Precipitation and land use alter soil respiration in an Inner Mongolian grassland

Grasslands are facing the threat of climate change and intensive land use. Soil respiration (Rs) in grassland ecosystems can be potentially altered by changes in precipitation and land use. We aimed to quantify the impact of changes in precipitation and common land use practices in an Inner Mongolia grassland, i.e., mowing and grazing, on soil respiration. We performed an in situ experiment with altered precipitation (+ 50%, ambient, and -50%) and land use (control or fencing, mowing, and grazing) to explore their impacts on soil respiration and its autotrophic (Ra) and heterotrophic (Rh) components. Altered precipitation had stronger impacts on abiotic and biotic drivers than land use, leading to stronger impacts on Rs and its components. Over the 3-year experiment, Rs, Ra and Rh decreased by 36%, 42% and 33% with reduced precipitation and increased by 29%, 36% and 25% with increased precipitation, respectively. Grazing and mowing caused relatively small decreases in Rs compared to fencing (generally < 10%). However, precipitation and land use interactively impacted abiotic and biotic drivers and thus Rs. The decrease in Rs with reduced precipitation was greater with grazing (38%) and mowing (37%) than with fencing (32%). Rs and its components may decrease under the projected decrease in precipitation and may further decrease with grazing and mowing compared to fencing. Therefore, land use should be considered when predicting grassland carbon cycling in response to future precipitation changes.


Introduction
Grasslands comprise approximately one-third of the Earth's land surface and account for 20% of the carbon in vegetation and soil globally (Piao et al. 2009;Zhou et al. 2017).In temperate and tropical grasslands, soil respiration (R s ) releases 390 and 601 g C m −2 annually to the atmosphere (Wang and Fang 2009), making grassland R s a substantial portion of global R s .R s comprises autotrophic (R a ) and heterotrophic (R h ) components, which are the result of plant roots and soil microbes converting organic carbon to CO 2 via biotic metabolic activities (Davidson and Janssens 2006;Bond-Lamberty and Thomson 2010;Yvon-Durocher et al. 2012).In global grasslands, biotic activities of plants and soil microbes are strongly limited by the typical water deficit in soil, resulting in grassland R s and its autotrophic and heterotrophic components being positively associated with precipitation (Liu et al. 2009(Liu et al. , 2016;;Geng et al. 2012).Therefore, the profound changes in grassland precipitation during the last decades (IPCC 2013) are expected to substantially impact grassland R s (Liu et al. 2009(Liu et al. , 2016)).In addition, most grassland ecosystems are subjected to diverse and intensive anthropogenic activities, such as grazing, mowing, and other land-use regimes (Liu et al. 2012;Han et al. 2012;Gossner et al. 2016).These intensive land uses have reduced the soil water content and decreased biomass production of plants and soil microbes in global grasslands (Gang et al. 2014;Gong et al. 2014;Wei et al. 2016;Tang et al. 2019;Bardgett et al. 2021), strongly influencing carbon cycling in grassland ecosystems.The altered grassland R s under changing precipitation and land-use regimes could in turn provide a feedback to climate change, resulting in significant global consequences.
In grassland ecosystems, grazing and mowing are widely practiced land-use regimes and may have strong impacts on R s and its components (Cao et al. 2004;Jia et al. 2006Jia et al. , 2007;;Zhou et al. 2019;Tang et al. 2019).The harvesting of aboveground plant tissues by intensive grazing and mowing can reduce the growth rates of plants and soil microbes (Bagchi and Ritchie 2010;Hou et al. 2014;Gong et al. 2014;Wang et al. 2020d), leading to a reduced amount of substrate supply to plant roots and soil microbes and thus decreasing R s and its components.In addition, vegetation removal can cause more solar radiation to reach the soil surface, resulting in an increased soil temperature and a decreased soil water content (Gong et al. 2014;Tang et al. 2019).It is also possible that vegetation removal may increase soil moisture, potentially due to the reduced soil water loss from transpiration (Peng et al. 2007).Altered soil microclimates can directly impact R s and its components or can indirectly impact them by influencing the growth rates of plants and soil microbes.For example, experimental studies and syntheses have shown that R s and its autotrophic and heterotrophic components often decrease with grazing intensity, potentially due to the reduced soil water contents (Cao et al. 2004;Wang et al. 2017Wang et al. , 2020d;;Tang et al. 2019).Similarly, studies have also shown that mowing can reduce the biomass production of plants and soil microbes by strengthening the soil water deficit (Zhou et al. 2007;Wei et al. 2016), leading to negative impacts on R a and R h and thus R s (Wan and Luo 2003;Wan et al. 2005;Koncz et al. 2015).Because grazing and mowing affect R s via climate-dependent processes, the effects of grazing and mowing may depend on climatic conditions.For example, the effects of grazing and mowing on R s and its components may be stronger in arid regions than in humid regions (Han et al. 2012;Wang et al. 2020d).As a result, understanding the grazing and mowing effects on R s in grasslands requires consideration of their potential interactions with climate (Xu et al. 2015a;Wang et al. 2020d).
Changes in precipitation and land use may interact because they both influence many of the same abiotic and biotic drivers of R s (Bagchi and Ritchie 2011;Hou et al. 2014;Wei et al. 2016;Li et al. 2018;Zhou et al. 2019).Changing precipitation and land uses such as grazing and mowing can interactively impact soil temperature and soil moisture (Hou et al. 2014;Wei et al. 2016;Li et al. 2018), potentially leading to interactive impacts on the biotic drivers of R s (Bagchi and Ritchie 2011;Wang et al. 2020d).For example, in dry years, low precipitation interacts with grazing and mowing, resulting in a further decrease in plant biomass production (Hou et al. 2014;Wei et al. 2016).A recent study showed that increasing precipitation could potentially alleviate the negative impact of intensive grazing on soil microbial biomass production (Li et al. 2018).The altered biomass production of plants and soil microbes can in turn affect R s by changing the substrate supply to plant roots and soil microbes (Xu et al. 2015a;Wang et al. 2020d).
Vol.: (0123456789) Disentangling how precipitation and land-use regimes affect these interacting abiotic and biotic drivers is key to understanding the effects of precipitation and land-use regimes on soil respiration and predicting soil carbon fluxes in the future.
The Inner Mongolian grassland, part of the Eurasian grassland biome (Han et al. 2009;Fang et al. 2015;Wu et al. 2015), is an ideal system to study the interactive effects of land use and precipitation on soil carbon cycling.Grazing and mowing are widely practiced land-use regimes (Lu et al. 2015;Fang et al. 2015;Wu et al. 2015;Wang et al. 2020a), and precipitation is often considered the most important climatic driver of R s in Inner Mongolian grasslands (Yang et al. 2020;Li et al. 2020).During recent decades, this region has been experiencing more intensive grazing and mowing (Wu et al. 2015;Li et al. 2018;Wang et al. 2020a) as well as reduced but more variable precipitation (Piao et al. 2010;Huang et al. 2015).While previous work in this region has examined the effects of land use and precipitation individually, it remains unclear how land use and precipitation may interact to affect soil respiration (Gong et al. 2014;Yang et al. 2020).Here, we performed an in situ experiment with altered precipitation amounts (-50%, ambient and + 50%) and land-use regimes (i.e., mowing and grazing, Fig. S1) to investigate their effects on R s and its autotropic and heterotrophic components.We hypothesized that decreased precipitation and land use both lead to a decrease in R s and that the negative effect of reduced precipitation on R s is exacerbated by land use.Finally, we quantified the potential abiotic and biotic drivers of R s and its components to identify the pathways through which changes in precipitation and land use affect R s .

