Seasonal Variation of Soil Respiration in the Mongolian Oak (Quercus mongolica Fisch. Ex Ledeb.) Forests at the Cool Temperate Zone in Korea

To investigate the variation in seasonal soil respiration (SR) as a function of soil temperature (Ts) and soil water content (SWC) in Mongolian oak (Quercus mongolica) forests in urban (Mt. Nam) and well-reserved (Mt. Jeombong) areas in South Korea, we conducted continuous field measurements of SR and other environmental parameters (Ts and SWC) using an automated chamber system. Overall, the SR rates in both stands were strongly correlated with the Ts variable during all seasons. However, abrupt fluctuations in SR were significantly related to episodic increases in SWC on a short time scale during the growing season. The integrated optimal regression models for SR using Ts at a depth of 5 cm and SWC at a depth of 15 cm yielded the following: the SR rate in Mt. Nam = SR(Ts) + ΔSR(Ts) = 104.87 exp(0.1108Ts) − 10.09(SWC)2 + 604.2(SWC) − 8627.7 for Ts ≥ 0 °C, and the SR rate in Mt. Jeombong = SR(Ts) + ΔSR(Ts) = 95.608 exp(0.1304Ts) − 33.086(SWC)2 + 1949.2(SWC) − 28499 for Ts ≥ 0 °C. In both cases, SR = 0 for Ts < 0 °C. As per these equations, the estimated annual total SRs were 1339.4 g C m−2 for Mt. Nam and 1003.0 g C m−2 for Mt. Jeombong. These values were quite similar to the measured values in field. Our results demonstrate that the improved empirical equation is an effective tool for estimating and predicting SR variability and provide evidence that the SR of Q. mongolica forests in the cool temperate zone of Korean Peninsula depends on Ts and SWC variables.


Introduction
Carbon dioxide (CO 2 ) efflux from the soil to the atmosphere (hereafter, referred to as soil respiration, SR) is one of the major pathways in the global carbon (C) cycle [1][2][3][4]. Potential SR rate changes in response to environmental conditions could either accelerate global warming driven by rising atmospheric CO 2 concentrations (via SR) or mitigate climate change by enhancing ecosystem C sequestration [5][6][7]. However, the high temporal-spatial variations in SR make it difficult to evaluate the soil C budget on a global scale and to quantify the feedback relationships with global warming and climate [8,9]. If we are to accurately predict ecosystem responses to climate change, it is crucial to quantify the processes of SR on the local scale by using high-resolution data and high frequency measurements [10][11][12].
Forests are regarded as significant CO 2 sinks [13][14][15][16], not only covering~30% (4.2 × 10 9 ha) of the global land area [17], but also containing over 80% of all aboveground vegetation C and~40% of all belowground soil C [18,19]. Given these large amounts and capacities in the forest C pool, small The specific objectives of this study were (1) to improve the empirical models for SR in Mongolian oak (Q. mongolica) forests derived from high-temporal frequency SR, temperature (T), and soil water content (SWC) measurements and (2) to investigate seasonal variations in SR in two Q. mongolica forest stands located in central western (Mt. Nam) and central eastern (Mt. Jeombong) Korea and their dependence on T and SWC.

Mt. Nam
The Mt. Nam ecological experimental site has a northeastern gentle slope and is located in the central western part of the Korean Peninsula (37°33′ N, 126°59′ E; 220 m a.s.l.; Figure 1). This region has a cool-temperate climate under the influence of the Asian monsoon and is characterized by mild springs and autumns, hot and humid summers, and cold and snowy winters. According to the climate database from the Seoul Weather Station of the Korea Meteorological Administration close to this site, the mean annual air temperature and precipitation are 13.3 °C and 1212.3 mm, respectively. The monthly mean temperatures in the coldest and warmest months are 0.4 °C in January and 26.5 °C in August, respectively. The vegetation at the study site is classified as a broadleaved deciduous forest (~49-55 years old) dominated by Mongolian oak (Quercus mongolica). The mean canopy height, average diameter at breast height (DBH), and vegetation density are 15.1 m, 23.2 cm, and 482 trees per hectare (ha −1 ), respectively. The vegetation is stratified into four layers composed of trees (>12 m), subtrees (4.5-8 m), shrubs (1-2.5 m) and herbs (<0.5 m), with coverages of

