating early-stage acceleration of N transformations with subsequent decline as substrates were depleted.
• Mycorrhizal transitions: Arbuscular mycorrhizal fungi (Glomeromycota), partners of Kobresia, peaked at S2 then declined, replaced by ectomycorrhizal partners (Thelephoraceae, Inocybaceae) toward S4; at S5, with Kobresia largely gone, ectomycorrhizal partners declined and initial AMF associated with pioneer plants appeared.
• Enzymatic activities: Hydrolytic enzyme activities (e.g., β-glucosidase, xylanase) increased from S0 to S3 and then declined at S4; N- and P-mobilizing enzymes (urease, alkaline phosphatase) declined from S3 to S4. In contrast, oxidative phenol oxidase activity increased from S2 to S4, consistent with enhanced degradation of complex SOM at advanced stages.
• Environmental drivers of community change: NMDS/CCA indicated that early-stage community shifts (S0–S3) were strongly linked to SOM quality (C/N, δ13C, N, P, lignin phenols), whereas later-stage shifts (S3–S5) were primarily driven by abiotic soil factors, notably increasing pH and decreasing clay content, with stronger effects on fungi than bacteria.
• Irreversible threshold: A pronounced shift in microbial community structure and function occurred between S3 and S4, coinciding with exhaustion of easily accessible N, complete topsoil loss through mineralization and erosion, and establishment of pioneer plants, indicating a transition to a new ecosystem state with low potential for recovery.

The study addressed the uncertainty around the relative contributions of erosion versus biogeochemical processes to SOC losses in degraded Tibetan Kobresia grasslands and tested the hypothesis of a critical shift in microbial functioning. By integrating a regional meta-analysis with field partitioning of SOC loss, the authors showed that while erosion dominates SOC decline (about two-thirds), reduced root C inputs and enhanced mineralization contribute substantially (about one-third). The strong linkage between SOC stocks and root density underscores the dependence of SOC accumulation on belowground inputs and rhizodeposition. Early degradation stages exhibit priming-driven decomposition of SOM due to high C/N root litter inputs in an N-limited system, stimulating hydrolytic enzymes and increasing nitrifier and denitrifier abundances. This accelerates N mineralization, nitrification, and denitrification, promoting N losses via leaching and gaseous emissions and altering plant–microbe symbioses. As degradation progresses to S3–S4, easily accessible nutrient pools are depleted, microbial communities shift toward taxa and enzymes specialized in degrading complex SOM (e.g., lignin), and abiotic soil changes (texture loss, higher pH) increasingly govern community structure. The observed δ13C and lignin phenol dynamics corroborate a transition from relative accumulation to subsequent breakdown of recalcitrant compounds at advanced degradation. The pronounced community and functional shift between S3 and S4 indicates a threshold beyond which the ecosystem enters a new state characterized by loss of topsoil, altered microbial functioning, and reduced capacity for nutrient cycling and C sequestration. These changes have broader implications for regional water holding capacity, water quality, and climate processes (e.g., monsoon dynamics) due to altered energy, water, and C fluxes, as well as for local food security reliant on pastoral systems.

This study quantifies SOC and N losses across degraded Kobresia grasslands of the Tibetan Plateau and disentangles the dominant processes responsible: approximately two-thirds of SOC loss stems from erosion of the C- and nutrient-rich topsoil, while one-third results from decreased root C inputs and accelerated mineralization. It demonstrates a progressive and then abrupt shift in microbial community composition and function along the degradation sequence, with a threshold between stages S3 and S4 marked by depletion of labile nutrient pools, enhanced oxidative degradation of recalcitrant SOM, and altered plant–fungal symbioses. Crossing this threshold likely renders ecosystem recovery improbable due to profound biotic and abiotic alterations, explaining the apparent irreversibility of severe degradation. Management implications include lowering livestock densities, increasing livestock mobility (e.g., community-based seasonal grazing) to prevent reaching critical stages, and recognizing that conventional restoration (e.g., reseeding) often fails after topsoil loss without substantial inputs and may risk nutrient leaching. Future research should refine quantification of coupled C–N process rates across degradation gradients, assess early-warning indicators of the S3–S4 transition, and evaluate landscape-scale management strategies that maintain root mat integrity and prevent erosion.

The partitioning of SOC losses into erosion versus mineralization/decreased input is based on explicit assumptions: (i) erosion-derived SOC losses are primarily associated with removal of topsoil; and (ii) declines in SOC in erosion-unaffected horizons are driven by mineralization and reduced root C input. The field study was conducted within a single 4 ha site near Nagqu with four replicates per stage and focused on the upper 30 cm of soil; while representative of widespread patterns, these constraints may limit generalizability across heterogeneous landscapes and deeper profiles. Microbial community analyses relied on marker-gene amplicons and enzyme assays, which infer function from composition and potential activity rather than direct process rates; although supported by literature and meta-analysis, causality should be interpreted with these considerations.