The study area is located in the Xinjiang Autonomous Region in northwest China (34°25’–48°10’N, 73°40’–96°18’E), which is deep inland and far from the ocean. It has a temperate continental climate with a large temperature difference, sufficient sunshine (annual sunshine time is 2500–3500 h), reduced precipitation, and an annual average precipitation of about 150 mm. The precipitation varies significantly from place to place. The temperature in southern Xinjiang is higher than that in northern Xinjiang, and the precipitation in northern Xinjiang is higher than that in southern Xinjiang. The average temperature in January in Junggar Basin is –20°C; the average temperature in July in Turpan is 33°C.

To construct a comprehensive database of grazing intensity effects on Xinjiang grasslands, we collected peer-reviewed publications from January 2000 to December 2024 using the Web of Science (http://apps.webofknowledge.com) and the China Knowledge Resource Integrated Database (http://www.cnki.net). The search term combinations were: “grazing or herbivory or fencing”, “Xinjiang grassland or Xinjiang grassland degradation”, and “grassland type”. Next, we screened the publications to identify appropriate studies based on the following criteria: (1) The selected article had to study the impact of grazing on grassland in Xinjiang. The study also had to provide the effects of grazing intensity, grazing mode, grassland degradation, and related variables on vegetation productivity, vegetation characteristics, plant productivity, diversity index, soil quality, carbon storage, and plant production. (2) The article had to provide information such as latitude and longitude coordinates, altitude, precipitation, test duration, and the degree of grassland degradation in the study area. (3) The article had to contain a control group (no grazing) and an experimental group (grazing), and data for the control group and the experimental group had to include sample size, mean, and standard deviation (SD) or standard error (SE). Some of the data could be directly extracted from the content and tables in the article, and some of the data in the figure needed to be extracted by using WebPlotDigitizer software. For data that provided only SE and sample size (n), we calculated the SD according to the following formula:

We identified 73 journal papers that met the noted criteria ( Figure 1 ). For the grassland type division and distribution map, see reference Zhang et al. (2020a). The distribution of the 73 sample plots collected was as follows: 14 temperate meadow steppes, 7 temperate steppes, 16 temperate desert steppes, 22 mountain meadows, and 14 alpine meadows. After extracting the data and removing duplicates or statistically insignificant data from each study, we created a database containing 1876 groups was created. The main indicators were as follows: bulk density (BD), soil water content (SWC), total carbon (TC), soil organic carbon (SOC), organic carbon density (OCD), soil organic matter (SOM), total organic matter (TOM), pH, total nitrogen (TN), total phosphorus (TP), total potassium (TK), available nitrogen (AN), available phosphorus (AP), available potassium (AK), carbon-to-nitrogen ratio (C:N), carbon-to-phosphorus ratio (C:P), nitrogen-to-phosphorus ratio (N:P), community vegetation height (CVH), community vegetation cover (CVC), community vegetation density (CVD), root-to-stem ratio (R:S), total biomass (TB), aboveground biomass (AGB), belowground biomass (BGB), aboveground standing crop (ASC), belowground standing crop (BSC), Shannon–Wiener index (SWI), Simpson index (SI), Pielous index (PI), species richness (SR), Patrick index (PAI), Margalef index (MI), Chao index (CI), carbon storage in live grassland biomass (CSLGB), litter carbon storage (LCS), root carbon storage (RCS), total carbon storage (TCS), and dry matter (DM). We divided grazing intensity grading standards into three levels according to Reynaud and Lanzanova (2017), as shown in Table 1 : light grazing, moderate grazing, and heavy grazing.

In this paper, we used the response ratio (RR) of the natural logarithm proposed by Hedges et al. (1999) to evaluate the effects of grazing intensity on soil physical and chemical properties and vegetation community characteristics of various grassland types in Xinjiang. We defined RR as the natural logarithm of the ratio of the mean value of the index in the test group (X_e ) to the mean value of the index in the control group (X_c ). The calculation formula is as follows:

The variance (v) of the RR is calculated as follows:

In the formula, N_e and N_c are the sample sizes of the experimental group and the control group, respectively, and SD_e and SD_c are the standard deviations of the indicators in the experimental group and the control group, respectively. The weighting coefficient (w) of each response ratio is the reciprocal of the variance (v). The calculation formula is as follows:

