Comparison of Soil Greenhouse Gas Fluxes during the Spring Freeze–Thaw Period and the Growing Season in a Temperate Broadleaved Korean Pine Forest, Changbai Mountains, China

Soils in mid-high latitudes are under the great impact of freeze–thaw cycling. However, insufficient research on soil CO2, CH4, and N2O fluxes during the spring freeze–thaw (SFT) period has led to great uncertainties in estimating soil greenhouse gas (GHG) fluxes. The present study was conducted in a temperate broad-leaved Korean pine mixed forest in Northeastern China, where soils experience an apparent freeze–thaw effect in spring. The temporal variations and impact factors of soil GHG fluxes were measured during the SFT period and growing season (GS) using the static-chamber method. The results show that the soil acted as a source of atmospheric CO2 and N2O and a sink of atmospheric CH4 during the whole observation period. Soil CO2 emission and CH4 uptake were lower during the SFT period than those during the GS, whereas N2O emissions were more than six times higher during the SFT period than that during the GS. The responses of soil GHG fluxes to soil temperature (Ts) and soil moisture during the SFT and GS periods differed. During the SFT period, soil CO2 and CH4 fluxes were mainly affected by the volumetric water content (VWC) and Ts, respectively, whereas soil N2O flux was influenced jointly by Ts and VWC. The dominant controlling factor for CO2 was Ts during the GS, whereas CH4 and N2O were mainly regulated by VWC. Soil CO2 and N2O fluxes accounted for 97.3% and 3.1% of the total 100-year global warming potential (GWP100) respectively, with CH4 flux offsetting 0.4% of the total GWP100. The results highlight the importance of environmental variations to soil N2O pulse during the SFT period and the difference of soil GHG fluxes between the SFT and GS periods, which contribute to predicting the forest soil GHG fluxes and their global warming potential under global climate change.


Introduction
Carbon dioxide (CO 2 ), methane (CH 4 ), and nitrous oxide (N 2 O) are recognised as the most important greenhouse gases (GHGs), as their contributions to the 100-year global warming potential (GWP 100 ), indicating the warming ability of each GHG over 100 years [1][2][3][4], have reached 60%, 15%, and 5%, respectively [5]. According to the fifth assessment report of the Intergovernmental Panel on Climate Change (IPCC), atmospheric GHG concentrations have reached their highest levels in the past their feedback on global climate changes, would contribute to improving our understanding of the mechanisms of soil GHG emissions, thereby enabling accurate estimations of C and N budgets.
The broad-leaved Korean pine mixed forest in the Changbai Mountains (CBF) is a typical zonal climax forest in the temperate zone of China [48][49][50]. Previous studies have shown that the forest soil in the CBF is a source of CO 2 and N 2 O as well as a CH 4 sink. In addition, soil GHG dynamics are mainly affected by environmental factors (e.g., temperature, precipitation, and soil moisture), C and N concentrations, and soil microorganisms [51][52][53][54]. However, previous studies were mainly focussed on CO 2 flux; few studies have assessed other GHGs (e.g., CH 4 and N 2 O) [8,55], especially during the SFT period [32,56]. Moreover, the relationships between soil GHG fluxes and their main environmental factors in the CBF during the SFT and GS periods remain poorly understood. Based on the static-chamber method, the present study continuously observed soil CO 2 , CH 4 , and N 2 O fluxes in the CBF during an SFT period and a subsequent GS in 2019. Our objectives are to (1) identify the temporal variation patterns of soil GHG fluxes and their relationships; (2) compare the responses of soil GHG fluxes to T s and VWC between the SFT and GS periods; and (3) to determine the relative contributions of soil GHG fluxes to the warming potential.

Site Description
The study was conducted in a broad-leaved Korean pine mixed forest in the Changbai Mountains (42 • 24 9 N, 128 • 05 45 E, 738 m a.s.l.) in Jilin Province, Northeastern China. The climate in this region is a temperate continental monsoon climate with long winters and short temperate summers. The snow cover period lasts for nearly five months, from late fall to early mid-spring [57,58]. The mean annual temperature is 3.6 • C (8.7 to 19.3 • C in July and −23.3 to −16.1 • C in January). The mean annual precipitation amount is approximately 713 mm, mainly occurring between June and August. The soil is classified as dark brown forest soil (cambisols, according to the World Reference Base for Soil Resources (WRB), 1998) [47,53,59]. The dominant species include Korean pine (Pinus koraieensis Sieb. et Zucc.), Amur linden (Tilia amurensis Rupr.), Mongolian oak (Quercus mongolica Fisch.), Manchurian ash (Fraxinus mandschurica Rupr.), and mono maple (Acer mono Maxim.) [60].

