Different grazing intensities affect soil nitrogen cycling by altering microbial nitrogen metabolism in alpine wetlands Authors: Shishi Feng, Qianmin Liu, Yuanyuan Chen, Di Kang, Shuzhen Zou Published: 2026 ABSTRACT SummaryGrazing significantly affects soil nitrogen cycling in eastern Qinghai-Tibet Plateau alpine wetlands. Grazing did not alter soil microbial α-diversity, but shifted community composition via metagenomic analysis. Moderate and heavy grazing reduced soil total and active nitrogen contents by 53.8%–92.0% vs. light grazing, significantly decreased abundances of nitrification genes (amoA and hao) and ammonium assimilation gene (glnA), while increased dissimilatory nitrite reduction to ammonium gene (nirB) by 142.1%. A nitrification bottleneck from impaired nitrification drove active nitrogen decline, and structural equation modeling identified nitrogen cycle gene abundance as the key driver. This study reveals microbial nitrogen cycling mechanisms and provides a scientific basis for sustainable grazing management in alpine wetlands. Introduction Wetlands play a crucial role in delivering key global ecosystem functions and services.1,2 In recent years, disturbances resulting from human activities, especially livestock grazing, have become a primary cause of wetland ecological degradation. Alpine wetlands in particular are susceptible to grazing disturbance due to their unique environmental characteristics. Alpine wetlands are characterized by widespread permafrost, low annual mean temperature, and a short growing season for vegetation.3 Such unique environmental conditions contribute to slow soil nitrogen transformation,4 limited microbial activity,5 and weak ecosystem buffering capacity.6 Consequently, their resistance and resilience to external disturbances are significantly lower than those of other wetland types. Affected by grazing, the area of alpine wetlands has diminished and their ecological functions have deteriorated,7,8 which in turn directly or indirectly alters biodiversity and food web composition.9,10,11 It also affects aboveground-belowground biological interactions, as well as the characteristics and functions of soil microbial communities, ultimately driving changes in soil nitrogen cycling processes. Soil nitrogen cycling is a core process in alpine wetland ecosystems.12 Grazing can weaken the soil’s capacity for nitrogen retention and transformation.13,14,15 Nitrogen limitation, in turn, threatens wetland health and biodiversity, inhibits soil microbial activity, and reduces nitrogen cycling efficiency.15,16 However, there is currently a lack of systematic understanding regarding the mechanisms by which grazing intensity regulates soil functions and microbial communities, particularly at the level of microbial functional genes and nitrogen cycling processes. Existing studies consistently show that soil microbial community composition, diversity, and assembly are highly sensitive to grazing intensity.17 Changes in microbial community structure and function act as central biological regulators of soil nitrogen cycling, a function widely acknowledged in the research field.18,19,20 Consistent with previous studies, grazing indirectly regulates key soil nitrogen transformation processes by modifying microbial community structure,21,22,23 generally suppressing nitrification while promoting denitrification.24 Conversely, grazing exclusion or reduced grazing intensity favors the accumulation of soil ammonium nitrogen (NH4+) and decreases nitrate nitrogen (NO3−) levels, further influencing greenhouse gas emissions.25 This phenomenon is closely linked to shifts in the structure and function of the wetland soil microbiome.26,27 Meanwhile, soil nitrogen enrichment in wetlands can also regulate microbial functions,28 influencing ammonium oxidation rates and nitrogen transformation efficiency by altering the abundances of functional groups such as ammonia-oxidizing archaea (AOA) and ammonia-oxidizing bacteria (AOB).29 More importantly, the soil nitrogen cycle exhibits a typical characteristic of functional redundancy: multiple microbial groups can jointly mediate the same nitrogen transformation process, and the same microbial group can also participate in multiple nitrogen metabolic pathways.30 Among these, ammonium assimilation is the most widespread core metabolic pathway and displays the highest functional gene abundance.31 Denitrification and dissimilatory nitrate reduction to ammonium (DNRA) are highly coupled at the molecular level,32 and functional genes including nar, nir, and nos collectively determine the reduction fate and transformation pathways of soil NO3−.33 Collectively, microbial nitrogen metabolic functions such as ammonification, denitrification, and nitrogen fixation serve not only as key indicators for deciphering the mechanisms of the soil nitrogen cycle, but also as core links connecting environmental disturbances, microbial communities, and ecosystem nitrogen balance.34 Although previous studies have established the key role of microbial communities in regulating nitrogen cycling, they have mostly focused on the observed correlations between microbial diversity or community composition and soil nitrogen content and fluxes, with insufficient exploration of the underlying causal mechanisms and regulatory pathways at the functional gene level. In recent years, with the advancement of molecular biology techniques, changes in the abundance of nitrogen functional genes have been confirmed to be closely related to the nitrogen turnover potential of soil microorganisms35 and can be used to determine the dominant active nitrogen cycling pathways in soil.36 These studies are crucial, as they initially confirm that changes in soil nitrogen processes induced by grazing activities are indeed associated with microbial functional genes. However, significant gaps remain in current studies: previous explorations of the relationship between “microbial community structure, functional genes, and nitrogen cycling processes” remain ambiguous, and this ambiguity constitutes the core scientific question of this study. Specifically, most current studies remain at the level of simple correlation analysis between microbial community structure and soil nitrogen content, and lack systematic analysis of how grazing-driven changes in functional microbial genes progressively alter soil nitrogen metabolic pathways and further reshape the distribution of soil nitrogen components. A tripartite mechanistic framework of “grazing intensity, microbial functional genes, and nitrogen cycling processes” has not yet been established. Furthermore, under the combined effects of permafrost, low temperature, a short growing season, and other environmental factors, soil nitrogen mineralization, transport, and availability exhibit significant nonlinear responses, which may further amplify the disturbance effect of grazing on the nitrogen cycle. However, this complex mechanism has not yet been fully elucidated. In particular, the regulatory role of microbial community succession in nitrogen cycling under freeze-thaw cycles remains poorly understood. These research gaps have left the core scientific question of “the mechanism underlying microbial-mediated nitrogen cycling processes in alpine wetlands under grazing disturbance” unresolved. They also provide a clear starting point for proposing the hypothesis of this study. The Qinghai-Tibet Plateau serves as both a vital livestock production base for China and the foundation for the survival of local herders. It functions as a crucial ecological security barrier for China and is also an ecologically sensitive region in the context of global climate change. This holds significant importance for safeguarding water resources, biodiversity, and climate stability not only in China but throughout Asia.37 Given its dual core values of “ecology and production,” clarifying the specific impacts of grazing activities on soil microorganisms can inform the development of sustainable grazing and land management policies. Ultimately, this provides critical support for preserving the ecological integrity of alpine wetlands and ensuring their long-term economic vitality. In recent years, metagenomics technology has been widely applied in environmental microbiology research. This technology enables high-throughput sequencing of total DNA from environmental samples without relying on microbial cultivation, while simultaneously resolving both the species composition and functional potential of microbial communities. Therefore, to gain a deeper understanding of the mechanisms of soil nitrogen cycling in alpine wetlands, this study collected and analyzed surface soil samples from areas with different grazing intensities in southwestern alpine wetlands. By integrating soil nitrogen component measurements with metagenomic sequencing data of nitrogen metabolism functional genes, we analyzed the association between soil microbial nitrogen metabolism functional potential (based on functional genes) and soil nitrogen component concentrations. This approach enabled us to explore the response mechanisms of soil microbial communities under different grazing intensities in a more comprehensive and in-depth manner, and revealed the dynamic mechanisms of soil nitrogen-metabolizing microorganisms in alpine wetlands under grazing activities. These research gaps reveal two key uncertainties regarding nitrogen cycling mechanisms in alpine wetlands under grazing disturbance. First, it remains unclear whether changes in nitrogen cycling caused by grazing are driven mainly by alterations in soil physicochemical properties or by microbial functional regulation. Second, the causal relationship between shifts in microbial community structure and changes in nitrogen cycling function remains unverified under the environmental constraints of alpine wetlands. To address these uncertainties, this study proposes the following scientific hypotheses: (1) grazing intensity indirectly drives the dynamics of key processes in the soil nitrogen cycle by regulating the coupling relationship between soil