Study site
We conducted this study at the Maodeng Grassland Ecosystem Research Station of Inner Mongolia University (44°10' N, 116°28' E, 1101 m a.s.l.) located in the Xilingol region of Inner Mongolia, China (Fig. S1a).This area has a hyper-continental climate.The short and cool growing season normally lasts from May to October and accounts for the majority of the annual precipitation.The non-growing season, in contrast, is characterized by low temperature, low precipitation and dormant plants (Bai et al. 2008;Fang et al. 2015;Wu et al. 2015).During the 3 years of the study (2017)(2018)(2019), the mean growing-season and annual air temperature were 10.7 °C and -1.2 °C, respectively.The annual growing-season precipitation ranged from 137 to 252 mm yr -1 and accounted for ~ 90% of the total annual precipitation (ranging from 168-278 mm, Fig. S2).The natural vegetation is a typical steppe dominated by perennial grasses such as Leymus chinensis, Stipa baicalensis and Cleistogenes squarrosa.The developed soil is a Calcic-Orthic Aridisol according to the US soil taxonomy classification system with a mean pH of 8.1 and a mean soil organic carbon content of 12.5 g kg -1 at the 0-20 cm soil depth (Wang et al. 2020b).

Experimental design
We established a field experiment with treatments of precipitation amount and land-use regime using a randomized complete block design with a split plot (Fig. S1b-c).Within each block, we arranged the land use treatments in the plots and the precipitation amount treatments in the subplots.Specifically, in 2011, we established three 100 m × 100 m blocks with 3 m in between.Each block was divided into nine 33.3 m × 33.3 m plots.Within each block, we randomly assigned three land use treatments to 3 plots.The three land use treatments were no grazing or mowing (wire fence enclosure), grazing (with 6 sheep in July and August, once per month until the residual height of the plants reached 6 cm), and mowing (to 6 cm, once per year in August).The grazing intensity and mowing frequency we implemented in the experiment reflect common practices in this region (Baoyin et al. 2014;Wu et al. 2015;Wang et al. 2020b).Five years after the construction of the land use treatments (in 2016), we established precipitation amount treatments at the end of the growing season.Briefly, we delineated three 3 m × 5 m subplots with 2 m in between within each land use treatment plot.Each subplot was randomly assigned to one of the three precipitation treatments.Around each subplot, we inserted metal plates to a depth of 15 cm with 5 cm remaining aboveground to prevent lateral movement of rainfall.Consistent with a network of worldwide precipitation manipulation experiments (www.droug ht-net.org), we implemented three levels of precipitation: 50% reduction (Dry), ambient (Amb), and 50% increase (Wet).The reduction in precipitation was achieved by installing ten transparent sheet channels over the subplots of the dry treatment (25 cm wide and 340 cm long).These channels were made of transparent Panlite (Zhuonier Corporate, Suzhou, China) with a high light transmission ratio (> 90%) to minimize shading.The channels were installed at a ~ 10° angle and covered 50% of the subplot so that 50% of the rainwater was intercepted (Fig. S1b).The intercepted rainwater was collected and immediately sprinkled to a wet treatment subplot within the same plot after the rainfall event, resulting in a 50% increase in precipitation for the wet treatment subplot.During the 3 years of the study (2017-2019), growing-season precipitation ranged from 137-252 mm/yr.The range of precipitation received by all treatments (68-378 mm/yr) represents 75% of the growing-season precipitation range observed in the last 10 years (136-454 mm/yr) (Fig. S3).

Soil respiration measurement
We manually measured R s and its components twice per month during the growing season (from May to October, Figs.S4-S6).Measurements were made between 8:00 AM and 12:00 PM on a sunny day of the 1st and 3rd weeks of each month using a LI-8100 Automated Soil CO 2 Flux system (Li-Cor Inc., Lincoln, NE, USA) on a shallow polyvinyl chloride (PVC) collar (20 cm in diameter and 10 cm in height) installed to a soil depth of 5 cm.We removed all aboveground vegetation inside the collar.We also measured respiration over a deep PVC collar (20 cm in diameter and 45 cm in height) installed to a soil depth of 40 cm with all aboveground vegetation inside the collar removed.These measurements represent heterotrophic respiration because more than 80% of the belowground biomass is distributed in the top 30 cm, and the installation depth was sufficient to exclude most organic matter input from adjacent plants (Fig. S7).We installed these deep collars 9 months prior to R h measurements to eliminate experimental artifacts arising from the pulse of dead root input after collar installation (Zhou et al. 2007;Wang et al. 2021).R a was calculated as the difference between R s and R h .While the collar method may result in artifacts such as changes in soil moisture and temperature due to vegetation removal (Hanson et al. 2000;Kuzyakov and Larionova 2005;Baggs 2006), it is a widely used standard method for soil respiration measurements (Zhou et al. 2007;Suseela et al. 2012;Suseela and Dukes 2013;Chen et al. 2016;Wang et al. 2018Wang et al. , 2021)).More importantly, the artifacts are likely similar across treatments and thus have minimal influence on the analysis of the treatment effects.
Concurrent with soil respiration measurements, we measured soil temperature (ST) and moisture (SM) at 5 cm depth (Figs.S8-S9) using 6000-09TC and GS-1 probes attached to the LI-8100.This soil depth was commonly used to measure ST and SM in previous studies (Feng et al. 2018;Wang et al. 2018Wang et al. , 2021) ) because temperature measured at this depth usually had the highest explanatory power with respect to R s (Phillips et al. 2011;Wang et al. 2014).Finally, we obtained hourly air temperature and precipitation data using ECT and ECRN-100 sensors attached to an EM50 datalogger (Decagon Devices Inc., Pullman, WA, USA) at a nearby weather station (~ 100 m).

