Soil Moisture but Not Warming Dominates Nitrous Oxide Emissions During Freeze–Thaw Cycles in a Qinghai–Tibetan Plateau Alpine Meadow With Discontinuous Permafrost

Large quantities of organic matter are stored in frozen soils (permafrost) within the Qinghai–Tibetan Plateau (QTP). The most of QTP regions in particular have experienced significant warming and wetting over the past 50 years, and this warming trend is projected to intensify in the future. Such climate change will likely alter the soil freeze–thaw pattern in permafrost active layer and toward significant greenhouse gas nitrous oxide (N2O) release. However, the interaction effect of warming and altered soil moisture on N2O emission during freezing and thawing is unclear. Here, we used simulation experiments to test how changes in N2O flux relate to different thawing temperatures (T5–5°C, T10–10°C, and T20–20°C) and soil volumetric water contents (VWCs, W15–15%, W30–30%, and W45–45%) under 165 F–T cycles in topsoil (0–20 cm) of an alpine meadow with discontinuous permafrost in the QTP. First, in contrast to the prevailing view, soil moisture but not thawing temperature dominated the large N2O pulses during F–T events. The maximum emissions, 1,123.16–5,849.54 μg m–2 h–1, appeared in the range of soil VWC from 17% to 38%. However, the mean N2O fluxes had no significant difference between different thawing temperatures when soil was dry or waterlogged. Second, in medium soil moisture, low thawing temperature is more able to promote soil N2O emission than high temperature. For example, the peak value (5,849.54 μg m–2 h–1) and cumulative emissions (366.6 mg m–2) of W30T5 treatment were five times and two to four times higher than W30T10 and W30T20, respectively. Third, during long-term freeze–thaw cycles, the patterns of cumulative N2O emissions were related to soil moisture. treatments; on the contrary, the cumulative emissions of W45 treatments slowly increased until more than 80 cycles. Finally, long-term freeze–thaw cycles could improve nitrogen availability, prolong N2O release time, and increase N2O cumulative emission in permafrost active layer. Particularly, the high emission was concentrated in the first 27 and 48 cycles in W15 and W30, respectively. Overall, our study highlighted that large emissions of N2O in F–T events tend to occur in medium moisture soil at lower thawing temperature; the increased number of F–T cycles may enhance N2O emission and nitrogen mineralization in permafrost active layer.

Large quantities of organic matter are stored in frozen soils (permafrost) within the Qinghai-Tibetan Plateau (QTP). The most of QTP regions in particular have experienced significant warming and wetting over the past 50 years, and this warming trend is projected to intensify in the future. Such climate change will likely alter the soil freezethaw pattern in permafrost active layer and toward significant greenhouse gas nitrous oxide (N 2 O) release. However, the interaction effect of warming and altered soil moisture on N 2 O emission during freezing and thawing is unclear. Here, we used simulation experiments to test how changes in N 2 O flux relate to different thawing temperatures (T 5 -5 • C, T 10 -10 • C, and T 20 -20 • C) and soil volumetric water contents (VWCs, W 15 -15%, W 30 -30%, and W 45 -45%) under 165 F-T cycles in topsoil (0-20 cm) of an alpine meadow with discontinuous permafrost in the QTP. First, in contrast to the prevailing view, soil moisture but not thawing temperature dominated the large N 2 O pulses during F-T events. The maximum emissions, 1,123. 849.54 µg m −2 h −1 , appeared in the range of soil VWC from 17% to 38%. However, the mean N 2 O fluxes had no significant difference between different thawing temperatures when soil was dry or waterlogged. Second, in medium soil moisture, low thawing temperature is more able to promote soil N 2 O emission than high temperature. For example, the peak value (5,849.54 µg m −2 h −1 ) and cumulative emissions (366.6 mg m −2 ) of W 30 T 5 treatment were five times and two to four times higher than W 30 T 10 and W 30 T 20 , respectively. Third, during long-term freeze-thaw cycles, the patterns of cumulative N 2 O emissions were related to soil moisture. treatments; on the contrary, the cumulative emissions of W 45 treatments slowly increased until more than 80 cycles. Finally, long-term freezethaw cycles could improve nitrogen availability, prolong N 2 O release time, and increase N 2 O cumulative emission in permafrost active layer. Particularly, the high emission was

INTRODUCTION
The global permafrost and seasonally frozen ground cover about 70% of all terrestrial ecosystems (Lawrence et al., 2012). In cold regions, soil freeze-thaw (F-T) events are a key natural process driving soil aggregate fragmentation (Chai et al., 2014), organic matter activization , root death (Kreyling et al., 2012), changes in microbial community structure and function Mao et al., 2019), and available nitrogen (N) transformation (Jiang et al., 2020;Mao et al., 2020). These process changes in a warmer world are further intensifying nitrous oxide (N 2 O) release from permafrost and seasonally frozen zones (Henry, 2008;Brooks et al., 2011;Risk et al., 2013;Chen et al., 2018;Lv et al., 2020).