Soil GHG Flux Measurements
Soil-atmosphere CO 2 , CH 4 , and, N 2 O exchanges were measured in situ with the static chamber method combined with the gas chromatography technique (GC, Agilent 5890A, Agilient Co., Santa Clara, CA, USA). The chamber was composed of an acrylic cylinder (25-cm diameter, 20-cm height) with an approximate volume of 10 L. The chamber was covered by a non-reflective thermal insulation layer to maintain a stable temperature inside. A mini fan was installed for mixing. Two tubes were installed on the top of the chamber to collect gas samples and equalise the air pressure between the chamber and the atmosphere. AK thermocouple thermometer (HH800A, Omega Engineering Inc., Norwalk, CT, USA) was inserted into the chamber to measure the air temperature (T a ). Eight soil collars (20-cm diameter) were inserted into the soil at 10-cm depth. The collars were approximately 2 m apart.
Soil gas sampling was conducted on sunny days. Four 30 mL gas samples were taken from the headspace of each chamber with gas-tight syringes 0, 7, 14, and 21 min after the chamber was closed. The procedure was repeated every 2 h from 8:00 a.m.-4:00 p.m. on the sampling days from July-September, and every 2 h from 9:00 a.m.-3:00 p.m. during other months. Gases were sampled twice a week during the SFT period and twice a month during the GS.
Soil CO 2 , CH 4 , and N 2 O concentrations were analysed within 24 h. Fluxes were calculated as follows [61]: where the fluxes of F CO 2 (µmol m −2 s −1 ), F CH 4 (nmol m −2 s −1 ), and F N 2 O (nmol m −2 s −1 ) are equal to the linear change in gas concentrations within the chamber multiplied by the ratios of chamber pressure P (kPa) to the sum of chamber temperature plus T 0 (T [ • C] +T 0 ) and chamber volume V (L) to the sectional area of the chamber bottom A (m 2 ); ∂C ∂t is the slope of the linear regression equation for GHG concentrations and time; P 0 and T 0 are the atmospheric pressure (101.325 kPa) and the thermodynamic temperature (273.15 K) under standard conditions, respectively, and; V 0 is the molar volume of the gas in the standard state (22.414 L mol −1 ). Positive fluxes represent soil gas emissions to the atmosphere, whereas negative fluxes represent the sink of gases by the soil from the atmosphere. Observations at 11:00 a.m. were selected to represent the daily fluxes, as it has been shown to be approximately equivalent to the average daily fluxes [62][63][64]. In the calculation of GHG fluxes, the R 2 value represented the quality of the linear fitting of GHG concentrations to time. In this study, fluxes with R 2 < 0.3 and fluxes which were >1.96 of the standard deviation (which shows the 95% confidence interval) were excluded [65].

Environmental Variable Measurements
Environmental factors, including T a , T s , VWC, precipitation, and snow depth were observed simultaneously from a meteorological tower within a 20 m distance from where the soil collars were installed. T a was measured using a temperature probe (HMP115A, Vaisala, Helsinki, Finland) at 2.5 m. Precipitation was measured using a rain gauge (TE525MM, Texas Electronics Inc., Dallas, TX, USA) at 62 m. The snow depth in the forest was measured using a snow-depth sensor (SR50A, Campbell Sci. Inc., Logan, UT, USA). The data were averaged at a 30-min scale using a data logger (CR1000X, Campbell Sci. Inc., Logan, UT, USA). Three T/VWC sensors (CS655, Campbell Sci. Inc., Logan, UT, USA) were installed next to the soil collars at the depths of 5 cm and 10 cm for measuring T s and soil VWC. We collected soil samples on the same day as the gas samples from three soil layers (2 cm, 5 cm, and 10 cm) at three 2 m × 2 m soil sampling areas that were evenly distributed around the eight collars; soil sampling was conducted once a week from March-April and once a month in other months. The soil gravimetric water content (WWC) was determined gravimetrically by drying the soil samples at 105 • C for 24 h.

Division of the SFT Period and the GS
The variations in environmental parameters were analysed for estimating the times of the SFT and GS periods. The lengths of the SFT period and the GS are determined based on the number of consecutive days on which the temperatures exceed the freeze criteria or a specific daily mean T a [66]. In snow-covered boreal ecosystems, the common range of winter T s is from −5 to −1 • C [67,68]. During the SFT period, the VWC drastically increases after snow and frozen soil melts. The T s fluctuates from −1 to 4 • C until the melted water is evaporated and the soil begins to dry [35]. This indicates that the VWC reaches its peak at the end of the SFT period. Therefore, the SFT period commences from the day when the daily mean T s reaches −1 • C and ends on the day when the daily mean VWC is no longer increasing. In the present study, we used the dates when the five-day running daily mean T a reached 7 • C as the beginning of the GS [66,69]. Previous studies [66,70] have defined the end of the GS as the first day on which T a reaches 0 • C in the fall. However, using a single T a threshold to define climatological GSs can sometimes be unreliable [66]. Thus, in the present study, the end of the GS should also meet other criteria, i.e., the first day when the average daily T a within five days of this day was <7 • C.

GWP 100 Calculation
GWP 100 is widely used to compare the ability of each GHG to trap heat in the atmosphere over time, relative to CO 2 gas (reference gas or CO 2 equivalent) [4]. Furthermore, CO 2 equivalent has been a unit for measuring the greenhouse effect potential of different GHG fluxes [4]. The CH 4 and N 2 O radiative forcing strengths per unit mass are 25 and 298 times greater than CO 2 over 100 years, respectively [2]. GWP 100 is the CO 2 quality of the corresponding effect within a 100-year time scale converted from the greenhouse effects of various GHGs as follows: where F CO 2 , F CH 4 , and F N 2 O are the total fluxes of CO 2 , CH 4 , and N 2 O, respectively (kg ha −1 ). We multiplied the daily mean soil CO 2 , CH 4 , and N 2 O fluxes by the number of days of the SFT and GS periods, respectively, in 2019, and calculated the GWP 100 (kg CO 2 eq ha −1 ) by adding them together.