Plant and soil sampling
We measured the aboveground (ANPP) and belowground (BNPP) net primary productivities once per year (Wang et al. 2020c).We used a harvest method for ANPP measurements in early August when plant biomass reached its annual peak (Xu et al. 2015b;Liu et al. 2018).Specifically, we prepared three 1 m × 1 m quadrats in each subplot and randomly selected one in each year without selecting the same quadrat in the previous year.For the grazing plots, three quadrats in each subplot were exposed to grazing before the establishment of the precipitation treatments (from 2011 to 2016).After the establishment of the precipitation treatments (from 2017 to 2019), one of three prepared quadrats was randomly selected each year and temporally protected with a cage, and the other two quadrats were exposed to grazing.Therefore, ANPP measured in the grazing plots represents the long-term effect of grazing on ANPP (Milchunas and Lauenroth 1993;McNaughton et al. 1996).In early August, all green plant tissues within the selected quadrats were harvested, oven dried and weighed to obtain ANPP.We used a root ingrowth-core method (Liu et al. 2018) instead of a harvest method to measure BNPP because the plant community in this region is dominated by perennial species (Bai et al. Vol.: (0123456789) 2004; Zhang et al. 2017;Wang et al. 2020c), causing belowground biomass to represent a multiyear accumulation of root biomass production.We first took three soil cores (7 cm in diameter) of 0-50 cm along the diagonal of each subplot at the end of the growing season in 2017, sieved (2 mm mesh) the soil to remove roots and refilled the soil cores with root-free soil collected from the same depth.During the following two years (2018 and 2019), we resampled soil from the same cores at the same time as the ANPP measurement, sieved (2 mm mesh) the soil samples to obtain roots, and oven dried and weighed the root samples to calculate BNPP.The residual root-free soil was returned for BNPP measurements the following year.
In early August 2019, we sampled three additional soil cores along the diagonal of each subplot and measured microbial biomass carbon (MBC) and nitrogen (MBN) using the chloroform fumigationextraction method (Vance et al. 1987).Specifically, we mixed three soil cores collected from the same subplot, sieved them through a 2 mm mesh to remove plant tissues, and then weighed 6 aliquots (3 g equivalent).Three of the aliquots were fumigated with ethanol-free CHCl 3 at 25 °C in darkness for 48 h, and the other 3 were unfumigated.The fumigated and unfumigated samples were extracted with 0.5 M K 2 SO 4 (12 ml) for 30 min on a shaker.Carbon and nitrogen in the K 2 SO 4 extracts were analyzed with a total organic C/N analyzer (Elementar vario TOC, Elementar Co., Germany), and the differences between fumigated and unfumigated samples were converted to MBC and MBN using a conversion factor of 0.45 (Brookes et al. 1985).

Statistical analysis
In our statistical analyses, the ST, SM, ANPP, BNPP, MBC and MBN data were directly analyzed, while the data of R s and its components (R h and R a ) were first log-transformed to conform to the assumption of normal residuals (Lloyd and Taylor 1994;Yvon-Durocher et al. 2012) and then analyzed.We analyzed the effects of precipitation and land-use treatments on ST, SM, R s and its components that were measured twice per month using repeated-measures analysis of variance (RMANOVA).To explore the average effect over the three years of the study and potential interannual variations, we performed this analysis separately in each year and using all three years of data combined.Similarly, we analyzed the treatment effect of precipitation and land use on ANPP and BNPP using data from each year separately or the three years combined.Because ANPP and BNPP were measured once per year, we used two-way ANOVAs when analyzing data from each year and RMANO-VAs when data from all three years were combined.Finally, we analyzed the treatment effects on MBC and MBN using two-way ANOVAs because MBC and MBN were only measured once during the experimental period.In these analyses, precipitation and land-use regime treatments and date (or year for ANPP and BNPP) of measurement were treated as categorical fixed effects, and block, plot and subplot were hierarchically arranged as random effects.
We used structural equation models (SEMs) to quantify the direct and indirect impacts of abiotic (ST and SM) and biotic (plant and soil microbial parameters) factors on autotrophic (R a ) and heterotrophic (R h ) respiration.We constructed initial SEMs including all potential paths based on theoretical knowledge (Liu et al. 2009;Geng et al. 2012;Xu et al. 2015a;Chen et al. 2016) of the drivers of R a and R h .We visually examined the bivariate relationships among abiotic and biotic drivers and R s and its components to ensure that the SEMs were generally reasonable for representing their relationships (see Figs. S10-S12 for details).Subsequently, we performed a model selection of SEMs using the chi-square ( 2 ) difference test, the goodness-of-fit index (GFI) and the root mean-square error of approximation (RMSEA) to obtain final SEMs that had a good fit (a small 2 with P > 0.1, GFI > 0.90 and RMSEA < 0.10) (Kline 2005).In the SEMs, we used the annual mean ST, SM, R h and R a in 2019 (27 observations for each variable) because MBC and MBN were measured only once in this year.We constructed separate SEMs for R a and R h because R a is calculated as the difference between R s and R h and thus is not independent of R h .In these SEMs, all variables were standardized for ease of interpretation and comparison.
All statistical analyses were conducted using R 3.6.3(R Core Team 2013).We used the lavaan package for the SEMs.All figures were produced using the ggplot2 package (Wickham 2009).Effects were considered statistically significant if P < 0.05 or marginally significant if 0.05 ≤ P < 0.10.