Global warming is profoundly altering soil F-T patterns by increasing the thickness of the permafrost active layer, reducing days of the frozen period, and increasing days of the thawing period (Henry, 2008;Elberling et al., 2013;Schuur et al., 2015;Lin et al., 2017). Moreover, the decreasing precipitation in winter (Gaëlle et al., 2011) would cause the topsoil to undergo more frequent F-T cycles because of a lack of insulation from snow cover (Wipf et al., 2015) and, consequently, accelerate soil organic matter decomposition (Bracho et al., 2016). All these changes would provide favorable microenvironments and sufficient feedstock for excessive use of N by soil microbial nitrification-denitrification (Teepe et al., 2001). However, the magnitude of F-T effects on the key process of N transformation, particularly N 2 O emissions, in high latitude and/or high-altitude regions remains highly uncertain. N 2 O is a product of nitrification and denitrification, and a potent greenhouse gas with about 265 times more warming potential than CO 2 over a 100-year period (IPCC, 2013). During recent F-T events, significant soil N 2 O emissions have been reported in different types of cold ecosystems (Wu et al., 2020;Li et al., 2021). We synthesized some existing studies and found that the amount of N 2 O released ranged from 0.7 to 27.2 kg N ha −1 during the nongrowing season (October-April), accounting for 10-80% of their total annual ecosystem soil emissions . Thus, in cold regions, the large N 2 O emissions during F-T cycles are not only a crucial part of the soil N pool loss but also a non-negligible global greenhouse gas source, with CO 2 release from widespread permafrost thawing (Wang et al., 2014;Schuur et al., 2015;Song et al., 2015;Lv et al., 2020).
Unfortunately, despite being an indicator of global climate change, N 2 O emissions during the F-T period from alpine grasslands in the Qinghai-Tibetan Plateau (QTP) are rarely reported. Moreover, the permafrost soil environment in the QTP has become more complicated under climate change. The overall warming rate of the QTP ranges from 0.16 to 0.67 • C decade −1 , and the temperature increase in winter (0.45 • C 10 a −1 ) is nearly twice that in summer (Qin, 2014). An in situ warming experiment showed that the thickness of the seasonally frozen ground would decrease by 14.8% when the temperature of surface soil in the alpine meadow increases 2.03-2.3 • C, the frozen period would decrease by 44-83 days, and the number of day-night F-T cycle days in the topsoil during spring would increase by 37-44 days (Lin et al., 2017). The annual precipitation is also increasing in most areas of the QTP, while some subregions are becoming drier (Kuang and Jiao, 2016). On account of the large specific heat capacity of water, the slight fluctuation of soil moisture will significantly change the process of soil heat exchange during the F-T process . Thus, it is worth studying the interaction effect of warming and altered soil moisture on N 2 O emission during freezing and thawing.
