Introduction

Soil organic carbon storage is a key function of soils that can not only mitigate climate change and ameliorate its consequences on a global scale, also has some influence on other soil functions1,2. As the largest terrestrial carbon pool, soil stores more carbon than flora and the atmosphere3,4.

Soil also provides various nutrients for the growth of vegetation. As an important part of the terrestrial ecosystem, grassland covers about 30% of the global terrestrial ecosystem area and stores 10% ~ 30% of the global SOC, which plays an important role in the terrestrial ecosystem carbon cycle5. Carbon sequestration in most grasslands is very sensitive to climate change, which has diverse effects on the accumulation and stability of soil organic carbon6.

The main use of grasslands is grazing, which has different effects on plant and soil properties due to livestock trampling, foraging, and excretion by different breeds of animals7. Scientific ways of grazing will enhance the sustainability of the grassland ecological system, it is important to maintain the ecological balance of the grass8,9. Relevant research shows that compared to HC grazing, Adaptive Multi-Paddock (AMP) grazing was found to increase soil organic carbon (SOC), average annual SOC increased by 7.5%10. Livestock grazing intensity (GI) is thought to have a major impact on soil organic carbon (SOC) storage and soil quality indicators in grassland agroecosystems. Grazing increases SOC for C4 but decreases it for C3 and C3-C4 mixed grasslands11. While low levels of grazing may promote SOC, studies from livestock systems show that intense grazing generally reduces SOC12. pasture management practices can increase soil carbon stocks13. Well-managed grazing is important to increase soil carbon14,15,16. Future climate change will reduce SOC stocks in grasslands if appropriate measures are not taken to maintain SOC stocks17,18. Therefore, under the background of "carbon peaking" and "carbon neutrality" proposed by the Chinese government, modeling and simulating the effects of grazing intensity on SOC density on the scale of grassland ecosystem is not only vital for the study of the regional carbon cycle19, but also of great significance for quantitative control of some management measures to promote regional sustainable development. At the same time, it can provide scientific basis and data support for regional carbon management of climate change mitigation and adaptation. In this study, the CENTURY model was used to study the changes of soil organic carbon density under different grazing intensities at each sampling point, while other factors remained unchanged. In the EVENT.100 program of the CENTURY model, four gradients were set for each sampling point: no grazing (CK), light grazing (LG), moderate grazing (MG) and heavy grazing (HG), and then the model results were analyzed.

Materials and methods

Study area

Zhangye is located in the northwest of Gansu Province and the middle of the Hexi Corridor, China. It has been the choke point of the Silk Road since ancient times. The geographical coordinates are 97°20′E ~ 102°12′E, 37°28′N ~ 39°57′N, with an east–west span of 4°52′and a north–south span of 2°29′. Bordering Qinghai Province in the south and Inner Mongolia in the north, with an average altitude of 3633 m. Bordering Qinghai Province to the south and Inner Mongolia to the north, this region has an average altitude of 3633 m. It has a temperate continental arid climate with hot summers and unevenly distributed, scarce rainfall. The annual average temperature is 7.8 °C, with significant temperature variations and annual precipitation of 174.9 mm. The total area of Zhangye is 38,600 km2, and the grassland is 19,286.3155 km2, accounting for 49.96% of the land area. Among them, the natural pasture is 7020.3738 km2, accounting for 36.40% of the total grassland area. The grassland is mainly distributed in Sunan Yugur Autonomous County, accounting for 69.09%.

