Potential effects of ultraviolet radiation reduction on tundra nitrous oxide and methane fluxes in maritime Antarctica

Stratospheric ozone has begun to recover in Antarctica since the implementation of the Montreal Protocol. However, the effects of ultraviolet (UV) radiation on tundra greenhouse gas fluxes are rarely reported for Polar Regions. In the present study, tundra N2O and CH4 fluxes were measured under the simulated reduction of UV radiation in maritime Antarctica over the last three-year summers. Significantly enhanced N2O and CH4 emissions occurred at tundra sites under the simulated reduction of UV radiation. Compared with the ambient normal UV level, a 20% reduction in UV radiation increased tundra emissions by an average of 8 μg N2O m−2 h−1 and 93 μg CH4 m−2 h−1, whereas a 50% reduction in UV radiation increased their emissions by an average of 17 μg N2O m−2 h−1 and 128 μg CH4 m−2 h−1. No statistically significant correlation (P > 0.05) was found between N2O and CH4 fluxes and soil temperature, soil moisture, total carbon, total nitrogen, NO3−-N and NH4+-N contents. Our results confirmed that UV radiation intensity is an important factor affecting tundra N2O and CH4 fluxes in maritime Antarctica. Exclusion of the effects of reduced UV radiation might underestimate their budgets in Polar Regions with the recovery of stratospheric ozone.

of vegetation litter in the Antarctic terrestrial ecosystem through the process of photodegradation [25][26][27] . In addition, they have the potential to affect the structure and function of Antarctic mosses, Ceratodon purpureus and Bryum subrotundifolium 28 and to influence indirectly the soil microbial populations and activities 26 . UV radiation is also a key regulator of vegetation morphology and genetic processes and is important in vegetation growth 27,29 . Vegetation growth and soil microbial activities are the main factors influencing plant respiration and N 2 O and CH 4 emissions from the tundra 28 . Sunlight could greatly affect N 2 O and CH 4 emissions from tundra ecosystem because of O 2 release via vegetation photosynthesis 30 . The UV-induced release of carbon from plant litter and soils might contribute to global warming 27 . Therefore, it is important to investigate the effects of UV intensity on tundra N 2 O and CH 4 fluxes and carbon and nitrogen cycles, in maritime Antarctica.
Currently, stratospheric ozone has recovered somewhat in Antarctica since the implementation of the Montreal Protocol in 1989 31 . The Antarctic ozone hole has shrunk by nearly 400,000 square miles since it was discovered around 30 years ago. The ozone layer in the Polar Regions is projected to recover to pre-1980 levels by 2048, thus less solar UV radiation will reach the earth's surface 32 . However, the effects of a reduction in UV radiation on N 2 O and CH 4 emissions to date have not been investigated in the Antarctic tundra. During the austral summers of 2011/2012, 2013/2014 and 2014/2015, we selected a tundra ecosystem in the maritime Antarctica as study area (Fig. 1) and for the first time, investigated tundra N 2 O and CH 4 fluxes under the conditions of simulated reduction in UV (UV-A and UV-B) radiation, to explore whether natural UV radiation reduction could stimulate tundra N 2 O and CH 4 emissions. This is an important attempt to increase the Antarctic GHGs data sets to reasonably evaluate the potential effects of UV radiation reduction on tundra N 2 O and CH 4 fluxes.

