Effects of Livestock Grazing on Interannual Variation of Soil Methane Uptake in an Inner Mongolian Meadow Steppe

Background and aims. This study aimed at identifying the effects of livestock grazing on interannual variation in soil CH 4 uptake and underlying mechanisms in a meadow steppe ecosystem. Methods. A multi-year grazing experiment subject to six stocking rates was conducted to quantify CH 4 uxes as well as the changes in driving factors: vegetation traits, soil physicochemical properties and climatic parameters. The closed static chamber technique and a gas chromatograph were used to measure methane uxes. Multivariate regression analysis was performed to explore empirical relationships. Results. With increasing stocking rate, the multi-year mean CH 4 uptake rate decreased in a sigmoid curve-shaped manner, with the threshold point appearing in the light grazing treatment. The interannual changes in soil CH 4 uptake were highly dependent on stocking rate, with increasing, leveling and decreasing trends detected with increasing grazing intensity. Major factors affecting CH 4 uxes included vegetation traits, soil moisture, and soil nitrogen content, with the soil NH 4+ -N content assuming the most important role. However, predominant factors regulating interannual changes in CH 4 uptake were rainfall, belowground biomass, and soil nitrogen regime. Conclusions. The steppe ecosystem acted as a CH 4 sink, irrespective of stocking rate and year. However, light grazing can be the threshold grazing intensity in terms of both the CH 4 uptake potential and primary production in this steppe ecosystem. Our ndings have important implications for further understanding magnitudes and regulations of CH 4 uptake in grassland soils worldwide.


Introduction
Methane (CH 4 ) is the second most important anthropogenic greenhouse gas after carbon dioxide (CO 2 ), the global warming potential of which is roughly 28 times that of CO 2 (IPCC 2013). Moreover, the atmospheric concentration of CH 4 has been increasing at an alarming rate (ca. 1% per annum) over the past several decades ( (Li 2021). A better understanding of the magnitude and trend over time for CH 4 uptake/release, as well as the underlying mechanisms, is therefore of great signi cance (Han et al. 1999).
Two most signi cant sinks of atmospheric CH 4 , respectively, are abiotic oxidation by tropospheric Previous studies on grassland soils have shown that CH 4 emission and uptake is site-speci c and may be affected to varying degrees by management practices such as grazing, cultivation, and mowing, with grazing having the most substantial impact on the CH 4 oxidation potential (Mosier et al. 1991(Mosier et al. , 1997(Mosier et al. , 2002 Geng et al. 2010). However, while some authors reported that grazing reduced soil CH 4 uptake, others found that it led to increases or no change. A matter of fact is that grazing may reduce the growth of vegetation and litter storage that in turn would affect soil organic matter although SOM is not the primary food supply for MOB in most cases (Liu et al. 2007;Dun eld, 2007). Grazing at the same time can enhance soil evaporation and reduce soil moisture, thus affecting the physical environment of MOB. In addition, trampling by livestock can substantially compact the surface soil, which reduces diffusion rates of CH 4 and oxygen. Unfortunately, most of the previous studies had been based on short-term observations (1-2 years; Liu et al. 2007Liu et al. , 2009 or conducted with only one stocking rate. As a result, one cannot: (1) assess long-term changes (including the potential legacy effects) and variation (e.g., intraand inter-annual variation) and thus (2) understand how grazing intensity may alter CH 4 uptake differently.
The Hulun Buir steppe of Inner Mongolia is a typical temperate meadow steppe of the Eurasian steppe. This ecosystem is characterized by the highest plant species diversity, net primary production (NPP), and carbon sequestration potential among all steppe types in China. Here, livestock grazing is the most common way of grassland utilization. Because the soils are mostly ne-textured and xeric, with waterlogging occurring frequently in wet years, the meadow steppe is assumed to be unique and signi cant in terms of soil CH 4 uptake. However, studies on soil CH 4 dynamics of this steppe have been rarely carried out.
We conducted a 9-year eld experiment to examine interannual changes and controls of soil CH 4 uptake in response to variable stocking rates. Our study objectives were: (1) to examine interannual variation in soil CH 4 uptake with respect to grazing intensity; and (2) to explore underlying mechanisms regulating CH 4 uptake. Grazing usually may result in a series of changes in plant community traits and soil properties that may more or less mediate CH 4 consumption by MOB. We selected canopy cover and height, litter biomass, aboveground biomass (AGB), belowground biomass (BGB), soil physical and microclimatic variables, soil microbial biomass (SMB), and soil nutrients as potential driving factors. We hypothesized that grazing may impose impacts on CH 4 uptake by the soil via three approaches: (1) trampling that affects soil thermal and water regimes and compact the soil; (2) herbivory that may decrease litter input and carbohydrate allocation into the soil and subsequently may in uence microbially-associated belowground processes; and (3) excretion of dung and urine on sward patches that may alter the soil chemical property thereof. All these may more or less in uence the abundance and/or activity of MOB.