physicochemical properties and microbial functions; (2) grazing intensity can alter key processes of soil nitrogen transformation by regulating the community composition of soil microbial genes involved in nitrogen cycling. If these hypotheses are verified, this study will clarify the impacts of different grazing intensities on soil microbial community composition and diversity, identify the key drivers of changes in soil microbial diversity and community structure in alpine wetlands, and reveal the underlying mechanisms of variations in soil nitrogen processes. This will provide theoretical support for the microbial ecological mechanisms of nitrogen cycling in alpine wetlands, and offer a scientific basis for the precise management and adaptive regulation of regional wetland ecosystems. Results Effects of grazing intensity on total nitrogen and nitrogen fractions Grazing intensity had a significant effect on soil nitrogen fractions. As grazing intensity increased from light grazing (LG) to heavy grazing (HG), the contents of soil total nitrogen (TN) and all nitrogen fractions showed a consistent overall decreasing trend (Figure 1A). The concentrations of TN, alkaline hydrolyzable nitrogen (AHN), amino acid nitrogen (AAN), acid hydrolyzable ammonium nitrogen (AN), ammonia nitrogen plus amino sugar nitrogen (NH3+ASN), and amino sugar nitrogen (ASN) were all significantly higher under the LG treatment than under the MG and HG treatments (p < 0.05). In terms of magnitude, nitrogen fractions decreased by 53.8%–70.4% from LG to MG, and further declined by 85%–92% from LG to HG. Compared with MG, most nitrogen fractions showed no significant difference under HG, except for AHN, which decreased significantly (p < 0.05), indicating that the effects of moderate grazing and HG on soil nitrogen fractions were relatively similar. Overall, LG maintained a relatively high soil nitrogen pool, whereas moderate grazing and HG resulted in substantial soil nitrogen loss. Non-metric multidimensional scaling (NMDS) analysis further demonstrated that the composition of soil nitrogen fractions differed significantly among grazing intensities (Figure 1B; R2 = 0.8890, p = 0.0060). The confidence ellipses of MG and HG overlapped extensively, suggesting similar nitrogen fraction structures. In contrast, the confidence ellipses of LG and HG showed almost no overlap, indicating distinct differences in soil nitrogen fraction composition between LG and HG.Figure 1Changes in soil nitrogen fractions across different grazing intensities(A) Concentrations of different nitrogen fractions under each grazing intensity. Data are presented as means ± standard deviation (n = 3 biological replicates per treatment). Different lowercase letters indicate significant differences among treatments (p < 0.05, one-way ANOVA followed by Tukey’s post hoc test).(B) NMDS ordination based on soil nitrogen fractions across grazing intensities. Ellipses represent 95% confidence intervals for each grazing treatment. Stress = 0.00469, R2 = 0.8890, p = 0.0030 (PERMANOVA test). Changes in soil nitrogen fractions across different grazing intensities (A) Concentrations of different nitrogen fractions under each grazing intensity. Data are presented as means ± standard deviation (n = 3 biological replicates per treatment). Different lowercase letters indicate significant differences among treatments (p < 0.05, one-way ANOVA followed by Tukey’s post hoc test). (B) NMDS ordination based on soil nitrogen fractions across grazing intensities. Ellipses represent 95% confidence intervals for each grazing treatment. Stress = 0.00469, R2 = 0.8890, p = 0.0030 (PERMANOVA test). Effects of grazing intensity on soil microorganisms and nitrogen-functional genes Soil microbial α-diversity was generally insensitive to grazing intensity (p > 0.05; see Figure S1). Compared with LG, only the Shannon evenness index was significantly higher under MG and HG (p < 0.05), whereas the Ace, Chao, Simpson, and Simpson evenness indices showed no significant differences among treatments. At the phylum level, dominant microbial groups exhibited distinct responses to grazing intensity (Figures 2C, 2E, and 2G). Among bacteria, the relative abundance of Pseudomonadota increased significantly with increasing grazing intensity (p < 0.001), that of Actinomycetota decreased significantly (p < 0.05), and that of Acidobacteriota changed significantly (p < 0.05). Among archaea, Euryarchaeota responded significantly to grazing intensity (p < 0.05), while Nitrososphaerota showed no significant response. Among fungi, Basidiomycota increased significantly with increasing grazing intensity (p < 0.05), whereas Mucoromycota and Ascomycota remained unchanged. Among non-dominant phyla, only a few groups (bacterial Nitrospirota, unclassified archaea, Thermoproteota, and fungal Olpidiomycota) showed significant or highly significant responses to grazing intensity; most other taxa exhibited no significant changes. Detailed results of multiple comparisons between groups are shown (see Table S1).Figure 2Soil microbial community composition under different grazing intensities(A) Overview of soil microbial community composition at the domain level (bacteria, archaea, fungi) across different grazing intensities.(B) Relative abundance of the three microbial domains across different grazing intensities.(C, E, and G) Circular plots showing the composition of dominant archaeal (C), bacterial (E), and fungal (G) phyla under different grazing intensities. (D, F, and H) Stacked bar charts showing the relative abundance of dominant archaeal (D), bacterial (F), and fungal (H) phyla across grazing treatments. Asterisks denote significant differences in relative abundance among the three grazing intensity groups (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Statistical differences were determined using one-way ANOVA followed by Tukey’s HSD post hoc multiple comparison test. Full results of post hoc multiple comparisons are provided (see Table S3). Soil microbial community composition under different grazing intensities (A) Overview of soil microbial community composition at the domain level (bacteria, archaea, fungi) across different grazing intensities. (B) Relative abundance of the three microbial domains across different grazing intensities. (C, E, and G) Circular plots showing the composition of dominant archaeal (C), bacterial (E), and fungal (G) phyla under different grazing intensities. (D, F, and H) Stacked bar charts showing the relative abundance of dominant archaeal (D), bacterial (F), and fungal (H) phyla across grazing treatments. Asterisks denote significant differences in relative abundance among the three grazing intensity groups (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Statistical differences were determined using one-way ANOVA followed by Tukey’s HSD post hoc multiple comparison test. Full results of post hoc multiple comparisons are provided (see Table S3). Redundancy analysis (RDA) showed that grazing intensity significantly drove the differentiation of microbial communities by altering soil physicochemical properties, and microbial community composition was strongly coupled with soil environmental conditions (Figure 3). Combined with Spearman’s rank correlation analysis (see Table S2), the HG group was driven by high soil pH. Pseudomonas exhibited significantly negative correlations with soil moisture content (SMC), organic matter (OM), and TN (r = −0.778 to −0.887, p < 0.01) and showed a positive trend with pH, indicating strong adaptability to high-pH and low-nutrient habitats. Nitrososphaerota showed no significant correlations with any environmental factor, but its distribution was also concentrated in the HG habitat. The LG group was shaped by high soil moisture, high nutrient levels, and high herb coverage, which was significantly negatively correlated with the HG environment, resulting in a distinctly separated microbial habitat. The MG group was in a transitional state; Actinomycetota and Acidobacteriota showed no significant correlations with environmental factors, suggesting their wide adaptability to heterogeneous transitional habitats. For fungi, Ascomycota, Basidiomycota, and Mucoromycota all displayed significantly negative correlations with SMC, OM, TN, and total phosphorus (TP) (r = −0.670 to −0.949, p < 0.05 or p < 0.01). These fungi generally favored low-moisture and high-pH habitats under HG, with no obvious habitat differentiation. In summary, the gradient changes in SMC, nutrient content, and pH induced by grazing intensity represent the core mechanism driving the divergence of soil microbial community structure.Figure 3RDA of soil microbial communities and soil physicochemical properties under different grazing intensitiesThe ordination is based on the relative abundance of dominant microbial phyla, with three biological replicates (n = 3) per grazing treatment. Green arrows represent environmental variables, including SMC, TP, HC (herb coverage), TN, SOM, TK (total potassium), and soil pH. Orange arrows represent microbial phyla. Black dots represent individual samples under light (LG), moderate (MG), and heavy (HG) grazing intensities. The percentages on the axes indicate the proportion of variance explained by each RDA axis (RDA1: 42.16%; RDA2: 29.73%) RDA of soil microbial communities and soil physicochemical properties under different grazing intensities The ordination is based on the relative abundance of dominant microbial phyla, with three biological replicates (n = 3) per grazing treatment. Green arrows represent environmental variables, including SMC, TP, HC (herb coverage), TN, SOM, TK (total potassium), and soil pH. Orange arrows represent microbial phyla. Black dots represent individual samples under light (LG), moderate (MG), and heavy (HG) grazing intensities. The percentages on the axes indicate the proportion of variance explained by each RDA axis (RDA1: 42.16%; RDA2: 29.73%) Nitrogen-metabolizing microbes mediate 12 classes of nitrogen metabolic pathways via functional genes (Figure 4A). Analysis of key nitrogen functional genes (p < 0.05; Figure 