Mt. Nam
The Mt. Nam ecological experimental site has a northeastern gentle slope and is located in the central western part of the Korean Peninsula (37 • 33 N, 126 • 59 E; 220 m a.s.l.; Figure 1). This region has a cool-temperate climate under the influence of the Asian monsoon and is characterized by mild springs and autumns, hot and humid summers, and cold and snowy winters. According to the climate database from the Seoul Weather Station of the Korea Meteorological Administration close to this site, the mean annual air temperature and precipitation are 13.3 • C and 1212.3 mm, respectively. The monthly mean temperatures in the coldest and warmest months are 0.4 • C in January and 26.5 • C in August, respectively. The vegetation at the study site is classified as a broad-leaved deciduous forest (~49-55 years old) dominated by Mongolian oak (Quercus mongolica). The mean canopy height, average diameter at breast height (DBH), and vegetation density are 15.1 m, 23.2 cm, and 482 trees per hectare (ha −1 ), respectively. The vegetation is stratified into four layers composed of trees (>12 m), subtrees (4.5-8 m), shrubs (1-2.5 m) and herbs (<0.5 m), with coverages of 82.8%, 57.3%, 30.9%, and 33.3%, respectively. The mid-and understory vegetation is primarily composed of Sorbus alnifolia, Styrax japonica, and Acer pseudosieboldianum ( Figure 2). Among these, S. alnifolia and S. japonica are flourishing species in extremely polluted areas such as industrial complexes [48,49]. These species reflect Mt. Nam's exposure to severe environmental pollution and human interference. The leaves of the dominant broad-leaved deciduous trees at the site begin to develop in April and fall from October to November. The topography of this stand has both ridge forms and middle slopes (18 • to 22.7 • ). The soil profile at the site ( Figure 3) includes a 1-4 cm soil depth organic layer (A0), a 12-25 cm topsoil layer (A1 and A2) overlaid on the subsoil layer (B1 and B2), and a parent material layer of granite or granite-gneiss (C and D). The soil is classified as a brown forest soil (Dystric Cambisols according to FAO-UNESCO [50]). Further description of the site was provided in [12]. human interference. The leaves of the dominant broad-leaved deciduous trees at the site begin to develop in April and fall from October to November. The topography of this stand has both ridge forms and middle slopes (18° to 22.7°). The soil profile at the site ( Figure 3) includes a 1-4 cm soil depth organic layer (A0), a 12-25 cm topsoil layer (A1 and A2) overlaid on the subsoil layer (B1 and B2), and a parent material layer of granite or granite-gneiss (C and D). The soil is classified as a brown forest soil (Dystric Cambisols according to FAO-UNESCO [50]). Further description of the site was provided in [12].  The Mt. Jeombong ecological experimental site is located on a gentle southeastern slope in a natural forest area in Inje county, Gangwon province, which is in the central eastern part of the Korean Peninsula (38°02′ N, 128°28′ E; 765 m a.s.l.; Figure 1). This mountainous region has well-  The Mt. Jeombong ecological experimental site is located on a gentle southeastern slope in a natural forest area in Inje county, Gangwon province, which is in the central eastern part of the Korean Peninsula (38 • 02 N, 128 • 28 E; 765 m a.s.l.; Figure 1). This mountainous region has well-preserved forests and is designated as a Biosphere Reserve by the United Nations Educational, Scientific and Cultural Organization (UNESCO)'s Man and Biosphere Project. Like Mt. Nam, this region also has a cool-temperate climate under the influence of the Asian monsoon. According to the climate database from the Inje Weather Station of the Korea Meteorological Administration close to this site, the annual mean air temperature and precipitation are 10.2 • C and 1135.7 mm, respectively. The monthly mean temperatures in the coldest and warmest months are −4.8 • C in January and 16.6 • C in August, respectively. The ground surface of this region is usually covered with snow from mid-December to early March. The vegetation at the study site is classified as a broad-leaved deciduous forest (~61-68 years old) dominated by Mongolian oak (Quercus mongolica). The mean canopy height, DBH, and density in the forest stand are 18.4 m, 23.6 cm, and 431 ha −1 , respectively. The vegetation is stratified into four layers composed of trees (>16 m), subtrees (7-12 m), shrubs (1-4 m) and herbs (<0.6 m), with coverages of about 78.2%, 36.2%, 66.6%, and 59.8%, respectively. The midand understory vegetation is composed of Acer pseudosieboldianum, Styrax obassia, Carpinus cordata, and Tilia amurensis ( Figure 4). The A. pseudosieboldianum and S. obassia species are typical sub-trees in Q. mongolica forests [51]. This species composition reflects that Mongolian oak forest established in Mt. Jeombong is a typical natural forest. The leaves of the dominant broad-leaved deciduous trees begin to develop in April and begin falling in October. The topography of this stand has both ridge forms and middle slopes (15 • to 24 • ). The soil profile ( Figure 5) includes a 1-5 cm soil depth organic layer (A0), a 27-42 cm topsoil layer (A1 and A2) overlaying the subsoil layer (B1 and B2), and a parent material layer of granite (C and D). The soil is classified as a brown forest soil (Dystric Cambisols, [50]). Further site descriptions were provided by Kim et al. [47].  Figure 4). The A. pseudosieboldianum and S. obassia species are typical sub-trees in Q. mongolica forests [51]. This species composition reflects that Mongolian oak forest established in Mt. Jeombong is a typical natural forest. The leaves of the dominant broad-leaved deciduous trees begin to develop in April and begin falling in October. The topography of this stand has both ridge forms and middle slopes (15° to 24°). The soil profile ( Figure 5) includes a 1-5 cm soil depth organic layer (A0), a 27-42 cm topsoil layer (A1 and A2) overlaying the subsoil layer (B1 and B2), and a parent material layer of granite (C and D). The soil is classified as a brown forest soil (Dystric Cambisols, [50]). Further site descriptions were provided by Kim et al. [47].