To improve the accuracy of the study, we also calculated the average weighted response ratio (RR₊₊ ) of each experimental group and the control group. The higher the accuracy of the study, the greater the weight assigned. The calculation formula is as follows:

In the formula, w_ij is the weighting coefficient of jth variable in i group, m is the number of groups, and k is the number of pairs of indicators in the i group (one value of the indicator in the experimental group is paired with one value of the target variable in the control group). The calculation formula of the weighted standard deviation is as follows:

We used the 95% confidence interval (CI) to test whether the weighted response ratio of the indicators to the treatment in the experimental group was significant. If the 95% CI overlapped with 0, the response ratio of the index was not significant; conversely, the response ratio of the indicator was significant (Xu et al., 2023). When the maximum value of the 95% CI was less than 0, this indicated that grazing had a significant negative effect on the effect value of typical grassland types in Xinjiang. If the minimum value of the 95% CI was greater than 0, this indicated that grazing had a significant positive effect on the effect value of typical grassland types in Xinjiang. The calculation formula is as follows:

We used the “Metafor” package in R 4.4.3 software to calculate the response ratio and 95% CI, and we used the “ggplot2” and “Forestplot” packages to draw the forest map. We selected some indices for a correlation analysis “Pearson” package. We established structural equation modeling (SEM) using the “lavaan” and “piecewiseSEM” packages to study the effects of grazing intensity on community species diversity (SWI, SI, PI and SR) and soil quality (SOC, TN, TP, TK, AN, AP, and AK.).

Grazing had significant negative effects on soil total carbon (TC) (−15.93% ± 0.0579 , the percentage of the response ratio, the same below), organic carbon density (OCD) (−14.4% ± 0.0312 ), soil total potassium (TK) (-7.06% ± 0.0284 ), community vegetation height (CVH) (−64.98% ± 0.0493 ), community vegetation coverage (CVC) (−32.5% ± 0.0401 ), community vegetation density (CVD) (−26.52% ± 0.048 ), aboveground biomass (AGB) (−57.32% ± 0.0599 ), total biomass (TB) (−69.42% ± 0.1461 ), belowground biomass (BGB) (−16.91% ± 0.0425 ), aboveground standing crop (ASC) (−69.11% ± 0.0799 ), belowground standing crop (BSC) (85.66% ± 0.1224 ), carbon storage in live grassland biomass (CSLGB) (−56.68% ± 0.0849 ), litter carbon storage (LCS) (−46.29% ± 0.0524 ), Shannon-Wiener index (SWI) (−10.21% ± 0.0483 ), Simpson index (SI) (−8.36% ± 0.0325 ) and species richness (SR) (−16.82% ± 0.0477 ). We identified a significant positive effect on soil available phosphorus (AP) (11.85% ± 0.0558 ) and dry matter (DM) (0.61% ± 0.002 ) ( Figure 2 ).

As shown in Figure 2 , compared with the control group (no grazing), soil TC decreased significantly by 10.21% ± 0.0287 and 23.25% ± 0.0973 in light and moderate grazing, respectively; SOC decreased significantly by 11.42% ± 0.036 in heavy grazing; OCD decreased by 6.52% ± 0.0188 and 27.2% ± 0.0464 in moderate and heavy grazing, respectively; soil TN increased significantly by 14.24% ± 0.0438 in moderate grazing; soil TK decreased significantly by 10.01% ± 0.0346 in moderate grazing; and soil AP increased significantly by 21.97% ± 0.0859 in heavy grazing. C:P decreased significantly by 16.3% ± 0.0823 and 14.24% ± 0.0697 in light and heavy grazing, respectively, and N:P decreased significantly by 36.18% ± 0.0892 and 25.26% ± 0.0977 in light and heavy grazing, respectively. CVH decreased significantly by 39.87% ± 0.0509 , 72.06% ± 0.1169 , and 92.8% ± 0.0873 in the three grazing intensities. CVC decreased significantly by 13.51% ± 0.0325 , 27.82% ± 0.0717 , and 56.13% ± 0.0828 in the three grazing intensities. CVD was significantly reduced by 23.01% ± 0.0497 and 42.96% ± 0.1155 in light and heavy grazing, respectively. AGB decreased significantly by 32.92% ± 0.061 , 49.25% ± 0.1495 , and 85.84% ± 0.1049 in the three grazing intensities. BGB was significantly reduced by 9.39% ± 0.0457 , 17.08% ± 0.0742 , and 22.4% ± 0.0822 in the three grazing intensities. TB was significantly reduced by 64.43% ± 0.1729 and 92.94% ± 0.3407 in moderate and heavy grazing, respectively. CSLGB was significantly reduced by 23.07% ± 0.0217 , 49.73% ± 0.0271 , and 97.45% ± 0.0644 in the three grazing intensities. LCS significantly decreased by 21.55% ± 0.0121 , 50.55% ± 0.0137 , and 66.95% ± 0.0449 in the three grazing intensities. RCS significantly increased by 39.23% ± 0.1039 in light grazing and significantly decreased by 87.74% ± 0.092 in heavy grazing. SWI significantly decreased by 21.67% ± 0.0948 in moderate grazing, and SI significantly decreased by 9.67% ± 0.0445 in moderate grazing. PI decreased significantly by 16.09% ± 0.0785 in moderate grazing, and SR decreased significantly by 38.51% ± 0.0989 and 19% ± 0.081 in moderate and heavy grazing, respectively, and CI decreased significantly by 13.68% ± 0.0295 in heavy grazing.