Data Analysis and Statistics
We applied linear equations (y = a + bx), quadratic equations (y = a + bx + cx 2 ), or exponential equations (y = ae bx ) to explore the responses of soil GHG fluxes to T s and VWC by using R software (version 3.5.2; R Foundation for Statistical Computing, Vienna, Austria) for Windows. The structural equation model using Amos Graphics software was fitted to infer the direct and indirect effects on GHG fluxes of T s and VWC during the SFT and GS periods. We reported the path coefficients as standardised effect sizes in the path correlation analysis.

Environmental Factors
Influenced by the monsoon climate, the environmental factors in the study site showed apparent seasonal variations (Figure 1a,b). The SFT period lasted for 56 days from March 6 to April 30 (days of the year (DOY): 65-120), and the average T s at 5-cm depth in this period was 0.17 • C. While the GS lasted for 167 days, from May 1 to October 14 (DOY: 121-287), during which the average T s at 5-cm depth was 13.92 • C. The T a and T s showed single-peak curves during the GS (Figure 1a). The study area was covered with snow during spring and winter, with a maximum snow depth of 10 cm. Furthermore, T s fluctuated slightly around 0 • C when the soil surface was covered by snow. The snow cover and frozen soil gradually melted when T a and T s reached 0 • C. During the SFT period, soil WWC variation exhibited an opposite trend compared to soil VWC ( Figure 1b). Total precipitation during the GS was 214.2 mm, concentrated from June-August. respectively [2]. GWP100 is the CO2 quality of the corresponding effect within a 100-year time scale converted from the greenhouse effects of various GHGs as follows: where ∑ , ∑ , and ∑ are the total fluxes of CO2, CH4, and N2O, respectively (kg ha −1 ). We multiplied the daily mean soil CO2, CH4, and N2O fluxes by the number of days of the SFT and GS periods, respectively, in 2019, and calculated the GWP100 (kg CO2 eq ha −1 ) by adding them together.

Data Analysis and Statistics
We applied linear equations (y = a + bx), quadratic equations (y = a + bx + cx 2 ), or exponential equations (y = ae bx ) to explore the responses of soil GHG fluxes to Ts and VWC by using R software (version 3.5.2; R Foundation for Statistical Computing, Vienna, Austria) for Windows. The structural equation model using Amos Graphics software was fitted to infer the direct and indirect effects on GHG fluxes of Ts and VWC during the SFT and GS periods. We reported the path coefficients as standardised effect sizes in the path correlation analysis.

Environmental Factors
Influenced by the monsoon climate, the environmental factors in the study site showed apparent seasonal variations (Figure 1a

Seasonal Patterns of Soil CO2, CH4, and N2O Fluxes
Seasonal variations of soil CO2, CH4, and N2O fluxes showed distinct differences (Figure 2), and their average fluxes and cumulative fluxes largely differed between the SFT and GS periods ( Table   Figure 1. (a) Variations in daily air temperature (T a ), soil temperature at 5-cm (T s _05) and 10-cm (T s _10) depths, and snow depth; and (b) daily precipitation, soil volumetric water content at 5-cm (VWC_05) and 10-cm (VWC_10) depths, and soil gravimetric water content at 2-cm (WWC_02), 5-cm (WWC_05), and 10-cm (WWC_10) depths. The grey shaded area indicates the spring freeze-thaw (SFT) period, and the area between the dashed lines indicates the growing season (GS).

