The study area was located in Muli Town, Tianjun County, Qinghai Province (38°09′34″N, 99°09′40″E), with an average elevation of 4,200 meters. The region experiences long winters without summers, with the cold season lasting 7–8 months, significant diurnal temperature variations, and a semi-arid plateau climate. The annual average sunshine duration is 2,160 h, the annual average temperature was −5.3 °C, and the annual average precipitation is 626 mm, with a growing season of only 120 days (Lyu et al., 2024; Hu et al., 2024). The soil is classified as alpine meadow soil, containing significant coal gangue and rock fragments. The soil total nitrogen content is 1.05 g kg⁻¹, total phosphorus content is 0.84 g kg⁻¹, organic matter content is 60.34 g kg⁻¹, electrical conductivity is 19,090 μS cm⁻¹, pH is 8.50, and moisture content is 15.0%. The original vegetation type is alpine meadow, dominated by Carex parvula O. Yano, Carex alatauensis S. R. Zhang, and Elymus nutans Griseb. An overview of the study site is shown in Figures 2a,b.

The microbial biofertilizer used was a composite effective microorganisms (EM) biofertilizer, containing key functional microbial groups such as Bacillus, photosynthetic bacteria, yeast, and lactic acid bacteria (effective viable count ≥ 1.0 × 10⁸ CFU g⁻¹). It was provided by Hubei Qiming Bioengineering Co., Ltd., and complied with the quality standards of Agricultural Microbial Fertilizers (GB 20287-2006). The bio-organic fertilizer, a mineral-derived humic acid type, contains organic matter ≥40% and N + P₂O₅ + K₂O ≥5.0%, supplied by Henan Liso Crop Protection Co., Ltd. The tested grass species included Qinghai Chinese fescue (Festuca sinensis Keng.), Qinghai Kentucky bluegrass (Poa pratensis L.), Qinghai alpine bluegrass (Poa crymophila Keng.), and Tongde short-awn wheatgrass (Elymus breviaristatus Keng f.). These were provided by the Grassland Institute of the Academy of Animal Science and Veterinary Medicine, Qinghai University, with purity ≥92% and germination rate ≥85%.

On May 5, 2022, the tested forage seeds were uniformly mixed and sown in rows at a seeding rate of 18.0 g m⁻². Fertilization was conducted in early June 2023 and repeated during the same period of 2024 for two consecutive years. Based on preliminary research findings from the team, fertilizer application guidelines, and baseline soil conditions (An et al., 2025; Yu et al., 2025a, 2025b; Zhang L. et al., 2025; Zhang X. L. et al., 2025; Zhang Y. F. et al., 2025), nine treatment groups were established for the combined application of organic fertilizer and EM biofertilizer: organic fertilizer application rates of 1.00 kg m⁻², 2.00 kg m⁻², and 3.00 kg m⁻²; EM biofertilizer application rates of 45.00 g m⁻², 60.00 g m⁻², and 75.00 g m⁻² (Table 1). Control group (CK) did not add organic fertilizer and EM bacterial fertilizer. Each experimental plot measures 20 m² (4 m × 5 m) with 2 m inter-plot spacing. Nine fertilization treatments and a CK (control) treatment form one block, with 30 m inter-block spacing. Three replicated blocks were established, totaling 30 experimental plots (see Figure 2c).

Vegetation height, coverage, density, and biomass were surveyed in September 2023 and the same period in 2024. Five randomly placed 0.5 m × 0.5 m quadrats were established within each experimental plot. Vegetation height was measured using a tape measure (1 mm precision) by averaging five individual plants per grass species. Plant density was calculated by counting all stems within each quadrat (the quadrat area is 0.5 m × 0.5 m). Community coverage was determined using the point-intercept method. Aboveground biomass was quantified by harvesting vegetation at ground level, followed by oven-drying at 105 °C for 30 min and drying at 65 °C for 48 h until constant weight. Plant community height, coverage, density, and biomass for the 10 treatments are presented in Table 2.

Soil samples were collected simultaneously, and 0–10 cm soil samples were drilled with 5 cm diameter soil by five-point mixed sampling method. One soil sample is mixed in each plot, which is divided into three parts: one is dry in the shade, which is used to determine the contents of soil organic matter (SOM), total nitrogen (TN) and total phosphorus (TP); One part of wet soil was stored in the refrigerator at −20 °C, and the soil mass moisture content, pH value and conductivity (SEC) were measured. A portion of wet soil was stored in an ultra-low temperature refrigerator at −80 °C for soil microbial diversity analysis.