Results
UV radiation and environmental variables between experimental treatments. In the summer of 2011/2012, UV radiation intensity showed similar temporal variation patterns between the control site and the sites covered by 0.03 mm and 0.06 mm filter membranes (Fig. 2a). The use of filter membrane between experimental treatments significantly decreased (analysis of variance (ANOVA) and least significant difference (LSD) test, P < 0.05) the amount of UV radiation penetrating into the chamber (Table 1). Compared with the control tundra site, the UV-A and UV-B through the sites with 0.03 mm and 0.06 mm filter membrane decreased by 20% and 50%, respectively (Fig. 2b). The highest mean UV-A and UV-B intensity occurred at the control site (14.4 ± 2.1 mW cm −2 and 4.7 ± 0.3 mW cm −2 , respectively), followed by the site covered by 0.03 mm membrane (11.4 ± 1.6 mW cm −2 and 3.8 ± 0.3 mW cm −2 , respectively) and the lowest at the site covered by 0.06 mm membrane (7.1 ± 1.0 mW cm −2 and 2.4 ± 0.2 mW cm −2 , respectively). However, no significant differences (ANOVA and LSD test, P > 0.05) were found in terms of chamber temperatures (CTs) between the different treatment groups (Table 1) and the CTs showed similar temporal variation patterns at different tundra sites (Fig. 2c). Thus, the use of filter membranes between experimental treatments did not significantly alter chamber micrometeorological conditions, except for the UV intensity. Therefore, the filter membranes could be used to stimulate various UV intensities and explore the effects of UV radiation on tundra N 2 O and CH 4 fluxes in maritime Antarctica. In addition, soil environmental properties, including pH, soil moisture, soil total organic carbon (TOC) and total nitrogen (TN) were similar to each other among the sites: AW1, AW2 and AW3 in the western tundra; AE1, AE2 and AE3 in the eastern tundra on Ardley Island; and GW1, GW2 and GW3 in the upland tundra on Fildes Peninsula. Detailed information about the climatic conditions and soil physiochemical properties is given in Supplementary Figures S1 and Tables S1 and S2.  Chamber temperature (°C) 6.3-20.0 11.1 ± 1.5 6.6-20.7 11.8 ± 1.6 6.6-19.7 11.9 ± 1.5 Table 1. Comparisons of UV radiation intensity and chamber temperature from the tundra observation sites with different thickness of UV radiation filter membrane. Note: The use of filter membrane between experimental treatments significantly decreased (ANOVA and LSD test, P < 0.05) the UV (UV-A and UV-B) radiation into the chamber, no significant differences (ANOVA and LSD test, P > 0.05) were found in terms of chamber temperatures between different treatment groups. ) under 20% reduction in UV radiation and the lowest was at the control site AW1 (mean fluxes were close to the detection limit) (Fig. 3a,b,c). Similarly, in the eastern tundra on Ardley Island substantial N 2 O emissions (mean 29.5 ± 2.6 μg N 2 O m −2 h −1 ) were observed at site AE3 under 50% reduction in UV radiation ANOVA and LSD tests on the N 2 O emission rates from all three sites showed a significant difference (P < 0.05) among the sites with different UV-radiation treatments (Fig. 4). Relative to the controls, the 20% reduction in UV radiation increased the tundra N 2 O emissions by more than 5 μg N 2 O m −2 h −1 , reaching as high as 14 μg N 2 O m −2 h −1 . The 50% reduction in UV radiation increased tundra N 2 O emissions by more than 9 μg N 2 O m −2 h −1 , reaching as high as 27 μg N 2 O m −2 h −1 during the observation periods (Table 2). Therefore, UV radiation intensity had an important effect on the N 2 O fluxes in maritime Antarctic tundra. Tundra N 2 O fluxes showed no significant correlations (Pearson correlation test, P > 0.05) with total organic carbon, soil moisture, total nitrogen, 0 cm soil temperature, 5 cm soil temperature, 10 cm soil temperature and NO 3 − -N and NH 4 + -N contents when the data at all the tundra sites were combined (Table S3) (Fig. 5a,b). Relatively strong CH 4 uptake occurred at the control site AW1, with a mean flux of −11.4 ± 41.2 μg CH 4 m −2 h −1 . The flux at site AW2 under 20% reduction in UV radiation ranged between a weak sink and a weak source, with the mean of 122.4 ± 33.9 μg CH 4 m −2 h −1 . The CH 4 flux at site AW3 under 50% reduction in UV radiation ranged between a weak sink (as low as −66.9 μg CH 4 m −2 h −1 ) and a strong source (up to 594.4 μg CH 4 m −2 h −1 ), with the greatest mean CH 4 emission rate (157.7 ± 40.9 μg CH 4 m −2 h −1 ) among all the sites. Similarly, the upland tundra acted as stronger CH 4 sink at the control site GW1 (mean −102.4 ± 88.3 μg CH 4 m −2 h −1 with the maximum uptake of 520.1 μg CH 4 m −2 h −1 ) compared with site GW2 (mean −14.3 ± 58.9 μg CH 4 m −2 h −1 ) under 20% reduction in UV radiation, whereas tundra site GW3 under 50% reduction in UV radiation showed weak CH 4 emission (mean 42.5 ± 94.5 μg CH 4 m −2 h −1 ) in summer 2014/2015 (Fig. 5c). Therefore, the reduction of UV radiation decreased tundra CH 4 uptake rates over all three sites and could even convert the tundra from CH 4 sinks into net sources in maritime Antarctica.