Seasonal Patterns of Soil CO 2 , CH 4 , and N 2 O Fluxes
Seasonal variations of soil CO 2 , CH 4 , and N 2 O fluxes showed distinct differences (Figure 2), and their average fluxes and cumulative fluxes largely differed between the SFT and GS periods ( Table 1). The soil was a source of CO 2 that varied from 0.18-2.98 µmol m −2 s −1 during the whole observation (WO) period ( Figure 2). During the SFT period, soil CO 2 emissions initially decreased and then gradually increased; however, the flux magnitude remained relatively low (0.32 µmol m −2 s −1 on average). After entering the GS, soil CO 2 emissions elevated tremendously with a characteristic of a bimodal curve ( Figure 2). The first CO 2 emission peak occurred 27 days before T a and T s reached their peaks on the same day. On DOY 251, the variation of CO 2 flux switched from a decreasing trend to an increasing trend, with the second CO 2 emission peak on DOY 266. The average CO 2 flux during the GS was 2.00 µmol m −2 s −1 . The cumulative flux of CO 2 during the GS was 18 times higher than the cumulative CO 2 flux during the SFT period, accounting for 94.8% of the total CO 2 flux during the WO period (Table 1). In contrast, the soil was a sink of CH 4 , which varied from −3.68 to −0.13 nmol m −2 s −1 . Specifically, the average CH 4 uptake was −0.50 nmol m −2 s −1 during the SFT period with small variations. In contrast, the CH 4 uptake during the GS strengthened initially, then decreased, and then increased again, which was similar to that of CO 2 emissions ( Figure 2). In addition, the total CH 4 uptake during the GS was around 13 times higher than the total CH 4 uptake during the SFT period, accounting for 93.0% of the total uptake during the WO period ( Table 1). The seasonal variation patterns of CO 2 emission and CH 4 uptake were similar throughout the WO period. Both CO 2 emission and CH 4 uptake peaked simultaneously during the GS ( Figure 2). Overall, the CO 2 flux reached the highest emission when the CH 4 uptake was approximately −2.5 nmol m −2 s −1 during the GS. The soil was a source of N 2 O, ranging from 0.016-0.89 nmol m −2 s −1 . The soil N 2 O emission during the SFT period was 1.8 times higher than that during the GS. Soil N 2 O emissions showed a continuous decreasing trend during the SFT period, which was consistent with the trend of the WWC (Figure 1b).
Forests 2020, 11, x FOR PEER REVIEW 6 of 18 gradually increased; however, the flux magnitude remained relatively low (0.32 μmol m −2 s −1 on average). After entering the GS, soil CO2 emissions elevated tremendously with a characteristic of a bimodal curve (Figure 2). The first CO2 emission peak occurred 27 days before Ta and Ts reached their peaks on the same day. On DOY 251, the variation of CO2 flux switched from a decreasing trend to an increasing trend, with the second CO2 emission peak on DOY 266. The average CO2 flux during the GS was 2.00 μmol m −2 s −1 . The cumulative flux of CO2 during the GS was 18 times higher than the cumulative CO2 flux during the SFT period, accounting for 94.8% of the total CO2 flux during the WO period (Table 1). In contrast, the soil was a sink of CH4, which varied from −3.68 to −0.13 nmol m −2 s −1 . Specifically, the average CH4 uptake was −0.50 nmol m −2 s −1 during the SFT period with small variations. In contrast, the CH4 uptake during the GS strengthened initially, then decreased, and then increased again, which was similar to that of CO2 emissions ( Figure 2). In addition, the total CH4 uptake during the GS was around 13 times higher than the total CH4 uptake during the SFT period, accounting for 93.0% of the total uptake during the WO period ( Table 1). The seasonal variation patterns of CO2 emission and CH4 uptake were similar throughout the WO period. Both CO2 emission and CH4 uptake peaked simultaneously during the GS ( Figure 2). Overall, the CO2 flux reached the highest emission when the CH4 uptake was approximately −2.5 nmol m −2 s −1 during the GS. The soil was a source of N2O, ranging from 0.016-0.89 nmol m −2 s −1 . The soil N2O emission during the SFT period was 1.8 times higher than that during the GS. Soil N2O emissions showed a continuous decreasing trend during the SFT period, which was consistent with the trend of the WWC ( Figure  1b).

Correlations between Soil CO2, CH4, and N2O Fluxes
The relationships among soil CO2, CH4, and N2O fluxes were complex (Figures 2 and 3). The negative correlation between CO2 emission and CH4 uptake during the GS appeared after DOY 251 ( Figure 2). In general, the seasonal pattern of N2O emission was different from those of CO2 emissions

Correlations between Soil CO 2 , CH 4 , and N 2 O Fluxes
The relationships among soil CO 2 , CH 4 , and N 2 O fluxes were complex (Figures 2 and 3). The negative correlation between CO 2 emission and CH 4 uptake during the GS appeared after DOY 251 ( Figure 2). In general, the seasonal pattern of N 2 O emission was different from those of CO 2 emissions and CH 4 uptake (Figure 3a). However, CH 4 uptake and N 2 O emissions showed a positive correlation during the GS, while their association was not evident during the SFT period (Figure 3b). The correlation between the N 2 O flux and the CO 2 flux was not evident during the GS. With the decrease in N 2 O fluxes during the SFT period, CO 2 fluxes changed from exhibiting a downwards trend to an upward trend when the N 2 O emission decreased to 0.42 nmol m −2 s −1 (Figure 3c). and CH4 uptake (Figure 3a). However, CH4 uptake and N2O emissions showed a positive correlation during the GS, while their association was not evident during the SFT period (Figure 3b). The correlation between the N2O flux and the CO2 flux was not evident during the GS. With the decrease in N2O fluxes during the SFT period, CO2 fluxes changed from exhibiting a downwards trend to an upward trend when the N2O emission decreased to 0.42 nmol m −2 s −1 (Figure 3c).