Data organization was conducted using Microsoft Excel 2016. For plant community characteristics and soil physicochemical properties, statistical analysis was performed using IBM SPSS Statistics 22.0 with one-way analysis of variance (ANOVA). Data are presented as mean ± standard deviation (SD), with statistical significance set at p < 0.05. Homogeneity of variances was tested using Levene’s test. Visualization analysis of soil bacterial communities (including Venn analysis, abundance analysis, LEfSe statistical analysis, diversity analysis, cluster analysis, and FAPROTAX functional prediction) was conducted through the Majorbio Cloud Platform.¹ QIIME 2 software was utilized to generate rarefaction curves, relative abundance plots of species, PCoA plots (Bray–Curtis), hierarchical clustering trees, and calculate α-diversity indices including OTUs, Shannon index, Simpson index, Pielou index, Chao1 index, and ACE index. R software (version 3.3.1) was employed to generate alpha diversity boxplots. To investigate differences in community structure among grouped samples, LEfSe statistical analysis was utilized to detect significant differences in species composition and community structure. The FAPROTAX tool was applied to predict and analyze microbial community functions in ecological samples. Network analysis was conducted based on Spearman correlation coefficients (|r| > 0.6, p < 0.05), and single-factor correlation network diagrams were constructed using NetworkX (version 1.11). R software (version 3.5.2) was used to generate correlation heatmaps between environmental factors, soil physicochemical properties, and microbial diversity. Redundancy analysis (RDA) of environmental factors, soil physicochemical properties, and soil microbial diversity was performed using Canoco 5.0.

The combined application of microbial biofertilizer and organic fertilizer significantly enhanced soil water content (SWC) and nutrient levels (carbon, nitrogen, phosphorus), while decreasing soil pH and electrical conductivity (SEC) (Table 3). In the second year (2023), the Y2E2 treatment produced a soil water content (SWC) of 26.27%, which was significantly higher than the CK treatment (p < 0.05), with an increase of 77.50%. The soil electrical conductivity (SEC) in the Y2E2 treatment was the lowest (818.33 μS cm⁻¹), showing a significant reduction of 58.53% compared to CK (p < 0.05). The pH value in the Y2E2 treatment decreased to 8.05, significantly lower than that of the CK treatment (p < 0.05). The soil organic matter (SOM), total nitrogen (TN), and total phosphorus (TP) contents all reached their peak in the Y2E2 treatment, with values of 253.09 g kg⁻¹, 7.50 g kg⁻¹, and 2.40 g kg⁻¹, respectively. These values represented significant increases of 72.72, 68.92, and 81.82% compared to CK (p < 0.05).

In the third year (2024), the trends in soil property changes remained consistent with the second year, and the improvement effects were even more pronounced. In the Y2E2 treatment, soil water content (SWC) further increased to 28.12%, and soil electrical conductivity (SEC) decreased to 714.00 μS cm⁻¹, both showing significant differences compared to the CK treatment (p < 0.05). The pH value significantly decreased to 7.84, further approaching neutral conditions. The soil organic matter (SOM), total nitrogen (TN), and total phosphorus (TP) contents in the Y2E2 treatment reached 261.44 g kg⁻¹, 7.74 g kg⁻¹, and 2.45 g kg⁻¹, respectively, representing significant increases of 100.60, 76.31, and 92.91% compared to the CK treatment (p < 0.05).

The application of microbial fertilizers introduces exogenous functional microbial communities into the soil, thereby revitalizing the soil ecosystem. When further combined with organic fertilizer application, it enhances organic matter decomposition processes, facilitates the conversion of organic nutrients into readily available forms, and thus improves soil nutrient cycling efficiency (Zhao et al., 2021; Pereg and McMillan, 2015). Multiple studies have demonstrated that the combined application of microbial fertilizers with organic/chemical fertilizers significantly increases vegetation biomass in degraded grassland restoration (Gao et al., 2025; Ren et al., 2021; XU et al., 2025; Ramakrishna et al., 2019). This study found that the combined application of EM bacterial fertilizer and organic fertilizer (Y2E2 treatment) consistently promoted vegetation growth during two consecutive growing seasons, with biomass increasing by 75.97 and 84.02% compared to the control (CK). Concurrently, soil water content (SWC), soil organic matter (SOM), and total nitrogen (TN) content were significantly enhanced. However, a distinct threshold effect was observed in application dosage—excessive application inhibited vegetation growth, which aligns with the findings of Song et al. (2015) and Cai et al. (2018).