There were significant differences (ANOVA and LSD test, P < 0.05) between the mean CH 4 fluxes under the different UV radiation intensities for all tundra sites (Fig. 4). Relative to the controls, the 20% reduction in UV intensity increased tundra CH 4 emissions by more than 77 μg CH 4 m −2 h −1 , reaching as high as 109 μg CH 4 m −2 h −1 . The 50% reduction in UV intensity increased tundra CH 4 emissions by more than 106 μg CH 4 m −2 h −1 , reaching as high as 150 μg CH 4 m −2 h −1 during the observation periods (Table 3). Therefore, UV radiation intensity had an impact on tundra CH 4 fluxes in maritime Antarctica. Except for 0 cm soil temperature, CH 4 fluxes showed no significant correlations (Pearson correlation analysis, P > 0.05) with total organic carbon, soil moisture, total nitrogen, 5 cm soil temperature, 10 cm soil temperature and NO 3 − -N and NH 4 + -N contents (Table S3), indicating that these environmental variables might not be the key factors affecting tundra CH 4 fluxes.

Discussion
In this study, no significant correlation (Pearson correlation analysis, P > 0.05) was found between tundra N 2 O fluxes and soil biogeochemical properties (Table S3). However, reduced UV radiation significantly (ANOVA and LSD test, P < 0.05) increased tundra N 2 O emissions in maritime Antarctica, confirming that the variability in UV radiation has an important effect on tundra N 2 O fluxes and a reduction in UV radiation might increase tundra vegetation N 2 O production. Some wetland plants can produce and release some N 2 O via the physiological reaction of plant tissues 33,34 . Generally nitrate reductase (NR), which is responsible for reducing nitrate into nitrite in some plants, plays a key role in the nitrogen metabolism pathway 26 . Furthermore, the reduction in UV radiation significantly stimulated the activities of NR and glutamine synthetase in plants 35,36 . In maritime Antarctica, tundra vegetation might also produce some N 2 O, which is probably related to the content of nitrate and the activity of    NR. Indeed, exposure to enhanced UV radiation caused a decrease in the growth rate of Deschampsia antarctica and the activities of NR in maritime Antarctica 26 . Therefore, the reduction in UV radiation might increase NR activity, thereby stimulating nitrate reduction and N 2 O formation in tundra vegetation, which would lead to an increase in N 2 O emissions from tundra vegetation. The increase in N 2 O emissions might also be caused by stimulation of tundra vegetation growth under reduced UV radiation. The response of tundra vegetation photosynthetic rates and vegetation-soil respiration rates to the change in light intensity was almost immediate in the static chambers 15,37 . Reduced UV radiation significantly increased photosynthesis, the leaf cross-section and the proportion of aerenchyma in most of wetland plants 34,36,38 . The growth of the two phanerogamic Antarctic plants, Deschampsia antarctica and Colobanthus quitensis, appeared to be affected by manipulated surface solar UV levels during the severe ozone depletion in field experiments 39 and leaf growth of Deschampsia antarctica decreased with elevated UV-B 40 . Plant growth affected the available nitrogen, soluble organic carbon and O 2 in the soil; and accelerated N 2 O production and release from the plant-soil system 14,38 . In addition, plants also serve as a conduit to transport the N 2 O produced in the soil to the atmosphere 14,36 . Therefore, the stimulation of tundra vegetation growth under reduced UV radiation might influence soil properties and further promote N 2 O emissions from the soil-vegetation system.