Effects of precipitation and land use on the abiotic environment
Precipitation treatment significantly affected growing-season ST when data from each year were analyzed separately or the data from all three years were combined (Fig. 1a-d, Table S1).The growing-season ST increased by 0.56 °C (P = 0.02) under the dry treatment and decreased by 0.49 °C (P = 0.03) under the wet treatment on average over the three years (Fig. 1d, Table S1).We did not observe significant effects of land-use regime or its interaction with the precipitation amount on ST (Fig. 1a-d, Table S1).SM was significantly impacted by the precipitation amount and land use as well as their interaction in the overall study period (Fig. 1e-h, Table S1).For the precipitation amount treatment alone, growing-season SM decreased by 22.3% (P < 0.001) and increased by 30.5% (P < 0.001) based on the 3-year average under dry and wet treatments, respectively (Fig. 1h).For land use treatment alone, the grazing and mowing treatments increased SM by 8.8% (P = 0.01) and 7.1% (P = 0.02), respectively, based on the 3-year average compared with fencing (Fig. 1h).The effect of mowing on SM was less pronounced under wet conditions.Compared with fencing, grazing and mowing increased SM under dry and ambient precipitation conditions, while under wet conditions, only grazing increased SM by 11.4% (P = 0.006), and mowing had no effect on SM (Fig. 1h).When separately  (not statistically significant) when P > 0.10.Different lower-case letters represent significant differences between precipitation amount treatments under the same land use and different upper-case letters represent significant differences between land uses under the same precipitation treatment (pairwise t test, P < 0.10) investigated in different years, SM was only significantly impacted by the precipitation amount in each year and was significantly or marginally significantly affected by land-use regime and its interaction with the precipitation amount in certain years (Fig. 1e-g, Table S1).

Effects of precipitation and land use on aboveground and belowground productivity
The precipitation amount significantly influenced ANPP, showing a decrease of 72.4% (P < 0.001) and an increase of 47.4% (P < 0.001) on average over the three years under the dry and wet treatments, respectively (Fig. 1l, Table S1).The direction of the precipitation effect was consistent across the 3 years and was statistically significant when data from each year were analyzed separately (Fig. 1i-k, Table S1).Landuse regime alone had significant effects on ANPP in certain years (Fig. 1i-k, Table S1) but had no effect on ANPP when integrated over the 3-year period (increased by 6.0% (P = 0.40) and decreased by 11.6% (P = 0.14) under grazing and mowing, respectively) (Fig. 1l).In addition, we detected a marginally significant interaction between the precipitation amount and land-use regime in 2018, as grazing and mowing had no effects on ANPP under dry and ambient precipitation conditions, while under wet conditions, grazing increased ANPP by 29.1% with marginal significance (P = 0.08), and mowing had no effect on ANPP (Fig. 1j).BNPP was significantly affected by the precipitation amount and land-use regime but not by their interaction when separately considered in different years or integrated over the study period (Fig. 1n-p, Table S1).Specifically, the dry and wet treatments resulted in a significant decrease of 16.1% (P < 0.001) and a significant increase of 14.3% in BNPP (P < 0.001), respectively, on average over the study period.Compared to fencing, grazing significantly reduced BNPP by 11.7% (P = 0.006), while mowing had no effect on BNPP (reduced by 1.0%, P = 0.66) based on the 3-year average (Fig. 1p).
The precipitation amount significantly impacted MBC and MBN, while the land-use regime significantly impacted only MBC (Fig. 2, Table S1).Specifically, the dry treatment significantly reduced MBC and MBN by 17.8% (P < 0.001) and 24.4% (P = 0.007), respectively, while the wet treatment had no effect on these variables (Fig. 2).For landuse regimes, grazing had no effect on MBC and MBN, and mowing increased MBC by 8.9% with marginal significance (P = 0.07) but had no effect on MBN (Fig. 3).We also detected significant or marginally significant interactive effects of precipitation and land use on MBC and MBN as more pronounced dry effects under grazing (Fig. 2).Specifically, the dry treatment decreased MBC and MBN by 33.0% (P < 0.001) and 48.8% (P = 0.002), respectively, under grazing and had no effects on these variables under fencing and mowing, while the wet treatment had no effects on either MBC or MBN under the different land-use regimes (Fig. 2a).indicated by *** when P < 0.001, ** when P < 0.01, * when P < 0.05, # when P < 0.10 and n.s.(not statistically significant) when P > 0.10.Different lower-case letters represent significant differences between precipitation amount treatments under the same land use and different upper-case letters represent significant differences between land uses under the same precipitation treatment (pairwise t test, P < 0.10) Vol:. ( 1234567890)

