Emissions of CO2, CH4, and N2O Fluxes from Forest Soil in Permafrost Region of Daxing’an Mountains, Northeast China

With global warming, the large amount of greenhouse gas emissions released by permafrost degradation is important in the global carbon and nitrogen cycle. To study the feedback effect of greenhouse gases on climate change in permafrost regions, emissions of CO2, CH4, and N2O were continuously measured by using the static chamber-gas chromatograph method, in three forest soil ecosystems (Larix gmelinii, Pinus sylvestris var. mongolica, and Betula platyphylla) of the Daxing’an Mountains, northeast China, from May 2016 to April 2018. Their dynamic characteristics, as well as the key environmental affecting factors, were also analyzed. The results showed that the flux variation ranges of CO2, CH4, and N2O were 7.92 ± 1.30~650.93 ± 28.12 mg·m−2·h−1, −57.71 ± 4.65~32.51 ± 13.03 ug·m−2·h−1, and −3.87 ± 1.35~31.1 ± 2.92 ug·m−2·h−1, respectively. The three greenhouse gas fluxes showed significant seasonal variations, and differences in soil CO2 and CH4 fluxes between different forest types were significant. The calculation fluxes indicated that the permafrost soil of the Daxing’an Mountains may be a potential source of CO2 and N2O, and a sink of CH4. Each greenhouse gas was controlled using different key environmental factors. Based on the analysis of Q10 values and global warming potential, the obtained results demonstrated that greenhouse gas emissions from forest soil ecosystems in the permafrost region of the Daxing’an Mountains, northeast China, promote the global greenhouse effect.


Introduction
Global warming has become increasingly prominent. The average surface temperature of the earth has increased by 0.85 • C compared to that before the industrial revolution. This trend is expected to continue and reach an increase of 0.3~4.8 • C by the end of this century [1,2]. The latest analysis of observations from the World Meteorological Organization (WMO) and Global Atmosphere Watch (GAW) program showed that the concentration of CO 2 was 405.5 ± 0.1 ppm, that of CH 4 was 1859 ± 2 ppb, and that of N 2 O was 329.9 ± 0.1 ppb in the atmosphere of 2017 [3]. Soil is the main source of greenhouse gas emissions. Almost 5~20% of CO 2 , 15~30% CH 4 , and 60~80% N 2 O are emitted from soils to the atmosphere every year [4], which comprises a key contributor to climate warming. The concentration of greenhouse gas is closely related to the carbon and nitrogen cycles of ecosystems, and its "source-sink" relationship directly affects the response and feedback of ecosystems to climate change [5]. Therefore, the dynamics of three greenhouse gas fluxes have become an important aspect of global climate change research.
Permafrost is the result of energy exchange between the lithosphere and the atmosphere and is also an important part of the cryosphere. Global climate change directly affects both the

Study Site
The study site was situated in the experimental area of the Mohe Forest Ecosystem National Positioning Observation and Research Station in Heilongjiang province (53 • 17 ~53 • 30 N, 122 • 06 ~122 • 27 E) (Figure 1). The experimental area is located in the permafrost area of the Daxing'an Mountains, with low hills at an average elevation of 300~500 m. It is controlled by a cold temperate continental monsoon climate with an annual average temperature of −3.3 • C and precipitation of 442.95 mm (monitored in 2016 and 2017). In this region, 66.47% of the precipitation is concentrated in summer (from June to August). Furthermore, the soil activity layer freeze-thaw period lasts for about half a year (from mid-April to mid-October), which is slightly longer than the snowpack cover time (from late October to early April). The annual sunshine hours are 2377~2625 h, with an annual total solar radiation of 402~448 kJ cm −2 , and ≥10 • C annual accumulated temperature of 1436~2062 • C. The zonal soil of the study site is dark brown forest soil. The vegetation in this area belongs to the southern extension of the Eurasian cold-temperate coniferous forest. The zonal vegetation is dominated species of Larix gmelinii. Other major species include Betula platyphylla, Pinus sylvestris var.mongolica, Populus davidiana, etc. Common plants include Ledum palustre var. dilatatum, Rhododendron dauricum, Vaccinium uliginosum, Vaccinium vitis-idaea, and Pyrola incarnate. Sample selection was based on representative and feasibility principles. After comprehensive investigation, the three most typical forest types in the permafrost region of the Daxing'an Mountains: Larix gmelinii forest (LF), Pinus sylvestris var.mongolica forest (PF), Betula platyphylla forest (BF) were selected. One 100 m × 100 m fixed experimental block with similar site conditions was set in each forest type (LF, PF and BF), and a total of 3 experimental blocks were set. All acronyms such as LF, PF, and BF which applicable for all tables and all figures.