In addition, N 2 O is produced naturally through nitrification and denitrification by soil microorganisms 41 . Although UV radiation cannot penetrate into the soil below 5 mm, enhanced UV radiation may impose indirect effects on the dynamics of microbial communities, mainly via its direct influence on vegetation growth and physiological metabolism, which in turn reduces the absorption of available N and affects root secretion 42 . Many studies have shown that reduced UV radiation significantly increased total abundance and activities of bacteria, such as nitrifiers and denitrifiers, in the rhizosphere soil of wetland vegetation 26,34,43 . Therefore, reduced UV radiation might increase the activities of tundra soil microorganisms associated with the nitrogen cycle in maritime Antarctica.
Similarly, the lack of a significant correlation (Pearson correlation analysis, P > 0.05) between tundra CH 4 fluxes and soil properties (Table S3) indicated that soil temperature, soil moisture and other soil properties had an insignificant effect on tundra CH 4 fluxes. In this study, the reduction of UV intensity could significantly (ANOVA and LSD test, P < 0.05) increase tundra CH 4 emission in maritime Antarctica, which was very similar to that observed at peatland sites in Finland 44 . Direct effects of UV radiation on CH 4 producing or oxidizing bacteria were not likely because solar radiation penetrates only a few centimeters into the ground 45,46 . However, there are some indirect effects between UV radiation and CH 4 emission, because the reduction of UV radiation induced changes in root exudates, which indirectly affect CH 4 production in the soil 42,47 . Unlike higher plants, lichens and mosses in Antarctica lack a well-developed root system; therefore, most C/N organic material entering the extracellular pools in polar soils probably comes from root and microbial turnover 48,49 . Vegetation root exudates provide carbon and energy sources for the growth of methanogens, thus promoting CH 4 production in the tundra 26,47 . Intense UV radiation might decrease the distribution of carbohydrates into the roots of vegetation in the Antarctic summer, which was thought to be the major reason why enhanced UV radiation inhibited CH 4 emissions in wetlands 29,50 . UV radiation induced changes in the contents of soil root exudates and decreased UV radiation led to an increase of 15.8% in the rate of CH 4 emissions from the wetlands 36 . Therefore, decreased UV radiation stimulated the secretion of root exudates, which might be an important mechanism underlying the effect of UV radiation on CH 4 emissions from tundra wetland.
By contrast, in general, ground vegetation might exhibit morphological changes under different ultraviolet intensities 34,51 . Outdoor species may be sensitive to an increase in UV and decreased UV radiation significantly increased the leaf cross section and proportion of aerenchyma in most wetland plants 44,51 . In our study area, tundra vegetation, including short mosses and lichens, grow very close to the ground and some of them were even buried in the tundra soils 14 , therefore aerenchymatous tissue of tundra vegetation might have an important role in transporting CH 4 from the soil to the atmosphere. In this experiment, the increased cross-sectional area of the plant aerenchyma caused by the reduction of UV radiation is one possible explanation for the stimulated transport of CH 4 from the soil to the atmosphere. However, it remains unclear whether the stomatal functioning controls CH 4 transport through the mosses or lichens. If the UV induces changes in the stomatal conductance of tundra plants, as shown in several studies with higher plants 44,52,53 , it could alter CH 4 emission rates. Therefore, the reduction in UV radiation might stimulate CH 4 emission by affecting tundra vegetation development.  Table 3. Tundra CH 4 fluxes under different experimental treatments in the summers of 2013/2014 and 2014/2015. Note: The ultraviolet radiation through the control site was not affected, the solar UV radiation through the site with 0.03 mm polyester filter membrane decreased by 20% and through 0.06 mm decreased by 50%. Analysis of variance (ANOVA) and the Least Significant Difference (LSD) test on the CH 4 emission rates from all three sites showed a significant difference (P < 0.05) between the sites with different UV-radiation treatments. The tundra CH 4 was not observed in 2011/2012 summer.