Data preparation and sources

Field data

Soil water content, soil bulk density, pH value, aboveground biomass, and underground biomass were obtained through field experiments in 2022 and 2023. According to the grassland type and its distribution, 33 sampling points (Fig. 1) were randomly set in proportion. At the same time, parameters used in the model localization process were calculated using the 15% accuracy requirement of VM0026 sustainable grassland management. The result is 8.62%, less than 15%, indicating sufficient sample points. GPS navigator was used to reach the sampling points to determine the latitude, longitude, and altitude of the sampling points. Each sampling point was set up with 3 repetitions. Soil samples of 0 ~ 10 cm, 10 ~ 20 cm, and 20 ~ 30 cm were taken from top to bottom with a ring knife. Each sample point was repeated three times. To avoid water evaporation, all the soil samples in the ring knife were transferred to the zipped bag and brought back to the laboratory to determine the soil water content, soil bulk density, and pH value20. The 50 g collected soil sample was put into a weighed clean aluminum box, and then the aluminum box was dried in an oven at 105℃ to constant weight, and then the soil moisture content was calculated. After the soil is taken by the weighed ring knife, the soil inside the ring knife is converted to the weight of the dried soil according to the soil moisture content. Soil bulk density is equal to the weight of the dried soil inside the ring knife divided by the volume of the ring knife. pH was determined by glass electrode method21. The topsoil at the sampling point was removed, and the soil drill was vertically inserted into the soil and collect 0–10 cm, 10–20 cm, and 20–30 cm soil samples from top to bottom with three replicates taken from each layer. The soil samples of the same level in each soil profile were mixed evenly, and 200 g of the mixed samples were first put into a zip lock bag and labeled, and then brought back to the laboratory to determine the SOC content by the potassium dichromate oxidation method. Soil organic carbon density is used in grams per square centimeter (g/cm2) to describe the organic carbon storage of surface soil per unit area, and is often used in ecological research, agriculture, and land management. Models that predict carbon dynamics and sequestration often use soil organic carbon density because it simplifies the integration of data from different sources and scales, making it easier to forecast changes in carbon storage under various scenarios.

Figure 1
figure 1

Location of Zhangye and distribution of experimental sampling points. Black marks the sampling points, and color indicates altitude.

Full size image

Basic data

Remote sensing maps of temperature and precipitation during 1970—2022 were downloaded from Google Earth engine and the WorldClim website (https://www.resdc.cn). ArcGIS was used to extract the monthly meteorological data of the research sites, including the monthly average maximum temperature, monthly average minimum temperature, and monthly average precipitation of each station, for the localized debugging of model parameters. Data on grassland vegetation types were downloaded from the Resources and Environmental Sciences and Data Center of the Chinese Academy of Sciences (https://www.resdc.cn) (Fig. 2).

Figure 2
figure 2

Vegetation type diagram of Zhangye grassland. The grassland types in Zhangye City mainly include desert, grassland and meadow, of which desert and grassland occupy a large area.

Full size image

The soil texture data of Zhangye grassland were obtained from the World Soil Database(https://www.fao.org/soils-portal/data-hub/soil-maps-and-databases/harmonized-world-soil-database-v12/en/), which contains information about soil type, phase, physical and chemical properties of each grid dot. ArcGIS was used to study sites and extract soil texture information, including clay, silt, sand content, and wilting point.

Methods

CENTURY model

The CENTURY model22 is a step for month-long biogeochemical Model, with continuous optimization, the model can not only be used to simulate the fluxes of C, N, P and S, but also for the simulation of farmland, grassland, forest and other ecosystems, has been verified and applied in many countries and regions in the world23,24,25,26,27. The CENTURY model divided the soil total organic carbon pool according to the decomposition rate of soil organic matter. This model could output the density of the total organic carbon pool and its three component pools (active organic carbon pool, slow organic carbon pool, and inert organic carbon pool).

The model includes 3 main programs and 12 parameter data files. The 3 main programs are FILE.100, EVENT.100 and LIST.100 respectively, and the 12 parameter files include meteorological data, site files, management measures, vegetation parameters and EVENT files, etc. The 12 parameter files are controlled by the FILE.100 program, which is mainly used to create and update input parameters; EVENT.100 is mainly used to create scheduling files, including setting vegetation types and debugging events that occur during the simulation, such as fires. Some of the parameter inputs of the CENTURY model are user-selected inputs, such as site information files, including geographical location, soil texture, etc. Meteorological data (.wth) includes monthly precipitation, monthly maximum and minimum temperature, etc. Some of them can choose the system default parameters. For unknown data and some data that is difficult to obtain, the system default values can be used. However, in order to reduce errors, local data should be found as much as possible to update parameters. In this study, based on the database of soil total organic carbon and its three components, combined with the collected data and the field data, we localized the parameters of the model. The main adjustment parameters include latitude and longitude of the site, climate parameters (monthly mean maximum temperature, monthly mean minimum temperature, monthly mean precipitation, etc.), soil parameters (soil bulk density, pH value, soil sand content, clay content, silt content, soil wilting point, field water capacity, etc.) (Table1) and default values used for crop management parameters (crop species, planting time, crop start time, crop end growth Time, harvest method, etc.)28.