Measurement of CO2, CH4 and, N2O Fluxes in Soil
Three 5 m × 5 m quadrats were randomly distributed along the diagonal of each fixed experimental block (100 m × 100 m) in each forest type, but each quadrat was separated by at least 20 m in order to collect greenhouse gas samples. A total of 9 quadrats were set in the three forest types. The static opaque chamber and gas chromatography method were adopted to monitor soil CO2, CH4, and N2O gas fluxes in situ. The static chamber was mainly composed of the chamber and a stainlesssteel base. Before sampling, the stainless-steel base with the water groove was buried in the quadrat one week in advance, and was kept fixed during the whole monitoring period to reduce interference by the surrounding environment. The outside of the chamber was affixed with insulation material to

Measurement of CO 2 , CH 4 and, N 2 O Fluxes in Soil
Three 5 m × 5 m quadrats were randomly distributed along the diagonal of each fixed experimental block (100 m × 100 m) in each forest type, but each quadrat was separated by at least 20 m in order to collect greenhouse gas samples. A total of 9 quadrats were set in the three forest types. The static opaque chamber and gas chromatography method were adopted to monitor soil CO 2 , CH 4 , and N 2 O gas fluxes in situ. The static chamber was mainly composed of the chamber and a stainless-steel base. Before sampling, the stainless-steel base with the water groove was buried in the quadrat one week in advance, and was kept fixed during the whole monitoring period to reduce interference by the surrounding environment. The outside of the chamber was affixed with insulation material to decrease the temperature disturbance by the environment inside the box. Three holes were reserved at the top of the chamber to connect the fan power cable, the thermometer probe, and the sampling port, respectively. A 12-V battery-powered fan was installed in the chamber to evenly mix the gas.
In sunny weather, gas samples were collected during mid-morning (9:00~11:00 a.m., local time), which was used to represent one day of average flux [25], and once per week during the growing seasons (from May to September in 2016-2017) and once or twice per month during the non-growing seasons (from October 2016 to April 2017 and from October 2017 to April 2018). The aboveground plant inside the stainless-steel base was cut off 1 day before each sampling during the growing season, while natural snow accumulation was not treated during winter greenhouse gas collection. A medical syringe (60 mL), equipped with a three-way stopcock, was used to collect gas 0, 10, 20, and 30 min after the opaque chamber was sealed with water. When the temperature was below 0 • C, we brought our incubator to store water (300 mL of water per static chamber to water seal), and removed ice or water from the stainless-steel base after each experiment to facilitate the next sampling for water sealing. The samples were immediately transferred to a 200 mL gas sampling bag (Delin gas packing co., Dalian, China) and shipped back to the laboratory for analyzing by 7890B Gas Chromatography (7890B GC System, Agilent, CA, USA).

