1. Due to complex root-soil interactions, the responses of carbon (C) dynamics in the rhizosphere to elevated nitrogen (N) deposition may be different from those in bulk soil. However, the potentially different response of C dynamics in the rhizosphere and bulk soils and their contributions to soil C sequestration under N deposition is still not elucidated.

2. We conducted an N addition experiment in an alpine shrubland dominated by Sibiraea angustata located on the eastern Qinghai-Tibet Plateau (QTP). We measured the soil organic C (SOC) contents and density fractions in the rhizosphere and bulk soils in the top 15 cm of mineral soil and then employed a numerical model based on the rhizosphere extent to evaluate how the rhizosphere modulates soil C sequestration under N addition. We also measured the microbial gene abundance and C-acquisition enzyme activities to assess microbial community responses to N addition.

3. The results showed that nitrogen addition had opposite effects on the rhizosphere and bulk-soil C stocks. Specifically, N addition decreased the rhizosphere SOC content through increasing bacterial abundance, β-glucosidase activity, and thus accelerating the loss of free light fraction C (FLF-C). However, N addition increased the bulk-soil C content, which was corresponding with the reduced oxidase activities and the accelerated accumulation of heavy fraction C (HF-C) under N addition. Numerical model analysis showed that the decrease induced by N addition in rhizosphere SOC stock ranged from 0.11 to 3.01 kg C m^-2 as root exudation diffusion distance extended from 0.5 mm to 2 mm, while the corresponding increase in the bulk-soil C stock ranged from 1.91 to 4.08 kg C m^-2. By synthesizing the dynamics of the SOC stocks in these two soil compartments under N addition, the SOC stock at the ecosystem level exhibited an increase in range of 0.73-2.44 kg C m^-2.

4. Synthesis Our results suggest that alpine shrublands on the eastern QTP have great potential for soil C sequestration under N deposition, and the magnitude of the sequestration would depend closely on the responses of rhizosphere microbial C processes and the rhizosphere extent. Our results highlight the importance of integrating rhizosphere processes into land surface models to accurately predict ecosystem functions in the background of elevated N deposition.

1. Study site: The Field Research Station of Alpine Shrubland Ecosystem is located in Songpan County, Sichuan Province, China. This region belongs to a typical alpine climate. The community composition of plants was dominated by temperate and cold-temperate vegetation at our study region. The dominant shrub species was Sibiraea angustata, followed by Salix oritrepha, Spiraea alpine, and Potentilla fruticosa (Wang et al., 2017). The average coverage, basal diameter and height of Sibiraea angustata are 53%, 0.62 cm and 1.03 m, respectively. Our previous study showed that the aboveground and belowground biomasses of Sibiraea angustata were 3.71 kg m^-2 and 4.39 kg m^-2, respectively, accounting for 96% and 97% of the aboveground and belowground biomass (Wang et al., 2017). These results suggest that both the aboveground- and belowground-biomass of Sibiraea angustata have an overwhelming superiority within this shrub community.                                                                                                             

2. The manipulated experiment of nitrogen addition: An experiment of nitrogen addition was initiated within this shrubland ecosystem in May 2012. A randomized block design with three replicated blocks of three treatments was established before the experiment; these treatments included the control (no N fertilizer), low-N addition (50 kg N ha^-1 year^-1), and HN (100 kg N ha^-1 year^-1). Each block included three plots in 5 m × 5 m, and each plot was surrounded by 10-m-wide buffer strips. According to the average increase of bulk N deposition across China (approximately 8 kg N ha^-1 yr^-1; Liu et al., 2013) and the level of N fertilizer experiments performed in neighbor sites on the QTP (10 and 350 kg N ha^-1 year^-1; Fu et al., 2017), we chose the doses of the N additions in our study equal to be nearly 3- and 5-times the value of the background N deposition. An ammonium nitrate (NH₄NO₃) solution has been sprayed onto the forest floor in 6 equal doses in the first week of every month from May to October (i.e., the growing season) since May 2012. In each N-addition event, the fertilizer was weighed, dissolved in 20 L of water, and sprayed evenly using backpack sprayers. The control plots received an equal volume of water.

3. Soil sampling protocol: We sampled the upper 15 cm of mineral soil in July 2017. After organic horizon removal, five replicate cores from each plot using a 6-cm diameter soil corer to ensure that fine roots would have a sufficient mass of adhering rhizosphere soil. The soil cores were stored on ice transported to the laboratory. The living roots of Sibiraea angustata in each core were empirically identified by features such as shape, color, and elasticity. The soil adhering to the roots was carefully separated from the roots using fine forceps; this fraction was operationally defined as rhizosphere soil, and non-adhering soil was considered to be bulk soil (Phillips et al., 2011). All soil samples were passed through a 2-mm mesh sieve and were divided into two parts: one part was used to determine the soil physicochemical properties and the soil carbon density fractionation; another part was stored at -20 °C for later analyses of microbial biomass carbon and enzymatic activity.