Effects of precipitation and land use on soil respiration and its components
The precipitation amount significantly impacted R s and its components over the 3-year study period (Fig. 3, Table S2).Specifically, R s and its autotrophic and heterotrophic components were reduced by 35.9% (P < 0.001), 41.7% (P < 0.001) and 33.1% (P < 0.001) under the dry treatment and were increased by 28.5% (P < 0.001), 36.0%(P < 0.001) and 25.0% (P < 0.001) under the wet treatment based on the 3-year average (Fig. 3d, h and l).In addition, land use impacted R s and its components (R a and R h ) across the three years (Fig. 3d, h and l, Table S2).In comparison to fencing, grazing reduced R a by 5.3% (P = 0.031) but had no significant effect on R s and R h , while mowing reduced R s , R a and R h by 7.2% (P = 0.006), 10.7% (P = 0.007) and 5.5% (P = 0.022), respectively.Under different precipitation treatments, grazing and mowing had neutral to negative impacts on R s and its components, with some negative impacts detected as significant or marginally significant (Fig. 3d, h and l).On average, over the three-year study period, precipitation and land use had no interactive impacts on R s and R a (Fig. 3d and h), but had a marginally significant interactive effect on R h (Fig. 3l).Compared with fencing, grazing significantly reduced R h by 5.3% (P = 0.023) under dry conditions and had no effect under ambient precipitation and wet conditions, while mowing significantly reduced R h under different precipitation conditions (ranging from 4.2%-7.1% and P < 0.05 for different precipitation conditions) (Fig. 3l).We also analyzed the impacts of precipitation and land use in each year.We found that the direction of precipitation in affecting R s and its components was consistent across the 3 years, and all effects were detected statistically significant (Fig. 3a-c, e-g and i-k, Table S2).Under the different precipitation treatments, the land-use regimes (not statistically significant) when P > 0.10.Different lowercase letters represent significant differences between precipitation amount treatments under the same land use and different upper-case letters represent significant differences between land uses under the same precipitation treatment (pairwise t test, P < 0.10) generally had neutral to negative impacts on R s and its components, with a few exceptions in 2018 (Fig. 3a-c, e-g and i-k, Table S2).
We then used SEMs to quantify the direct and indirect effects of abiotic (ST and SM) and biotic (plant and soil microbial parameters) factors on R a and R h separately (see Figs. S10-S12 for results of regressions used to construct the initial SEMs as well as Fig. S13 and Tables S3-S4 for further details of the initial SEMs).The SEM for R a showed that SM positively influenced R a both directly and indirectly by impacting BNPP (Fig. 4a, Table S5).Furthermore, BNPP was positively correlated with ANPP, which was positively associated with SM (Fig. 4a, Table S5).The SEM for R h showed that SM positively influenced R h either directly or indirectly by impacting MBN (Fig. 4b, Table S6).In addition, MBN was positively correlated with MBC, which was positively associated with SM and negatively associated with ST (Fig. 4b, Table S6).