Measuring Soil Respiration with an Automated Chamber System
An AOCC based on the closed dynamic method was used for continuous SR measurements from the forest soil surface at multiple locations. This system consists of eight automated chambers, an 8-channel gas sampler, an infrared gas analyzer (IRGA; Li-840; Li-COR, Lincoln, NE, USA), a data logger (CR1000; Campbell Scientific Inc., Logan, UT, USA), and a notebook computer. The chamber (cylindrical with a 30 cm internal diameter, 20 cm height, and 0.5 cm wall thickness) has a wall made of transparent polycarbonate (PC) with a hinged lid on the side wall. The transparent PC lid (5 mm thick) can be opened and closed automatically by a fixed 12-V DC motor. To ensure a gas-tight seal between the cylindrical chamber body and the closed lid, a soft rubber gasket was attached to the top edge of the chamber. While the chamber is closed, a mixing fan (KMFH-12; Nihon Blower, Tokyo, Japan) inside each chamber maintains air movement at a speed of 0.1 m s −1 .
One study plot sized in 30 m × 30 m was established on the forest floor of the Q. mongolica forest stand beneath the eco-towers at Mt. Nam and Mt. Jeombong, respectively. The eight chambers were randomly installed at least 1 m away from the surrounding overstory trunks in the forest stand study plots and inserted 3 cm below the top of the litter layer. With the exception of the deep snow-coverage season, the SR rates in the forest stands at Mts. Nam and Jeombong were measured continuously from January to December 2011. All green plants were periodically removed from inside each chamber. When the chamber was open (during non-measuring times) a section of the chamber lid was raised vertically to allow rainfall, snow, leaves, litter, and twigs to reach the soil surface inside the chamber.
Teflon tubes (20 m long) were used to sample the air from each chamber. While the chamber air was withdrawn sequentially from all eight chambers, only the air from the closed chamber was supplied to the multichannel gas sampler system by an air pump (CM-50 with a maximum flow rate of 5 L min −1 ; EMP, Tokyo, Japan). The flow rate in each chamber was monitored and balanced by a mass-flow controller (SEF-21A; STEC, Kyoto, Japan). The 5 L min −1 airstream from the outlet of the chamber was divided by two mass-flow meters; one airstream line of 0.7 L min −1 flowed to the IRGA to measure the CO 2 concentration, while the other bypassed the IRGA (4.3 L min −1 ). The air exhausted from the IRGA went to a bypass tube, with the total remaining air continuously flowing into the chamber inlet. To avoid any pressure changes, the same amount of air supplied to the chamber inlet was simultaneously withdrawn by an air pump from the chamber outlet. Over the course of a half hour, the chambers were closed sequentially and continuously by an 8-channel relay driver controlled by a program in the data logger. We set a sampling time of 225 s per chamber. The data logger acquired output from the IRGA at one second intervals, and the output was averaged and recorded every ten seconds. The raw signal data for the CO 2 concentration from the IRGA were used to calculate the SR rates (Equation (1)). The data collected from the eight automated chambers were averaged for each 30 min cycle. The IRGA was calibrated with CO 2 zero gas (pure nitrogen), and two span gasses with different CO 2 concentrations at least once every three months. In both study sites, the precipitation was measured at hourly intervals using a weighing bucket rain gauge (Model-R301, Weather Tec. Seoul, Korea) installed and mounted on the top of eco-tower at a height of 30 m above ground.