4B) showed that core genes involved in the entire nitrification pathway, including ammonia oxidation, hydroxylamine oxidation, and nitrite oxidation, exhibited a significant downregulation trend with increasing grazing intensity. Compared to LG, the downregulation of the hao gene in HG reached 64.6%, that of the amoA gene reached 44.2%, and that of the NxrA/NxrB genes ranged from 26.8% to 28.4%, representing the most critical downregulated pathways in the nitrogen cycle response to grazing disturbance. The denitrification pathway exhibited a differential response: the core functional genes of the pathway—NorB, NorC, and NosZ—were significantly downregulated as grazing intensity increased, with the maximum decrease in the HG group reaching 72.5% compared to the LG group; only the nirS gene within the pathway showed a slight, non-significant upward trend. The dissimilatory nitrite reduction to ammonium (DNiRA) pathway was the only nitrogen cycle pathway that showed a significant positive response to grazing. The abundance of the core gene nirB increased significantly with increasing grazing intensity, with a 142.1% increase in the HG group compared to the LG group (p < 0.001); in contrast, there were no significant intergroup differences in the nirD gene within the same pathway. In terms of abundance, ammonium assimilation genes showed the highest abundance, followed by mineralization genes. Denitrification and aerobic nitrate reduction to nitrite (ANRN) genes were intermediate, while ammonia oxidation genes were the lowest. In terms of the number of genes, the ammonium assimilation pathway contained the largest number of key genes, indicating that this pathway is the core process of soil nitrogen cycling in the study area. Collectively, grazing alters the soil nitrogen metabolic pattern mainly by inhibiting nitrification-related processes and promoting denitrification and dissimilatory ammonium reduction.Figure 4Abundance of nitrogen functional genes under different grazing intensities(A) Difference in nitrogen functional gene abundance before screening.(B) Difference in abundance of key nitrogen functional genes after screening. Data are presented as means ± standard deviation (n = 3 biological replicates per treatment). Asterisks denote significant differences in gene abundance among grazing treatments (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Statistical differences were determined using one-way ANOVA followed by Tukey’s HSD post hoc multiple comparison test. Abundance of nitrogen functional genes under different grazing intensities (A) Difference in nitrogen functional gene abundance before screening. (B) Difference in abundance of key nitrogen functional genes after screening. Data are presented as means ± standard deviation (n = 3 biological replicates per treatment). Asterisks denote significant differences in gene abundance among grazing treatments (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Statistical differences were determined using one-way ANOVA followed by Tukey’s HSD post hoc multiple comparison test. Relationships between nitrogen-functional microorganisms and individual nitrogen fractions under the influence of grazing intensity Correlation analysis between key functional genes of different nitrogen transformation pathways and major soil nitrogen fractions (Figures 5A and 5B) revealed strong pathway-specific regulatory patterns. Multiple key genes involved in hydroxylamine oxidation, assimilatory nitrite reduction to ammonium (ANiRA), DNiRA, and others were significantly correlated with soil nitrogen fraction contents (p < 0.05), indicating that these are critical pathways regulating soil nitrogen transformation and nitrogen pool composition. In contrast, key genes in the nitrogen fixation pathway showed no significant correlations with any nitrogen fraction (p > 0.05), suggesting a negligible influence on soil nitrogen dynamics under the experimental conditions. Collectively, the regulation of soil nitrogen fractions by key genes differed significantly among nitrogen transformation pathways. Hydroxylamine oxidation, ANiRA, DNiRA, and others were the core pathways driving soil nitrogen transformation, whereas nitrogen fixation made no obvious contribution.Figure 5Mantel analysis of key nitrogen functional genes under the key nitrogen metabolic pathway and each nitrogen component(A) Metabolic pathway containing only one key gene.(B) Metabolic pathway containing multiple key genes. Asterisks indicate significant correlations between key nitrogen functional genes and soil nitrogen fractions (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Line thickness represents the Mantel correlation coefficient, and color gradient indicates the significance level of the correlation. Mantel analysis of key nitrogen functional genes under the key nitrogen metabolic pathway and each nitrogen component (A) Metabolic pathway containing only one key gene. (B) Metabolic pathway containing multiple key genes. Asterisks indicate significant correlations between key nitrogen functional genes and soil nitrogen fractions (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). Line thickness represents the Mantel correlation coefficient, and color gradient indicates the significance level of the correlation. Soil N metabolic pathways under grazing intensity regulation Soil nitrogen metabolic pathways differed significantly in their response to grazing intensity (Figure 6A). The DNiRA pathway was the most sensitive to grazing disturbance and was the key pathway driving differences in nitrogen transformation.In contrast, the ammonium assimilation pathway showed no significant response to changes in grazing intensity. Regression analysis (see Figure S2) further indicated that the abundance of nirB, a key gene in the DNiRA pathway, was significantly negatively correlated with all soil nitrogen fractions, with the highest explanatory power for ASN and AHN, suggesting that these two fractions are key targets through which the DNiRA pathway regulates soil nitrogen transformation. Structural equation modeling (Figure 6B) and generalized linear modeling (see Table S3) collectively demonstrated that microbial function represents the core driver regulating soil nitrogen pool dynamics in alpine wetlands, exerting a significant positive effect on nitrogen fractions. Grazing indirectly altered the soil nitrogen pool by negatively affecting microbial function and soil physicochemical properties. By contrast, soil microbial diversity had a weak direct effect on nitrogen fractions, further highlighting the dominant role of functional microorganisms in soil nitrogen cycling.Figure 6Differences in soil nitrogen metabolic pathways and regulatory mechanisms under different grazing intensities(A) Differences in nitrogen metabolism pathways under LG intensity and HG intensity conditions. Asterisks denote significant differences after correction for multiple comparisons (∗p < 0.05, ∗∗p < 0.01). Non-significant differences in ammonium assimilation pathways and in comparisons involving moderate grazing intensity (MG vs. LG, MG vs. HG) are not shown.(B) Pathway analysis of how grazing intensity influences soil microbial nitrogen metabolism through nitrogen functional genes. Asterisks denote significant standardized path coefficients and indicator loadings (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). The model meets acceptable fit criteria (VIF = 5.95, SRMR = 0.075). Red arrows indicate positive effects, and blue arrows indicate negative effects; numbers on arrows represent standardized path coefficients. Differences in soil nitrogen metabolic pathways and regulatory mechanisms under different grazing intensities (A) Differences in nitrogen metabolism pathways under LG intensity and HG intensity conditions. Asterisks denote significant differences after correction for multiple comparisons (∗p < 0.05, ∗∗p < 0.01). Non-significant differences in ammonium assimilation pathways and in comparisons involving moderate grazing intensity (MG vs. LG, MG vs. HG) are not shown. (B) Pathway analysis of how grazing intensity influences soil microbial nitrogen metabolism through nitrogen functional genes. Asterisks denote significant standardized path coefficients and indicator loadings (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001). The model meets acceptable fit criteria (VIF = 5.95, SRMR = 0.075). Red arrows indicate positive effects, and blue arrows indicate negative effects; numbers on arrows represent standardized path coefficients. Discussion The effect of grazing on soil nitrogen processes has been a widely studied topic in ecological research. Grazing affects microbial processes in plants and soils mainly through trampling, foraging, and manure, which in turn affects soil nitrogen dynamics. Most studies have focused on the effects of grazing on nitrogen retention,38 nitrogen use,39 and nitrogen mineralization40 in grassland ecosystems. Although the microscopic mechanisms by which grazing intensity influences nitrogen metabolism processes in alpine wetlands remain unclear, extensive research indicates that their soil nitrogen pools and turnover processes are highly sensitive to grazing disturbance.41,42,43 Thus, this study of soil nitrogen metabolism under grazing is crucial for understanding how to maintain the ecological functions of alpine wetlands. Effects of grazing intensity on soil nitrogen fractions and microorganisms The low-temperature flooded environment of alpine wetlands inherently restricts nutrient cycling in plant roots,44 making these ecosystems particularly sensitive to grazing disturbances. This study confirms that grazing exacerbates this sensitivity through three pathways: direct removal of plant aboveground biomass forces plants to reallocate nutrients drastically45; reduced litter return cuts off the primary source of soil nitrogen replenishment46; weakened microbial immobilization reduces nitrogen conversion efficiency.47 These three factors mutually reinforce each other, forming a positive feedback loop of degradation that collectively leads to the depletion