Variation of livestock grazing intensity modified the magnitude of carbon sequestration and flow within the 1 plant-soil system of a meadow steppe ecosystem 2

30 Aims: Livestock grazing, one of the principal utilization patterns, usually exerts a substantial effect on the carbon 31 allocations between the above- and belowground components of a grassland ecosystem. The major aims of this 32 study were to evaluate the proportions of 13 C allocation to various C pools of the plant-soil system of a meadow 33 steppe ecosystem in response to livestock grazing intensity. 34 Methods: In situ stable 13 C isotope pulse labeling was conducted in the plots of a long-term grazing experiment 35 with 4 levels of grazing intensities. Plant and soil materials were sampled at on eight occasions (0, 3, 10, 18, 31, 56 36 and 100 days after labeling) to analyze the decline in 13 C over time, and their composition signature of 13 C were 37 analyzed by the isotope ratio mass spectrometer technique. 38 Results: We found a significantly larger decline in assimilated 13 C for the heavily grazed swards compared to other 39 grazing intensities, with the relocation rate of 13 C from shoots to belowground C pool being the highest. In contrast, 40 light grazing significantly allocated 13 C assimilates in the belowground pool, especially in the live root and topsoil 41 C-pools. 42 Conclusions: The effects of livestock grazing on the carbon transfers and stocks within the plant-soil system of the 43 meadow steppe were highly intensity dependent, and different carbon pools differed in response to gradient changes 44 in grazing intensity. 45

Grazing is the most widely adopted land use of grassland ecosystems, being among the most critical factors 58 affecting their phytomass production and matter circulation (Zhou et al. 2019). In terms of scale, the degradation of 59 approximately 35% of the world's total area of grasslands can be attributed by overgrazing, the impacts of which far 60 exceed any other land use type (WRI 2018). To date, the impacts of grazing on the carbon cycling of grassland 61 ecosystems have mostly focused on the aspects of net primary production, root biomass, soil stock, and soil 62 respiration, whereas the effects of grazing on the carbon flow transfers within the plant-soil system, especially 63 8 To guarantee a uniform distribution of 13 CO2, four 5-volt fans were installed in each upper corner of the chamber. 130 Consequently, the plants assimilated the labeled 13 CO2 for 4 hours during the 3 days of treatments, generally from 131 8:00 am to 12:00 am, after which the chamber was removed. In addition, a cooling system was run during the time 132 when 13 CO2 labeling was conducted to prevent the plants from being affected by high temperatures that may inhibit 133 photosynthesis inside the chamber. 134 135

Tissue sampling 136
After the pulse labeling had been done, plant and soil samples were collected on eight occasions (0, 3, 10, 18, 31, 56 137 and 100 days after the treatment) during the entire study period. To obtain the combined mean residence time (MRT) 138 of carbon in the aboveground vegetal tissues, living shoots were additionally sampled two more times (316 and 394 139 days). At each sampling time, the plant-soil system was separated into the following components: shoots (live and 140 standing dead), roots (living and dead, separated by different specific gravity in water), and rhizospheric soil (fine 141 soil, bulk soil sieved at 2mm). One sample point was collected in each chamber at every sampling time, and there 142 were 3 chambers in each plot. Above-ground tissues were cut from a small round area with a diameter of 10 cm, and 143 the litter in the spot was collected after above-ground tissues were removed, and then rhizospheric soil and root 144 samples were immediately collected with the 10cm diameter soil core from three soil layers (0-10, 10-20 and 20-30 145 cm) that were sampled separately. Carbonate in the rhizospheric soil was removed by by adding 10 M HCl for 3 146 days, then neutralized by adding deionized water and dried (Harris et al. 2001). All plant materials were dried, 147 weighed and ball milled. Three replicate plant samples (unlabeled spots) were taken from each block and were used 148 as a natural reference. The C isotope composition (δ 13 C, ‰) signature and the carbon contents of shoot, root, soil 149 9 and control samples were measured by an isotope ratio mass spectrometer (IsoPrime 100, Isoprime, UK) coupled 150 with an elemental analyzer (Elementar, Vario Pyro Cube, Germany). 151 152