Table 1 Localization of some key parameters of the CENTURY model.
Full size table

Statistical analysis

The common Kriging interpolation method in geostatistics was applied to analyze the spatial variation structure of grassland SOC by spatial interpolation, which was operated in GS + 9.0 and ArcGIS 10.8.

Results

Model validation

The CENTURY model was verified by the measured SOC density from the field experiment in 2022 and 2023. Linear regression analysis was performed on the simulated data and the measured data (Fig. 3).

Figure 3
figure 3

Linear regression of measured and simulated value in the study area. The black dashed line is the linear regression line between the measured value (Y-axis) and the simulated value (X-axis), and its equation is: y = 0.8737x + 687.35(R2 = 0.8998), and the red dashed line is the 45° auxiliary line.

Full size image

Simultaneously performing cross-validation, we obtained the following results (Table 2): RMSE = 1333.568, indicating an average difference between the predicted and actual values of approximately 1333.568, which accounts for about 7.41% of the data range. MD = 562.945, indicating that the predicted values are on average 562.945 higher than the actual values, accounting for 3.13% of the data range. R = 0.9541206, indicating a strong positive correlation between the predicted and actual values. RE = 0.06933035, indicating that the average prediction error accounts for about 6.93% of the actual values. The relatively low relative error suggests that the model performs well in controlling errors. LOF = 0.1079755, indicating that the model fits the data well and can explain most of the data's variability. These results indicate that t the CENTURY model fitted well after parameter adjustment and correction, and effectively controlled the error, and could be used to simulate the SOC density in Zhangye Grassland.

Table 2 Results of cross-validation of CENTURY model results model root-mean-square error, mean difference, correlation coefficient, relative error lack of fit, and bias.
Full size table

Temporal changes of SOC density under different grazing intensities

Relative change of SOC density in different years

Firstly, the relative changes between different grazing intensities and no grazing were studied. Figure 4 shows the interannual variation trend of the relative changes of soil total and each component library organic carbon density in Zhanye grassland under different grazing intensities. It can be seen that different grazing intensities will cause a different degree decrease in soil organic carbon density. Under the same grazing pressure for 53 years, the average soil organic carbon decreased by 195.114 g·m−2 and 1.91%, compared with the initial grazing. Moderate and heavy grazing decreased by 533.387 g·m−2 and 1811.11 g·m−2, respectively, with a loss of 5.21% and 17.69%. It can be found that long-term heavy grazing will cause a serious reduction of soil total organic carbon. In the short term, the total SOC under mild and moderate grazing will first decrease and then increase, even higher than that under no grazing, and the increased range of moderate grazing is greater than that under light grazing. However, with the increase of grazing years, the decrease of soil total organic carbon density under moderate grazing is faster than the light grazing, and shows a decreasing trend relative to no grazing. In the short term, heavy grazing can also make soil total organic carbon decrease first and then increase slightly, but compared with no grazing, it has been in a decreasing trend, and the reduction is significant.

Figure 4
figure 4

Relative changes of SOC density in Zhangye under different grazing intensities. (a–d) are the changes of total soil organic carbon, active soil organic carbon, slow soil organic carbon and inert soil organic carbon during 1971–2022, respectively, LG, MG, and HG represent light, moderate, and heavy grazing treatments.

Full size image

According to the sub pool of SOC, the fluctuation of active SOC density was obvious (Fig. 4b). Light grazing and moderate grazing increased the density of active SOC in the first 10 years, and the increase of light grazing was larger, which could increase the density of active soil organic carbon by 22.11% and 7.77%, respectively. However, as grazing continues, the SOC density will decrease rapidly, and then it will increase slightly after that, but it is always in a decreasing state compared with the non-grazing state, and its fluctuation amplitude is relatively weakened after 20 years of grazing. Compared with the initial grazing, the loss of active SOC under light and moderate grazing in 2022 was 3.65% and 17.07%, respectively. After 5 years of heavy grazing, active SOC showed an obvious downward trend, and after 10 years of grazing, it would decrease sharply. 53 years of continuous heavy grazing resulted in 37.46% loss of active SOC.