Measuring Temperature and Soil Water Content
The air temperature (Ta) 1.5 m above the ground surface, the soil surface temperature (Tss; 0 cm depth), and soil temperatures (Ts) at 5 and 10 cm depths were obtained with copper-constantan thermocouple probes, and the volumetric soil water content (SWC, 0-100 vol% = [V w m 3 (the volume of water)/V t m 3 (the bulk volume of soil)] × 100) at the 15 cm depth was obtained with a time-domain reflectometry (TDR) probe (CS-616; Campbell Scientific Inc., Logan, UT, USA). These measurements were taken near the automated chambers simultaneously with the SR rate measurements. All sensors were calibrated at six month intervals. All micrometeorological data were continuously measured at one second intervals, and the selected data were averaged and recorded once every ten seconds by a datalogger (CR1000, Campbell Scientific Inc., Logan, UT, USA). After removing erroneous raw data, 30 min averages for the Ta, Ts, Tss, and SWC data were calculated and used to analyze the mean 30 min SR rates.

Data Analysis
The half-hourly measurements made by the eight automated chambers were averaged to obtain representative 30 min data for the Q. mongolica forest stands on Mts. Nam and Jeombong. Over the course of the field measurements, some SR data could not be obtained due to electrical malfunctions, IRGA calibrations, instrument failures, heavy snowfall, or heavy snow accumulation, mostly during the winter season.
The soil respiration rate (mg CO 2 m −2 s −1 ) was calculated as where a is the change rate of the CO 2 concentration in the chamber (µmol mol −1 s −1 ); p is the CO 2 density in the air (mg m −3 ); V is the chamber volume, including the volume of the tubes; and A is the soil surface area (m 2 ) enclosed by the chamber. Thirty minute averages of the SR rate values from the eight automated chambers were used to obtain the empirical models.
Optimal regression equations were derived to examine the relationships among SR, Ta (1.5 m above ground), Ts (5 and 10 cm below ground), Tss (soil surface), and SWC. The dependence of the SR rate on temperature (Ta, Ts and Tss) was modeled using the following regression equation: where the SR rate is the measured SR (mg CO 2 m −2 s −1 ); t is the measured Ta ( • C), Tss ( • C), or Ts ( • C); α is the coefficient of the basic respiration rate at a reference temperature of 0 • C, and b is the sensitivity of the SR to temperature (Ta, Ts and Tss). The sensitivity (b) is related to Q 10 , which is derived as follows: where Q 10 is the SR rate increase for a 10 • C rise in temperature (T). The a, b, and Q 10 parameter values derived from the observed field data reflect the effects of T and other variables on SR. Thus, Q 10 is generally used to describe the temperature dependence of SR.

Seasonal Temperature and Soil Water Content Variation
Both Q. mongolica forest stands (Mts. Nam and Jeombong) showed significant seasonal variation (   The stand at Mt. Nam showed a temporary increase in the daily mean volumetric SWC at the 15 cm depth in late February 2011 ( Figure 6). The SWC tended to decrease continuously from early May to mid-June, and then had relatively high values during the rainy season from late June to July. The   The stand at Mt. Nam showed a temporary increase in the daily mean volumetric SWC at the 15 cm depth in late February 2011 ( Figure 6). The SWC tended to decrease continuously from early May to mid-June, and then had relatively high values during the rainy season from late June to July. The The stand at Mt. Nam showed a temporary increase in the daily mean volumetric SWC at the 15 cm depth in late February 2011 ( Figure 6). The SWC tended to decrease continuously from early May to mid-June, and then had relatively high values during the rainy season from late June to July.
The SWC then decreased up to early September, after which it maintained constant values without any significant changes. The daily mean SWC ranged from 22.0% to 29.2% at the Mt. Nam site, and 64% of the total annual precipitation fell in July and August.
Except for relatively high values from mid-April to late August at the Mt. Jeombong site, there were no clear seasonal changes in the daily mean volumetric SWC (range: 24.6% to 28.9%) at the 15 cm depth (Figure 7). Fifty percent of the total annual precipitation fell in July and August at this site.

