Nitrogen deposition may enhance soil carbon storage via change of soil respiration dynamic during a spring freeze-thaw cycle period

As crucial terrestrial ecosystems, temperate forests play an important role in global soil carbon dioxide flux, and this process can be sensitive to atmospheric nitrogen deposition. It is often reported that the nitrogen addition induces a change in soil carbon dioxide emission in growing season. However, the important effects of interactions between nitrogen deposition and the freeze-thaw-cycle have never been investigated. Here we show nitrogen deposition delays spikes of soil respiration and weaken soil respiration. We found the nitrogen addition, time and nitrogen addition×time exerted the negative impact on the soil respiration of spring freeze-thaw periods due to delay of spikes and inhibition of soil respiration (p < 0.001). The values of soil respiration were decreased by 6% (low-nitrogen), 39% (medium-nitrogen) and 36% (high-nitrogen) compared with the control. And the decrease values of soil respiration under medium- and high-nitrogen treatments during spring freeze-thaw-cycle period in temperate forest would be approximately equivalent to 1% of global annual C emissions. Therefore, we show interactions between nitrogen deposition and freeze-thaw-cycle in temperate forest ecosystems are important to predict global carbon emissions and sequestrations. We anticipate our finding to be a starting point for more sophisticated prediction of soil respirations in temperate forests ecosystems.

winter 7 ; (2) increased CO 2 emissions may be due to enhancing microbial metabolism by substrate supply in the FTC period 21 ; (3) increased substrates leaching from the litter layer accumulated during the winter might lead to CO 2 burst 24 . Currently, there are still many uncertainties in the mechanisms of these increased CO 2 fluxes 25 . The first objective of our study was to examine the impact of spring FTC on the soil CO 2 emissions in the temperate forest, and then to investigate the mechanisms potentially inducing FTC period CO 2 emissions.
In addition, atmospheric nitrogen (N) deposition is another important factor to soil C cycle, because the cycles of soil C and N are closely coupled [26][27][28] . Some previous studies showed that simulated N addition had significantly increase release of CO 2 29 . Nevertheless, other studies found controversial affecting soil CO 2 .fluxes in terrestrial ecosystem [30][31] . The different responses of soil CO 2 fluxes to N addition have been reported in the different ecosystem, including increases 26 , decreases 32 , and no significant differences [33][34] . Summary, most of the high concentration N deposition may limit CO 2 release, and low concentration may promote or no changes. Several potential mechanisms have been proposed to clarify the N-induced change of CO 2 emissions: (1) N inhibition of lignin degradation largely resulted from change of microbial composition 35 ; (2) change of CO 2 emissions may be due to ecological shifts in the soil microbiota under N deposition 36 ; (3) the coupling of soil carbon and nitrogen was broken due to N deposition, which might lead to change of CO 2 emission 37 . Although the effect of FTC on C cycles and the effect of atmospheric N deposition on C cycles have been investigated, respectively 28,38,39 , the effect of FTC together with atmospheric N deposition on C cycles has never been reported. We hypothesized that soil respiration (Rs) could have a special response pattern to N deposition due to the changes of soil physicochemical properties and microbial characteristic in the FTC period. The second objective of our study was to examine the impact of simulated N deposition together with spring FTC on soil CO 2 fluxes in temperate forest.
The major objective of this paper was to evaluate the change quantities of CO 2 due to N deposition addition in FTC period in temperate forests which cover 9.7% of the earth's continental surface 40 . We hypothesized that N deposition would inhibit CO 2 emissions via delay burst or decrease fluxes in spring FTC period in temperate forest. In addition, previous studies did not show an understandable mechanism regarding impact of FTC and N deposition on CO 2 fluxes. The FTC and N deposition could affect the soil biological and physicochemical processes leading to C dynamic change. Therefore, we conducted a simulated N deposition experiment from May 2010 to present, and investigated the interactive effects of N deposition and spring FTC on soil C fluxes in a temperate forest and the potential mechanisms in 2015.