Most single-factor studies suggest that large soil N 2 O emissions are significantly affected by thawing temperature (Yao et al., 2010), and the emissions occur over a few F-T cycles (Teepe et al., 2001;Bollmann and Conrad, 2004;Wu et al., 2019). However, our previous in situ study showed that soil temperature had no significant correlation with the N 2 O fluxes during spring thawing because the snow-cover thawing increased soil water content and led to the surface soil remaining at about 0 • C for a long time . The short pulses of N 2 O emission resulted from the combined impact of high soil moisture and flush available nitrogen (Yao et al., 2010;Lu et al., 2015a,b;Jiang et al., 2020;Wu et al., 2020). In view of the synchronous change in temperature and precipitation on the QTP, the first question we would like to explore was how N 2 O emissions respond to different moisture and thawing temperature gradations during the F-T process.
Our second aim was to determine whether soil N 2 O emissions in the QTP permafrost region are closely related to the number of F-T cycles. It is still difficult to accurately capture N 2 O emissions during soil freezing or thawing in the field without real-time online monitoring instruments. These conditions are unfavorable for assessing greenhouse gas emissions during nongrowing seasons in high latitudes or elevation regions. Thus, we aimed to create a model of N 2 O flux pattern under the coupled conditions of varying temperature and moisture gradients. Based on the above objectives, this study used long-term F-T cycle laboratory experiments to test QTP alpine meadow soil N 2 O dynamics and determine the relationship between N 2 O flux and soil F-T environmental factors. Ultimately, the results of this study would provide guidance for further discussion of the ecological effects of F-T on the soil N cycle in the QTP permafrost under climate change.

Site Description
Soil samples were collected from Daban Mountain (37 • 20 16.93 N, 101 • 23 47.21 E, 3,705 m a.s.l.), located in the eastern part of the Qilian Mountains in Qinghai Province. The area has a typical plateau continental climate, with a long cold winter and short warm summer. The annual average temperature was −1.6 • C, with the maximum monthly mean temperature in July (10.1 • C) and the minimum monthly mean temperature in January (−15.0 • C). The historical extreme maximum and minimum temperatures were 26.8 and −37.1 • C, respectively. The number of days with a daily minimum temperature below 0 • C during the year was as high as 280 days. The annual precipitation was 560 mm, on average, of which 85% was concentrated in May-September. The regional annual mean evaporation was 1,238.0 mm (Li et al., 2004;Dai et al., 2019). The dominant species is Kobresia pygmaea, and it is associated with Saussurea, Potentilla, Leontopodium, Gentian, Saxifraga, Poa, Oxytropis, and Polygonum. The average height of vegetation was < 10 cm, and the aboveground dry biomass averages 210 g m −2 .
Soil samples were collected in a transition zone of seasonal permafrost and spot (island) permafrost, and the soil was classified as Inceptisol or Cambisol. The topsoil (0-10 cm) organic matter, bulk density, pH, and soil volume water content in K. humilis meadow are 138.52 ± 13.82 g kg −1 , 0.75 ± 0.05 g cm −3 , 7.50 ± 0.22, and 32.7% ± 5.17%, respectively . The site froze from late October to mid-November. A stable thin permafrost layer began to form in late November, and the thickness of the permafrost continued to increase and reached the maximum freezing depth of 0.5-1 m in mid-February of the following year. The soil began to enter the thawing period in early March, and the thaw depth continued to increase until thawing was complete by late April (Lin et al., 2017;Dai et al., 2019).

Experimental Design
We used soil VWC to express the level of soil moisture. Three VWC values (W 15 -15%, W 30 -30%, and W 45 -45%) and three thawing temperature (T 5 -5 • C, T 10 -10 • C, and T 20 -20 • C) levels were applied in a randomized complete block design, for nine total treatments and three replications for each treatment. The topsoil of the study site usually begins to thaw from early-March to mid-April (Lin et al., 2017;Dai et al., 2019), and during this time, the day air temperatures fluctuate between 0 • C and 13 • C. Therefore, the 5 • C of thawing temperature in our study was used as the control temperature. The 10 • C of thawing temperature could represent the century-scale warming trend (2-7 • C) of the QTP (Qin, 2014). The 20 • C means extreme high temperature. According to our pre-experiment, the maximum volumetric water content was about 45%, and field study showed that the soil moisture in the growing season (July-August) were maintained at about 25%-35% VWC. Thus, we take the 30% VWC as the medium soil water. The difference in moisture gradient was controlled by 15% VWC. The existing precipitation control test also set treatments with natural and 50% reduction or increase in rainfall in an alpine grassland on the northeastern Tibetan Plateau (Xu et al., 2017).