By integrating the similarities and differences of the effects of different grazing intensities on various grassland types ( Table 2 ), grazing reduced the vegetation community height of various grassland types as a whole. In addition, and heavy grazing reduced the vegetation community height, coverage, and aboveground and underground biomass of various grassland types. Light grazing decreased the SOC of temperate steppe and increased the SOC of temperate desert steppe. Moderate grazing reduced the AGB of mountain meadow and increased the AGB of alpine meadow. Moderate grazing decreased SOC in temperate meadow steppe and increased SOC in temperate desert steppe.

The interaction effects of grassland type and grazing intensity on all key indicators were significant (P < 0.05; Table 3 ), confirming that the impact of grazing had a strong ecosystem dependence. The temperate meadow steppe was highly sensitive to the loss of soil organic carbon (SOC: β = −0.42, P = 0.01) and aboveground biomass (AGB: β = −1.2, P < 0.001). The temperate steppe showed significant changes in soil physical properties, including increased pH under light grazing (β = 0.3, P = 0.03) and increased bulk density under heavy grazing (β = 0.15, P = 0.04). The temperate desert steppe was the only ecosystem with a significant increase in SOC under moderate grazing (β = 0.28, P = 0.02), but its biodiversity (SI: β = −0.8, P = 0.001) was severely damaged. AK in mountain meadows was significantly accumulated under light and moderate grazing (β = 0.35, P = 0.01), while heavy grazing induced the allocation of resources to the underground (R/S: β = 0.25, P = 0.03). The alpine meadow showed a contradictory response: although moderate grazing increased aboveground biomass (AGB: β = 0.4, P = 0.04), it decreased total biomass (TB: β = −0.6, P = 0.02), which suggested a transformation of community structure.

According to the correlation analysis of vegetation characteristics and diversity index, a significant positive correlation existed between CVH and CVC (P < 0.001), and there was a significant positive correlation existed between CVC and AGB (P < 0.01). BGB was significantly positively correlated with SI (P < 0.001) and SR (P < 0.05). SWI was significantly positively correlated with PI (P < 0.001) and SR (P < 0.01). SI was positively correlated with SR (P < 0.001) ( Figure 8 ). These results demonstrated that an increase in grassland vegetation height promoted the expansion of coverage and then an increase in AGB. The increase in BGB promoted the increase in SR and diversity index.

The influence path of different grazing intensities on soil quality and species diversity in Xinjiang grassland is shown in Figure 9 . Grazing intensity is significantly negatively correlated with CVC (P < 0.05), and CVC is significantly positively correlated with species diversity (P < 0.001). Therefore, grazing intensity affected species diversity by affecting vegetation coverage. The greater the grazing intensity, the smaller the vegetation coverage, which resulted in a decrease in species diversity. In addition, grazing intensity was significantly negatively correlated with AGB (P < 0.05) and soil quality (P < 0.001). Increasing grazing intensity led to a decrease in AGB and a decrease in soil quality.

Grazing intensity significantly affected the physical and chemical properties of the soil. With the increase in grazing intensity, the BD of the soil surface (0–10cm) usually increased, and the soil became compact. The results of this study showed that light and moderate grazing reduced soil BD in Xinjiang grassland, whereas heavy grazing increased soil BD, which was similar to the results of Zhang et al. (2002). SWC decreased significantly under moderate and heavy grazing conditions. It is possible that livestock frequently trampled the grassland, compacted the soil, reduced the porosity, reduced the storage space of air and water in the soil, and hindered the infiltration and retention of water, thus leading to a decrease in SWC.