Materials and Methods
Site description. This study was conducted at the Fenglin Natural Reserve of Lesser Khingan Mountains in Heilongjiang province, Northeast China (48°02′ -48°12′ N, 128°58′ -129°15′ E). The climate is continental monsoon climate, with dry, cold winters, and humid, warm summers. The forests had a mean annual temperature of − 0.5 °C from 1959 to 2013, with the lowest and highest monthly mean air temperature being − 25.6 °C in January to 23.8 in July, respectively. The mean annual precipitation is 728 mm, of which approximately 75% falls between July and August. The snowpack lasted for 148 days with the snow depth ranging from 0 to 42 cm during the measurement years (Nov, 2014-Mar, 2015. The soil is classified as a dark brown forest soil 41 . The vegetation type is a cold-temperate spruce-fir Korean pine forest with the age of exceeded 200 years. The community is dominated by Picea koraiensis, Abies nephrolepis and Pinus koraiensis. The mean stand density is 972 ± 96 trees ha −1 , the mean diameter at breast height is 13.7 ± 7.5 cm and the mean tree height is 16.7 ± 5.3 m. The major species in the canopy layer are Pinus koraiensis, Abies nephrolepis, Picea koraiensis, Picea jezoensis var. microsperma, Larix gmelini, Betula platyphylla, Acer mono, Fraxinus mandshurica and Betula costata. Experimental design. To investigate changes in soil CO 2 fluxes (Rs) following N application, we established three random blocks in May 2010, and each consisting of four research plots measuring 20 m × 20 m. The plots were separated by 10 m wide buffer strips to avoid horizontal movement of the soil N. The simulated N deposition was initiated at the onset of this experiment and included four treatments, control (no added N), low-N (5 g N.m −2 .yr −1 ), medium-N (10 g N.m −2 .yr −1 ) and high-N (15 g N.m −2 .yr −1 ), with three replicates randomly distributed at each treatment. The N was applied as ammonium nitrate (NH 4 NO 3 ) solution and was distributed on six occasions during annual growing season applied to the forest floor every half a month during the growing season (May to October) from 15th May 2010. In each plot, the NH 4 NO 3 was mixed with 32 L of water (equal to 0.08 mm annual precipitation), and applied by using a backpack sprayer below the canopy. Two passes were made across each plot to ensure an even distribution of the fertilizer. The control plots received 32 L water without N addition. The simulated N deposition was applied from May 2010 to the present. Soil CO 2 fluxes measurements during spring FTC period. The spring FTC period soil CO 2 fluxes (Rs) were measured every day from April 1 st to May 5 th 2015. For each of 12 plots, three polyvinyl Chloride (PVC) collars (20 cm inside diameter and 12 cm in height) were randomly inserted approximately 9 cm into the soil, with 3 cm left above the ground surface for Rs measurements, one week before N addition in 2010. A total of 36 soil collars were installed. The collars were left in the same place throughout the entire study period for exploring the change of the spring FTC period in Rs. The Rs was measured with a Li-8100 automated soil CO 2 flux system (Li-Cor Inc, Lincoln, NE, USA) between 10:00-14:00 in spring FTC period. Each measurement was repeated 3 times for each collar to produce a collar's mean Rs rate. Rs were calculated using exponential regression model with the LI-8100 file viewer application software (LI-8100/8150 Instruction Manual).
Soil physical and chemical properties and microbial characteristic measurements. The soil temperature at 5 cm depth (T 5 ) and soil volumetric water content at the 5 cm depth (W 5 , % v/v) were monitored simultaneously with the measurement of Rs by using a soil temperature probe (Omega Engineering Inc. USA) and soil moisture probes (Deltat Devices Ltd., Cambridge, England) connected to Li-8100. The continuous soil Scientific RepoRts | 6:29134 | DOI: 10.1038/srep29134 temperature at 5 cm depth (T 5cm ) was monitored hourly by Em-50 data logger (Decagon Devices, Inc. USA) The air temperature (T a ) was same measured hourly by Em-50 data logger (Decagon Devices, Inc. USA).
During the measurement of Rs, because of the difficulty in collecting soil samples from frozen soil, all soil samples were collected days nearly the soil collars from a depth of 0-10 cm using a specially designed auger (2.5 cm in diameter). Three soil cores were collected and pooled to one composite sample at each plot. All of the visible extraneous materials (such as roots, stones, etc.) were removed by hand, and then divided the composite sample into three sub-samples. One sub-sample was air-dried at ambient temperature, and then sieved (2 mm) and ground for the analysis of soil total C and total N by using an automated TOC/TN analyzer (multi N/C3100, Analytikjene AG, Germany). In addition, soil pH values were measured by a pH meter (SX7150, China) with soil: water ratio of 1:2.5. The second sub-sample was maintained original state, and taken back to laboratory. Thawed soils were mixed, whereas frozen soil was reduced to small pieces, with the pieces being homogenized to the extent possible 42 . Immediately following, the inorganic N concentrations were determined by extracting fresh soil with K 2 SO 4 . The extractable NH 4 + -N concentrations were measured by using the indophenol blue method, followed by the colorimetric analysis. The NO 3 − -N content was determined by using the copper-cadmium reduction method. The third sub-sample was also maintained original states, and taken back to the laboratory immediately to assess microbial biomass C (MBC) and N (MBN). The MBC and MBN were measured by using a fumigation-extraction method 43 . The extracts of N and C from fumigated and unfumigated samples were analysed by an automated TOC/TN analyzer (multi N/C3100, Analytikjene AG, Germany). The MBN and MBC were calculated from the difference between extractable N and C contents in the fumigated and the unfumigated samples using conversion factors (kEN and kEC) of 0.45 and 0.38, respectively 43 . All extraction for NO 3 -N, NH 4 -N, MBC and MBN was done with K 2 SO 4 of 0.5 mol l −1 in 25 °C, and the duration of extraction was half an hour.