Seasonal Variations in Soil Respiration
The SR rates in the forest stands at Mts. Nam and Jeombong were measured continuously from January to December 2011 using the AOCCs (Figures 8 and 9). The mean annual SR was 549.8 ± 185.6 mg CO 2 m −2 h −1 at Mt. Nam and 539.5 ± 148.1 mg CO 2 m −2 h −1 at Mt. Jeombong. For both forest stands, the daily mean SR showed significant seasonal changes, similar to those seen for temperature (Figures 6 and 7). The daily mean SR increased markedly from the spring (DOY 60 to The seasonal SR variations were significantly correlated with T in both forest stands throughout the entire experimental period (r 2 = 0.78-0.91, p < 0.01), but had no significant correlation with the SWC (r 2 = 0.13-0.27, p > 0.05). However, the SR rate abruptly increased with an increase in the SWC due to precipitation during the growing season (Figures 6-9). These results show that SWC can cause abrupt fluctuations in SR on short time scales following rainfall events, but it is the constant influence of T that reflects the seasonal time scale.  (Figure 7). Fifty percent of the total annual precipitation fell in July and August at this site.

Seasonal Variations in Soil Respiration
The SR rates in the forest stands at Mts. Nam and Jeombong were measured continuously from January to December 2011 using the AOCCs (Figures 8 and 9). The mean annual SR was 549.8 ± 185.6 mg CO2 m −2 h −1 at Mt. Nam and 539.5 ± 148.1 mg CO2 m −2 h −1 at Mt. Jeombong. For both forest stands, the daily mean SR showed significant seasonal changes, similar to those seen for temperature ( Figures  6 and 7). The daily mean SR increased markedly from the spring (DOY 60 to DOY 151: 75-399 mg  Figures 8 and 9). The seasonal SR variations were significantly correlated with T in both forest stands throughout the entire experimental period (r 2 = 0.78-0.91, p < 0.01), but had no significant correlation with the SWC (r 2 = 0.13-0.27, p > 0.05). However, the SR rate abruptly increased with an increase in the SWC due to precipitation during the growing season (Figures 6-9). These results show that SWC can cause abrupt fluctuations in SR on short time scales following rainfall events, but it is the constant influence of T that reflects the seasonal time scale.

Relationships between Soil Respiration and Temperature
To elucidate the dependence of SR on the temporal variabilities of temperature in the forest stands, optimal regression equations using the exponential form of Equation (2) were derived from the relationships between the daily mean SR rates and the temperatures (Ta, Tss and Ts) observed at several different depths (1.5 m above ground, soil surface, and 5 and 10 cm soil depths) for the entire experimental period (Figures 10 and 11). The Ts measurements at the 5 cm depth were found to be highly correlated with the SR for both forest stands.

Relationships between Soil Respiration and Temperature
To elucidate the dependence of SR on the temporal variabilities of temperature in the forest stands, optimal regression equations using the exponential form of Equation (2) were derived from the relationships between the daily mean SR rates and the temperatures (Ta, Tss and Ts) observed at several different depths (1.5 m above ground, soil surface, and 5 and 10 cm soil depths) for the entire experimental period (Figures 10 and 11). The Ts measurements at the 5 cm depth were found to be highly correlated with the SR for both forest stands.

Relationships between Soil Respiration and Temperature
To elucidate the dependence of SR on the temporal variabilities of temperature in the forest stands, optimal regression equations using the exponential form of Equation (2) were derived from the relationships between the daily mean SR rates and the temperatures (Ta, Tss and Ts) observed at several different depths (1.5 m above ground, soil surface, and 5 and 10 cm soil depths) for the entire experimental period (Figures 10 and 11). The Ts measurements at the 5 cm depth were found to be highly correlated with the SR for both forest stands.    (4) and (5) using the soil temperature at the 5 cm depth. The best regression equations in Figures 12 and 13 reflect the seasonal trends of the observed daily mean SR rates for both Q. mongolica forest stands well; however, they cannot explain the peak SR values observed. The abrupt short-term fluctuations in the observed SR rates seem to be related to the SWC fluctuations in both forest stands.
Equation (4) (r 2 = 0.91, Q 10 = 3.0) and Equation (5) (r 2 = 0.84, Q 10 = 3.7) are valid for the Ts at 5 cm depth, which is above 0 • C, and the rate of SR (Ts) for Ts < 0 • C is near-zero value. Figures 12 and 13 show the seasonal variations in the observed daily mean SR rates at Mt. Nam ( Figure 8) and Mt. Jeombong ( Figure 9) along with the values predicted by Equations (4) and (5) using the soil temperature at the 5 cm depth. The best regression equations in Figures 12 and 13 reflect the seasonal trends of the observed daily mean SR rates for both Q. mongolica forest stands well; however, they cannot explain the peak SR values observed. The abrupt short-term fluctuations in the observed SR rates seem to be related to the SWC fluctuations in both forest stands.  (4) and (5) using the soil temperature at the 5 cm depth. The best regression equations in Figures 12 and 13 reflect the seasonal trends of the observed daily mean SR rates for both Q. mongolica forest stands well; however, they cannot explain the peak SR values observed. The abrupt short-term fluctuations in the observed SR rates seem to be related to the SWC fluctuations in both forest stands.