of the soil nitrogen pool.48 This mechanism refines the regulatory theory of soil nitrogen cycling in alpine wetlands under grazing disturbance, addressing the existing research gap regarding the unique characteristics of this habitat. As grazing intensity increases, the content of all nitrogen components in the soil decreases significantly, consistent with the trend observed in most studies of high-altitude grasslands and wetlands.6,49,50 Long-term grazing has been shown to suppress plant carbon and nitrogen inputs to the soil, thereby weakening the soil’s nitrogen sequestration capacity.51 The specific mechanism is as follows: as grazing intensity increases, nutrient allocation in plants shifts toward above-ground parts, while the distribution of photosynthetic products to below-ground parts decreases.45 Meanwhile, decreased litter return halts soil nitrogen supply,52 and weakened microbial immobilization reduces nitrogen retention efficiency.53 These three factors form a positive feedback loop of degradation, accelerating nitrogen loss. To quantify the net effects of grazing, we constructed a literature-based ungrazed baseline table (see Table S4). The comparison reveals that the response patterns of soil nitrogen components to grazing intensity in this study are fully consistent with our discussion. During the LG period, AHN and AN concentrations were significantly higher than the ungrazed baseline range (based on integrated literature results). This confirms that LG may stimulate the release of plant root exudates,54 providing additional carbon sources for microorganisms that promote microbial nitrogen fixation55 and maintain a relatively stable nitrogen cycle. Although this study did not specifically focus on this effect, the overall trend is consistent with this observation. This suggests that moderate grazing may activate positive feedback mechanisms between plants and soil.56 However, as grazing intensity increased to MG and HG levels, the concentrations of all nitrogen components declined sharply, falling far below the ungrazed baseline levels (based on integrated literature results). This threshold effect clearly demonstrates that when grazing intensity exceeds a certain limit, negative feedback mechanisms replace self-regulation, leading to accelerated nitrogen loss.57 Based on the ungrazed microbial baseline established in this study (see Table S5), dominant microbial groups exhibited distinct responses to grazing: grazing intensity significantly drove the differentiation of microbial community structure, consistent with the general principle that “grazing achieves environmental filtering by altering soil physicochemical properties.”58 However, this study presents two unique findings: first, microbial diversity showed no significant response to grazing, and second, dominant fungal phyla exhibited no clear habitat preferences. This contrasts with the conclusions drawn in most studies that “grazing significantly alters diversity”59 and “fungal communities show marked differentiation.”60 The differentiated response mechanisms of bacterial communities are as follows. Under HG, soil pH and total potassium content significantly increased. The abundance of Pseudomonadota increased significantly to 45%, approaching the upper limit of the natural baseline range. This indicates that HG substantially promotes the enrichment of this eutrophic group, confirming the pattern that “HG drives alkalization and enriches alkaline-tolerant groups.”61 This study further clarifies the synergistic regulatory effect of pH and total potassium. LG maintains higher moisture content and nutrient levels, and the Euryarchaeota phylum shows significant responses. Water-nutrient coupling homeostasis is key.62 Moderate grazing, as a transitional type, showed a significant reduction in Actinomycetota abundance, yet it remained markedly higher than the natural baseline level; Acidobacteriota exhibited significant change. This differs from the “classic moderate disturbance hypothesis”63 (where diversity peaks). This may stem from the low-temperature, high-humidity baseline of alpine wetlands, where microbial responses rely more on overlapping environmental gradients than interspecific competition release.64 This is the key distinction between this study and traditional theories. Fungal communities exhibit a distinct disturbance buffering mechanism: Although the abundance of Basidiomycota increased significantly with grazing intensity, Ascomycota showed no significant change, and neither group exhibited distinct habitat preferences. RDA analysis indicates that fungi exhibit broad adaptability to soil pH, moisture, and nutrients.65 Their mycelium network structure can buffer local environmental fluctuations.66 Additionally, the metabolic inertia of alpine wetlands (slow growth and delayed response)67 reduces their sensitivity to grazing. This provides new insights into the disturbance response of fungi in alpine wetlands. In summary, this study validates the core hypothesis: grazing intensity indirectly drives key processes in nitrogen cycling through the coupling relationship between soil physicochemical properties and microbial function.68 This mechanism aligns with existing research on “physicochemical factors driving microbial differentiation” while exhibiting distinct characteristics in terms of “diversity non-response” and “fungal non-preference,” providing microbial ecological support for precise management of grazing in alpine wetlands. Based on the threshold effects described above, this study identified two threshold levels for grazing management in the study area: 0.6–1.0 yaks per hectare represents the optimal grazing intensity for the study area, while 1.4 yaks per hectare marks the critical threshold for degradation; exceeding these thresholds leads to irreversible damage to the soil nitrogen pool and microbial homeostasis. By incorporating the seasonal patterns of the nitrogen cycle,69 we have established a precision management system that thoroughly integrates threshold values with seasonal variations. Pre-growing season (April–June, critical sensitive period): strictly limit grazing to the lower limit of the LG range (≤0.6 yaks per hectare) to retain the positive stimulating effects of LG while avoiding high-intensity disturbances that could disrupt the annual nitrogen pool baseline; active growing season (July–August): maintain the upper limit of the LG range (≤1.0 yaks per hectare), leveraging vegetation’s compensatory growth advantage to strengthen the positive feedback loop between plant and soil nitrogen cycling; Dormant season-Winter and spring (September–March of the following year): maintain stocking rates at the lower to middle levels of the LG range (≤0.7 yaks per hectare) to prevent a decline in ground cover that could exacerbate nitrogen loss and ensure nitrogen pool accumulation for the following year. This threshold system directly links responses in nitrogen cycling to quantitative indicators of grazing, providing a practical, quantifiable basis for precise grazing management in alpine wetlands. Characterization of functional genes involved in nitrogen metabolism in microorganisms Building upon the established patterns of microbial community differentiation, this study further employed metagenomic analysis to characterize nitrogen metabolism functional genes, thereby elucidating the systemic regulation of grazing on nitrogen cycling functional potential. Results indicated (Figures 4A and 4B) that the abundance of functional genes involved in nine key processes, including ammonium assimilation, mineralization, DNiRA, ANiRA, denitrification, and nitrogen fixation, was significantly regulated by grazing intensity. The overall pattern exhibited high abundance of ammonium assimilation, mineralization, and denitrification genes, coupled with low abundance of nitrogen fixation and ammonium oxidation genes. This pattern is largely consistent with the nitrogen cycle gene characteristics observed in temperate grasslands,70 agricultural ecosystems,71 and Arctic permafrost,72 and reflects the universal patterns of nitrogen metabolism in terrestrial ecosystems.73 However, the uniqueness of alpine wetlands lies in the direction and intensity of functional gene regulation in response to grazing disturbance. Unlike the pattern observed in temperate grasslands where “LG promotes while HG completely suppresses,”74 this study found that HG primarily exerts negative interference with significant gene specificity. Ammonium assimilation genes maintained high abundance, while ammonium oxidation genes exhibit extremely low abundance, which theoretically should lead to AN accumulation. However, actual measurements of nitrogen components revealed an overall decline in available nitrogen, exposing the paradoxical pattern of “relative abundance of high-ammonium assimilation genes—low nitrogen availability.” It is speculated that the suppression of ammonia-oxidizing gene expression in cold-anaerobic environments creates a “nitrification bottleneck” (metabolic bottleneck) rather than substrate limitation,75 which represents the key mechanism distinguishing this study from temperate grassland research. The degradation of organic nitrogen and the decline in synthetic gene abundance weakened the efficiency of nitrogen transformation. The abundance of core genes in the ANiRA pathway decreased significantly, whereas the abundance of core genes in the DNiRA pathway increased significantly with increasing grazing intensity; the two pathways exhibited divergent responses. This significantly enhanced DNiRA’s ability to compete with the denitrification pathway for nitrate. This finding differs from the pattern of “synergistic responses between DNiRA and denitrification under anaerobic conditions” observed in paddy soils76 and coastal wetlands.77 The study reveals that, under the low-temperature,78 anaerobic conditions of alpine wetlands,79 DNiRA functional bacteria exhibit greater adaptability to cold environments and preferentially occupy ecological niches. Ultimately, it led to a decrease in the effective nitrogen content, becoming