Study site
We conducted the eld experiment in a Leymus chinensis meadow steppe ecosystem in the Hulun Buir steppe region, on the northeastern Inner Mongolia Plateau. In brief, the landform is dominated by hills, lowlands and table lands, with elevations mostly varying between 600 and 800 m. The region has a temperate semi-humid climate, with an annual precipitation mostly between 350 and 400 mm that is highly seasonally variable. The mean annual air temperature spatially varies between − 5 and − 2°C, with a growing period of around 110 days. Chestnut soil is the predominant soil type, which corresponds to Castanozems in the soil taxonomic system of the FAO (Li et al. 2020). The vegetation is composed of perennial grasses, sedges, and forbs, of which Leymus chinensis, Stipa baicalensis, Carex duriuscula, Galium verum, Bupleurum scorzonerifolium, and Filifolium sibiricum are the dominant species. There was a more than 100-year history of free-range livestock in the region, up until the 2000s, after which prescribed ranging was implemented.

Experimental design
The eld study plot is located at the Hulun Buir Grassland Ecosystem Observation and Research Station of the Chinese Academy of Agricultural Sciences (CAAS) (49°19′349′′N, 119°56′521′′E; 670m a.s.l.) (Fig.   1S). The steppe had been under heavy grazing since the 1980s. The grazing experiment with ve stocking rates and one control unit was initiated in 2009 and has been continuously run till nowadays. Therefore, the control and all grazing treatments except the heavy grazing treatment (0.69, G0.92) represented a stand sequence with variable degrees of recovery, while the heavy grazing treatment represented a stand under continuous long-term disturbance. The treatments were arranged in a randomized block design. Each treatment had three replicated units, with each unit being 5 ha in size. As such, a total of 18 units of six treatments were established. The units were separated by fences. The stocking rates were set as 0.00, 0.23, 0.34, 0.46, 0.69, and 0.92 AU ha − 1 , where 1 AU = 500 kg of adult cattle, corresponding to 0, 2, 3, 4, 6, and 8 young cattle (with ~ 250-300 kg) per unit (Fig. 1S). Grazing lasted for 120 days between June and September in each growing season from 2009 to 2018. The grazing cattle were kept in each unit for 24 h each day during the entire grazing period. CH 4 measurements and data analysis CH 4 uxes were measured in situ using the closed static chamber technique (Hutchinson and Mosier 1981). The static chambers were made of stainless steel and consisted of two parts: a square base frame (0.5 m × 0.5 m × 0.1 m) and a removable lid (0.5 m × 0.5 m × 0.5 m). Three frames were inserted in each unit at a soil depth of 10 cm, and those remained xed during the whole study period (2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018). The frames were only 3 cm above the ground, so they did not interfere with the movement of grazing cattle. Movable protection was used during sampling to prevent staff trampling from impacting vegetation and ground conditions. A fan powered by a 12V battery was installed on the top wall of each static chamber to mix the air in the chamber. When placing the chambers on the frames, the vegetation within the frames was kept as intact as possible. CH 4 in the chambers was collected using a 60 ml airtight plastic syringe at 0, 10, 20 and 30 minute intervals for opaque chambers after manually closing the chamber between 9:00 to 11:00 h am. Our previous studies indicated that CH 4 emission during the time interval of 9:00-11:00 in the morning is well representative of the average rate over a 24 h cycle (Wang et al. 2005b). Gas samples were pumped into 50 ml airbags and sealed tightly. These bags were then transported to the lab within one week for CH 4 concentration measurement using an Agilent 7890A series gas chromatograph (Agilent 7890A, Agilent Technologies, Ltd., Co., USA). CH 4 uxes were measured biweekly from June to September in 2010 and weekly in the following years. CH 4 uxes were calculated by least squares regression of concentrations over time and expressed as µg CH 4 -C m − 2 h − 1 , after being corrected for air pressure, volume, and surface area.
The greenhouse gas CH 4 ux was calculated from the concentration change over the sampling intervals by using the following expression: 1 where F means gas ux (mg m − 2 h − 1 ); ρ is the gas density inside the chamber (ρ = P/RT, P is air pressure at the sampling site, R refers to the gas constant, and T is temperature inside the chamber); V is the volume of the measuring chamber (m 3 ); is the linear slope change of CH 4  Vegetation and soil properties Five 1 m × 1 m quadrats were randomly placed within each grazing unit in August of each year. The canopy height was derived in light of the individual shoot heights of all major species, with 5 randomly selected individuals per species being measured. A 1 m × 1m point frame with 100 crosshairs using a grid was used to measure canopy coverage. The aboveground vegetation was subsequently clipped to the ground, separated into living and dead materials, and dried at 65°C for 48 h to determine the aboveground biomass (AGB) and litter mass. Below-ground biomass (BGB) samples were taken at the same time at 10 cm-interval soil depths of 0-60 cm by a 30 cm × 30 cm cross-section, with three replications for each unit being sampled ).
Soil samples of the top 10 cm layer were collected using a soil drill with an inner diameter of 5cm from 10 random locations per unit in early August of each year, with the ten samples being mixed to get an average for each unit. The combined samples were divided into two parts. One part was immediately screened by a 2 mm soil sieve, so it was maintained fresh and intact for determining soil microbial biomass and carbon, nitrogen, ammonium nitrogen and nitrate nitrogen contents. The other part was wind-dried, crushed and passed through a 0.15 mm sieve and a 2 mm sieve, respectively, for analysing soil properties. Soil organic carbon (SOC) was measured by the dichromate oxidation method; total soil nitrogen (TN) was measured by semi-micro Kjeldahl determination (Bao 2000); soil available nitrogen (SAN) was measured using the distillation method (Bao 2000). Soil ammonium (NH 4 + -N) and nitrate (NO 3 − -N) contents were determined using a ow injection auto-analyser (Bao 2000); soil microbial biomass carbon (MBC) and microbial biomass nitrogen (MBN) was measured using the fumigationextraction method. Soil pH (in water) was determined by using the electrode method. The soil moisture (Ms) was measured by the oven-drying method, and soil bulk density (SBD) was measured with the ring knife method by oven-drying the soils at 105°C for 24 hours (Bao 2000). Soil temperature was measured daily by using portable thermometers in the early years and automatic datalog sets in later years.