Discussion
Based on a three-year manipulative experiment representing commonly practiced land use (Baoyin et al. 2014;Wu et al. 2015;Wang et al. 2020b) and natural interannual variation in precipitation (Fig. S3), we investigated their impacts on R s and its components as well as abiotic (e.g., soil moisture) and biotic (e.g., plant productivity and soil microbial biomass) drivers of respiration.We found that precipitation had more substantial impacts on autotrophic, heterotrophic and total R s than land use.In addition, precipitation and land use interactively impacted the abiotic and biotic drivers of respiration, leading to interactive impacts on R s and its components in certain years and a further decrease of ~ 5% in R h based on the 3-year average when decreasing precipitation was integrated with grazing or mowing.Despite previous studies on the precipitation effect (Liu et al. 2009(Liu et al. , 2016;;Hashimoto et al. 2015) or the land use effect (Liu et al. 2012;Han et al. 2012), quantification of their interactive Details of these SEMs can also be found in Table S5-S6.Details of initial SEMs can be found in Fig. S13 and Table S3-S4 effect remains rare.Our study highlighted the complex interactive effects of precipitation amount and land use on soil carbon fluxes through their influence on abiotic and biotic factors.We suggest that precise prediction of the consequences of climate change and land use should incorporate these interactions.
Effects of precipitation amount and land-use regime on soil respiration and its components In grasslands, water is the most important limiting factor of biological activities (Liu et al. 2009(Liu et al. , 2016;;Suseela et al. 2012;Suseela and Dukes 2013;Wang et al. 2021), and grazing and mowing are the most common land uses (Baoyin et al. 2014;Zhou et al. 2019;Tang et al. 2019;Wang et al. 2020b;Bardgett et al. 2021).The current study found that precipitation positively impacted R s and its components, leading to substantial decreases under a 50% decrease in precipitation and increases under a 50% increase in precipitation (Figs. 3 and 4).Land uses such as grazing and mowing reduced R s and its components to a much lesser extent than the impacts of 50% changes in precipitation (Fig. 3).On the one hand, the detected precipitation and land use impacts are in line with previous grassland R s investigations (Liu et al. 2009;Geng et al. 2012;Wei et al. 2016;Tang et al. 2019;Wang et al. 2020d).On the other hand, the much stronger precipitation impacts on R s and its components resulted from the more substantially impacted abiotic and biotic driving factors by the 50% changes in precipitation (Figs. 1 and 2).In our manipulative experiment, the implemented 50% changes in precipitation reflect its natural interannual variation (Fig. S3); the grazing intensity and mowing frequency represent the most common land-use practices of the study region (Baoyin et al. 2014;Wu et al. 2015;Wang et al. 2020b).Thus, our results indicate that interannual variation in precipitation has a much stronger impact on R s and its components than land use in this region.
We also found evidence for the interactive effects of the precipitation amount and land use on R s and its components.This is particularly the case for R h .Averaged over the 3-year study period, grazing neutrally and mowing negatively impacted R h under ambient and increasing precipitation, while under decreasing precipitation, both grazing and mowing negatively impacted R h (Fig. 3).The detected interactive impacts are consistent with a recent manipulative experiment combining increased precipitation and grazing in a meadow steppe of northeastern China (Wang et al. 2020d).This experiment showed that increased precipitation alleviated the negative impacts of grazing on R s and its components over a 2-year study period (Wang et al. 2020d).These results suggest that the precipitation and land use interaction may be a general phenomenon in Inner Mongolian grasslands.More importantly, the current study moves forward from previous findings by further showing the underlying biotic mechanisms of this pattern, i.e., changes in precipitation and land use interact in influencing the biological activities of plants and soil microbes, leading to interactions with R s and its components (Fig. 4).During recent decades, the Inner Mongolian grassland has been experiencing a pronounced reduction in precipitation (Piao et al. 2010;Huang et al. 2015) and an increasing intensity in land use (Wu et al. 2015;Li et al. 2018;Wang et al. 2020a).