Relationships between Soil Respiration and Soil Water Content
To identify the dependence of SR on the temporal variability of SWC in the Q. mongolica forests, we ran a time series of the difference [ΔSR(Ts)] between the observed and predicted (regressions using Equations (4) or (5)) values, the SWC measured at a 15 cm depth, and the daily total precipitation (mm d −1 ) in each forest stand (Figures 14 and 15). The SWC was markedly correlated with the precipitation rate in both forest stands, and the peak ΔSR(Ts) values were closely associated with the dramatic increase in SWC during the forest growing season. Figure 16

Relationships between Soil Respiration and Soil Water Content
To identify the dependence of SR on the temporal variability of SWC in the Q. mongolica forests, we ran a time series of the difference [∆SR(Ts)] between the observed and predicted (regressions using Equations (4) or (5)) values, the SWC measured at a 15 cm depth, and the daily total precipitation (mm d −1 ) in each forest stand (Figures 14 and 15). The SWC was markedly correlated with the precipitation rate in both forest stands, and the peak ∆SR(Ts) values were closely associated with the dramatic increase in SWC during the forest growing season. Figure 16

Relationships between Soil Respiration and Soil Water Content
To identify the dependence of SR on the temporal variability of SWC in the Q. mongolica forests, we ran a time series of the difference [ΔSR(Ts)] between the observed and predicted (regressions using Equations (4) or (5)) values, the SWC measured at a 15 cm depth, and the daily total precipitation (mm d −1 ) in each forest stand (Figures 14 and 15). The SWC was markedly correlated with the precipitation rate in both forest stands, and the peak ΔSR(Ts) values were closely associated with the dramatic increase in SWC during the forest growing season. Figure 16 Equations (6) and (7) indicate that SR increases at Mts. Nam (r 2 = 0.74) and Jeombong (r 2 = 0.70) with increases in SWC up to 29.2% and 28.9%, respectively. But SR tended to remain at a stable state in further increase in SWC.
Equations (6) and (7) indicate that SR increases at Mts. Nam (r 2 = 0.74) and Jeombong (r 2 = 0.70) with increases in SWC up to 29.2% and 28.9%, respectively. But SR tended to remain at a stable state in further increase in SWC.

Seasonal Variation of Soil Respiration Estimated by Soil Temperature and Soil Water Content
To quantify the seasonal variability in the SR rates in the Q. mongolica forest stands at Mts. Nam and Jeombong, optimal integrated regression equations were constructed with both the Ts and SWC variables driving the SR rates.
The for Mts. Nam and Jeombong, respectively. The SR rate = 0 for Ts < 0 °C Equations (8) and (9) were quite effective for use at all temperature (T) ranges in the two forest stands, whereas the SWC variables were more effective for the Q. mongolica forest growing season (Figures 18 and 19).