The soil pH of the temperate desert steppe in Xinjiang increased and decreased significantly under light and moderate grazing, respectively. The reason may be that light grazing reduced CVC and accelerated soil moisture evaporation. The soluble salts in the soil may have gradually accumulated on the surface, thus increasing soil pH. Moderate grazing accelerated the decomposition of SOM and released more organic acids. These organic acids would neutralize the alkaline substances in the soil, thereby reducing the soil pH (An and Xu, 2013). SOC, TC, and TN decreased significantly under heavy grazing. SOC decreased under light and moderate grazing conditions in two types of grassland in Xinjiang temperate meadow steppe and temperate steppe, but the SOC content in temperate desert steppe and mountain meadow increased significantly under light and moderate grazing. This may have been the result of a decrease in CVC under light and moderate grazing conditions in temperate meadow steppe and temperate steppe, which resulted in a decrease in plant residue input, a corresponding decrease in SOC accumulation, a destruction of soil structure, and a decrease in porosity. This, in turn, may have affected the soil microbial activity and organic carbon fixation (Ding et al., 2014). The vegetation types of temperate desert steppe and mountain meadow usually have strong adaptability and can better cope with grazing pressure. Li et al. (2008) have shown that plants such as Stipa breviflora in desert steppe can maintain a good growth state under light and moderate grazing, and their root exudates and litter input contributed to the accumulation of SOC. TP in alpine meadow soil increased significantly under light grazing, which was similar to the results of Wu et al. (2024b). It is possible that grazing changed the growth strategy of plants, further increasing plant root exudates. These exudates promoted the dissolution and release of phosphorus in the soil, which increased the TP content in the soil.

Studies have shown that grazing is an important factor in the dominant community structure when the climatic conditions are basically the same (He et al., 2021). The results of this study showed that the CVH, CVC, and CVD in various grassland types in Xinjiang decreased significantly under different grazing intensities, which was consistent with the research results of Zhang et al. (2021) on mountain meadow in Xinjiang that is, CVD, CVH, and CVC under grazing conditions were lower than those in enclosed areas. The study of S. purpurea grassland in northern Tibet by Duan et al. (2010) also found that high-intensity grazing behavior significantly reduced CVC, which indicated that the trampling and grazing of livestock under grazing conditions affected the growth of CVD, CVH, and CVC. The results of this study also showed that grazing reduced the AGB, BGB, and TB of different grassland types in Xinjiang, which was consistent with the results of previous studies (Wang et al., 2009; Yu et al., 2015). We determined that grazing reduced the number of plants and the total photosynthetic area of plants, thus resulting in a decrease in the efficiency of organic matter accumulation. Compared with no grazing, the R:S of heavy grazing was greater, which indicated indicating that grazing gradually reduced the that TB of the community and the TB of the community gradually trended downward (Zhang et al., 2020b). Litter biomass also decreased with the increase in grazing intensity, because grazing accelerated the decomposition and removal of litter (Zhang et al., 2020b).

Light grazing had higher community species diversity than heavy grazing. The results of this study showed that light and moderate grazing increased SR and SWI in temperate steppe of Xinjiang, whereas moderate grazing in mountain meadow and alpine meadow decreased SR, SWI, and SI. It is possible that moderate grazing disturbance broke the single advantage of vegetation and provided growth space for grazing-tolerant or less palatable plants, thereby increasing species diversity. This occurred because intermediate disturbance not only inhibited the competitive exclusion by dominant species but also provided survival opportunities for some rare species (Xiang, 2017). Mountain and alpine meadow ecosystems are more sensitive to grazing disturbance. Moderate grazing was sufficient to have had a greater impact on vegetation community structure, which resulted in a decrease in species diversity. Mountain and alpine meadows usually have lower vegetation productivity and slower recovery ability. Therefore, even moderate grazing may have had a negative impact on species diversity (Cai et al., 2024). In the study of alpine meadow in eastern Qilian Mountains, Wang et al. (2019) found that the SR and SWI were significantly higher in light grazing than those under other grazing intensities, while moderate and heavy grazing reduced these diversity indexes.