an important molecular basis for the reorganization of microbial communities and the function of the nitrogen cycle.80 The abundance of key genes involved in nitrification and denitrification decreased synchronously, blocking the production and conversion of nitrate nitrogen. Traditional theory posits that nitrification and denitrification exhibit an inverse relationship.79 However, in this study, the synergistic inhibition of both factors creates a “metabolic deadlock” in the nitrogen cycle. This “dual-inhibition pattern” originates from substrate co-limitation in alpine wetlands: Inhibition of ammonium oxidation prevents AN from being oxidized to nitrate, leading to a simultaneous shortage of substrates for denitrification. Unlike the linear relationship of “negative correlation between grazing intensity and gene abundance” observed in most studies, this study found that nitrogen-fixing gene abundance was higher under moderate grazing than under LG. Moreover, the DNiRA pathway marker genes showed a significant positive correlation with grazing intensity, exhibiting highly synchronized trends in their changes. This pattern aligns with findings from studies on the Qinghai-Tibet Plateau meadows81 and alpine mountain ecosystems82 that moderate disturbance promotes functional gene expression, though the underlying mechanisms differ. The latter is attributed to the release of resources through the moderate disturbance hypothesis,83 while this study posits that it stems from the functional coupling between nitrogen fixation and DNiRA, The former provides nitrogen sources, while the latter regulates nitrogen forms. Together, they maintain the internal cycle of nitrogen between ammonium and organic forms, reducing gaseous losses.84 However, this study also found that the relative abundance of nitrogen-fixing genes was not significantly correlated with the content of nitrogen components in the soil, presenting a feature where the changes in the relative abundance of genes were decoupled from the dynamics of the nitrogen pool. This is speculated to be due to the transcriptional expression of nitrogen-fixing genes and the activity of nitrogenase being strongly limited by environmental factors such as low temperature and the effective carbon content of the soil. Their potential nitrogen supplementation function was not fully realized, which also emphasizes the multi-level regulatory characteristics of soil nitrogen cycling-environment-gene-function, rather than a simple determinism of gene abundance.85 In summary, this section reveals three unique characteristics of high-altitude cold and humid regions compared to other ecosystems at the level of nitrogen metabolism functional genes: first, low temperature-anaerobic coupling inhibits nitrification, forming a metabolic bottleneck; second, DNiRA-denitrification competition imbalance weakens nitrogen retention; third, nitrogen fixation-DNiRA synergistic response constructs a buffering cycle. This lays the groundwork for subsequent research focusing on core nitrogen metabolism pathways and elucidating micro-regulatory mechanisms at the functional potential level. Functions of soil microbial nitrogen metabolism under grazing intensity regulation Building upon the identified patterns of functional gene differential responses, this study further elucidated the systemic regulatory mechanisms of nitrogen metabolism under grazing intensity by quantifying the abundance of 11 nitrogen metabolism pathways and employing multi-method validation. In this study, the DNiRA pathway was identified as the most significantly responsive nitrogen metabolic pathway to grazing disturbance, which stands in stark contrast to the conclusion that denitrification pathways dominate in studies of temperate grasslands.86 This reveals the unique nitrogen cycling regulation patterns in alpine wetlands. A global meta-analysis87 indicates that moderate grazing to HG significantly reduces grassland N2O emissions (by 22%–25%) and is accompanied by a decrease in the abundance of denitrification functional genes (NirS and NarG), suggesting denitrification is the primary pathway for nitrogen loss in these ecosystems. However, in perennial or seasonal anaerobic alpine wetlands, extreme anaerobic conditions inhibit the complete denitrification process (N2O→N2), making DNiRA a more competitively advantageous nitrogen retention strategy.88 This represents a fundamental difference from the nitrogen cycling patterns in temperate grasslands (aerobic-anaerobic alternation) and tropical wetlands (high-temperature anaerobic conditions). Compared with other permafrost regions worldwide, the alpine wetland in this study features a thinner active layer of permafrost.89 Additionally, grazing disturbance exacerbates soil compaction, which further reduces soil aeration and intensifies anaerobic conditions. Concurrently, low temperature persistently inhibits enzyme activity. These factors collectively lead to highly distinctive responses of nitrogen cycling pathways, providing an important case for establishing a unified framework for nitrogen cycling in alpine wetlands globally. Arctic permafrost is dominated by ammonification and nitrification, with a minor contribution from the DNiRA pathway.90 In contrast, the DNiRA pathway acts as the core regulatory pathway in this study, which also provides an important case for constructing a unified framework of nitrogen cycling in alpine wetlands worldwide. Although alpine meadows (non-wetland) on the Qinghai-Tibet Plateau also experience low temperatures, denitrification remains the dominant pathway of nitrogen loss due to relatively high soil aeration and weak anaerobic conditions.5 This is in sharp contrast to the pattern of “silenced denitrification but activated DNiRA” observed in this study, which also provides an important case support for establishing a unified framework of nitrogen cycling in alpine wetlands worldwide. This difference at the pathway level reflects the shift in the advantages of functional gene niches: unlike the sensitivity of the nirS gene to grazing in temperate grasslands,91 nirB (the DNiRA marker gene) emerged as the key biomarker distinguishing grazing gradients in this study, marking a shift from “denitrification gene sensitivity” to “DNiRA gene sensitivity.” This finding holds significant implications for regional soil nitrogen management: in temperate grassland ecosystems, reducing grazing intensity can effectively suppress denitrification processes and decrease gaseous nitrogen losses. In alpine wetlands, the DNiRA process is the key regulatory link that controls the soil nitrogen availability and retention capacity. This study reveals the response characteristics of the DNiRA pathway to grazing disturbance at the gene level in alpine wetland ecosystems, providing new molecular ecological evidence for understanding nitrogen transformation processes in alpine wetlands. From an internal genetic perspective (see Figure S3), key genes within the DNiRA pathway exhibit functional differentiation characteristics. As a key biomarker for distinguishing grazing gradients, nirB expression is primarily regulated by soil pH and redox conditions, exhibiting significant negative correlations with SMC and TP. This pattern aligns with the ecological adaptation of nirB gene-carrying bacteria, such as E. coli, in alkaline soils.92 This suggests that the functional expression of nirB in alpine wetlands is more sensitive to changes in enzyme activity and electron transport efficiency caused by pH variations. It is noteworthy that the negative response of nirB to SMC seems to contradict the anaerobic nature of DNiRA. However, this can be explained as follows: under extreme anaerobic conditions, such as those in heavily compacted soils resulting from intensive grazing, there may be competitive inhibition of nitrate reduction by sulfate reduction or methanogenesis.93 Alternatively, high humidity can cause diffusion limitations of the substrate (NO3−).94 In contrast, the NrfH gene in the same pathway showed significant positive correlations with SMC, OM, and TP, indicating sensitivity to resource availability rather than redox conditions. This discrepancy can be attributed to the function of NrfH as a membrane-bound cytochrome c subunit. As such, it relies more heavily on the integrity of the cell membrane and the continuous supply of electron donors (organic carbon).95 Under grazing disturbance, the elevated contents of soil organic matter (SOM) and TP, which are likely due to livestock manure input and plant litter accumulation, directly enhance the metabolic activity of microorganisms carrying the nrfH gene.96 This division of labor, wherein nirB responds to soil environmental stresses (pH and redox changes) while nrfH responds to resource availability (carbon and phosphorus availability), shares similarities with the DNiRA gene differential expression mechanism reported in marine sediments,97 but has not yet been systematically documented in alpine terrestrial ecosystems. Further highlighting the unique regulatory role of alpine wetland habitats in the functional differentiation of nitrogen metabolic genes. In conclusion, the DNiRA pathway achieves coordinated adaptation to multiple stressors (soil compaction, nutrient input, and pH changes) caused by grazing through the differential environmental response strategies of NirB and NrfH. The formation of this adaptive mechanism is inseparable from the core habitat of “low temperature-anaerobic-permafrost” in alpine wetlands, thereby becoming the core pathway driving soil nitrogen transformation in alpine wetlands. This mechanism not only explains the microbial-driven basis for the changes in nitrogen components observed in this study, but also provides a theoretical basis for predicting the nitrogen cycle responses of alpine wetlands under global change scenarios (such as permafrost thawing and rising temperatures). If global warming leads to an intensified thawing of permafrost, a weakened anaerobic state of the soil, and a significant release of organic carbon, it may disrupt the current “DNiRA-dominated” nitrogen cycle pattern and shift toward denitrification dominance, thereby increasing the gaseous loss of nitrogen. However, metagenomic analysis revealed a seemingly contradictory phenomenon: despite the DNiRA pathway exhibiting high sensitivity to grazing disturbance, the samples simultaneously possessed a complete and highly potent genetic basis for denitrification (see Figure S4). This decoupling phenomenon between genetic potential and metabolic function98 represents a key characteristic of nitrogen cycling in alpine wetlands, indicating that microbial communities possess denitrification potential but fail to achieve functional expression. Combining gene set enrichment analysis (GSEA) results (see Figure S4) with the habitat characteristics of alpine wetlands in the western Sichuan Plateau, this decoupling phenomenon may stem from triple environmental constraints. Low temperatures significantly inhibit denitrification functions: Holten-Hartwig et al.99 observed in a 0°C–20°C temperature gradient experiment that denitrification rates at 0°C were substantially lower than predicted values, with N2O reductase activity exhibiting the most pronounced suppression. This indicates that low temperatures selectively inhibit the expression of denitrification genes. Carbon source limitation further restricts denitrifying enzyme activation: glucose addition can increase denitrifying enzyme activity by 2.3- to 2.5-fold.100 Studies on Shewanella model strains confirm that DNRA dominates under carbon-rich conditions, while incomplete denitrification pathways switch under carbon-limited conditions.101 Global regulation ultimately determines pathway fate: CRP1 subunits selectively activate DNRA genes while suppressing denitrification genes under high-carbon conditions.101 The above analysis reveals a dual-pathway regulatory model for nitrogen cycling in alpine wetlands: at the genetic level, both denitrification and DNiRA pathways possess full potential; however, at the functional level, environmental stressors (cold temperatures, anaerobic conditions, and carbon limitation) differentially inhibited denitrification while selectively promoting the DNiRA pathway. This seesaw effect of “denitrification suppression-DNiRA activation” explains why soil nitrogen is retained more as AN rather than lost as gaseous nitrogen under HG conditions. This study provides a new perspective for understanding nitrogen retention mechanisms in alpine wetlands and lays a theoretical foundation for predicting nitrogen cycle responses in these ecosystems under global change. Overall, the mechanism analysis based on genomics has revealed the differentiated response strategies of the DNiRA pathway. Comparisons with other global ecosystems indicate that the unique habitat characteristics of alpine wetlands have reshaped the competitive landscape of the nitrogen cycling pathway, forming a unique regulatory pattern dominated by DNiRA. To quantitatively verify the regulatory model of “functional genes driving rather than overall diversity being dominant,” this study further analyzed the causal path of grazing intensity-functional genes-nitrogen components through structural equation modeling (SEM) and generalized linear regression (GLM), confirming the key driving role of the core nitrogen metabolism pathway in the nitrogen cycling system of wetlands, and providing important case support for the construction of a unified framework for nitrogen cycling in global alpine wetlands. Limitations of the study At the same time, this study has four limitations, which provide directions for future research. First, this study conducted soil sampling only in June (prior to the peak growing season), failing to examine the seasonal dynamics of soil nitrogen cycling. Seasonal variations in hydrothermal conditions, plant growth, and microbial potential may jointly regulate soil nitrogen cycling processes alongside grazing intensity. Future research should employ continuous seasonal sampling to elucidate the seasonal patterns of soil nitrogen cycling under different grazing intensities and their underlying mechanisms, thereby enhancing the generalizability and accuracy of research conclusions. Second, this study solely reflects microbial nitrogen cycling potential based on functional gene abundance without verifying the activity of the corresponding encoded enzymes, thus unable to directly quantify the actual functional contribution and expression efficiency of genes. Future research should integrate measurements of key nitrogen cycle enzyme activities to validate the actual expression levels of functional genes. Concurrently, nitrogen isotope labeling techniques should be employed to precisely quantify microbial-mediated nitrogen transformation pathways and rates, thereby further elucidating the mechanisms by which grazing influences soil nitrogen cycling. Third, there is a lack of quantitative metabolic evidence. Since no in situ nitrogen isotope labeling tracer experiments have been conducted, it is difficult to accurately and quantitatively determine the regulatory effects and relative contributions of different grazing intensities on microbial-mediated nitrogen metabolic pathways under real-world field conditions, thereby limiting an in-depth analysis of the underlying mechanisms driving grazing-induced changes in soil nitrogen cycling. Future research could utilize in situ 15N isotope tracing and in situ microcosm experiments to accurately quantify the rates and relative contributions of various nitrogen transformation pathways under in situ conditions in high-altitude wetlands, thereby further elucidating the microscopic metabolic mechanisms by which grazing disturbances influence soil nitrogen cycling. Fourth, the grazing management plan proposed in this study is based primarily on the specific habitat conditions of the high-altitude, cold, and wet wetlands in the study area (Figure 7), as well as on empirical data regarding traditional grazing patterns and soil nitrogen pool responses prior to the growing season. As such, it is clearly tailored to the specific region and provides valuable practical guidance for local implementation. However, its generalizability across larger spatial scales and under different climatic and vegetation conditions still requires systematic verification through future multi-regional collaborative field trials to avoid overinterpretation of the study’s conclusions and their exaggerated application across regions.Figure 7Spatial distribution map of soil sample collection points under different grazing conditions in eastern Qinghai-Tibet Plateau Spatial distribution map of soil sample collection points under different grazing conditions in eastern Qinghai-Tibet Plateau Resource availability Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Di Kang (kangdi@cwnu.edu.cn). Materials availability This study did not generate new unique reagents. Data and code availability •The raw sequence data used in this study can be found at GSA: CRA033662 (https://ngdc.cncb.ac.cn/gsa/browse/CRA033662).•This paper does not report original code.•Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request. •The raw sequence data used in this study can be found at GSA: CRA033662 (https://ngdc.cncb.ac.cn/gsa/browse/CRA033662). The raw sequence data used in this study can be found at GSA: CRA033662 (https://ngdc.cncb.ac.cn/gsa/browse/CRA033662). •This paper does not report original code. This paper does not report original code. •Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request. Acknowledgments We are grateful for the financial support from the 10.13039/501100018542Natural Science Foundation of Sichuan Province (2023NSFSC0143); the 10.13039/501100001809National Natural Science Foundation of China (31800458); and the Innovation and Entrepreneurship Training Program for College Students at 10.13039/501100007433China West Normal University (cxcy2024282 and S202410638110). We thank the editors and reviewers for their comments. Author contributions Conceptualization, S.S.F.; methodology, D.K. and S.Z.Z.; investigation, S.S.F., Q.M.L., and Y.Y.C.; writing – original draft, S.S.F.; funding acquisition, D.K. Declaration of interests The authors declare no competing interests. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERDeposited dataRaw sequence dataGSA: CRA033662https://ngdc.cncb.ac.cn/gsa/browse/CRA033662Software and algorithmsSPSS 27IBM Corphttps://www.ibm.com/products/spss-statisticsOrigin 2024OriginLab Corporation, 2024https://www.originlab.com/STAMP 2.1.1.0102Parks et al., 2014STAMP: statistical analysis of taxonomic and functional profiles. https://github.com/donovan-h-parks/STAMP/releases/tag/v2.1.1SmartPLS-v 4.1.1.2SmartPLS GmbHhttps://www.smartpls.com/Majorbio103Han et al., 2024Majorbio Cloud 2024: Update single-cell and multiomics workflows. https://www.majorbio.com/tools REAGENT or RESOURCESOURCEIDENTIFIERDeposited dataRaw sequence dataGSA: CRA033662https://ngdc.cncb.ac.cn/gsa/browse/CRA033662Software and algorithmsSPSS 27IBM Corphttps://www.ibm.com/products/spss-statisticsOrigin 2024OriginLab Corporation, 2024https://www.originlab.com/STAMP 2.1.1.0102Parks et al., 2014STAMP: statistical analysis of taxonomic and functional profiles. https://github.com/donovan-h-parks/STAMP/releases/tag/v2.1.1SmartPLS-v 4.1.1.2SmartPLS GmbHhttps://www.smartpls.com/Majorbio103Han et al., 2024Majorbio Cloud 2024: Update single-cell and multiomics workflows. https://www.majorbio.com/tools Method details Site selection This study focuses on a typical ecosystem in the eastern fringe of the Tibetan Plateau (Figure 7), with specific geographic coordinates of 32.39° to 32.73°N, 102.34° to 102.41°E. This region is a typical example of the cold-temperate monsoon climate found on the continental plateau. The synergistic effects of high-altitude terrain and monsoon circulation create a unique climatic pattern: there is no period throughout the year that meets the meteorological criteria for summer, and winter lasts for over eight months, exhibiting the distinctive feature of “long winters with no summer.7 The growing season is limited