In early July 2019, we set three 1-m × 1-m plots in the experiment site, then cut off the aboveground biomass, collected a sufficient amount of top soil (0-20 cm) with a drill (diameter 10 cm), and brought soil back to the laboratory the same day. After removing dead grass litter, plant roots, and gravel (grain diameter > 2 mm), 500 g of soil was weighed and transferred to a 1-L glass jar; then, the bottle was shaken to ensure that the soil contacted the bottle. The height of the soil in the bottle was 8-10 cm, and the soil surface was 11-13 cm away from the bottle plug. The plastic plugs of each bottle had three holes, two of which were connected to a rubber tube and a three-way valve. One of the three-way valves was linked with sampling needle tubing (10 ml), and the other valve was connected with the needle tubing (100 ml) to balance the bottle pressure. The third hole was used to hold the probe thermometer for recording the air temperature in each bottle. The interfaces of the holes were treated with latex to ensure airtightness. A time-domain reflectometer was used to measure soil moisture, which was adjusted in every treatment to the required VWC by adding ultrapure water. Each bottle was measured every 5 days during the whole test, and according to the change in mass, ultrapure water was added to keep the soil moisture stable.
Because the daily maximum temperature in the study area was nearly 15 • C when we collected the samples, in order to make the test soil microbial community as consistent as possible with the early cold season, all samples were successively incubated at 15 • C, 10 • C, and 5 • C for 1 week. After the cold acclimation, samples were put in three different incubators, and day and night F-T treatment was initiated. The freezing temperature was set as −20 • C from 20:00 to 8:00 the next day, and thawing temperatures were set as 5 • C, 10 • C, and 20 • C from 08:00 to 20:00. On the day of sampling, the freezing time were reduced by 90 min (21:30 to 08:00 the next day). The F-T cycles lasted about 165 days.

N 2 O Flux Measurements
Gas collection and measurement were similar to the static chamber-gas chromatography method. Gasses were sampled once a day during the first 7 days, then once every 3 days from the 8th to 22nd day, once every 5 days from the 23rd to 48th day, and once every 7 days from the 49th to 165th days. The gas collection began at 20:00, and after the bottle was closed, a 10ml gas sample was taken from the bottle with a plastic syringe at 0, 30, 60, and 90 min. Every time before the gas sample was drawn, the syringe piston was quickly pushed and pulled twice to stir and mix the air in the bottle. The three-way valves of the air intake channel were kept closed between the sampling intervals. We recorded the air temperature inside and outside of the bottle at every sampling, which were used to correct the N 2 O flux. The soil surface temperatures of each treatment were measured with an infrared detector.
The N 2 O concentration of the gas sample was analyzed within 24 h using gas chromatography (Agilent 7890A, Agilent Technologies, Sta Clara, CA, United States). The N 2 O flux was calculated according to the following equation: where F (µg m −2 h −1 ) is the N 2 O flux; dc/dt is the slope of the linear regression for the N 2 O concentration gradient as a function of time; M (g mol −1 ) is the molecular mass of N 2 O; P (Pa) is the atmospheric pressure; T (K) is the absolute temperature during sampling; V 0 (L), T 0 (K), and P 0 (Pa) are the gas mole volume, absolute air temperature, and atmospheric pressure under standard condition, respectively; and H (m) is the height of the soil surface to the bottle plug .

Data Analysis
All data were assessed for normality of variance before analysis. First, the effects of VWC, temperature, number of F-T cycles, and their interactions on mean N 2 O flux were investigated using general linear-model multivariate analysis. In the model, the significant differences in the mean N 2 O fluxes between the treatments based on a least significant difference (LSD) test (IBM SPSS Statistics 20.0, SPSS Inc., Chicago, IL, United States). Second, a quarter of the measured N 2 O flux was used to represent the daily mean emission rate, and the cumulative emissions were estimated using a linear interpolation method. Finally, linear, polynomial, exponential, and peak functions were used for fitting N 2 O flux and soil moisture and temperature in OriginPro 2020b (OriginLab Corp., Northampton, MA, United States). The models with the highest fitting degree were selected to represent the regression relationship between N 2 O flux and other environmental factors. All significant differences were at 0.05 (α).