The changing trend of slow SOC (Fig. 4c) under different grazing pressures was similar to that of total SOC. The density of slow SOC increased slightly in the early stage of mild and moderate grazing; After 13 consecutive years of grazing under moderate grazing pressure, slow SOC showed a downward trend compared with the initial period; however, under the condition of heavy grazing, the change was not obvious at the beginning, but decreased sharply after 20 years of continuous grazing. Compared with the initial grazing stage, continuous grazing for 53 years under different grazing pressures resulted in 4.24%, 11.96% and 24.37% loss of slow SOC, respectively, the degree of loss is relatively severe. Inert SOC (Fig. 4d) showed different trends under different grazing pressures. In general, light grazing would increase the inert SOC density, and the increasing trend has become more obvious since 1990. Compared with the initial grazing, light grazing increased the inert SOC density by 0.255%; The change trend of moderate grazing was similar to that of light grazing, but its relative change showed a decreasing trend in general, and the inert SOC density lost 0.014% compared with the initial grazing; Under heavy grazing, the relative changes showed a downward trend, and with the increase of grazing years, the loss of inert SOC density became more serious. Compared with the initial grazing, the loss of inert SOC density was 0.328%.

Relative change of SOC density in different seasons

Since there are differences between the growing season and the aging season of various crop types, different grazing intensities and grazing times also have an impact on SOC density, the seasonal changes of SOC density in Zhangye grassland were studied. Figure 5 shows the histogram of seasonal differences of soil total organic carbon and its component library densities under different grazing intensities.

Figure 5
figure 5

Seasonal variation of SOC density at different grazing intensities. (a–d) were the seasonal changes of soil organic carbon under different grazing intensity, and LG, MG and HG represented light, moderate and heavy grazing treatment, respectively. SOMC, SOM1C, SOM2C and SOM3C stand for total soil organic carbon, active soil organic carbon, slow soil organic carbon and inert soil organic carbon respectively.

Full size image

As can be seen from Fig. 5a, soil total organic carbon density under different grazing intensities was higher in the first and fourth quarters, and lower in the second and third quarters. In the first quarter, different grazing intensities resulted in the decrease of soil total organic carbon density by 4.38%(LG), 6.36%(MG) and 18.29%(HG) respectively; In the second quarter, soil total organic carbon density decreased by 3.21%(LG), 6.67%(MG) and 17.26%(HG) respectively; In the third quarter, soil total organic carbon density decreased by 3.38%(LG), 6.87%(MG) and 17.49%(HG) respectively; And in the fourth quarter, soil total organic carbon density decreased by 2.28%(LG), 5.89%(MG) and 16.73%(HG) respectively. It can be seen that under each grazing intensity, the loss of total organic carbon density was the lowest in the fourth quarter, the difference in losses between the second and third quarters is not significant, while the loss was the largest in the first quarter under light grazing and heavy grazing.

Figure 5b shows that the active organic carbon density under different grazing intensities was the highest in the fourth quarter and the lowest in the second quarter. The highest loss of organic carbon density in active soil was 21.43 g·m−2 (LG), 72.82 g·m−2 (MG) and 149.08 g·m−2 (HG) under different grazing intensities in the first quarter; The lowest losses were 12.54 g·m−2 (LG), 65.85 g·m−2 (MG), and 142.69 g·m−2 (HG) in the fourth quarter. Under different grazing intensities, both slow SOC density and inert SOC density increased gradually from the first quarter to the fourth quarter, but the inert SOC density increased slightly. The maximum increment of slow SOC (FIG. 5c) under different grazing intensities was in the third quarter, which increased by 80.59 g·m−2 (CK), 42.21 g·m−2 (LG), 39.29 g·m−2 (MG) and 30.86 g·m−2 (HG) compared with the second quarter, respectively.