to July to September, and there is a clear seasonal differentiation with significant alternation between rainy and dry periods.104 This zone has an average elevation of about 3507 meters above sea level, an average annual temperature of 1.4°C, and a total annual precipitation of 860.8 millimeters, creating an alpine and humid environmental substrate. This region is underlain by a permafrost layer with a maximum depth of 72–101 centimeters. Its water-retaining properties maintain surface water accumulation, forming marshy wetlands. However, over the past 30 years, the lower boundary of the permafrost has risen by 25–100 meters,105 causing groundwater levels to drop and increasing the risk of wetland shrinkage. Grassland covers as much as 91.8% of the total area in Hongyuan County, where the research site is located. Dominant species includ Carex alatauensis: averaging approximately 30 cm in height, it forms dense, low-growing patches in wetland meadows with a community cover exceeding 95%; Elymus nutans: Culms 50–70 cm tall with deep root systems providing both drought resistance and soil stabilization; C. muliensis: Triangular stems reaching about 65 cm, forming clumpy grass mounds in marsh areas; E. dahuricus: Morphologically similar to E. nutans but with erect panicles, exhibits exceptional cold tolerance (withstands -38°C), expanding communities on dry slopes at wetland margins; and Poa annua L.: Low-growing (10–20 cm), tolerant of trampling and short-term flooding, forming carpet-like communities in wetland grazing areas. Soils in the study area were acidic (pH < 7.0), with a deep peat layer with an average thickness of 4.5 m, and significant accumulation of organic matter. Soil physical indicators showed an average bulk density of 0.2 g/cm3. Field survey and sample collection Animal activity traces were recorded based on historical remote sensing image analysis and field research. According to the classification criteria for grazing intensity proposed by Yin,87 light grazing (proper grazing, normal grassland vegetation composition, no furrows trampled by hooves, <40% forage utilization), moderate grazing (slightly heavier grazing, furrows with animal hoof trampling, 40%-50% forage utilization), and heavily grazed (overgrazed, extensive occurrence of furrows trampled by animal hooves, >50% forage utilization rate, significant reduction in pasture and increase in weeds). To ensure the accuracy and reproducibility of grazing intensity classification, this study employed a two-step approach for plot selection: NDVI remote sensing preliminary screening combined with field indicator verification. After initial screening was conducted by calculating the Normalized Difference Vegetation Index (NDVI) from remote sensing imagery of the study area, corresponding threshold ranges were established as follows: LG (0.20–0.23), MG (0.18–0.23), and HG (0.09–0.14). Plots with NDVI values falling within these respective ranges were designated as candidate sample plots (For the spatial distribution of NDVI and spatial differentiation characteristics under different grazing intensities in the study area, see Figure S5). Subsequently, five 1 m × 1 m survey plots were randomly established within each candidate site to conduct field measurements of hoofprint density and aboveground biomass, using the following specific methods:(1)Hoofprint Density: Count the number of clearly visible yak and Tibetan sheep hoofprints within each small sample plot and calculate the average as quantitative hoofprint density data. Based on the hoofprint trampling grading standards,106,107 the corresponding quantitative thresholds, trampling characteristics, and specific stocking rate parameters for each grazing intensity are as follows: LG (hoofprint density 0–2 per m2, no animal hoof ruts); MG (hoofprint density 3–5 per m2, animal hoof ruts present); HG (hoofprint density ≥6 per m2, abundant animal hoof ruts). (2) Above-ground biomass: Samples were measured using the 1 m × 1 m plot mowing method.106 They were then dried in a 65°C oven to constant weight before weighing, serving as an auxiliary indicator of vegetation utilization. (3) Stocking Rate: Determined based on the natural carrying capacity of the study area and local traditional grazing levels, and quantified with reference to relevant studies on alpine wetlands in the Qinghai-Tibet Plateau.108 LG (0.6–1.0 yak/ha, equivalent to 3–5 Tibetan sheep/ha); MG (1.4–1.8 yak/ha, equivalent to 7–9 Tibetan sheep/ha); HG (2.2–2.6 yak/ha, equivalent to 14–18 Tibetan sheep/ha). Under a non-fixed rotational grazing system, grazing occurs continuously throughout the year. Livestock roam freely within the study area to feed and complete their foraging activities within their natural foraging range.109 All plots are naturally grazed by local herders following traditional regional grazing management practices, with no artificially established rotational grazing systems. The grazing intensity level for each plot was ultimately determined by integrating multiple indicators including remote sensing NDVI, hoofprint density, aboveground biomass, and stocking rate. This study did not conduct actual measurements to calculate forage utilization rates. The aforementioned forage utilization rate thresholds were directly adopted from Yin87 classification system and were used solely to define the levels of grazing intensity. (1)Hoofprint Density: Count the number of clearly visible yak and Tibetan sheep hoofprints within each small sample plot and calculate the average as quantitative hoofprint density data. Based on the hoofprint trampling grading standards,106,107 the corresponding quantitative thresholds, trampling characteristics, and specific stocking rate parameters for each grazing intensity are as follows: LG (hoofprint density 0–2 per m2, no animal hoof ruts); MG (hoofprint density 3–5 per m2, animal hoof ruts present); HG (hoofprint density ≥6 per m2, abundant animal hoof ruts). (2) Above-ground biomass: Samples were measured using the 1 m × 1 m plot mowing method.106 They were then dried in a 65°C oven to constant weight before weighing, serving as an auxiliary indicator of vegetation utilization. (3) Stocking Rate: Determined based on the natural carrying capacity of the study area and local traditional grazing levels, and quantified with reference to relevant studies on alpine wetlands in the Qinghai-Tibet Plateau.108 LG (0.6–1.0 yak/ha, equivalent to 3–5 Tibetan sheep/ha); MG (1.4–1.8 yak/ha, equivalent to 7–9 Tibetan sheep/ha); HG (2.2–2.6 yak/ha, equivalent to 14–18 Tibetan sheep/ha). Under a non-fixed rotational grazing system, grazing occurs continuously throughout the year. Livestock roam freely within the study area to feed and complete their foraging activities within their natural foraging range.109 All plots are naturally grazed by local herders following traditional regional grazing management practices, with no artificially established rotational grazing systems. The grazing intensity level for each plot was ultimately determined by integrating multiple indicators including remote sensing NDVI, hoofprint density, aboveground biomass, and stocking rate. This study did not conduct actual measurements to calculate forage utilization rates. The aforementioned forage utilization rate thresholds were directly adopted from Yin87 classification system and were used solely to define the levels of grazing intensity. Hoofprint Density: Count the number of clearly visible yak and Tibetan sheep hoofprints within each small sample plot and calculate the average as quantitative hoofprint density data. Based on the hoofprint trampling grading standards,106,107 the corresponding quantitative thresholds, trampling characteristics, and specific stocking rate parameters for each grazing intensity are as follows: LG (hoofprint density 0–2 per m2, no animal hoof ruts); MG (hoofprint density 3–5 per m2, animal hoof ruts present); HG (hoofprint density ≥6 per m2, abundant animal hoof ruts). (2) Above-ground biomass: Samples were measured using the 1 m × 1 m plot mowing method.106 They were then dried in a 65°C oven to constant weight before weighing, serving as an auxiliary indicator of vegetation utilization. (3) Stocking Rate: Determined based on the natural carrying capacity of the study area and local traditional grazing levels, and quantified with reference to relevant studies on alpine wetlands in the Qinghai-Tibet Plateau.108 LG (0.6–1.0 yak/ha, equivalent to 3–5 Tibetan sheep/ha); MG (1.4–1.8 yak/ha, equivalent to 7–9 Tibetan sheep/ha); HG (2.2–2.6 yak/ha, equivalent to 14–18 Tibetan sheep/ha). Under a non-fixed rotational grazing system, grazing occurs continuously throughout the year. Livestock roam freely within the study area to feed and complete their foraging activities within their natural foraging range.109 All plots are naturally grazed by local herders following traditional regional grazing management practices, with no artificially established rotational grazing systems. The grazing intensity level for each plot was ultimately determined by integrating multiple indicators including remote sensing NDVI, hoofprint density, aboveground biomass, and stocking rate. This study did not conduct actual measurements to calculate forage utilization rates. The aforementioned forage utilization rate thresholds were directly adopted from Yin87 classification system and were used solely to define the levels of grazing intensity. The grazing animals comprise a mixed herd of yak (Bos grunniens) and Tibetan sheep (Ovis aries). The mixed herd is primarily composed of non - pregnant adult females (≥90%), with a small proportion of sterile males (<10%).110,111,112,113 Yak accounting for 60% of the mixed herd. The average age of the yak is 3.5 years,111 and they prefer to graze on the aboveground parts of C. alatauensis and C. muliensis. Tibetan sheep (40% of the mixed herd), with an average age of 1 year,114 exhibit complementary foraging preferences to yak, favoring the leaves and stems of grasses such as E. nutans and P. pratensis.115 During the summer, natural grazing lasts approximately 15 hours