N 2 O Flux Pattern in Freeze-Thaw Cycles
The N 2 O emissions of W 30 treatment were higher than W 15 and W 45 , and its patterns were similar even at different thawing temperatures. Figures 1A-1 Variance analysis also indicated that at the same thawing temperature, the average fluxes of W 30 treatment were significantly higher than those of W 15 , and the W 45 treatment had the lowest fluxes (Figures 1A-2,B-2,C-2). At the same moisture, the average fluxes were not significantly different between different thawing temperatures in the W 15 and W 45 groups, but the mean flux of W 30 T 5 (747.02 µg m −2 h −1 ) was significantly higher than that of W 30 T 20 (363.99 µg m −2 h −1 ), and W 30 T 10 (220.41 µg m −2 h −1 ) was the lowest.

The Pattern of N 2 O Cumulative Emissions During Long-Term Freeze-Thaw Cycles
The three soil moisture levels showed three cumulative emission patterns. The total cumulative emissions of the W 30 group were the highest (102.04-366.6 mg m −2 ), followed by the W 15 group (24.96-34.18 mg m −2 ) and the W 45 group (0.81-2.96 mg·m −2 ). In the W 15 group, the N 2 O cumulative emissions continued increasing over all F-T cycles (average accumulation rate, 0.19 mg m −2 day −1 ) (Figures 2A-C). The accumulations of the W 30 group were mainly reflected at the beginning of F-T cycles (average accumulation rate, 8.3 mg m −2 day −1 ) and increased minimally after 60 cycles (Figures 2D-F). However, N 2 O fluxes of the W 45 group were low in early F-T cycles, only increasing after 80 cycles, primarily in W 45 (Figures 2G-I).

Dynamics of the Ammonium and Nitrate Contents
During the early freeze-thaw cycle, ammonium increased significantly in all treatments, and the peak value range from 57.42 to 188.80 mg kg −1 (Figures 3A-1,B-1,C-1). The increase in soil ammonium contents coincided with N 2 O large emission in W 15 and W 30 . Although the peak ammonium contents in the water-saturated soil were higher than other treatments at the early freeze-thaw stage, there was no significant N 2 O emissions. The nitrate content increased sharply in W 30 T 5 during early freeze-thaw cycle, and the peak value of nitrate content was nearly 24.11 mg kg −1 (Figure 3A-2). Meanwhile, only low thawing temperature (T 5 ) improved nitrate accumulation after long freeze-thaw cycles (Figures 3A-2,B-2,C-2).

Relationship Between N 2 O Flux and Freeze-Thaw Environmental Factors
The results showed that soil moisture, thawing temperature, and the number of F-T cycles all had significant impacts on N 2 O flux, and their interactions were significant (P < 0.01) ( Table 1). Figure 4 shows that the flux response to soil moisture during F-T cycles conformed to the Gauss Amp function at the three different thawing temperatures (P < 0.01). In the range of VWC 15-45%, N 2 O flux variations were "unimodal." The peak emission values of T 5 , T 10 , and T 20 treatments were at VWC 27%, 25%, and 30%, respectively, and the 95% confidence intervals were, respectively, 22-32%, 17-32%, and 22-38% (Figure 4). These results highlighted that the medium level of soil water was essential in controlling large N 2 O emissions during F-T cycles.
There were no correlations between the N 2 O flux and thawing temperature in the W 15 and W 45 groups (Figures 5A,C). The response of flux to temperature showed a polynomial relationship (P = 0.02) in the W 30 treatment, and it increased with the temperature increase in the range of 0 • C to 5 • C, then decreased in the range of 5-10 • C and increased again from 10 • C to 20 • C ( Figure 5B). This suggested that the N 2 O flux was mostly   influenced by a medium moisture level in soil, and the large N 2 O emissions during F-T cycles had complicated nonlinear relationships with the thawing temperature.