Spatial distribution of SOC density under different grazing intensities

Compared with light grazing, moderate grazing and heavy grazing, the total SOC density decreased significantly. With the upgrading of grazing intensity, the decrease in SOC density becomes greater. Under light grazing (Fig. 6a), the relative change of soil total organic carbon density in Zhangye showed a higher distribution pattern in the southeast and a lower distribution pattern in the middle and northwest. The decrease in total soil organic carbon density in the southern of Gaotai County is relatively small, and the decrease in Shandan County and the southeast part of Sunan county was also small, but larger than that in the southern part of Gaotai County; The soil total organic carbon density decreased significantly in the northwestern and central areas of Sunan, Linze County and Ganzhou district, while the decrease in most areas of Minle County is moderate. The soil total organic carbon density decreased by 10.5866 g·m−2 at the lowest level under light grazing, and the least decreased area was in the middle and northern part of Sunan County bordering Gaotai County; The area with the largest decrease was 349.364 g·m−2, which was distributed near Qifeng village in the northwest of Sunan County. The relative change of soil total organic carbon density under moderate grazing and heavy grazing showed similar spatial distribution (Fig. 6b,c), showing a distribution pattern of high in the northwest and low in the southeast; However, the relative change of high value area under moderate grazing was more than that under heavy grazing. Under moderate grazing, the soil total organic carbon density in Dacha pasture area in southern Sunan County decreased seriously, with a maximum loss of 894.062 g·m−2; the decrease of total organic carbon density in Gaotai County, Shandan County and the northwestern area of Sunan is relatively small, by about 110 g·m−2. For 53 consecutive years, heavy grazing would seriously reduce soil total organic carbon density, and the decrease was between 125.314 ~ 2521.56 g·m−2, which was about 7 times of that under light grazing. The most declining areas are in the southern and southeastern parts of Sunan County, the areas with the least decline are in the northwest of Gaotai County and Sunan County.

Figure 6
figure 6

Spatial distribution of relative changes in SOC density at different grazing intensities. (a–c) are the spatial distribution of relative changes of total soil organic carbon under light, moderate and heavy grazing conditions, respectively, with red representing larger values and green representing smaller values.

Full size image

Different grazing intensities have different effects on different grassland types(Fig. 7).

Figure 7
figure 7

Spatial distribution of relative changes in total SOC density under different grazing intensities of different grassland types. (a,c,e) respectively represent the relative changes of soil total organic carbon density under mild, moderate and heavy grazing of grassland. (b,d,f) respectively represent the relative changes in soil total organic carbon density in desert grassland under mild, moderate and heavy grazing.

Full size image

Under light grazing, there was little difference in the relative changes of soil total organic carbon between grassland and desert grassland. The decreased range of soil total organic carbon density in grassland was 10.587 ~ 349.364 g·m−2 (Fig. 7a), and in desert was 10.697 ~ 331.958 g·m−2 (Fig. 7b). With the increases in grazing intensity, the decrease of SOC density in steppe was greater than in desert grassland. Under moderate grazing, the total SOC density of grassland decreased by 843.706 g·m−2 (Fig. 7c), while the desert decreased by 648.911 g·m−2 (Fig. 7d); The maximum reduction of total SOC density in grassland soil under heavy grazing was about 1.6 times than that in desert soil (Fig. 7e,f).

For grassland, the decrease of light grazing was relatively moderate in other areas except the south and west of Sunan County; The spatial distribution of their relative changes under moderate grazing and heavy grazing was similar, with a small decrease in the northwest and a large decrease in the southeast. However, compared with moderate grazing, the range of high values for the decrease in heavy grazing is wider, especially in the middle and southwest of Sunan County.

For desert grassland, the relative change pattern of soil total organic carbon density under light grazing and heavy grazing was similar, with a small decrease in the middle and a large decrease in the two sides. The decrease rate in the southern part of Gaotai County is the smallest, while the decrease rate in other areas is relatively large and the difference is not obvious. Under moderate grazing, the decrease of soil organic carbon density in the eastern of Linze and western and southern of Ganzhou was the largest, while the decrease in other areas was relatively small, and the decrease of soil organic carbon density was still concentrated in Gaotai County.

Effects of different grazing intensities on SOC

To investigate whether grazing intensity significantly affects total soil organic carbon (SOC) from 1970 to 2022, ANOVA was conducted to analyze the impact of different grazing intensities on SOC content. The multiple comparison test (e.g., Tukey's HSD) was used to identify specific group differences. The ANOVA result yielded a P-value of 0, indicating that grazing intensity has a significant effect on SOC content. Tukey's HSD test categorized no grazing, light grazing, and moderate grazing into group 'a,' while heavy grazing was placed in group 'b.' It also showed significant differences between heavy grazing and no grazing, light grazing, and moderate grazing. As seen in Fig. 8, heavy grazing is significantly different from no grazing, light grazing, and moderate grazing.

Figure 8
figure 8

Box plot with significance, The horizontal axis represents the grazing intensity, the vertical axis represents the soil organic carbon content, and the horizontal line above represents the significance of the difference in the effects of different types of grazing intensity on soil organic carbon.