Soil Sampling and Analysis
In each quadrat (5 m × 5 m), five soil samples (0-15 cm) were randomly collected by removing the surface layer of the soil, and uniformly mixed into one subsample. Each time, 9 soil subsamples were collected from 3 fixed experimental blocks (100 m × 100 m), and total of 216 soil subsamples were collected during the observation period. The atmospheric pressure (BY-2003P barometer, Xieya Electronics Co., Beijing, China), air temperature (Ta), 10 cm soil temperature (T 10 ) (Delta TRAK portable thermometer, Delta TRAK, CA, USA) were measured simultaneously with soil sampling. All soil samples were shipped back to the laboratory in ziplock bags for the determination of soil moisture, bulk density (Bd), pH, ammonium nitrogen (NH 4 + -N), nitrate nitrogen (NO 3 − -N), total nitrogen (TN), and total organic carbon (TOC) ( Table 1). Soil moisture was determined by the dry-weighing method. Soil bulk density was determined by the ring-knife method. Soil pH was determined in suspensions composed of 1:5 ratio of air-dried soil and deionized water using a PHSJ-3F pH meter (PHSJ-3F pH, Shanghai, China). Soil NH 4 +− -N and NO 3 − -N were determined by extraction with potassium chloride solution-spectrophotometric method using a flow injection auto-analyzer (Skalar SAN++, The Netherlands) [26]. Soil TN was determined spectrophotometric using a flow injection auto-analyzer (Skalar SAN++, The Netherlands) [27]. Soil TOC was determined using the dry combustion method by TOC/TN analyzer (Multi C/N 3100, Jena, Germany) [28]. Table 1. The surface soil (0~15 cm) physicochemical properties in LF, PF, and BF (mean ± SD, n = 216).

Statistical Analysis
The gas flux was calculated according to the following equation [29]: where F (mg·m −2 ·h −1 ) is the gas flux, dc/dt is the slope of the linear regression for the gas concentration gradient over time, M (g·mol −1 ) is the molecular mass of gas, P (Pa) is the atmospheric pressure in sampling site, T (k) is the temperature inside the chamber during sampling, H (m) is the height of chamber, V 0 (m 3 ·mol −1 ), P 0 (Pa) and T 0 (k), are the gas mole volume, atmospheric pressure under standard conditions, and absolute air temperature, respectively. A positive F-value means that there is a net emission of gas, and a negative value is the opposite. The global warming potential (GWP) was calculated according to [30]: where F CO2 , F CH4 and F N2O represent the greenhouse gas emission flux during the monitoring period (t·hm −2 ), 25 and 298 are the conversion factors of CH 4 and N 2 O, respectively (for a 100 timeframe), to present GWP in t CO 2 Eq·ha −1 .
The following equation was established to calculate the temperature sensitivity of gas flux to the changes of T 10 [31,32]: where F is the gas flux, T is the soil temperature, coefficient a is the intercept of soil respiration when temperature is zero, coefficient b represents the temperature sensitivity of soil respiration, Q 10 is the temperature sensitivity. The differences in soil environment factors among all forests were compared using one-way ANOVA and Tukey's tests. The relationships of the soil greenhouse gas fluxes with soil temperature and moisture were assessed by regression analysis. The correlations between soil greenhouse gas fluxes and physical and chemical properties were analyzed by Pearson correlation with two tails. SPSS 20.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis. All figures were drawn using OriginPro 2016 software (OriginLab Corp., Northampton, MA, USA.).

The Temporal Variation of Soil N2O Fluxes across LF, PF, and BF
The soil N2O fluxes of the three forest types were largely consistent during the monitoring period of 2016~2018 (Figure 2c). The soil N2O fluxes ranged from −3.87 ± 1. 35

The Temporal Variation of Soil CH 4 Fluxes across LF, PF, and BF
During the test, the three types of forest soil all showed CH 4 absorption fluxes (Figure 2b), with apparent seasonal fluctuations. The soil CH 4 fluxes had similar temporal variation in LF, PF, and BF, which showed increasing fluctuation from June to January of the following year, followed by a gradual decreased until June. The soil CH 4 fluxes ranged from −57.71 ± 4.65 ug·m −2 ·h −1 to 32.51 ± 13.03 ug·m −2 ·h −1 . During the growing season, all three types of forest soil absorbed CH 4 ; however, positive CH 4 fluxes were found in winter. The highest absorption for CH 4 fluxes from LF and PF were observed in June (−47.76 ± 3.9 ug·m −2 ·h −1 , −52.18 ± 8.36 ug·m −2 ·h −1 ), while it was August for BF (−57.71 ± 4.65 ug·m −2 ·h −1 ). The peak fluxes from PF and BF occurred in February (20.43 ± 8.15 ug·m −2 ·h −1 , 23.46 ± 6.35 ug·m −2 ·h −1 ), and in January for LF (32.51 ± 13.03 ug·m −2 ·h −1 ). The average annual flux of CH 4 in LF, PF, and BF were −19.33 ug·m −2 ·h −1 , −24.59 ug·m −2 ·h −1 , and−30.46 ug·m −2 ·h −1 , respectively. The soil CH 4 absorption fluxes in broad-leaved forest were higher than in coniferous forest, and BF soil CH 4 cumulative absorption fluxes were significantly higher than LF (Table 2).