Seasonal Variation of Soil Respiration Estimated by Soil Temperature and Soil Water Content
To quantify the seasonal variability in the SR rates in the Q. mongolica forest stands at Mts. Nam and Jeombong, optimal integrated regression equations were constructed with both the Ts and SWC variables driving the SR rates.
The for Mts. Nam and Jeombong, respectively. The SR rate = 0 for Ts < 0 • C Equations (8) and (9) were quite effective for use at all temperature (T) ranges in the two forest stands, whereas the SWC variables were more effective for the Q. mongolica forest growing season (Figures 18 and 19). Figures 20 and 21 show scatter plots of the observed SRs vs. those predicted by the integrated regression equations using the Ts and SWC data. The seasonal SR rates predicted by Equations (8) and (9) simulated the values of the observed SRs quite well, with a slope of 1.02 (r 2 = 0.90) for Mt. Nam and a slope of 0.96 (r 2 = 0.89) for Mt. Jeombong. This indicates that including the Ts and SWC variables improved the optimal regression equations for the quantitative estimations of the SR rates in the forest stands.   (8) and (9) simulated the values of the observed SRs quite well, with a slope of 1.02 (r 2 = 0.90) for Mt. Nam and a slope of 0.96 (r 2 = 0.89) for Mt. Jeombong. This indicates that including the Ts and SWC variables improved the optimal regression equations for the quantitative estimations of the SR rates in the forest stands.   (8) and (9) simulated the values of the observed SRs quite well, with a slope of 1.02 (r 2 = 0.90) for Mt. Nam and a slope of 0.96 (r 2 = 0.89) for Mt. Jeombong. This indicates that including the Ts and SWC variables improved the optimal regression equations for the quantitative estimations of the SR rates in the forest stands.     Tables 1 and 2 show the seasonal and annual total SR rates estimated from the observed SRs in Figures 8 and 9, the optimal regression Equations (4) and (5) constructed using only soil temperature (Ts), and the integrated regression Equations (8) and (9) constructed from both the Ts and SWC data. For both Q. mongolica forest stands, the total annual SR rates obtained from the measured data were slightly lower than those predicted by the regression equations. The measured results may be underestimated due to 'missing days' (no field data due to various problems) at each forest stand site during the course of the experiment. The total annual SR rates estimated from the observed data were~1207.3 g C m −2 at Mt. Nam and~975.9 g C m −2 at Mt. Jeombong, whereas those predicted by the integrated regression equations using both Ts and SWC were~1339.4 and~1003.0 g C m −2 , respectively (Tables 1 and 2). The total annual total SR values estimated by Equations (4) and (5) that used only Ts and by Equations (8) and (9), which also included SWC, were~1.2 and~1.3 times lower for Mt. Jeombong than for Mt. Nam, respectively.