Our results indicate that these changes can interact and cause further decreases in R s and its components.Thus, a precise prediction of ecosystem carbon cycling in response to climate change in this region should consider the context and increasing intensity of land use.
Potential drivers of soil respiration and its components R s is composed of autotrophic and heterotrophic components, which are the results of root and soil microbial activities, respectively (Zhou et al. 2007;Bagchi and Ritchie 2010;Hou et al. 2014;Wang et al. 2020d).Thus, it is not surprising that plant primary productivity and soil microbial biomass C and N are significant predictors of R s .We also found that soil moisture was closely correlated with plant primary productivity and soil microbial biomass C and N. Since both precipitation and land use treatments can influence soil moisture, we speculate that soil moisture might be the nexus through which precipitation and land use interact to influence soil respiration.This speculation is consistent with previous evidence.Soil moisture has been shown to substantially affect R h in many previous works (Liu et al. 2009(Liu et al. , 2016;;Suseela et al. 2012;Suseela and Dukes 2013;Chen et al. 2016;Tang et al. 2019;Wang et al. 2021), potentially by influencing either the growth of soil microbes or the diffusion of substrates (Schmidt et al. 2004;Hungate et al. 2007;Wang et al. 2021).A previous precipitation manipulation experiment in our region found that increased soil moisture under high precipitation amounts can improve the biomass of soil microbes and therefore stimulating R h and R s (Liu et al. 2009).This pattern is consistent with the current study.Moreover, our SEM further showed a direct impact of soil moisture on R h (Fig. 4b), suggesting the importance of substrate diffusion in regulating the R h responses in this region.
R s can also be strongly regulated by plant productivity, as a productive plant community can allocate more photosynthates belowground, stimulating R a and thus R s .At global and regional scales, previous studies have shown that precipitation and/or soil moisture can impact R s by influencing plant productivity (Geng et al. 2012;Peng et al. 2013), which is consistent with the current study (Fig. 4a).In the current study, the detected regulation of R a by BNPP may result from the capacity of plants to physiologically adapt to water stress by adjusting stomatal conductance (Chaves et al. 2002) and changing the vertical distribution of roots to use groundwater from deeper soils (Jackson et al. 2000).For example, a recent 5-year precipitation manipulation experiment showed that decreasing the precipitation amount by 50% resulted in more plant roots distributed in deeper soils (Liu et al. 2018).Changing precipitation amount and land-use regime can influence plant community productivity by chronically shifting its structure in the long term (Xu et al. 2015a;Liu et al. 2018).The discovered regulation of R a by plant productivity suggests that the long-term impacts of these changes on R a and R s may be different from the short-term impacts.
It is worth noting that we constructed SEMs only with the 2019 measurements because MBC and MBN data were only available for this year.This means that the relationships between soil respiration and its potential drivers were determined based on spatial variations in these variables across treatments and plots.As a result, the findings from the SEMs may not be applicable to describe how these drivers influence soil respiration over time.This is because 1) the range of variation in these biotic and abiotic drivers might be different over years than across plots within the same year, or 2) the responses of soil respiration to these drivers might vary over time.However, we speculate that the findings of the SEMs might be robust.First, the detected impacts of abiotic (ST and SM) and biotic (plant productivity and soil microbial biomass) factors on R a and R h in this study are consistent with those in previous studies in this region (Liu et al. 2009;Xu et al. 2015b;Zhang et al. 2017;Wang et al. 2020d).For example, the negative effects of ST and positive effects of SM on plant productivity, soil microbial biomass, R a and R h are consistent with previous manipulative experiments in this region (Liu et al. 2009;Xu et al. 2015b;Zhang et al. 2017).Second, regressions between variables that were measured multiple times over the three years showed that the results were qualitatively consistent over the three years .Taken together, the findings from our SEMs are potentially applicable more generally.