Soil nutrients, as the material basis for plant growth, directly govern vegetation development through their effective availability (Jyoti et al., 2024). Data from this study reveal that the Y2E2 treatment synergistically enhanced the physicochemical properties of the 0–10 cm topsoil layer via the combined application of microbial fertilizer and organic fertilizer. Over 2 years, this treatment significantly increased soil water content (SWC), soil organic carbon (SOC), total nitrogen (TN), and total phosphorus (TP) compared to the control, while markedly reducing soil electrical conductivity (SEC) and pH levels. This improvement effect stems from functional bacterial strains enhancing nitrogen and phosphorus availability by accelerating the mineralization process of organic fertilizer (Chen et al., 2025). Additionally, organic acids produced by soil bacterial metabolism promote organic matter decomposition and neutralize alkaline substances through proton release, creating a microenvironment conducive to nutrient activation (Simon et al., 2024). These findings align with studies by Song et al. (2019) and Ku et al. (2018), which confirmed that co-application of microbial and organic fertilizers enhances soil nitrogen reserves and organic matter accumulation. Notably, a threshold effect governs the combined application: excessive use (e.g., Y3E3 treatment) elevates risks of nutrient leaching, reduces nutrient turnover efficiency, and triggers salinization and pH instability (Kong et al., 2013). In this study, Y3E3 exhibited slightly higher pH and SEC than Y2E2, indicating that soil buffering capacity becomes compromised under excessive nutrient inputs (Wang S. et al., 2025; Wang Z. et al., 2025). Chu (2014) further observed that surpassing critical application rates increases soil conductivity, corroborating the physicochemical degradation observed here due to over-application. Therefore, precise regulation of the microbial organic fertilizer ratio is critical for optimizing nutrient conversion efficiency and preserving soil ecological stability (Yang et al., 2024; Xu, 2011; Zhong et al., 2020).

Among the 10 treatments in the Muli mining area, Pseudomonadota and Actinomycetota were the two dominant bacterial phyla with the highest relative abundances in the soil, indicating their critical functional roles in the bacterial communities of this artificial grassland (Yang et al., 2023; Xu, 2011; Zhong et al., 2020). Numerous studies have demonstrated that variations in soil nutrient content significantly reshape bacterial community composition (Yang et al., 2023; Lian et al., 2025; Xu et al., 2021). In the Y2E3 treatment group, Actinomycetota and Acidobacteriota exhibited a synergistic enrichment pattern, with their abundances significantly surpassing those in the CK treatment. The primary reason lies in the functional complementarity of soil bacteria: Actinomycetota bacteria drive soil carbon cycling by decomposing lignin and other complex organic matter, while Acidobacteriota bacteria stabilize the soil carbon reservoir. This functional complementarity between Actinomycetota and Acidobacteriota within soil bacteria enhances carbon conversion efficiency in soils (Lian et al., 2025). Xu et al. (2021) observed that habitats with high phosphorus content and healthy vegetation communities harbor higher relative abundances of Sphingomonas (a genus within Pseudomonadota) and Arthrobacter (a genus within Actinomycetota). This correlation likely stems from their dual nutritional traits: plant growth promotion capabilities; Optimization of soil phosphorus transformation, with their abundances positively correlated with soil phosphorus levels. Consistent with these findings, the current study revealed significantly elevated relative abundances of Sphingomonas and Arthrobacter in fertilized treatments compared to CK, further validating their role in phosphorus cycling and ecosystem health.

Soil bacteria, as the core components regulating soil ecosystem functions, play a pivotal role in governing material cycling and energy flow within plant–soil systems through their community structure and functional traits (Yang et al., 2024). This study found that the combined application of EM bacterial fertilizer and organic fertilizer significantly influenced soil bacterial community diversity. In the Y2E2 treatment group, compared to the control treatment (CK), the number of OTUs (operational taxonomic units) and the Ace index increased by 8.37 and 6.44%, respectively. These findings align with the conclusions of Xu (2011) regarding the combined application of microbial and organic fertilizers. The mechanism is that organic fertilizer provides complex carbon sources, while nitrogen-fixing bacteria and phosphate-solubilizing bacteria in EM microbial fertilizer accelerate organic matter decomposition by secreting extracellular enzymes. These effects create diversified soil niches, thereby enhancing the diversity of bacterial communities in artificially restored grassland soils (Zhong et al., 2020). Further single-factor correlation network analysis revealed that the rhizosphere bacterial network complexity in the Y1E2 treatment group was significantly enhanced compared to CK, with a positive correlation ratio of 67.11%. This indicates that most bacterial populations in this treatment formed a tightly collaborative network through strategies such as mutualistic symbiosis (e.g., nutrient exchange) and cooperative commensalism (e.g., metabolic complementarity). Such a highly connected bacterial structure optimizes rhizosphere nutrient turnover efficiency and accelerates the synchronized metabolic activities of nitrogen-fixing and phosphorus-solubilizing bacteria (Che et al., 2024; Zhang et al., 2025; Zhang et al., 2025; Zhang et al., 2025).