daily, following a pattern of early morning departure and late evening return.116,117,118 During winter, natural grazing, daily grazing duration is reduced to 8–10 hours, adopting a late-out, early-return model.119,120,121 Soil sampling was conducted in June (the early stage of the vigorous growing season) in this study. During this period, the plant community had completed spring green-up, and root growth and nutrient uptake activity had not yet reached their annual peak, which could minimize the perturbation of short-term plant growth activities on the background values of soil nutrients. This early growing-season sampling protocol has been widely adopted in studies related to grazing disturbance and soil biogeochemical cycling in alpine ecosystems.122 Monitoring plots were established for three grazing intensity conditions: light grazing (LG), medium grazing (MG) and heavy grazing (HG), where three replications of each grazing intensity were made, and each grazing plot was 100 m × 100 m in size. After the initial sample selection, grazing intensity was further verified based on soil indicators. Ultimately, three sample plots were selected for each grazing intensity level, resulting in a total of nine sample plots. three subplots of 20 m × 20 m were established in each plot, and three replicate sampling points were randomly selected in each sample plot for soil sampling. Sampling was done avoiding animal feces and after removing visible plant and animal debris and apomictic material, top soil was collected to a depth of 15 cm using an earth auger. Composite soil samples were obtained by thoroughly mixing each sample soil sample and removing fine roots and larger plant fragments from the composite soil samples. Soil samples were mixed well and divided into two portions. One portion was left to dry naturally and preserved for the determination of soil nutrient content, while the other was quickly brought back to the laboratory and stored in a refrigerator at -80°C for microbial sequencing. Total nitrogen (TN) content was determined using a TOC analyzer (MultiN/C3100TOC, Analytik, Jena, Germany). Soil moisture (SM) was analyzed by weighing the soil and calculating the mass loss after oven drying at 105°C. Soil pH was determined using a pH meter (FE20, Mettler-Toledo Instruments, China) at a soil water ratio of 1:2.5 (w/v). Soil organic nitrogen fractions-aminosugar nitrogen (ASN), amino acid nitrogen (AAN), ammonia and aminosugar nitrogen (NH3+ASN), and alkaline hydrolyzable nitrogen (AHN) were determined according to the method described by Zhu.123 Macrogenome assembly, classification In this study, each sample was sequenced to a depth of 10 Gb of clean data. The minimum length threshold for overlapping clusters retained for subsequent analysis was set at 300 bp. All bioinformatics analyses were performed on the Majorbio Cloud Platform124 (https://cloud.majorbio.com/). Fastp125 (v0.20.0) was used for quality control of the raw sequencing data: the adapter sequences at the 3′ and 5′ ends of the reads were precisely trimmed, and low-quality reads with a length of <50 bp after trimming and an average base quality value of <20 were removed. After quality control, the Q20 ratio of all samples was ≥95% and the Q30 ratio was ≥90%, and the retained high-quality clean sequences were used for subsequent bioinformatics analyses. Megahit126 (v1.2.9) was used to perform co-assembly of the quality-controlled sequencing data from multiple samples, generating initial contigs and filtering sequences with a length of ≥300 bp. To improve the capture efficiency of low-abundance species, Bowtie127 (v2.1.0) was used to map the quality-controlled reads back to the initial assembly results. Unmapped reads were separated as new input, and secondary assembly was performed using SPAdes128 (v3.13) to obtain high-quality assembled sequences. Prodigal129 (v2.6.3) was used to identify open reading frames (ORFs) with a length of ≥100 bp from the assembly products and translate them into amino acid sequences. CD-HIT130 (v4.7) was used to cluster and remove redundancy from the predicted gene sequences of all samples; parameters were set at 90% identity and 90% coverage, and the longest gene in each cluster was selected as the representative sequence to construct a non-redundant gene catalog. Salmon131 (v1.5.0) was used to construct gene-specific indexes. Accurate quantification of gene abundance was achieved using a two-phase mapping strategy and systematic bias modeling, and normalized abundance values for each sample were calculated by correcting for gene length. Detailed statistics of metagenomic sequencing depth and core contig assembly parameters are provided (see Table S6). The NR (non-redundant)132,133 database was used for taxonomic annotation of the microbiome. DIAMOND134 (v2.0.13) was employed to perform BLASTP alignment between the amino acid sequences from the non-redundant gene catalog (obtained after gene prediction and redundancy removal) and the NR database. To ensure the reliability of annotation, strict filtering criteria were applied: E-value < 1e-5 and score >60, and high-quality taxonomic annotation results were obtained. We would like to clarify that the present study performed conventional metagenomic analysis without binning or MAG reconstruction. BUSCO135 is designed for assessing the completeness of individual microbial genomes or MAGs. In this study, we only constructed a non-redundant gene set from mixed soil communities without MAG reconstruction; therefore, BUSCO is not applicable for quality control of such a mixed gene set. In this study, we used sequencing depth (clean data size) and contig length cutoffs as the standard quality control criteria for metagenomic functional analysis, which are widely recognized in soil metagenomic studies. These QC metrics are sufficient to ensure the reliability of functional gene analysis. Screening, classification and annotation of nitrogen metabolism genes This study utilized the Kyoto Encyclopedia of Genes and Genomes (KEGG), a resource that systematically integrates genomic, chemical, and biosystem functional data for functional annotation analysis.136 Specific operations were as follows: The full set of reference genomic protein sequences compiled from the KEGG database was first obtained. The DIAMOND (v0.8.20) alignment tool was used to align the target microbiome protein sequences with the KEGG database, and with screening thresholds set as follows: E-value<1e-5, bit score>60. Through this process, we successfully mapped the target sequences to the KEGG Orthology (KO) identifiers. The obtained KO numbers were further associated with the corresponding Pathway and Module via the built-in association mapping system of KEGG. This enabled the functional categorization of target gene sequences in the biometabolic network, with a focus on identifying genes involved in the microbial nitrogen transformation process (see Table S7). Based on the results of this functional annotation, this study quantified the abundance of gene sequences encoding specific microbial functions to provide a basis for subsequent functional resolution. Quantification and statistical analysis Data were processed using Microsoft Excel 2019, analyzed using SPSS 27 (IBM, United States), and plotted using Origin 2024, STAMP 2.1.1.0102, and SmartPLS v4.1.1.2. In this study, online analytical tools from the MajorBio103 cloud platform (https://www.majorbio.com/tools) were used to calculate soil microbial α-diversity indices, including the Simpson index, Shannon index, Ace index, Chao1 index, Simpson evenness index, and Shannon evenness index. The composition of soil microbial communities was also analyzed, focusing on the top 10 dominant phyla of archaea, bacteria and fungi in terms of relative abundance. Data normality was tested using the Shapiro-Wilk test, and homogeneity of variance was assessed via Levene’s test. After the prerequisites were satisfied, one-way analysis of variance (ANOVA) combined with Tukey’s HSD multiple comparison test was applied to examine differences in soil nitrogen fractions among different grazing intensities. Non-metric multidimensional scaling (NMDS) based on Bray-Curtis dissimilarity matrices was used to visualize the similarity of soil nitrogen fractions across grazing intensities. Permutational multivariate analysis of variance (PERMANOVA) was used to test the significance of compositional differences among treatments. The results are reported as R2 and p values. Mantel analysis based on Bray-Curtis dissimilarity matrices was performed to explore the relationships between nitrogen functional genes and soil nitrogen fractions. Meanwhile, generalized linear model (GLM) analysis was conducted to provide auxiliary support for subsequent structural equation modeling. To explore the regulatory mechanism of grazing on soil microbial nitrogen metabolism, partial least squares path modeling (PLS-PM) was established to analyze the direct and indirect pathways among grazing conditions, soil physicochemical properties, microbial diversity and nitrogen metabolic functions. The model was constructed using SmartPLS v4.1.1.2 with a path weighting scheme and 5000 iterations. Model fit indicators, including the variance inflation factor (VIF) and standardized root mean square residual (SRMR), were adopted to evaluate model quality. To quantify the differences in key nitrogen metabolic pathways under different grazing intensities, the abundance variations of functional genes in each pathway were compared. Each grazing intensity group in this study contained n = 3 biological replicates, representing independently collected soil samples. Central tendency of the results is presented as the mean, and data dispersion is expressed as standard deviation (SD) or 95% confidence intervals, with specific presentation forms indicated in the corresponding figure legends. All relevant statistical details are described in detail in the figure legends (Figures 1, 2, 3, 4, 5, and 6). Significance levels are marked with asterisks in the text as follows, ∗: p < 0.05; ∗∗: p < 0.01; ∗∗∗: p < 0.001.