Responses of Soil CO2, CH4, and N2O Fluxes to Environmental Factors
The responses of CO2, CH4, and N2O fluxes to Ts and VWC during the SFT and GS periods differed (Figure 4). During the WO period, soil CO2 emissions increased linearly with the increase in Ts (Figure 4a). In general, Q10 is a quotient of the change in Rs by a 10 °C increase in temperature, which indicates the sensitivity of Rs to temperature [55]. The CO2 emission of the SFT and GS periods had the same sensitivity to Ts, with the same Q10 value of 2.70. However, the response of CO2 flux to VWC was not significant over the WO period. During the SFT period, the variation range of VWC was large (0.05-0.45 m 3 m −3 ); however, the CO2 flux during this period was relatively low, showing a slow linear upward trend with the increase in VWC. During the GS, CO2 emissions increased rapidly as VWC increased and then decreased to a low level after the VWC reached 0.38 m 3 m −3 , which can be expressed approximately as a quadratic function (Figure 4b). Path coefficient analysis indicated that both Ts and VWC promoted CO2 emissions during the SFT period, with a greater contribution of VWC (Figure 5a). Nevertheless, VWC showed a net inhibited CO2 flux during the GS. In general, the main factor regulating CO2 emissions during the SFT period was VWC, whereas the main factor controlling CO2 emission during the GS and WO periods was Ts (Figure 5b,c).
During the WO period, CH4 uptake increased exponentially with the increase in Ts with a Q10 value of 2.18. Similar to the WO period, CH4 uptake in the SFT period also increased exponentially with the increase in Ts with a higher Q10 value of 4.53. The CH4 uptake during the GS increased with the increase in Ts; however, the Q10 value was as low as 1.13. In addition, the effect of Ts on CH4 was unclear (Figure 4c). Throughout the WO period, the responses of soil CH4 flux to the changes in VWC distinctly differed between the SFT and GS periods (Figure 4d). During the SFT period, CH4 uptake decreased linearly with the increase in VWC. Conversely, soil CH4 uptake during the GS drastically decreased as the VWC increased. However, as the CH4 uptake was correlated inversely with the VWC during the SFT and GS periods, the response of CH4 fluxes to the VWC was not significant over the WO period. As shown in the path coefficient analysis (Figure 5), the influences of Ts and VWC on CH4 uptake during the WO period were consistent with those during the SFT period. Specifically, the CH4 flux was mainly regulated by Ts during the SFT and WO periods. While during the GS, both Ts and VWC inhibited the uptake of CH4, and the dominant factor was VWC.
During the SFT period, soil N2O emissions decreased sharply with the increase of Ts (Figure 4e).

Responses of Soil CO 2 , CH 4 , and N 2 O Fluxes to Environmental Factors
The responses of CO 2 , CH 4 , and N 2 O fluxes to T s and VWC during the SFT and GS periods differed (Figure 4). During the WO period, soil CO 2 emissions increased linearly with the increase in T s (Figure 4a). In general, Q 10 is a quotient of the change in R s by a 10 • C increase in temperature, which indicates the sensitivity of R s to temperature [55]. The CO 2 emission of the SFT and GS periods had the same sensitivity to T s , with the same Q 10 value of 2.70. However, the response of CO 2 flux to VWC was not significant over the WO period. During the SFT period, the variation range of VWC was large (0.05-0.45 m 3 m −3 ); however, the CO 2 flux during this period was relatively low, showing a slow linear upward trend with the increase in VWC. During the GS, CO 2 emissions increased rapidly as VWC increased and then decreased to a low level after the VWC reached 0.38 m 3 m −3 , which can be expressed approximately as a quadratic function (Figure 4b). Path coefficient analysis indicated that both T s and VWC promoted CO 2 emissions during the SFT period, with a greater contribution of VWC (Figure 5a). Nevertheless, VWC showed a net inhibited CO 2 flux during the GS. In general, the main factor regulating CO 2 emissions during the SFT period was VWC, whereas the main factor controlling CO 2 emission during the GS and WO periods was T s (Figure 5b,c).
During the WO period, CH 4 uptake increased exponentially with the increase in T s with a Q 10 value of 2.18. Similar to the WO period, CH 4 uptake in the SFT period also increased exponentially with the increase in T s with a higher Q 10 value of 4.53. The CH 4 uptake during the GS increased with the increase in T s ; however, the Q 10 value was as low as 1.13. In addition, the effect of T s on CH 4 was unclear (Figure 4c). Throughout the WO period, the responses of soil CH 4 flux to the changes in VWC distinctly differed between the SFT and GS periods (Figure 4d). During the SFT period, CH 4 uptake decreased linearly with the increase in VWC. Conversely, soil CH 4 uptake during the GS drastically decreased as the VWC increased. However, as the CH 4 uptake was correlated inversely with the VWC during the SFT and GS periods, the response of CH 4 fluxes to the VWC was not significant over the WO period. As shown in the path coefficient analysis (Figure 5), the influences of T s and VWC on CH 4 uptake during the WO period were consistent with those during the SFT period. Specifically, the CH 4 flux was mainly regulated by T s during the SFT and WO periods. While during the GS, both T s and VWC inhibited the uptake of CH 4 , and the dominant factor was VWC.
During the SFT period, soil N 2 O emissions decreased sharply with the increase of T s (Figure 4e). The N 2 O emissions decreased to approximately 6.7% when T s increased by 10 • C. In contrast, the N 2 O flux did not change significantly with the increase in T s during the GS (Figure 4e). Consequently, N 2 O emission showed a negative correlation to T s during the WO period with a Q 10 value of approximately 0.33, due to the evident negative correlation between the N 2 O flux and T s during the SFT period. During the WO, SFT, and GS periods, the N 2 O flux showed a conspicuous downward trend with the increase in VWC (Figure 4f). Overall, T s and VWC were negatively correlated with N 2 O emissions ( Figure 5). During the SFT period, N 2 O emissions were not only regulated by T s but also influenced by the VWC. Compared to T s , VWC had greater control over the N 2 O emissions during the GS and WO periods. emission showed a negative correlation to Ts during the WO period with a Q10 value of approximately 0.33, due to the evident negative correlation between the N2O flux and Ts during the SFT period. During the WO, SFT, and GS periods, the N2O flux showed a conspicuous downward trend with the increase in VWC (Figure 4f). Overall, Ts and VWC were negatively correlated with N2O emissions ( Figure 5). During the SFT period, N2O emissions were not only regulated by Ts but also influenced by the VWC. Compared to Ts, VWC had greater control over the N2O emissions during the GS and WO periods.