Calculations and statistical analyses 153
The carbon isotope abundance ratio ( 13 C/ 12 C) of a given sample is expressed as Rsample relative to the δ 13 C and 154 international standard Pee Dee Belemnite (PDB). The carbon isotopic abundance ratio of any sample is thus 155 expressed as shown in Eq. (1): 156 (1) 157 where RPDB is the isotope abundance ratio of PDB (RPDB = 0.01112333). The enrichment value of 13 C as 13 C atom%, 158 excess (% of total C atoms) is shown in Eq. (2-3): 159 (2) 160 (3) 161 where 13 Catom%, sample (% of total C atoms) and 13 Catom%, nature (% of total C atoms) are the amounts of 13 C in the sample 162 and natural abundance, respectively. To determine the amount of 13 C incorporated into various plant and soil pools, 163 equation (4) was used: 164 where 13 Ct, pool is the mass of 13 C (g/m 2 ) in the considered pool at time t after labeling, 13 Ct, atom%, excess is the 166 increase in 13 C at t time in the considered pool. The percentage of 13 C recovered in each pool ( 13 Ct, rec) is defined by 167 the sum of the total assimilated 13 C mass of day 1 in the plant-soil system pools (see Eq. (5)) and we also assume 168 that the decreased amount of assimilated 13 C mass in plant-soil system over time is attributed to ecosystem 169 10 respiration. 170 The first-order exponential decay function was used to simulate assimilated 13 C decreased in shoot C pool, and 172 the expressed was shown in Eq. (6): 173 where 13 Ct, shoot is the percentage of 13 C recovered at t time in shoot, 13 C0, shoot is the percentage of 13 C recovered at 175 day 1 in shoot, a is the intercept of the model, and k is the coefficient of decay rate. The MRT was calculated by 176 reciprocal of k. 177 For the comparison of C sequestration among the four grazed treatments, C stocks (Mg C ha -1 ) of the above-178 and below-ground mixed vegetation biomass and of the soil were calculated. Carbon stocks in the soil layers 0-10, 179 10-20 and 20-30 cm were calculated using the following equation: 180 where Ti (cm) is the thickness, BDi (g cm -3 ) is the bulk density and SOCi (%) is the C content of the i th soil layer. 182 The significance test of differences among grazing treatments with respect to biomass, C pool, and 13 C mass 183 was carried out by a linear mixed model, in which block and sampling time were considered random effects. To 184 conduct a multiple comparison between every two grazing intensity levels, a post hoc Tukey HSD test was used. 185 The first-order exponential decay function was fitted by the nonlinear simulation model (NLS). All graphics and 186 statistical analyses were carried out by R software (R Core Team 2019). 187 188

Dynamics of recovered 13 C in shoots 191
Recovered 13 C in the shoots under all grazing intensities followed a first-order exponential decreasing trend with the 192 time of labeling (Fig. 2), with the MRT of 13 C in shoots averaging 19 days. Among the treatments, a significantly 193 larger decline trajectory in 13 C recovery was detected only in the heavily grazed swards (G0.92), which occurred 194 most notably during the period of the 10 th to the 56 th days after labeling. The allocation rates of 13 C from shoots to 195 the belowground C pools and shoot respiration as of the 56 th day after labeling were calculated as 66.43%, 52.81%, 196 50.03% and 63.32% of the assimilated 13 C for the 4 treatments, respectively, with that of the G0.00 and G0.92 197 treatments being significantly higher than the rest of the treatments. The remaining amounts of 13 C were largely 198 retained in the standing dead and litter pools. 199 200

Recovered 13 C allocated to below-ground pools 201
The recovery rates of 13 C to the belowground pools were much less temporally variable compared with those in the 202 vegetation during the first 100 days for all the individual pools of each treatment (Fig. 3). 203 Of the individual below-ground components, approximately 5.04% to 18.08% of 13 C was allocated to the live 204 root pools of all treatments after pulse labeling. Of special note, significantly lower proportions were measured in 205 the heavily grazed subplots on most of the sampling occasions, whereas significantly higher values were measured 206 in the lightly grazed subplots (G0.23) on three sampling occasions out of the total seven occasions. 207 In contrast, a range of 0.84% to 4.73% of recovered 13 C was measured in dead roots of all treatments during the 12 chase period. The temporal patterns were much less variable and highly comparable among the treatments, and no 209 significant differences in either the trend or the bulk were found (Fig. 3). 210 The 13 C recovery in the 0-30 cm soil pools was basically comparable to that in the live roots but much higher 211 than that in the dead roots of all treatments and on any sampling occasion during the chase period (Fig. 3). In 212 addition, the temporal trajectories were fairly level and comparable among the treatments. Overall, compared to 213 ungrazed, grazing led to a lower 13 C recovery rate in all individual soil layers, and at the entire soil profile. On the 214 whole soil 13 C recovery rate was highest under light grazing (G0.23) followed be high grazing (G0.92) and moderate 215 grazing (G0.46). 216 When all measurements on all sampling occasions were taken together until 31 st day after labeling, the 217 relationships between grazing intensity and recovered 13 C in live roots, soil and the total belowground pool, 218 displayed bump-shaped curves where recovered 13 C increased with grazing intensity until a maximal point and then 219 declined. This maximal point was reached under moderate grazing (G0.46) which had the largest enhancing effects 220 on recovered 13 C in soil and total belowground pools, but under light grazing (G0.23) in live root pool (Fig. 4). 221 When considering the relation between recovered 13 C in live roots to soil at all sampling dates, we found that more 222 percentage of assimilated 13 C was found in soil compared to roots under heavy grazing, where more 13 C was found 223 in roots compared to soil in the ungrazed and lightly grazed subplots, whereas almost equal % of recovered 13 C was 224 found under moderately and lightly grazed subplots (Fig. 5). 225