4. Basic physicochemical properties: For soil bulk density (SBD) determination, a bulk-density corer with 5-cm diameter stainless steel rings as an inner sleeve was manually inserted into the upper 15-cm depth of the mineral soil (three corers for each plot) (Davidson et al., 2004). The SOC was determined by dichromate oxidation and titration with ferrous ammonium sulfate (Walkley and Black, 1934). The DOC was extracted with 0.5 M K2SO4 and determined using a TOC analyzer (Vario TOC, Elementar Corp., UK). The total N concentration (SON) in soil was determined on a TOC analyzer (Vario TOC, Elementar Corp., UK). Soil NH₄⁺ and NO₃^- were extracted by 2 M KCl (soil:solution = 1:5) and then determined on a continuous flow injection analyzer (SEAL Analytical, Germany). The dissolved inorganic N (DIN) was the sum of NH₄⁺ and NO₃^-. The soil organic N concentration (SON) in soil was the difference between TSN and DIN. Available phosphorus (Av.P) was extracted with Bray-I solution (0.03 M NH4F - 0.025 M HCl) (Bray and Kurtz, 1945) and was determined by molybdenum antimony colorimetry. Soil pH was measured in slurry with a soil-to-water ratio of 1:2.5 using a pH meter (Mettler-Toledo Instruments Co., Ltd., Shanghai, China).

5. Soil carbon fractionations: The SOC fractions were separated using a density fraction method (McLauchlan and Hobbie, 2004). Briefly, a 15 g of air-dried rhizosphere or bulk soil (passed through 2mm-mesh sieve) was weighed and placed in a 100 mL centrifuge tube and dispersed in 50 mL of NaI (with a density of 1.8 g cm⁻³). The tubes were centrifuged at 3000 rpm for 20 min. The suspended materials (FLF) were decanted into a vacuum filter unit with 0.40 µm polycarbonate filter. This process was repeated 2-3 times until no floating material remained. The materials remaining at the bottom (HF) of the centrifuge tube were then rinsed into the vacuum filter unit. All samples on the filter paper were washed with 75 mL 0.01 mol L⁻¹ CaCl₂, followed by at least 75 mL of distilled water. The light and heavy materials were dried at 60°C for 48 h and weighed. All samples were passed through a 0.25 mm mesh sieve and analyzed for SOC as previously described.

6. Microbial gene abundance: For microbial gene abundance, DNA was extracted from 0.25 g of soil using the MoBio Power Soil DNA isolation kit (Mobio Laboratories, CA, USA). DNA quality and concentration were measured using a nanodrop spectrophotometer (NanoDrop, DE, USA) and electrophoresis in agarose gels (1% w/v in TAE), then stored at -20 °C prior to amplification. Quantitative PCR (qPCR) was used to quantify the gene copy numbers of bacterial 16S rRNA and fungal ITS using the primer pairs 515F/909R and ITS7F/ITS4R, respectively (Li et al., 2014; Schulz et al., 2018). Each 10-µL reaction contained 5 µL of SybrGreen (2×) PCR Master Mix (Bio-Rad, USA), 0.5 µL of each primer (10 pM), 2 µL of DNA templates and 2.5 µL of sterilized water. Bacterial 16S rRNA and fungal ITS conditions were 5 min at 95 °C, followed by 40 cycles of 95 °C for 30 s, 30 s at 55 °C, and 30 s at 72°C, with a final extension cycle of 8 min at 72°C. The qPCR standards for quantification were prepared from PCR products of target genes from environmental DNA with each primer set using the method described by Kou et al. (2017). Four replicates were performed for each sample. The amplification efficiencies of the 16S rDNA gene and ITS gene were 90% and 92%, respectively, with R² values higher than 0.99, and no signals were observed in the negative controls.

7. Carbon-acquisition enzyme activities: Two grams of sieved soil was suspended in 125 mL of 50 mM sodium acetate buffer (pH = 5.0) in a homogenizer for 1 min to form slurry. Black 96-well microplates were used for fluorometric analysis. The microplates were assigned to six parts, including the sample assay, sample control, quench control, reference standard, negative control, and blank wells. First, 200 μL of buffer was pipetted into the blank, reference standard and negative control wells. Second, 50 μL of buffer was pipetted into the blank and sample control wells. Third, 200 μL of the soil slurry was pipetted into the sample assay, sample control, quench control wells, and then 50 μL of 10 μM 4-methylumbelliferyl (MUB) was pipetted into the reference standard and quench standard wells. Finally, 50 μL of 200 μM fluorogenic substrate (4-methyl-umbelliferyl β-D-glucopyranoside) were pipetted into into the negative control and sample assay wells. Plates were incubated for 4 h in the dark (25°C) and then scanned on a Varioskan Flash multiplate reader (Thermo Scientific, USA) at 365 nm excitation and 450 nm emission wavelengths. The unit for BG activity was expressed as μmol MU g^-1dry soil (dry weight) h^-1.

Phenol oxidase and peroxidase activities were measured spectrophotometrically using _L-3, 4-dihydroxyphenylalanine (_L-DOPA, Sigma, St. Louis, USA) as the substrate. A total of 200 μL soil suspension (see above) and 50 μL of 25 mM DOPA were added to each sample well. The wells of peroxidase assays additionally received 10 μL of a 0.3% H₂O₂ solution. The microplates were incubated in the dark at 20°C for up to 8 h. Absorption was measured at 450 nm and expressed in units of μmol DOPA g⁻¹ h⁻¹. The background absorbance of DOPA was measured, and an extinction coefficient was calculated using a standard curve of DOPA degraded with mushroom tyrosinase.