Full size image

The year can encompass many potential influencing factors, such as climate change and changes in soil management practices. By including the year as a random effect, we can control for these confounding variables, allowing the model to focus more on the direct impact of grazing intensity on SOC. Thus, we established a mixed-effects model with the year as a random effect and grazing intensity as a fixed effect. The model results showed that the baseline category (no grazing) had an average SOC value of 10,236.3 g·m−2. Light grazing increased SOC by 685.9 g·m−2, moderate grazing decreased SOC by 240.7 g·m−2, and heavy grazing decreased SOC by 3112.9 g·m−2 compared to no grazing. These results indicate that light grazing positively affects SOC, while moderate grazing has a slight negative impact that is not significant. Heavy grazing significantly negatively affects SOC. The random effects part showed the impact of the year on SOC, with a variance of 4,063,693 and a standard deviation of 2016. The residual variance was 474,717, with a standard deviation of 689, indicating significant variation between years affecting SOC changes.

Figure 9 shows the impact of different grazing intensities on SOC using the fixed effects results from the model. Different colors represent different grazing intensities. The SOC content is higher in the no grazing and light grazing groups, while it is lowest in the heavy grazing group. This suggests that grazing intensity significantly affects SOC, with heavy grazing likely reducing SOC.

Figure 9
figure 9

Result of fixed effects in mixed effects model. The dots in the chart represent individual data points, and different colors indicate different grazing intensification.

Full size image

Figure 10 displays the relationship between actual SOC and fitted SOC from the model. Red and green points are mostly near the reference line, indicating accurate predictions under no grazing and light grazing conditions. Blue points are near the reference line but slightly scattered, indicating some prediction error under moderate grazing. Purple points are mostly in the low-value area and deviate from the reference line, suggesting larger prediction errors under heavy grazing.

Figure 10
figure 10

Fit value to actual value diagram. This diagram shows the relationship between the Actual SOC and the Fitted SOC of a mixed-effect model. The black dashed line in the figure is the 45°reference line.

Full size image

Figure 11 shows the relationship between fitted SOC and residuals, with different colors representing different grazing intensities. Red and green points are mainly near the reference line but show some positive bias at high fitted values. Blue points are evenly distributed near the reference line but show some bias at low and high fitted values. Purple points show significant bias, indicating larger prediction errors under heavy grazing. The differing residual distributions across grazing intensities indicate heteroscedasticity, suggesting inconsistent model fits across different grazing intensities. To address this, we could consider introducing more fixed or random effect variables, such as soil type and climate conditions, or trying other models like generalized linear mixed models (GLMM) or nonlinear mixed models (NLMM).

Figure 11
figure 11

Residual plot of mixed effects model. This residual plot shows the relationship between Fitted SOC and Residuals, with different colors representing different grazing intensities.

Full size image

Comparison of SOC density under different management practices

Since 2017, different management measures have been adopted in some sampling sites in Zhangye, which can be divided into three types: grassland degradation improvement, grazing rest and rotational grazing. A total of 7 sampling sites implemented degraded grassland improvement measures, 3 sampling sites implemented grazing measures, and 2 sampling sites implemented rotational grazing measures. This study divided the study period into two periods: 1970–2016 and 2017–2022, compared and analyzed the early and late periods of the implementation of management measures.

Figure 12 shows the comparative changes in the total SOC density and the density of each sub-pool at each sampling point before and after the implementation of the degradation grassland improvement. It can be seen from Fig. 12a that the implementation of degraded grassland improvement can improve the soil total organic carbon density at each sampling site, especially at the SN06 sampling site, which increases by 6047.24, while ML01 has the smallest increment of 12.479. The change trend of active SOC, slow SOC and total SOC was the same, but the inert SOC was slightly different. As can be seen from Fig. 12d, after the implementation of the degraded grassland improvement measures, the inert SOC of five sampling sites showed a decreasing trend, and only GZ02 and SN06 still increased.

Figure 12
figure 12

Changes of SOC density before and after improvement of degraded grassland at each sampling point.

Full size image

Figure 13 shows the changes in soil total organic carbon density and each component library density of each sampling point before and after the grazing measures. As can be seen from Fig. 13a, the total SOC density of SD06 and GT02 increased by 3809.89 and 968.151, however, the soil total organic carbon density of GT05 decreased slightly, with an average reduction of 403.66. The slow SOC showed a consistent trend with the total SOC. Resting grazing increased the slow SOC density of SD06 and GT02 by 762.845 and 142.221, respectively, but decreased the SOC density of GT05 by 276.64. The change in active SOC density was similar to that of inert SOC density.