GWP100 of Soil CO2, CH4, and N2O Fluxes
The GWP100 of these three soil GHG fluxes during the WO period was 56.81 kg CO2 eq ha −1 yr −1 . Among three GHGs, soil CO2 flux contributed to 98.0% of the GWP100, followed by N2O flux (3.1%), whereas CH4 flux offset 1.0% of the GWP100. The GWP100 during the GS was 12.54 times higher compared to that during the SFT period, indicating that the GWP100 during the GS dominated the GWP100 during the WO period.

CO2
Soil CO2 flux was relatively stable with low emission during the SFT period, whereas it showed a bimodal pattern during the GS, with higher emission. Low microbial activity limited by the low temperature during the SFT period tended to reduce soil respiration [3]. Compared to the SFT period, the increasing physiological activity of plants during the GS allocated more photosynthetically fixed

GWP 100 of Soil CO 2 , CH 4, and N 2 O Fluxes
The GWP 100 of these three soil GHG fluxes during the WO period was 56.81 kg CO 2 eq ha −1 yr −1 . Among three GHGs, soil CO 2 flux contributed to 98.0% of the GWP 100 , followed by N 2 O flux (3.1%), whereas CH 4 flux offset 1.0% of the GWP 100 . The GWP 100 during the GS was 12.54 times higher compared to that during the SFT period, indicating that the GWP 100 during the GS dominated the GWP 100 during the WO period.

CO 2
Soil CO 2 flux was relatively stable with low emission during the SFT period, whereas it showed a bimodal pattern during the GS, with higher emission. Low microbial activity limited by the low temperature during the SFT period tended to reduce soil respiration [3]. Compared to the SFT period, the increasing physiological activity of plants during the GS allocated more photosynthetically fixed carbohydrates to roots and microorganisms, which promoted the metabolic activity of the soil microbes to produce CO 2 [71,72].

CH 4
During the WO period, soil acted as a sink for atmospheric CH 4 , which is consistent with most previous results [8,32,40,73]. In the present study, the average soil CH 4 flux was −0.50 nmol m −2 s −1 during the SFT period and −2.24 nmol m −2 s −1 during the GS (Table 1). At the beginning of the SFT period, the slow decrease in CH 4 uptake may be partly explained by the gradual release of CH 4 from the melting soils, which offsets a portion of the soil CH 4 uptake [22,40]. On the contrary, CH 4 uptake showed an increasing trend at the beginning of the GS, caused by the hydrothermal conditions and aerobic conditions suitable for methanotrophs, together with the inactive methanogenic bacterial community. When entering the rainy season, the soil anaerobic conditions promoted an increase in CH 4 production due to precipitation; therefore, CH 4 uptake was reduced [40].

N 2 O
The soil N 2 O flux during the SFT period was higher than that in the GS. The pulse emission of soil N 2 O detected at the beginning of the SFT period is consistent with the results of Peng et al. [47]. Many studies have reported high soil N 2 O emissions during the SFT period in forest ecosystems, which played an important role in regulating the annual N 2 O budget [74,75]. For example, N 2 O emissions during the FTC accounted for approximately 88% of the annual N 2 O emission in a broad-leaved spruce forest in Germany [32]. The average N 2 O flux during the SFT period in the present study was 0.42 nmol m −2 s −1 (Table 1). Studies have found that low temperatures had little influence on the denitrification under anoxic conditions, while net N mineralisation and nitrification within frozen soil could still produce N 2 O [34,76]. Thus, a substantial release of N 2 O could occur during the soil melting process after a long freezing period [47,77,78]. N 2 O emissions showed a gradually decreasing trend during the SFT period ( Figure 2). The increasing freeze-thaw cycles gradually consumed the contents of C and N, which lowered the potential of producing N 2 O [38]. The N 2 O flux during the GS in the present study is consistent with the average global N 2 O flux in temperate forests (0.07 nmol m −2 s −1 ) [79]. The possible reasons for the low N 2 O flux during the GS may be the substrate limitation caused by the high consumption of substrates during the SFT period and the inhibition of N 2 O production caused by the high VWC during this period. At the end of the GS, N 2 O emissions tended to increase as precipitation decreased and the soil became dry [80].

Relationships among Soil GHG Fluxes
The strong coupling effect of soil GHG fluxes increases the uncertainty in estimating the impact of forest soil GHG fluxes on greenhouse effects and global climate change [51,81]. Therefore, it is necessary to observe GHG (CO 2 , CH 4 , and N 2 O) fluxes synchronously. The variation patterns of CO 2 emissions and CH 4 uptake were similar, showing higher fluxes during the GS with two peaks. Under aerobic conditions, the final product of CH 4 oxidation was CO 2 , while CH 4 was generated by methanogens under anaerobic conditions by consuming CO 2 [7,82,83]. During the second part of the GS, both the CO 2 emissions and CH 4 uptake were low, which was possibly due to the intensity of precipitation from late June to early August decreased T s and inhibited soil CO 2 emissions and CH 4 uptake. The decomposition of soil organic matter provided substrates for nitrifying and denitrifying bacteria and promoted the production of soil N 2 O flux [84,85]. Soil respiration also affected soil nitrification and denitrification by changing soil aerobic/anaerobic conditions, which consequently affected soil N 2 O flux [86]. An evident positive correlation between CH 4 uptake and N 2 O emission in the middle of the GS was observed, due to the production or consumption processes related to CH 4 and N 2 O fluxes controlled by redox reactions [51,81]. In previous studies, the rates of CH 4 uptake and N 2 O emissions were higher under aerobic conditions, and N in the substrate under anaerobic conditions was the common limiting factor that led to a positive correlation between CH 4 uptake and N 2 O emissions [81,87].