Statistical analyses
Mean CH 4 uptake rates were calculated by arithmetically averaging individual ux values on a given sampling occasion. Differences in mean and seasonal cumulative CH 4 uptake rates among treatments were determined by analysis of two-way ANOVAs followed by Duncan multiple range test, with effects of p < 0.05 being signi cant. Because the effect of grazing differed among the study years, repeatedmeasures ANOVAs were applied to determine the main and interactive effects of measurement year and grazing intensity on CH 4 uptake rate, respectively. Linear regression was used to determine the mean variation of CH 4 uptake in response to variations in vegetation properties and soil parameters. All statistical analyses were carried out using the SPSS software package (v24.0, SPSS, Inc.). Concurrently measured biotic or abiotic variables were assumed to cause variations in CH 4 uptake when the correlation was signi cant (P < 0.05).
Interestingly, we found no signi cant difference among these four treatments (P > 0.05). The long-term average CH 4 uptake with stocking rate followed a clear sigmoid pattern (Fig. 1). However, when examined at annual scale, stocking rates had different effects on the CH 4 uptake rate, depending on the study year ( Fig. 2A). Overall, there were no signi cant differences in the CH 4 uptake rate between the treatments in the rst three study years (2010-2012). In contrast, signi cant negative effects at the highest stocking rate (G0.92) were detected in all the subsequent years, while with the exception of the lowest stocking rate (G0.23), all the other stocking rates showed signi cant negative effects in 3-4 of the following years (i.e., CH 4 uptake signi cantly decreased with increasing stocking rate) ( Fig. 2A) (Fig. 2B). It is worth noting that the variation among the treatments of the same year also changed over time during the 9-year study period (Fig. 3B). The variation among the treatments, as measured by con dence interval (CV), exhibits a logistical change with year, ranging from the lowest value (4.84%) in 2010 to 24.93% in 2016.