Cumulative Soil Greenhouse Gas Emissions and GWP
This study showed that the CO 2 accumulation fluxes of three forest types in the permafrost region of the Daxing'an Mountains dominated, followed by N 2 O and CH 4 as absorption fluxes (Table 2). However, LF, PF, and BF showed the "source" of greenhouse gases. The global warming potential is to evaluate the relative impact of greenhouse gases on global climate change using CO 2 as a reference gas on a 100-year time scale. The radiation effects of CH 4 and N 2 O were 25 times and 298 times that of CO 2 , respectively [4]. The greenhouse gas GWP value of three forest types showed: PF > LF > BF ( Table 2). The study clarified that the release of greenhouse gases from forest soils in the permafrost region of the Daxing'an Mountains has a positive effect on global warming, and the greenhouse gas GWP value of coniferous forests was higher than that of broad-leaved forests.

Effects of Environmental Factors on Soil Greenhouse Gas Fluxes
Soil temperature and moisture directly or indirectly affected the production and release of greenhouse gases. The correlations between gas fluxes and soil temperature were higher for CO 2 (0.84 < R 2 < 0.89, P < 0.001; Figure 3a) and CH 4 (0.62 < R 2 < 0.76, P < 0.001; Figure 3b) than for N 2 O (0.22 < R 2 < 0.35, P < 0.001; Figure 3c). Soil moisture was positively correlated with CH 4 fluxes (0.45 < R 2 < 0.7, P < 0.001; Figure 3e), while CO 2 (0.23 < R 2 < 0.42, P < 0.001; Figure 3d) and N 2 O (0.14 < R 2 < 0.24, P < 0.005; Figure 3f) were negatively correlated. During the study period, the soil CO 2 and N 2 O fluxes of all forest types showed a significant exponential relationship with T 10 (P < 0.001), the coefficient of variation of the regression model ranged between 0.22 and 0.89. The Q 10 values of CO 2 fluxes in LF, PF, and BF were 5.47, 3.67, and 4.06, while those for N 2 O were 2.23, 1.82, and 1.49, respectively.
The Person's correlation analysis between the greenhouse gas fluxes and various environmental factors is showed in Table 3. Three forest-type soil CO 2 fluxes had a significant negative correlation with TN (P < 0.05). The soil CH 4 fluxes exhibited significant correlation with TOC in LF and BF (P < 0.01) and had a significant positive correlation with TN in LF (P < 0.05). The soil N 2 O fluxes showed a significant positive correlation with NH 4 + -N and NO 3 − -N in LF, NO 3 − -N in PF (P < 0.05), and significant positive correlation with soil pH, NO 3 − -N and TN in BF (P < 0.01).
1  Note: ** Correlation is significant at the 0.01 level, * correlation is significant at the 0.05 level.