CO 2
T s generally affected soil respiration by changing soil microbial activities [3,88]. During the GS, T s was the main driving force of CO 2 emission, with the majority of the CO 2 flux occurred at high T s . In the present study, the responses of soil CO 2 emissions to soil VWC varied greatly between the SFT and GS periods. The dynamics of soil CO 2 emission was dominated by soil VWC during the SFT period, showing a positive correlation. During the GS, soil CO 2 emissions initially increased and then decreased with the increase in soil VWC, which was consistent with the observation in a mid-temperate forest [34]. In the early stage of the GS, precipitation promoted CO 2 production through increasing available substrates, which was due to the enhanced infiltration of precipitation releasing the previously accumulated CO 2 in the soil pores [89,90]. Soil respiration was sensitive to the intensity and frequency of precipitation. Precipitation events during the early stage of the GS could promote soil CO 2 emissions but had an inverse impact during the late stage of the GS. Soils with higher accumulation of precipitation in the soil pores inhibited the diffusion of soil CO 2 as well as the soil oxygen content, thereby reducing the soil respiration and ultimately resulting in the decreased CO 2 emissions [40,91].

CH 4
Soil CH 4 flux is determined by the balance between methane production from methanogenic archaea and consumption by methanotrophs [7,8,92]. As the aeration condition of forest soil is conducive to methane oxidation [93], forest soils often serve as sinks of CH 4 [8]. The increase in precipitation reduces soil methane uptake by increasing soil VWC [94]. In the present study, CH 4 uptake showed a positive response to T s (Figure 4). During the SFT period, T s acted as the main regulator of CH 4 flux. The increase in T s may stimulate the activity, abundance, and community composition of methanotrophs, thereby promoting CH 4 uptake [50]. In addition, enhanced T s caused the rise of CH 4 transmission by decreasing soil VWC under warming conditions, thereby enlarging soil pores [25]. A previous study showed that T s within the range of 5-30 • C had a promoting effect on CH 4 uptake [95], which was consistent with what the present study found during the SFT period. However, no evident correlation between CH 4 uptake and T s was determined during the GS, which may be related to the strong substrate limitation of methanotrophs [7,40]. It showed that VWC affected soil CH 4 uptake quite differently between the SFT and GS periods. Compared to T s , soil VWC explained 84% of the variation in the CH 4 flux [96,97]. During the SFT period, CH 4 uptake was generally positively correlated with soil VWC, which may be related to increasingly unstable C and N contents [92]. During the GS, soil VWC played as a key regulator of CH 4 flux and could serve as a good predictor of soil CH 4 uptake. The soil VWC restricted oxygen transmission from the atmosphere to the soil, which reduced methanotroph activity. Meanwhile, soil VWC provided necessary anaerobic conditions for methanogenic bacteria [98][99][100]. Precipitation reduced CH 4 uptake by improving the soil VWC, thereby accelerating the concentration of CH 4 into the atmosphere [94].

N 2 O
Previous studies have demonstrated that T s and soil VWC are key controllers of N 2 O emissions in the absence of limitations by other factors. For instance, Kitzler, et al. [101] reported that T s and soil VWC can explain >95% of the variation in soil N 2 O emissions. However, the influence of T s on N 2 O emissions in the present study was more complicated. Soil N 2 O emissions showed poor correlation with the variation in T s during the GS, whereas a higher N 2 O emission during the SFT period decreased exponentially with increasing T s . Soil nitrification is an aerobic process, whereas denitrification is an anaerobic process [102]. Denitrification was the dominant process of N 2 O production in soils during the SFT period as the enzyme activity of N 2 O production within the denitrification process was not distinctly affected by low temperatures; however, the nitrification process was more inhibited by low temperatures [26,34,103]. This indicates that other factors, such as the reduced amounts of labile substrate availability for microbial metabolism, could be responsible for decreased N 2 O emissions with the increase in T s during the SFT period [33,104]. Soil VWC was the key factor determining the soil N 2 O flux during the GS by influencing the N 2 O emission inversely. The strict anaerobic condition caused by the limitation of soil oxygen utilisation under high soil VWC triggered the reaction from N 2 O to N 2 in the denitrification process, ultimately restricting N 2 O emissions [29,105]. Moreover, low microbial activity and N 2 O transmission from the soil into the atmosphere under high soil VWC were the inhibited factors for soil N 2 O emissions. During the SFT period, N 2 O emissions increased drastically. One of the possible reasons was that the soil water molecules expanded after soil freezing, which accelerated the rupture of soil aggregates and subsequently released more soil labile organic matter [40,47]. The labile substrates available for microbial metabolism, which entered different layers of soil with soil water after thawing, were used by the denitrification process to produce N 2 O. However, because of substrate supply limitations provided by the nitrification in the microbial denitrifying process, the tremendous increase of N 2 O emission is usually discontinuous. On the other hand, nitrification and denitrification are processes that depend on suitable oxygen availability that produces N 2 O [8]. Nevertheless, the soil N 2 O flux decreased during the late stage of the SFT period caused by the decrease in oxygen concentrations due to the increasing soil VWC. Conversely, the variation trend of soil N 2 O emissions was consistent with the WWC trend. The reason may be that the frozen water content of soil rapidly decreased, which only distinctly affected soil WWC, whereas the available water content in soil (VWC) rapidly increased during the SFT period.