Model of N 2 O Flux During Short-Term Freeze-Thaw Cycles
Based on the relationship between soil N 2 O dynamics and soil water and thawing temperature in the first 48 F-T cycles (large emission period), we drew the diagram of N 2 O flux model (Figure 6). We speculated that the medium moisture controlled the large production and release of N 2 O over F-T cycles. Because of appropriate soil, porosity tends to result in abundant aerobic and anaerobic microenvironments. These microzones were especially common at low thawing temperatures when the ice crystals in deep soil do not melt completely and coexist with soil air and free water, simultaneously promoting nitrification and denitrification. However, under dry conditions, the vast aerobic environment was beneficial to nitrification, and some N 2 O was released as the nitrification by-product. For the high soil water content, N 2 O production was dominated by denitrification, and the large intermediate products of N 2 O were completely converted to N 2 under an anaerobic environment. Moreover, under high moisture, the large interstices of deep soil were filled with water or ice, and the upper layer was covered with water. These physical barriers would block N 2 O emissions and greatly reduce N 2 O flux when frozen soil was thawing. The effect of thawing temperatures on the redox conditions of dry and waterlogged soil was limited, so the response of soil N 2 O flux to thawing temperature tended to be consistent in these two states.

The Effects of Abiotic and Nitrogen Transformation on N 2 O Flux During the Freeze-Thaw Cycles
We found that no matter how the thawing temperature changed, the N 2 O fluxes had significant peak functions with soil VWC, and the 95% confidence intervals ranged from 17 to 38% (VWC). These findings show that the relationship between soil moisture and N 2 O flux during F-T cycles was not linear, and a medium   level of moisture is more likely to promote N 2 O emissions. In contrast, when the soil was dry or waterlogged, N 2 O emissions were sharply reduced during F-T. Our previous in situ test of seasonal frozen arable soils in northeast China also indicated that soil moisture explained 32% of the N 2 O flux variation, and high soil moisture (30-55%) triggered the N 2 O burst during the spring thaw . Quick warming of the bare soil after snow melt resulted in rapidly decreased water content that returned to normal along with N 2 O flux . However, in our earlier research, the soil moisture (55%) at maximum emissions was higher than in this study (17%-38%), which may be related to the high water-holding capacity of arable black soil (mollisols) . Teepe et al. (2004) explained that N 2 O emissions during thawing decreased in the order 64% > 55% > 42% [water-filled pore space (WFPS)], but the flux at 76% WFPS was less than that at 55% WFPS. If our soil moisture were converted into WFPS, the peak values corresponding to the regression curve were 61% (T 5 ), 55% (T 10 ), and 66% (T 20 ). This means that the flux was positively correlated with the soil moisture when WFPS was lower than 55%, while it was negatively correlated with moisture when WFPS was higher than 66%. Thus, we assumed FIGURE 6 | Model of N 2 O flux response to soil moisture and temperature during short-term freeze-thaw cycles. The horizontal axis represents soil moisture, increasing from left to right. The diagram below the abscissa shows the soil environment under different soil water contents, in which the black lines represent plant roots, the gray irregular geometric shapes represent soil aggregates, the white gaps represent interstices of soil, and the blue parts represent soil free water. The photo of vegetation landscape from left to right is alpine steppe, alpine meadow, and alpine wetland. All the photos were taken by the first author in the Qinghai-Tibetan Plateau. that the soil moisture had a threshold value to determine the large N 2 O release during F-T. Other laboratory experiments also support our inference. For instance, Lu et al. (2015b,a) studied meadow-steppe grassland soil and found that N 2 O emissions in F-T processes are 30 µg kg −1 h −1 when soil moisture was about 8% VWC; however, the flux plunged to 8.4 µg kg −1 h −1 in waterlogged soil. If soil water is a crucial factor that controls N 2 O emissions during F-T, it is easy to understand why the current results of global N 2 O variation are so large (0.7-27.2 kg N ha −1 ) . If more informed studies of in situ data of N 2 O fluxes and soil moisture in the F-T period were integrated by meta-analysis, more meaningful and accurate conclusions may be obtained.