Conclusion
Our manipulative experiment combined 50% changes in precipitation and land uses such as grazing and mowing, which represent the natural interannual variation in precipitation and the commonly practiced land uses in the study region, respectively.We found that R s and its components were more substantially affected by the 50% changes in precipitation and were interactively impacted by precipitation and land use, especially R h .These results indicate that, in this region, interannual variation in precipitation has a much stronger impact on R s and its components than land use, and consideration of the land use context and intensity should be helpful to precisely predict grassland carbon cycling in response to future precipitation changes.We also quantified abiotic and biotic factors driving the autotrophic and heterotrophic components of R s and found that soil moisture can impact them directly or indirectly by influencing the biomass production of plant roots and soil microbes.Changes in precipitation and land use can chronically shift the structure of the plant community, leading to long-term impacts on plant community productivity that differ from those in the short term.The detected regulation effect of plant and soil microbial biomass production on respiration responses suggests a bias in predicting the long-term respiration responses with short-term results, highlighting the importance of long-term manipulative experiments.

Fig. 1
Fig.1Effects of altered precipitation amount (P) and landuse regime (L) on soil temperature (a-d) and moisture (e-h) at 5 cm depth, aboveground (ANPP, i-l) and belowground (BNPP, n-p) net primary productivities.The box plots showed the mean and median (solid and dashed black lines in the boxes), interquartile ranges (boxes) and 10th and 90th percentiles (short black lines).Black cycles represent actual mean values.Results (F values) of the analysis of variance

)Fig. 2
Fig. 2 Effects of altered precipitation amount (P) and landuse regime (L) on soil microbial biomass carbon (MBC, a) and nitrogen (MBN, b).The box plots showed the mean and median (solid and dashed black lines in the boxes), interquartile ranges (boxes) and 10th and 90th percentiles (short black lines).Black cycles represent actual mean values.Results (F values) of the analysis of variance are shown in figure and

Fig. 3
Fig. 3 Effects of altered precipitation amount (P) and landuse regime (L) on soil respiration (R s , a-d) and its autotrophic (R a , e-h) and heterotrophic (R h , i-l) components.The box plots showed the mean and median (solid and dashed black lines in the boxes), interquartile ranges (boxes) and 10th and 90th percentiles (short black lines).Black cycles represent actual mean values.Results (F values) of the analysis of variance are shown in figure and indicated by *** when P < 0.001,

Fig. 4
Fig. 4 Diagrams of final structural equation models (SEMs) for relating the autotrophic (a) and heterotrophic (b) components of soil respiration to their biotic and abiotic driving factors.These SEMs were constructed by using observations only in 2019 (n = 27 observations for each variable).All arrows are scaled in relation to the strength of the relationship, with numbers showing the standard path coefficients and indicated by *** when P < 0.001, ** when P < 0.01, * when P < 0.05, # when P < 0.10 and n.s.(not statistically significant) when P > 0.10.R 2 values are proportions of variance explained by dependent variables in the model.Model-fit statistics such as χ 2 -test, RMSEA and GFI are shown in each panel.Details of these SEMs can also be found in TableS5-S6.Details of initial SEMs can be found in Fig.S13and TableS3-S4