Contributions of Soil GHG Fluxes to the Greenhouse Effect
Determining the GWP 100 of forest soil CO 2 , CH 4 , and N 2 O fluxes would help assess the impacts of soil GHG exchanges on global warming [106]. In the present study, there were distinct differences in GWP 100 of the temperate forest soil between the SFT and GS periods; the GWP 100 during the GS was approximately 12.54 times higher than that during the SFT period. The soil GHG fluxes during the GS were of greater significance to global warming due to the higher rates and total amounts of CO 2 flux in this period; the GWP 100 of the CO 2 flux was 11,727.03 kg CO 2 eq ha −1 yr −1 during the GS and 681.74 kg CO 2 eq ha −1 yr −1 during the SFT period. Although the CO 2 flux made the largest contribution to GWP 100 , the warming capacities of CH 4 and N 2 O are much stronger than that of CO 2 [2]. Compared to the GS, the SFT period had higher N 2 O emissions and lower CH 4 uptake, and therefore, their GWP 100 values during the SFT period cannot be ignored. The GWP 100 of the N 2 O flux during the SFT period (263.55 kg CO 2 eq ha −1 yr −1 ) contributed more than that during the GS (127.19 kg CO 2 eq ha −1 yr −1 ), whereas the GWP 100 of the CH 4 flux during the SFT period (−9.60 kg CO 2 eq ha −1 yr −1 ) offset less than the GWP 100 during the GS (−122.04 kg CO 2 eq ha −1 yr −1 ). Similar results have been found in previous studies, where the CO 2 flux contributed >90% of the GWP 100 [51,55]. In particular, the total GWP 100 (12.46 t CO 2 eq ha −1 yr −1 ) for the over-mature forest in Northeastern China was very close to the total GWP 100 (12.67 t CO 2 eq ha −1 yr −1 ) found in the present study [55]. However, the GWP 100 of N 2 O in Wu and Mu [55] was four times higher than that reported in this present study, which was mainly due to the spatial heterogeneity of the N 2 O flux. In addition, the incomplete non-growing season observation of GHG fluxes in the present study led to an underestimation of the GWP 100 of the N 2 O flux due to the high soil N 2 O emissions during the freeze-thaw period in the temperate forest [34,47,78].

Conclusions
Overall, we comprehensively examined the soil GHG (CO 2 , CH 4 , and N 2 O) fluxes between the SFT and GS periods in a temperate broad-leaved Korean pine mixed forest and investigated their connections with the major environmental variables (T s and soil VWC). Our results indicate that the cumulative fluxes of both CO 2 emissions and CH 4 uptake during the GS were far higher than those during the SFT period. However, the cumulative N 2 O fluxes in the SFT period accounted for 64.7% of the total emissions in the WO period, which was probably attributed to a substantial release of N 2 O after a long freezing period and the decomposition of soil nitrification and denitrification resulted from the increased soil microbial activities and available substrates during the SFT period. Furthermore, CO 2 emissions and CH 4 uptake showed similar variation trends throughout the WO period. Meanwhile, N 2 O emissions and CH 4 uptake presented consistent variation trends during most of the GS. These results imply that the effect of T s and soil VWC on the mechanisms of soil GHG emissions/uptake can be similar in certain periods throughout the year.
Affected by the East-Asia monsoon climate, the T s and soil VWC in this temperate forest differed drastically between the SFT and GS periods. We found that soil GHG fluxes responded differently to T s and soil VWC during the SFT and GS periods. N 2 O emissions decreased drastically with the increases in T s and soil VWC during the SFT period but varied little with T s during the GS. While soil N 2 O emissions and CH 4 uptake were dominated by the soil VWC during the GS, soil CO 2 emissions and CH 4 uptake during the WO period were mainly controlled by the T s . These results stress the role of T S and soil VWC in shaping soil GHG fluxes during the SFT and GS periods, respectively, thereby helping to improve the accuracy in estimating the soil GHG budgets under global climate change.
The results also show that soil CO 2 flux contributed predominantly for GWP 100 followed by soil N 2 O flux, whereas soil CH 4 flux was an offsetting factor for the total GWP 100 . Taken together, our findings suggest that the soil freeze-thaw cycle is important to characterise the dynamics of soil GHG fluxes and highlight the importance of the N 2 O pulse during the SFT period. Further studies are required for the exploration of the mechanisms among soil GHG exchanges and the accurate estimation of total annual GHG fluxes and their global warming potential by analysing soil GHG fluxes, soil properties, and soil microbial characteristics in a longer time series, especially during the SFT period.