Interannual change in CH 4 uptake
We also examined our eld data for interannual changes and variations (Fig. 3A, Fig. 4). The changes in CH 4 uptake rate over time appear to depend on grazing intensity (Fig. 3A). Nevertheless, an obvious trend in this pattern occurred with increasing stocking rate. The interannual variation in the CH 4 uptake rate show an obvious increasing trend under no-grazing (G0.00) and light grazing (G0.23) treatments, but this uptake shows no obvious increase at G0.34, and this increasing trend was shifted to a decreasing one under G0.46, G0.69 and G0.92. Interestingly, we found that the interannual variation (i.e., CV values) show a positive linear relationship with an increasing stocking rate (Fig. 4), which ranged from 13.63% under G0.00 to 27.98% under G0.92.

Regulation of CH 4 uptake
By pooling all eld data from 9 years and 6 grazing treatments, we detected signi cant positive relationships between CH 4 uptake rate and major vegetation parameters, including canopy height, canopy cover, AGB, BGB, and litter biomass (Fig. 5). Of all the microclimatic and soil parameters, signi cant negative relationships between CH 4 uptake rate and precipitation, soil moisture, and soil NH 4 + -N content were ascertained. However, signi cant positive relationships between CH 4 uptake rate and soil organic carbon, total nitrogen, and microbial biomass nitrogen were found (Fig. 6).
We performed forward stepwise regression analysis by including all potential biophysical variables as independent variables (  Two-way ANOVA analysis indicated that treatment had a signi cant effect on CH 4 uptake rate and vegetation parameters except BGB (Table 2). For soil properties, grazing produced signi cant effects only on soil moisture and soil NO 3 − -N (Table 2). Interestingly, the interannual variation was signi cant for all variables. The interactive effects from treatment and year were signi cant on CH 4 uptake and vegetation parameters except BGB, but nonsigni cant on most soil properties, except soil moisture, total nitrogen, and nitrate.

Discussion
Grazing effects on CH 4 uptake We found two contrasting patterns of CH 4 ux in response to increasing stocking rate. In the rst two years of grazing, CH 4 uptake rate increased with stocking rate ( Fig. 2A) with no signi cant differences detected among the treatments, suggesting that grazing effects on CH 4 ux are not immediate. Indeed, the meadow steppe is more resistant to grazing than other steppe types in Inner Mongolia, due to its rich soil ( ne textured, well-drained) and vegetation (height, diversity). The dominant plant species (i.e., Leymus chinensis) is more resistant to feeding and trampling than the Stipa spp. that dominate other steppes. Previous studies show that soil moisture and aeration conditions of the L. chinensis steppe are much less prone to grazing trampling, and its regrowth of biomass is also less affected by short-term herbivory, mainly due to the rhizomatous growth form of L. chinensis per se, in striking contrast to the bunch growth form of Stipa species (Li et al., 2020). As a result, the activity of MOB and CH 4 uptake were also less affected by grazing in the initial few years in this steppe ecosystem.
An obviously decreasing trend in CH 4 uptake with increasing grazing intensity appeared in the following years, with the highest stocking rate showing the greatest negative effects on CH 4 uptake ( Fig. 2A, 2B).
Overall, grazing reduced CH 4  Three mechanisms may be responsible for the changes and differences in CH 4 uptake under grazing in our study grassland. First, they may have resulted from changes in vegetation traits due to variable grazing. The CH 4 uptake rate signi cantly positively correlated with the vegetation traits examined in our study (Fig. 5), indicating that the effects of grazing on methane uptake was partially mediated by vegetation. CH 4 oxidation in the rhizosphere is the most important process for CH 4  The second mechanism may be that grazing changes soil NH 4 + -N content and thus affects methane uptake. A signi cant negative relationship between the CH 4 uptake rate and soil NH 4 + -N content was found in this study (Fig. 6). Our long-term data showed that the multi-year mean soil NH 4 + -N content had increased by all grazing treatments, particularly at the highest stocking rate (Zhang et al. 2021). For some years, one-to two-fold increases in soil NH 4 + -N content were observed under the highest stocking rate compared to the reference unit (G0.00). Increases in available nitrogen content, soil NH 4 + -N in particular, Thirdly, altered soil moisture by grazing may be another mechanism that regulates CH 4 uptake. Our longterm data showed that soil moisture was higher in most cases under various stocking rates of grazing than the no grazing (G0.00) treatment, although no linearly increasing trend of soil moisture with increasing stocking rate was found (Zhang et al. 2021). Stocking-resultant increases in soil moisture were likely related in part to the corresponding decreases in ANPP which consumed less soil water. In addition, we observed that dicotyledonous forbs with deep taproots were signi cantly more abundant in the nograzing treatment than in all grazing treatments except the intermediate-grazing treatment, which could also be a cause in this regard. Given the signi cant negative relationship between the CH 4 uptake and soil water content (Fig. 6), the decrease in CH 4 uptake with stocking rate can be partially explained through the mediating role of soil moisture and increases soil bulk density. Soil submersion allows the development of the methanogenic activity and reduces methanotrophic activity by reducing the size of the oxidized zone. In a typical steppe, CH 4 uptake was primarily determined by soil temperature and soil moisture of the topsoil (7 cm) (Wang et al. 2005a). Negative correlation between CH 4 uptake and soil moisture was reported in an alpine steppe (Wei et al. 2012) and temperate steppes (Tang et al. 2013). At Swiss grassland sites, the soil methanotrophic activity is related to its water content. With the increase of water content, it gradually decreases and is close to the eld capacity (Imer et al. 2013). Negative correlations between CH 4 consumption and soil moisture were also reported in Canadian forests and in a Massachusetts forest (Le Mer and Roger 2001). Upland soils, when temporarily submerged, may become CH 4 sources. This was also observed in a Canadian grassland with well drained soils (Wang and Bettany 1995).
A positive correlation is often observed between the methanogenic potential and the SOM content in rice eld soils, and almost all in situ studies have shown that organic matter incorporation markedly increased CH 4 emission. However, a positive relationship between CH 4 uptake and soil SOC content was detected in this study (Fig. 6), which may re ect the fact that grassland soils per se have rather low methanogenic potentials. This can be further corroborated by the fact that CH 4 production is often positively correlated with SOM content in soils exhibiting a high methanogenic activity (Le Mer and Roger 2001).