We suspected that the effects of soil moisture on N 2 O emissions during F-T cycles were largely related to nitrification and denitrification, both of which generate N 2 O and are regulated by soil O 2 partial pressure. Studies show that, in dry soil, the N 2 O emissions of nitrification increase as soil water content increases, and the release reaches a maximum when soil moisture is in the range of 45-55% maximum water holding capacity (WHC) (Bollmann and Conrad, 2004). N 2 O emissions of denitrification in moist soils increase as soil water content increases, and maximum N 2 O emissions were observed at 65-80% WHC (Bollmann and Conrad, 2004). Other studies found that denitrification is the major contributor of N 2 O emissions under F-T environments, accounting for 80-90% of total N 2 O emissions Öquist et al., 2004;Yanai et al., 2007). Soil moisture in these studies was mostly maintained at 60% WFPS, similar to our W 2 group. In waterlogged soil, the N 2 O fluxes maintained at a low level might be caused by an increased ratio of N 2 /N 2 O in the anaerobic environment  moisture. We also observed that the water level of the soil surface in W 45 increased about 0.5-1.0 mm compared with the initial value (Figure 7). Previous studies have indicated that unfrozen soil water would be transferred from deep soil to upper layers during freezing (Dai et al., 2019). Thus, the increased water layer enhances the physical barrier and impedes soil N 2 O outflow. It is inferred that when the soil water was below the threshold value, the positive correlation between N 2 O flux and moisture was due to the accumulation of N 2 O, as the by-product of nitrification and the product of nitrate are reduced by denitrifying bacteria (Ji et al., 2015). However, complete denitrification accelerates N 2 O consumption, and aqueous barriers alleviate N 2 O emissions when the moisture is high. The soil N 2 O emissions with moderate moisture under low thawing temperature (T 5 ) were much higher than in normal (T 10 ) and high temperature (T 20 ) treatments. Surprisingly, the peak flux of W 30 T 5 was as high as 5,849.54 µg·m −2 ·h −1 , which was the highest compared to all previous studies. Teepe et al. (2001) confirmed that frequent F-T cycles create abundant aerobic, anaerobic, or intermediate microzones, which are favorable for nitrification and denitrification. Consistent with Teepe et al. (2001), we observed that the frozen soil of W 30 T 10 and W 30 T 20 samples were thawed completely after 12 h, but W 30 T 5 only thawed about 1/4 volume. Although we did not have direct evidence of the soil porosity changes under different thawing temperatures, it is not inconceivable that the air, free water, and ice crystals coexist in semithawed soil (Figure 6). We speculate that under moderate moisture and low thawing temperature, F-T cycles allowed for unique redox conditions favorable to N 2 O production in nitrification and denitrification, and the semithawed state also provides large soil voids and ice openings for N 2 O to move to the atmosphere.
In this study, the N 2 O emissions of the W 15 and W 30 groups, respectively, peaked at the 3rd/4th cycles and 13th cycles, similar to previous studies (1st-9th cycles) (Teepe et al., 2001;Bollmann and Conrad, 2004;Wu et al., 2019). Nutrient availability is widely regarded as the main reason for the massive release of N 2 O during short F-T events (Risk et al., 2013;Congreves et al., 2018). This hypothesis holds that fragmented soil aggregates (Chai et al., 2014) and microbial mortality during F-T can release small molecular organic N (NH 4 + , NO 3 − , and amino acid) (Yanai et al., 2011;Yu et al., 2011). The abundant metabolic substrates stimulate microbial activities and accelerate N 2 O production. Our previous study also demonstrated that the available N explained 30% of the variation in N 2 O fluxes during spring thawing . Subsequent studies also confirmed that dissolved organic matter and inorganic N jointly regulate N 2 O flux from soils with different moistures during a freeze-thaw period (Wu et al., 2019).