226
Effects of grazing on carbon allocation 227 Fig. 6 shows that grazing substantially lowered the total carbon fixation rate (9.8% compared to 17.4%), the 228 13 decreased degree of which apparently increased with grazing intensity, whereas the efficiency of conversion from 229 photosynthesis to shoot biomass as well as from shoot biomass to standing dead and litter was reduced by heavy 230 grazing to variable extents. However, light and moderate grazing indeed slightly enhanced the conversion rates from 231 photosynthesis to shoot biomass. In addition, the efficiency of conversion values from shoots to live roots was 232 enhanced by light grazing but lowered by moderate and heavy grazing, while those for roots and aboveground litter 233 to the soils were promoted to certain degrees in all grazing treatments. Concerning transfer to soil pool, grazing 234 treatments were comparable, whereas ungrazed treatment was the lowest 235 236

Changes in carbon stocks with increasing grazing intensity 237
An analysis of the plant and soil C stock data showed that the majority of the total carbon stock was contained in the 238 belowground pool, varying between 95.73% and 98.80% among the grazing intensity treatments. Of the 239 belowground carbon stock (live root, dead root and soil) , the soil component accounted for a much higher 240 proportion than the root compartment, which ranged from 93.70% to 96.11% among the treatments. However, 241 significant changes indeed occurred in standing dead material and living root carbon pool (0-30 cm soil layer), 242 showing for both variables the lowest amount under heavy grazing followed by moderate and light grazing and 243 grazing exclusion, respectively (Table 1). In regard to soil C pool, no significant changes in the total carbon stock 244 (0-30cm depth) were detected with increasing grazing intensity except for heavy grazing which had a higher soil C 245 stock than light grazing (Table 1). 246 Among the components of the above-ground carbon stock, the live shoots contained the largest fraction in 247 all the treatments, followed by the litter and standing dead vegetation. In stark contrast to the case for the 248 belowground carbon stock, significant decreases in both the total aboveground carbon stock and its components 249 were engendered by grazing, all of which displayed a decreasing trends with increasing grazing intensity. . Grass species usually have the ability to adapt their C allocation patterns to grazing activity 257 (Fig. 6). The partitioning of 13 C was almost completed in shoots within 19 days after labeling. Thus, measurements 258 of 13 C lasting for 31 days after labeling should have reflected a steady state of the recovered 13 C in the plants (Wu et 259 al. 2010). It is crucial to discriminate the meanings of 13 C content and 13 C proportion. The former represents the 260 amount of 13 C mass remaining in the considered pools, while the latter mainly indicates the 13 C allocation amount. 261 Grazing reduced the quantities of total C fixation, and the amounts of carbon in the plant-soil C pools 262 exhibited obvious decreases to varying degrees with increasing grazing intensity, except for the SOC of the light 263 grazing treatment (Fig. 6)

Effects of grazing intensity on the dynamics of 13 C recovered in shoots 277
The dynamics of recovered 13 C in shoots generally reflect transfers of assimilated C into belowground pools and C 278 under this grazing intensity. A further analysis revealed that this increased turnover was due mainly to shifts in the 287 plant species composition of the sward at this grazing intensity. Indeed intense grazing was characterized by 288 16 significant increases in the abundance of annuals and a few forbs or C4 plant species. Zhao et al. (2015Zhao et al. ( , 2017 found 289 that C lost by shoot respiration was significantly greater for annual than perennial herbs. 290 291

Effects of grazing intensity on carbon transfers to below-ground pools 292
We observed significant increases in the 13 C allocation proportion from shoots to the live root pool at G0.23 and 293 slight increases at G0.46 but significant decreases at G0.92 compared with the allocation proportions in the grazed 294 plots over time (Fig. 3). The enhanced nutrient requirements of grasses under moderate grazing may often increase 295 C allocation from the canopy to the roots. In addition, the necessity for defoliated plants to allocate more C 296 belowground to maintain enhanced activities of live roots or as storage for regrowth after grazing may also be 297 partially responsible for the larger belowground C transfer from the canopy to the roots in this study. This was 298 consistent with the results reported by Hafner et al. (2012), who found that 13 C allocation to the below-ground pool 299 was significantly larger at the moderately grazed site than heavy grazed in Tibetan meadow grassland. Furthermore, 300 several studies reported that light grazing stimulated more photosynthetically fixed C inputs to roots, leading to 301 Pausch and Kuzyakov (2018) noted that heavy grazing tended to stimulate C transfer from the roots to the soil. 308