Characteristics of Soil CO 2 Emissions from LF, PF, and BF
This study showed that the seasonal variation of CO 2 emission fluxes in typical forest LF, PF, and BF in the Daxing'an Mountains were similar, and reached a peak during the growing season (from July to August, Figure 2a), which is consistent with the results of the Song et al. [33] and Li et al. [34]. Soil CO 2 is mainly derived from autotrophic respiration (mainly vegetation root respiration), heterotrophic respiration (mainly soil microbial respiration), and mineralization decomposition of organic matter. The vegetation in this study area is mostly a shallow root system. During summer, with gradually increasing temperature, the appropriate combination of water and heat in the soil drives root respiration and soil microbial activity, which promotes soil CO 2 to reach emission peaks. Previous studies reported that the seasonal change of CO 2 emissions were dominant controlled by soil temperature [32,35]. This study found a consistent result, whereby soil temperature was significantly correlated with CO 2 fluxes as the dominant environmental variables in affect CO 2 emissions (Figure 3a). During the non-growing season, soil CO 2 fluxes fluctuated within a lower range. Analysis of the reasons indicated that root respiration and microbial activity were weak due to low temperature. On the other hand, when the soil was frozen, the active nutrient substrate, which can be directly used by microorganisms was reduced [36], thus resulting in the reduction of soil CO 2 fluxes. Two years of observations revealed that the temperature in the study area suddenly decreased due to the influence of low environment temperatures at the end of August each year, and the soil CO 2 fluxes of the three forest types decreased. Then, the soil CO 2 emission flux value increased, accompanying the increasing temperature back to the normal level of the same period [35]. This study found that the soil CO 2 fluxes in coniferous forests were higher than that in broad-leaved forests. The PF soil CO 2 cumulative fluxes were significantly higher than BF. Analysis of the reason found that the PF soil with high temperature had strong microbial metabolic activity, and the humus layer under the forest decomposed faster, releasing a large amount of CO 2 into the atmosphere. Q 10 is widely used to assess the sensitivity of soil or ecosystem respiration to temperature changes [37]. The Q 10 value of forest soil respiration in China is 1.33~5.53 [38]. The LF soil CO 2 fluxes (Q 10 = 5.47) are more sensitive than BF (Q 10 = 4.06) and PF (Q 10 = 3.67) to temperature changes, which is consistent with the research results of Zheng et al. [39]. The significant correlation between soil temperature in the study area and soil CO 2 fluxes indicates strong positive short-term feedback between climate warming and soil CO 2 fluxes.

Characteristics of Soil CH 4 Emissions from LF, PF, and BF
CH 4 is produced by a biogeochemical cycle. It is oxidized by methanotrophs at the soil-water interface or rhizosphere aerobic environment, and the remainder is released into the atmosphere [40]. During this process, methanotrophs and methanogens play a key role. Previous studies have shown that soil CH 4 as trace gas has both absorption [41,42] and emission [43][44][45]. Soil hydrothermal conditions directly or indirectly change the community characteristics of anaerobic methanogens and aerobic methanotrophs to affect soil CH 4 fluxes; therefore, the difference in hydrothermal conditions under different ecosystems leads to different soil CH 4 fluxes [46]. This study observed that the three types of forest soil CH 4 showed overall absorption. Soil CH 4 emission, with maximum emission fluxes, only occurred in winter (January or February), and the other seasons showed absorption with the maximum absorption fluxes appearing in June (Figure 2b). After the soil was completely frozen in winter, the soil formed a better anaerobic environment, which was beneficial for the metabolic activity of methanogens and promoted the release of CH 4 . At the beginning of the growing season, the thickness of the permafrost active layer further increases with increasing temperature, which provides a large place for the survival of the methanotrophs [47]. The release of C and N from the frozen microorganisms that were killed in winter provides an important matrix for methanotrophs to accelerate the oxidative absorption of CH 4 [48]. At the same time, the soil has a short drought period before the rainy season, which is conducive for the spread of atmospheric CH 4 and O 2 to the soil and increase the absorption of CH 4 [49]. During summer, the soil moisture increases due to precipitation, and the soil CH 4 absorption decreases. The study also found that soil temperature and moisture were significantly related to soil CH 4 fluxes (Figure 3b,e). Soil CH 4 average annual absorption flux of three forest types showed BF > PF > LF, and the soil CH 4 absorption fluxes in the broad-leaved forest were higher than in the coniferous forest, which was consistent with the results of Steudler et al. [50]. The reason for this could be that LF and PF are loamy soils, which are tight and have poor aeration, while the soil bulk density of BF is relatively small, and the gravel content is high. The BF loose soil is conducive to oxygen transport, which enhances the activity of methane oxidase and methane oxidizing microorganisms in soil, which improves the absorption capacity of CH 4 in BF soil.