Accordingly, it is easy to understand the sharp increase in emissions during the 1st-13th F-T cycles of W 15 and W 30 treatments. However, previous studies showed that available N increased rapidly only in the first to fourth F-T cycles, and then, the contents stabilized after multiple repeated cycles (Grogan et al., 2004;Yu et al., 2011). This means that the promotion of N 2 O emissions due to the increase in effective N over long-term F-T cycles is limited. However, the emission period of our study lasted until the 27th (W 15 ) and 48th (W 30 ) cycles. Moreover, the cumulative emissions of W 15 and W 45 were still increasing after 80 F-T cycles. Contrasting with the short-term effects of F-T on other seasonally frozen soil, frequent F-T disturbance may cause longer N transformation in alpine grasslands. Therefore, we suspected that the nutrient release in dry soil persisted during the long-term F-T cycles, and the N conversion pattern was different from that in soil with high moisture. The effect of F-T on the decomposition of soil organic matter in alpine grasslands under different water gradients needs further study.

The Model of N 2 O Flux During Freeze-Thaw Cycles Under Global Change
This model (Figure 6) can provide some guidance for predicting the response of N 2 O emissions in the nongrowing season under climate change, especially during spring thaw or snow melt. As our results showed, F-T disturbances drive N conversion and accelerate greenhouse gas (N 2 O) emissions. Therefore, exploring the response of the soil C and N cycles to F-T events under different ecosystems of hydrothermal coupling is also necessary to predict the impacts and feedback of climate change in cold regions. Under the current climate, Wang et al. (2013) found that N 2 O emissions of alpine meadow soil during spring thaw were 6.79 mg m −2 , accounting for 11% of emissions for the year. This in situ site is 30 km away from our sampling area. Although the value is much lower than our results, the warmer climate results in the topsoil undergoing many more F-T cycles and consequently prolongs the N 2 O emission period in spring thawing and increases the proportion of emissions during the nongrowing season.
The current research shows that the climate of the QTP has become warmer and wetter, and the increased precipitation is mainly concentrated in the growing season . This means that if the rainfall/snowfall increased in late September or early March, as our results suggested, this change would probably aggravate soil N 2 O emissions in dry ecosystems, such as meadow and steppe. The large gaseous N loss may further reduce N availability and even limit N supply during the wet season. However, the annual total evapotranspiration (522.28 mm) of this region was larger than precipitation (447.30 mm), and in winter, almost all snowfall was returned to the atmosphere directly via evapotranspiration. Moreover, the temperature increase in winter (0.45 • C 10 a −1 ) is nearly twice that in summer (Qin, 2014). Thus, topsoil would become drier in the nongrowing season. According to our prediction, dry soil would have reduced N 2 O loss during F-T compared with wet soil. N 2 O release from alpine grasslands may thus be a negative feedback to global warming. However, it cannot be ignored that the lower water levels of the alpine swamp in winter in the QTP possibly accelerated the succession of the swamp into a meadow or steppe ecosystem (Levy, 2008;Liu et al., 2018). This consequently exposed more peat soil and converted waterlogged soil to soil with a moderate moisture level.
It is estimated that the alpine meadow, steppe, and swamp in the QTP have approximately 133.52 Pg of organic C in the 0-0.75 m profile soils (Wang et al., 2002) and 16.08 Pg total N (Liu et al., 2012). Once the barrier of water and/or ice layer was lost in swamp soil, we detected that more than 1,000 times the normal level of N 2 O would be released during F-T cycles. The large N pool size together with significant F-T cycles suggests a high risk of N 2 O emissions and positive climate feedback across the alpine swamp. A past work has established that precipitation overrides warming in mediating soil N pools in an alpine grassland ecosystem on the QTP (Lin et al., 2016), but these studies are based on the growing season, and the mechanisms of F-T cycles on the soil N pools when the soil moisture and temperature change simultaneously in alpine ecosystems are still not clear.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
ZC, WW, and HZ: conception and design of study. PL and YZ: acquisition of data. SG, ZZ, YD, BY, and HX: analysis and/or interpretation of data. All authors contributed to the article and approved the submitted version.