Dividing the year into spring FTC period and Statistical analyses. The spring FTC period was defined as the period that starts when soil surface snow is start to melt (the maximum T a is above 0 °C) and ends when daily minimum T 5cm is above 0 °C 16 . The spring FTC period lasted for 35 days (DOY 90-124 in 2015) in this study.
To assess the quantity of Rs under different N addition level in the FTC periods, Rs-T models were constructed. Compared to the several commonly used models, such as the modified van't Hoff 's model 44 , the sigmoid-shaped Lloyd-Taylor 45 and logistic models 46 , the Gamma model performed either better or as good as the other models 47 .
In addition, Gamma model were tested across a wide Ts range (− 18-35 °C) and can also be expanded, using simple mathematics to help researchers analyse the Rs-T relationship in the context of other environment factors, such as soil nutrients 47 . The Gamma model was adopted based on R 2 and the Akaike Information Criterion (AIC). Therefore, Gamma model used to assess the impact of different quantities of N additions on Rs during the FTC period.
Gamma model was expressed as following: where T is (T 5cm + 40), a, b and c are regression coefficients. T 5cm is measured soil temperature under 5 cm below surface. 40 °C is added to T 5cm because negative T 5cm results in negative or imaginary Rs (or non-meaningful Rs), and 40 °C has been chosen as the lowest T 5cm where Rs continues has been measured at − 39 °C. The natural logarithm (Ln) transformed version of the Rs data was applied to alleviate the heteroscedasticity problem. Two-ways analysis of variance was used to examine the impacts of different quantities N deposition, spring FTC and their interactions on soil total C, total N, NH 4 + -N, NO 3 − -N, soil pH values, MBC, MBN. Fisher's LSD followed the two-way analysis of variance between the N treatments. Tukey's HSD tests were used to reveal the significant pairwise differences of the N addition. Pearson's correlation analysis was used to determine the correlations between Rs and soil properties or microbial characteristics. Statistically significant differences were accepted at p < 0.05. All statistical analyses were performed using R 3.2.2 Version Software (R Development Core Team 2015).

Effects of spring FTC, N deposition and their interaction on Rs.
At the beginning of the spring FTC period, the daily maximum T a was above 0 °C, but all of T 5cm were below 0 °C ( Fig. 1(a)), and the snow was melting. However, the Rs remained at a low level ( Fig. 1(b)), and the Rs under medium-N and high-N treatments was significantly lower than control and low-N treatments at the early stage of the spring FTC period ( Fig. 1(b)). The significant differences in Rs were observed on next period of time, and temporal peaks of Rs occurred. The ephemeral burst of Rs observed from DOY 97 to DOY 102 under control treatments and lasted for 6 days, with the maximum Rs of 0.83 μ mol m −2 s −1 (Fig. 1(b)). Simultaneously, we observed the high Rs occurred from DOY 98 to DOY 102 under low-N treatments and lasted for 5 days, with the maximum Rs of 0.76 μ mol m −2 s −1 (Fig. 1(b)). During the period, the daily mean of air temperature and the mean of soil temperature in 5 cm depth increased continuously ( Fig. 1(a)). The snowpack had melted completely. But, the ephemeral enhancement of Rs occurred at later stage of the spring FTC and lasted for 5 days (DOY 107-111) under medium-N and high-N treatments ( Fig. 1(b)). The Rs pulse lasted for a short time period and after that the rate decreased to the normal status during the spring FTC period. During most of observation period, the Rs increased with temperature.