Characteristics of Soil N 2 O Emissions from LF, PF, and BF
Soil nitrification and denitrification are two important links in the nitrogen cycle of ecosystems and form an important source of atmospheric N 2 O. The monitoring results showed that N 2 O fluxes of the three forest types were all sources of emissions. In June (during the growing season), soil N 2 O fluxes showed peak emission periods. Thomas et al. [51] also reached a similar conclusion. The main reason may be that the outer surface of the soil particles is covered by an ice layer and the inner layer is wrapped with a tightly bound liquid water film, which forms a better anaerobic environment and remains high active nutrients after the soil completely frozen in winter [52]. The anaerobic environment provides a good place for denitrification to produce N 2 O, which is sequestered by the frozen soil [53]. The N 2 O enclosed in the soil is burst out into the atmosphere after thawing. In summer, plants grow vigorously and absorb a large amount of nitrogen. Competition between vegetation and microbes causes soil microbes to utilize substrate reduction, and frequent rainfall during the summer causes the shallow soil to alternate between wet and dry conditions, all of which affects the soil N 2 O emission rate [54]. Permafrost soils, which are characterized by cold temperatures, have low net N mineralization rates and availability of mineral nitrogen [36]. Consequently, the available nitrogen in the soil (predominantly ammonium nitrogen and nitrate nitrogen) is poor in the forests of northern China [55]. This study found that PF soil N 2 O fluxes were higher than LF and BF. The available nitrogen (NH 4 + -N and NO 3 − -N) contents of different forest types showed that BF was significantly higher than both PF and LF (P < 0.05), and the NH 4 + -N content was significantly higher than the NO 3 − -N content (P < 0. 05) (Table 1). Therefore, denitrification may be the main source of soil N 2 O fluxes in this region. However, BF soil pH was significantly lower than LF and PF (P < 0.05) ( Table 1). According to Struwe et al. [56], the optimal pH range for denitrification in soil is between 6 and 8. Soil denitrification is inhibited under acidic conditions and the rate of denitrification decreases with increasing soil acidity. Therefore, the N 2 O emission rate of LF was lower than that of PF and BF. Our study found that the temperature sensitivity of N 2 O in LF, PF, and BF were 2.23, 1.82, and 1.49, respectively, and LF soil N 2 O responded most to temperature rise. However, the difference of N 2 O fluxes between the three forest types were not significant, and the forest type was not the main factor affecting soil N 2 O fluxes.

Conclusions
(1) The typical forest soil of the Daxing'an Mountains permafrost region was a "source" of CO 2 and N 2 O and a "sink" of CH 4 . The three greenhouse gas fluxes showed strong temporal variety, while the fluxes varied depending on the different sites. The PF soil CO 2 fluxes were significantly higher than BF. At the same time, the soil absorption fluxes of CH 4 in BF were significantly higher than that in LF. (2) Soil temperature and moisture were key environmental factors that correlated the CO 2 and CH 4 fluxes of different forest types in this high-latitude permafrost region. Q 10 values showed that LF soil greenhouse gas fluxes were more sensitive to temperature. The N 2 O fluxes were mainly correlated by the soil nitrogen content. (3) Against the background of climate warming, the CO 2 and N 2 O emission rates of the three forest types increased with increasing temperature, and the CH 4 absorption rate decreased, thus enhancing the atmospheric greenhouse effect. On a 100-year time scale, the greenhouse gas GWP of the three forest soil systems in the Daxing'an Mountains permafrost region was positive, which had positive feedback on global warming.
The Daxing'an Mountains permafrost region is extremely sensitive to climate change. The CO 2 , CH 4 , and N 2 O fluxes in the study area had significant emission potential. Temperature alteration leads to complex changes in hydrothermal conditions that either directly or indirectly affect forest ecosystems in the cold region, which transforms ground-gas exchange ratios to affect greenhouse gas fluxes. In the future, stable isotope tracing and microbial high-throughput sequencing technologies will be comprehensively used to analyze the mechanism of soil greenhouse gas accumulation, conversion, and transmission in permafrost region.