Figure 13
figure 13

Changes of SOC density before and after grazing at each sampling site.

Full size image

As can be seen from Fig. 14, rotational grazing can increase the soil total organic carbon density and its component library density at each sampling point. The increased range of SOC density in SN05 was greater than that in SN11, and the increased range of SOC density in slow soil was the largest in each component library. Compared with without rotational grazing, the density of slow SOC in SN05 increased by 6705.359 and the total SOC density increased by 7369.658 after rotational grazing.

Figure 14
figure 14

Changes of SOC density before and after rotational grazing at each sampling site.

Full size image

Figure 15 shows the average soil total organic carbon density before and after the implementation of different management measures and the changes in its component library densities. As can be seen from Fig. 15a, the implementation of degraded grassland improvement, grazing rest and rotational grazing all increased the soil total organic carbon density, and rotational grazing increased the soil total organic carbon density by 6717.542 on average, while degraded grassland improvement increased the soil total organic carbon density by 2090.396 on average. Rest grazing caused the smallest increase, with an average increase of 1458.128. As can be seen from Fig. 15b, rotation grazing and improvement of degraded grassland increased the density of active soil organic carbon, while resting grazing reduced the density, with an average decrease of 26.979. Slow SOC density (FIG. 15c) was consistent with total SOC density under the three management measures, while inertSOC density (FIG. 15d) was consistent with active soil organic carbon density.

Figure 15
figure 15

Average SOC density changes before and after different management measures.

Full size image

Discussion

The study showed that long-term grazing resulted in a decrease in organic carbon density compared with no grazing, however, when grazing management at the study ranch was changed from the current adaptive multi-paddock (AMP) grazing to hypothetical light continuous grazing, simulated average annual SOC decreased 2.6% decline28. Adaptive Multi-Paddock (AMP) Grazing is an advanced grazing management strategy that involves dividing a large pasture into multiple smaller paddocks and rotating livestock among them. light grazing and moderate long-term grazing can increase the accumulation of organic carbon29,30, especially the density of active SOC, during the early grazing period, moderate grazing has a small impact on organic carbon about five years before grazing. The conclusion that rotational grazing increased soil organic carbon compared with other grazing measures was consistent with our results, while the increase in soil organic carbon caused by moderate continuous grazing was contradictory to our results, which may be because the study of moderate continuous grazing was conducted in temperate grassland, while Zhangye region is in a continental arid climate. The effects of climate on grazing management are not fully understood. With the increase in grazing years, organic carbon density will decrease significantly29,31, while heavy grazing will lead to a sharp decline in organic carbon density. This is basically consistent with the results of this study. The results of this study are similar to those of Li Dong32. Increasing grazing intensity had a negative effect on SOC accumulation, and the response of soil organic carbon to grazing intensity was different in different grassland types. Reasonable grazing intensity is beneficial to SOC accumulation33. Soil organic carbon storage in temperate grasslands increased significantly during short-term grazing (less than 5 years)34. However, different environmental and geographical factors cause different effects of grazing intensity on soil organic carbon in different study areas, so there is no unified standard to specify the grazing intensity of increasing soil organic carbon. Globally, light grazing showed minimal negative effects and was even able to promote soil organic carbon storage, while moderate and heavy grazing consistently reduced soil carbon storage35,36,37. The environment determines the extent and direction of the impact of grazing on soil carbon sequestration, and changes with climate, soil conditions, vegetation characteristics, grazing strategies, livestock type, and grazing intensity and timing,35,38. Short-term grazing increases organic carbon density mainly because grazing increases soil water infiltration through animal trampling, and also enhances soil microbial activity. However, long-term grazing will lead to a decrease in vegetation coverage, reducing the above ground litter and underground root biomass, thus reducing the input source of soil carbon39. A large number of studies40 have shown that grassland degradation can be prevented when grazing is light or moderate. At this time, various indexes of grassland, such as biomass, richness and evenness, will reach the optimum41, and heavy grazing will significantly reduce SOC density. Therefore, reasonable and moderate grazing must be adopted to prevent grassland degradation. At the same time, the formulation of corresponding policies should be based on the local demand and the actual situation of grassland, and comprehensive indicators such as grazing intensity, species composition, and the advantages and disadvantages of pasture42.