Discussion
Soil respiration rates in forested regions depend on changes in primary environmental factors such as Ta, Ts, and SWC [29,31,32,52] as well as other factors such as the soil C substrates, vegetation, and forest management practices [28,30,53]. Several studies have focused on investigating the effects of the Ta and Ts variables driving the SR rate in temperate and cool-temperate forest regions under the Asian monsoon climate [10,12,26,45]. In these studies, Ta and Ts typically accounted for more than 70% of the SR variability on the seasonal and annual time scales. Our results show that SR rates normally fluctuate in parallel with the seasonal changes in Ta and Ts throughout the year (Figures 6-9).
In particular, among all T variables, the Ts at the 5 cm soil depth showed the highest correlation with SR at both Q. mongolica forest stands (Figures 10 and 11). This was the most important environmental parameter, accounting for more than 84% of the seasonal SR variability (Figures 12 and 13). These results are in agreement with previous studies on cool-temperate Q. mongolica forest regions (e.g., [12,45]). At both experimental sites, the soil layers at the 5 cm depth were classified as topsoil (A-layer), with large amounts of organic matter and plant roots, which have much more soil C and nutrients than the layers below and abundant soil microbes (bacteria and fungi) and fauna. Fierer et al. [54] reported that high SR rate values for the topsoil layer are largely attributable to the rapid responses of soil microorganisms to changes in Ts. Wang et al. [55] showed that the rate of SR is especially affected by the replenishment and availability of soil substrates from the bulk organic C pool under favorable conditions of T and SWC. Eberwein et al. [56] also highlighted that the response of SR to N enrichment and changes in T is dependent on the C availability of soil substrates.
The Q 10 exponential function for the SR sensitivity to T has been widely used in SR studies in various forest ecosystems to provide an effective empirical equation for estimating annual SR rates and their T-dependence [25,26,39,54,57]. The annual Q 10 values derived from the Ta variables for our forest stands are similar to the 1.3 to 3.3 range reported by a literature review of SR studies on various forest soils [57]. The annual Q 10 values derived from the Ts variables increased with increasing soil depth, from 3.0 to 3.1 at Mt. Nam and from 3.7 to 4.6 at Mt. Jeombong (Figures 10 and 11). These increasing values with soil depth are within the 3.8 to 5.5 range reported by Mo et al. [26] for cool-temperate broad-leaved deciduous forests. Uchida et al. [58] suggested that the high Q 10 values in cool-temperate forests can be partly explained by the strong T-dependence of microbial respiration under low Ts conditions. In contrast, Pumpanen et al. [59] reported that the annual Q 10 value in a Scots pine forest during an extremely dry, hot year was almost half of that during a normal year.
Soil water content is known to be one of the environmental factors that influences SR, because it affects the respiratory processes of the living plant roots and soil microbial community [29,60,61]. A high SWC in soils can delay soil CO 2 diffusion [62], whereas a low SWC can restrain root and soil microbial respiration [63]. Thus, it is known that moderate SWC is the most effective condition for SR [38]. In our study, the temporal changes in the Ta and Ts variables described the seasonal variability in SR for both forest stands well, but the SWC values did not follow the seasonal SR trends (Figures 6-9). Liang et al. [10] also found no significant relationship between SR and SWC in temperate and cool-temperate forests under the influence of the Asian monsoon. Indeed, with the exception of during short-term drought and saturation periods, the effects of SWC on SR are rarely observed in the field and forest soils of these regions [26]. Accordingly, the SWC at our forest stand sites fluctuated little throughout the year, because the sites also have abundant precipitation during the rainy season, even frequent rainfall, and the typhoons of the monsoon climate regime [12]. However, for both forest stands, there was an episodic increase in SWC that coincided significantly with the peak SR rates during the summer from June to August (Figures 14-17). It seems that the rapid SR rate increase during the peak forest growing period is linked to the resulting moderate soil conditions by rainfall events and may be partly affected by the enhanced plant photosynthetic capacity and assimilation as well as by accelerated soil microbial activities [33,35,64].
The results of our study show that AOCCs with short-and long-term measurements can provide sequential continuous high-quality data on the SR rates associated with environmental parameters (Ta, Ts, SWC, rainfall, etc.), ensure the effects of the main parameters on SR, and improve empirical models for estimating seasonal and annual SR in dominant Mongolian oak forest stands in the cool-temperate regions of the Korean Peninsula. Our results suggest that the seasonal SR variation at both forest stands is strongly correlated with the 5-cm depth Ts throughout all seasons, while the abrupt fluctuations in SR during the forest growing season from June to August are closely correlated with SWC at the 15-cm depth. We found that the SR rate increased with increases in Ts above 0 • C and increased SWC (up to 29.9% at Mt. Nam and 29.5% at Mt Jeombong) enhanced the SR rate on a short time scale. Further increases in the SWC would probably lead to a plateau in the SR rate due to reductions in soil aeration and gas diffusivity.
Based on the integrated optimal regression equations (Equations (8) and (9)) that use the measured Ts and SWC values in the Q. mongolica forest stands, the total annual SR rates were estimated to be 1003.0 (Mt. Jeombong) to 1339.4 g C m −2 (Mt. Nam) and. These values are near or within the ranges reported for temperate mixed oak forests (610−1414 g C m −2 ) [57], mixed oak forests in southern Europe (511−2660 g C m −2 ) [65], deciduous forests in eastern North America (800−1207 g C m −2 ) [13], and the cool-temperate deciduous (Q. mongolica) forests of central Korea (1107−1246 g C m −2 ) [45]. These equations, with their two main parameters (Ts and SWC) based on field measurements, provide a useful empirical tool that improves the estimation and prediction of SR rate variability in response to environmental change in the Q. mongolica forests in the cool-temperate region of the Korean Peninsula.
In this study, the precipitation and vegetation density in the Q. mongolica forest stands at Mt. Nam were slightly higher than those at Mt. Jeombong, which agrees with previous studies in the same areas (e.g., [12,47]). There was no apparent difference in the SWC values between the Mt. Nam and Mt. Jeombong sites. However, the Ts and Ta variables exhibited higher variability in Mt. Nam compared to Mt. Jeombong, and their temporal fluctuations exerted a strong influence on the seasonal variation in SR. The total annual SR at Mt. Nam, which is artificially disturbed by urbanization, was higher than that at Mt. Jeombong, which is surrounded by a natural landscape. In fact, the urban forest at Mt. Nam is located at the center of metropolitan Seoul and has been more consistently affected by heat island effects, air pollution, and direct or indirect human disturbances than the well-preserved natural forest at Mt. Jeombong [47][48][49]. Urbanization can change ecosystem functions by altering the biogeochemical cycles and driving forces (e.g., temperature, precipitation, C dynamics) [66,67].
Further detailed studies are required to examine other SR controlling factors such as the vegetation phenology, photosynthate allocation, growth of plant roots, activities of soil microorganisms and fauna, soil physicochemical characteristics, vegetation composition, and disturbance intensity in various urban and natural forest stands in order to understand and predict potential changes to the C cycle and budget in forest ecosystems in the cool temperate region of the Korean Peninsula.