Complexity of grazing regulation on CH 4 uptake
Our multivariate linear regression analysis indicated that livestock grazing had substantially changed the interannual variation of both CH 4 uptake and its biophysical regulators in this steppe ecosystem (Fig. 3A,   3B; Table 1). The increasing trends of CH 4 uptake with year in the no-grazing (G0.00) and light grazing treatments (G0.23) may re ect a progressive successional process of the ecosystem under these treatments. In fact, this steppe had been under heavy grazing for ages prior to our experiment. As such,  We found some inconsistent and even contradictory results with previous reports on the impacts of livestock grazing on grassland soil CH 4 uptake, partially due to the inconsistency in the number of study years and grazing intensity. Most previous studies were conducted in a single season or over 1-3 years. These studies did not have a broad range of stocking rates, either (i.e., limited to 1-3 grazing intensities).
Clearly, long-term treatments with multiple stocking rates should be studied in other worldwide grasslands.
Our hypothesis that livestock grazing imposes impacts on methane dynamics in grassland soils through trampling, selective feeding, and excretion of dung and urine is generally accepted. These impacts are

Conclusion
We showed that the meadow steppe ecosystem acted as a signi cant CH 4 sink, irrespective of stocking rate and year. With increasing grazing intensity, CH 4 uptake rate decreased in a sigmoid curve-shaped manner, with the threshold grazing intensity (in ection point) being at the light grazing (G0.34). The interannual variation of CH 4 uptake depended on grazing intensity. An increasing CH 4 uptake was found under the no-grazing (G0.00) and light grazing (G0.23) treatment, while a stable uptake was apparent under G0.34 treatment and a decreasing trend was seen for all other treatments. Major factors affecting CH 4 uptake among the treatments include vegetation characteristics, soil moisture and soil nitrogen regime, with soil NH 4 + -N content assuming the most importance role. Interannual variations of CH 4 uptake were dominated by rainfall, belowground biomass (BGB), and soil nitrogen regime. Continued efforts are needed to see if our ndings hold at longer time scales, as well as in other steppe ecosystems.
We recognize that multi-year studies across a wider range of stocking rates are extremely necessary to test the effects of grazing on CH 4 uptake in global grassland ecosystems.

Declarations
The authors declare that they have no known competing nancial interests or personal relationships that could have appeared to in uence the work reported in this paper. Figure 1 Change in annual mean CH4 uptake (i.e., negative values) under different grazing treatments from 2010 through 2018. Bars represent the standard errors of the seasonal means. A simple logistic model is used to predict CH4 uptake with grazing intensity. The letters indicate signi cant differences between the treatments in one-way ANOVA at P<0.05.   Linear increase of interannual variations (CVs) in CH4 uptake strength with grazing intensity. CV represents the con dence interval of CH4 uptake over the 9-year study period.  Empirical dependency of the seasonal CH4 uptake rate on soil moisture (A), soil carbon (B), total soil nitrogen (C), ammonium nitrogen (D), and soil microbial biomass nitrogen (E).