The effects of different quantity of N addition on Rs were highly variable during spring FTC period. During the measurement period, the mean of Rs was 0.58, 0.57, 0.47, 0.48 μmol m −2 s −1 for different quantity of N addition, i.e., control, low-N, medium-N, high-N treatments, respectively (Table 1). Our results found that the simulated N deposition had significantly impact on the Rs due to inhibiting CO 2 fluxes or delaying outburst event (Table 1; Fig. 1(b)). Likewise, the FTC also had a significantly impact on Rs (Table 2), which varied from 0.32 to 1.06 μ mol m −2 s −1 and showed the high fluctuations under natural status (control plots) ( Fig. 1(b)). In addition, Rs was also significantly affected by the interaction of the simulated N deposition and spring FTC (p < 0.001) ( Table 2). In general, the two-way ANOVA analysis showed that the simulated N deposition, the spring FTC and their interaction exhibited significant effects (p < 0.001) on the Rs during the whole measurement period ( Table 2).
Spring FTC period contribution of Rs to the winter and annual budget and assessing the future C dynamic in temperate forest. Applying the empirical Rs-T models assessed the quantities of Rs under different N addition levels (i.e., control, low-N, medium-N, high-N) during the spring FTC period. The ordinary least squares was used to calculate the coefficients (i. e., a, b, c), similar to what was performed in the Khomik 2009 Gamma model paper ( Table 3). The predicated spring FTC period Rs was 17.53 ± 0.43 g C m −2 yr −1 in this temperate forest without N addition (Table 1). Low-N treatment exerted negative effects on spring FTC   Rs, and its value was 16.44 ± 0.58 g C m −2 yr −1 ( Table 1). The cumulative Rs during the spring FTC period were 10.67 ± 0.75 g C m −2 yr −1 in medium-N plots and 11.24 ± 0.69 g C m −2 yr −1 in high-N plots (Table 1). In general, the N addition exerted a negative impact on spring FTC Rs and decreased it by 6% (low-N), 39% (medium-N) and 36% (high-N) compared with the control. The predicted annual Rs was 974.3 ± 67.1 g C m −2 yr −1 without N addition treatment; the values of Rs in winter were 46.8 g C m −2 yr −1 (control), 35.7 g C m −2 yr −1 (low-N), 41.89 g C m −2 yr −1 (medium-N) and 62.35 g C m −2 yr −1 (high-N) 48 . Under different quantities of N addition, the cumulative Rs during spring FTC period contributed 37.49% (control), 46.88% (low-N), 25.50% (medium-N) and 18.03% (high-N), respectively, to the winter Rs and contributed 1.80% (control), 1.69% (low-N), 1.10% (medium-N) and 1.15% (high-N), respectively, to the annual Rs (Fig. 2).
The Fenglin Natural Reserve (our study site) covered an area of 18165 hm 2 . We hypothesized that whole Reserve was used to simulate the impact of N addition on Rs. The Rs in the study area was reduced by 1.97 × 10 −4 Tg C yr −1 (low-N), 1.25 × 10 −3 Tg C yr −1 (medium-N) and 1.14 × 10 −3 Tg C yr −1 (high-N) during the spring FTC period. The temperate forest covers 9.7% of the earth's continental surface 40 . The temperate forest covered an area of 14.5 million km 2 . The Rs in whole temperate forest would be reduced by 15.81 Tg C yr −1 (low-N), 99.47 Tg C yr −1 (medium-N) and 91.21 Tg C yr −1 (high-N) during the spring FTC period. Global total CO 2 emission (excluding Land-use Change and Forestry) cumulative value was 33843.05 Mt (about 9229.92 Tg C) in 2012 49 . The decrease values of Rs under medium-and high-N treatments during spring FTC period in temperate forest would be approximately equivalent to 1% of global annual C emissions.

Relationships between spring FTC Rs and soil biochemical property.
The mean values of soil biochemical property were summarized in Table 4. The correlation analyses between Rs and soil biochemical property were performed to attempt to explain the observed changes in Rs during spring FTC period. But we only found the Rs and the soil NH 4 + -N, the soil NO 3 − -N, the soil MBC and the soil MBN are related. The Rs was positively correlated with the soil MBC and MBN during the spring FTC period (Fig. 3a,b). But, the Rs was positively correlated with the lower concentrations of the soil NH 4 + -N and the soil NO 3 − -N, and negatively correlated with the higher concentrations of the soil NH 4 + -N and the soil NO 3 − -N during spring FTC period (Fig. 3c,d). We made the best fitting equation of the Rs and the soil biochemical property. The Rs increased linearly with the soil MBC (y = 0.47x − 0.21, R 2 = 0.75, p < 0.01, y was defined as Rs values, x was defined as MBC values) and the soil MBN (y = 4.07x − 1.83, R 2 = 0.74, p < 0.01, y was defined as Rs values, x was defined as MBN values). The Rs decreased exponentially with the soil NO 3 − -N concentrations (y = 0.70e −0.03x , R 2 = 0.63, p < 0.01, y was defined as Rs values, x was defined as soil NO 3 − -N concentrations). The Rs changed irregularly with the soil NH 4 + -N concentrations (y = 139.55x 2 − 139.89x + 64.82, R 2 = 0.33, p < 0.05, y was defined as Rs values, x was defined as soil NH 4 + -N concentrations).  Table 3. Regression models of Rs against soil temperature at the 5 cm depth (T 5 ) for the FTC period.