Li et al. (2025) demonstrated that alterations in soil bacterial community structure led to distinct functional characteristics. This study corroborates their findings through FAPROTAX-based predictions of bacterial functions across different fertilization treatments, revealing significant functional divergence among the 10 treatments. Notably, the Y1E1 and Y2E2 treatments exhibited markedly higher manganese oxidation functional intensity compared to other groups. This enhancement likely stemmed from the abundant photosynthetic bacteria in EM fertilizer, which establish a unique photoenergy-manganese oxidation coupled metabolic pathway via light-driven CO₂ fixation (Wang S. S. et al., 2025; Wang Z. X. et al., 2025; Cai et al., 2025; Cohen et al., 2022). The concurrent significant increase in chloroplast-related functional intensity in these treatments further supports this hypothesis. Critically, the Y3E3 treatment group showed suppressed functional intensities in chlorate reduction, iron respiration, and phototrophic autotrophy, underscoring that excessive combined application may inhibit specific bacterial metabolic functions, reinforcing the existence of a threshold effect (Zhong et al., 2020; Yang et al., 2023). The functional intensity of photosynthetic autotrophic-related functions in soil bacteria under the control treatment (CK) was significantly lower than that under other fertilization treatments. The potential reason may be attributed to the chronic deficiency of organic matter input in the in situ soil of the Muli mining area (Qinghai Province, China), which has resulted in a severe shortage of substrates (e.g., sulfides, low-molecular-weight organic compounds) required for photosynthetic bacteria growth, thereby constraining their metabolic activity.

Mantel test analysis revealed that soil pH exhibited an extremely significant positive correlation with SEC (p < 0.001), while both variables demonstrated extremely significant negative correlations with other soil physicochemical properties and vegetation community characteristics (p < 0.001). These findings align with research by Gao et al. (2025), which posits that higher soil nutrient and moisture levels promote superior plant community growth. Redundancy analysis (RDA) further demonstrated that soil SEC was negatively correlated with bacterial community diversity indices (Shannon, Ace, and Chao1), indicating that soil salinity stress significantly inhibits the recovery of bacterial diversity (Luo et al., 2025). In contrast, vegetation community traits and soil fertility indicators soil organic matter (SOM), total nitrogen (TN), and total phosphorus (TP) exhibited significant positive correlations with bacterial diversity indices (except the Simpson index). This suggests that organic matter and mineral nutrients (N, P) enhance bacterial community structural succession and functional differentiation by supplying energy and metabolic substrates, thereby fostering diversity restoration. In summary, the vegetation restoration process in the Muli mining area of Qinghai achieves bidirectional enhancement of soil fertility improvement and bacterial community optimization through the “vegetation-soil-microbe” interaction mechanism, ultimately establishing a self-reinforcing ecological virtuous cycle.

Soil physicochemical properties exert a significant shaping effect on the diversity and structure of soil bacterial communities in artificial grasslands within the Muli mining area of Qinghai Province. Unlike traditional studies that prioritize soil nutrient availability as the core driver of bacterial community distribution, this study highlights the dominant regulatory role of pH (Liu et al., 2018). This finding is highly consistent with the “rugby ball model” proposed by Xun et al. (2015). The model posits that when soil pH falls within the neutral range (6.5–7.5), soil nutrient conditions primarily govern changes in bacterial community structure and functional dynamics. However, when pH shifts to acidified (5.5–6.5) or alkalized (7.5–8.5) ranges, pH value becomes the primary controlling factor influencing bacterial community structure and functional dynamics. In the study area, soil pH ranged from 7.84 to 8.82 (characteristic alkaline conditions), consistent with their model. Thus, pH not only influences rhizosphere microbial communities in grasslands through soil fertility but also regulates bacterial community stability via plant-microbe interaction networks. For ecological restoration practices, adjusting soil pH to the neutral range (6.5–7.5) can maximize bacteria-driven ecological services such as carbon cycling and nutrient transformation, optimizing ecosystem functionality.