The regression models are of the form: Rs = T a × exp (b + cT), where T is (T 5 + 40), a, b and c are regression coefficients, and determination coefficient(R 2 ) and Akaike Information Criterion (AIC) are given.

Discussion
We found that the ephemeral spikes of Rs occurred in control plots (without N addition) at the early stage of the spring FTC period, which was consistent with some previous observation 16,19,25 . At the beginning of the outburst, the snow was completely melted. And, the soil microbes can recover rapidly from disturbance resulting from   freezing 50 . Therefore, we conjectured that microbial activity and biomass may increase via enhanced substrate supply and available water (liquid water) result in Rs emission pulses. Wang et al. 16 also suspected that Rs pulses might be related to the soil hydrology changes in the spring FTC period. Priemé and Christensen 51 pointed out that the mechanisms for Rs pulses during the FTC period were stimulated by microbial metabolism via the enhanced substrate supply, which was also consistent with our results. During the outburst period, we measured the MBC and the MBN that were significantly difference among all treatments (Table 4), and the results supported our conjecture. However, the rate of Rs gradually decreased to a normal level after short time pulses. We considered that the soil microbial activity or biomass were higher in the early stage of spring FTC and decreased in the following cycles, indicating that successive FTCs might lead to the decrease of the microbial biomass in the soil examined. There are some consistent explanations for the phenomenon 10,51,52 . Schimel and Clein's 53 study shown that the successive FTCs might lead to the lysis of microbial cells, and followed by the decrease of Rs. In addition, Haei et al. 54 suggested that the change of dissolved organic carbon (DOC) might also impact on Rs during FTCs, and the decreased rate of Rs after the pulses could be due to change of DOC utilization.
We also found the nitrogen addition exerted the negative impact on the soil respiration of spring freeze-thaw periods due to delay of spikes and inhibition of soil respiration. The mechanisms for impact of N addition on Rs during the spring FTC period are complicated. In our study, the mechanism for FTC-induced enhancement of Rs is not consistent with the conjecture of Elberling and Brandt 7 that a pulse during the FTC period resulted from the release of trapped CO 2 in the winter. We suggested that a relatively high microbial biomass is more likely to release a pulse of CO 2 during FTC than a relatively low microbial biomass. With respect to the delay under N addition, we hypothesized N and salt in high concentrations inhibited microbial activity and biomass during the early period of FTC so that pulse of Rs did not occur in this period. After this period of FTC, the pulse of Rs was observed, because the continuous FTC promoted N leaching losses 55 , which decreased the inhibition of microbial activity and biomass. Simultaneously, when most of the extrinsic inhibitor can be removed, the microbial activity and the biomass may rapidly increase resulting in the Rs emission pulses in treatments plots. The hypothesis is also supported by the results of others [56][57][58][59][60][61] .
The contributions of Rs during the spring FTC period to the annual Rs were 1.80%, 1.69%, 1.10% and 1.15% for control, low-N, medium-N and high-N treatment, respectively (Table 1, Fig. 3). Our results suggested that response of Rs to simulated N deposition in temperate forests is a decline, and it may vary depending on the level of N deposition during the spring FTC periods. In the previous studies, the decrease in Rs occurred in the warm and wet growing season in the N addition plots 62-64 , and not occurred in winter among treatment 63 . In addition, our results also suggested that contribution of Rs during the spring FTC period to the annual Rs will vary when the global N deposition are greatly altered with the atmospheric N levels rise 64 . We suspected that the decline of Rs due to N addition may be an improvement in the C use efficiency of the soil microbial community, and might impact on the global C cycle. However, N deposition may enhance soil carbon storage via decrease of Rs during spring FTC period. Therefore, more attention should be paid to the impact of N deposition on soil respiration in the spring FTC period.

Conclusions
The simulated N addition delayed the outburst of Rs compared with control (no N addition). The soil spring FTC decreased the soil C that releases into atmosphere under N deposition. The relative diminution of Rs induced by N addition may potentially affect C cycle in temperate forest. In general, the effects of N addition and spring FTC on Rs are very important to accurately predict soil CO 2 flux in cold region forest ecosystems under a changing climate.