In this study, atmospheric photochemical reactions in the chamber should also be considered. The UV-induced photolysis of N 2 O comprises approximately 90% of the global N 2 O sink 54 and it is very likely that the enhanced N 2 O emissions under lower UV intensity were caused by reduced photolysis of N 2 O. In addition, an important atmospheric sink for CH 4 is the reaction between OH and CH 4 55 and less OH might be generated when UV radiation is reduced in the chambers, thus the "apparent" CH 4 flux from the tundra sites might also be enhanced when the chambers are covered by the thicker filter membranes. More research is needed to test these hypotheses in the future. In general, our results indicated that a reduction of natural UV radiation significantly (ANOVA and LSD test, P < 0.05) increased tundra N 2 O and CH 4 emissions compared with the control under ambient UV levels (Tables 2 and 3). Solar UV radiation might have an important effect on N 2 O and CH 4 budgets in the maritime Antarctic tundra. Although strong solar UV radiation still exists in maritime Antarctica, recovery of stratospheric ozone has occurred since the implementation of the Montreal Protocol in 1989 and the amount of solar UV radiation reaching the earth's surface would be decreased 31,32 . The effects of UV radiation on tundra N 2 O and CH 4 fluxes and their budgets, should be evaluated in the Arctic and Antarctic regions. The exclusion of its effects might underestimate N 2 O and CH 4 budgets in the tundra ecosystem of Polar Regions. To assess the regional N 2 O and CH 4 budget precisely, long-term measurements of GHG fluxes should be designed in the Antarctic or Arctic tundra ecosystems to show effects of UV radiation intensities on N 2 O and CH 4 fluxes.

Methods
Study area and investigation sites. One research area was located on Ardley Island (62° 13′ S, 58° 56′W; an area of 2.0 × 1.5 km) (Fig. 1). This island is recognized by the Scientific Committee of Antarctic Research as an area of special scientific interest. The western region of this island is a costal lowland tundra marsh and the vegetation cover was around 95% 14 . The middle on this island is a non-level, hilly and relatively dry upland tundra, with vegetation coverage of 90-95% 14 . The middle upland and western lowland tundra are free of active penguin populations. The active penguin populations only concentrate in the east of this island 12 and tundra patches have formed in the marginal zones of penguin nesting sites and are almost totally (90-95%) covered by mosses, algae and lichens in the east 15 .
Another research area was situated on Fildes Peninsula (61° 51′−62° 15′S, 57° 30′−59° 00′W; an area of 30 km 2 ) in the southwestern area of King George Island (Fig. 1a,b). Communities of mosses and lichens represent the vegetation on this peninsula. An upland tundra was well-developed in the northwest of the Chinese Great Wall Station, at a distance of about 500 m from the station. The upland tundra was nearly dry, with an elevation of around 40 m a.s.l. The sampling ground was totally covered by mosses (Bryum Pseudotriquetrum and Bryum muelenbeckii) and lichens (Usnea sp.), with a depth of around 5-10 cm for the vegetation layer. Under the vegetation cover is an organic clay layer, with the depth of around 10-15 cm. A more detailed description about the study area was given by Zhu et al. 15 .
During the summers of 2011/2012, 2013/2014 and 2014/2015, three observation sites were set up in the western tundra marsh on Ardley Island, equipped with three chamber collars each. The chambers were covered by special polyester filter membranes (Mylar-D, 0.03-mm/0.06-mm thick; DuPont Co., Wilmington, DE, USA), which removed part of the UV-A and UV-B wavelengths and had no effect on other wavelengths of light 56 , to simulate the effect of natural UV-radiation reduction on tundra GHG fluxes: (1) the control site AW1 had transparent chambers; (2) site AW2 had transparent chambers covered by a 0.03-mm filter membrane; and (3) site AW3 had transparent chambers covered by a 0.06-mm filter membrane (Fig. 1c). In addition, during the summer of 2012/2013, three other observation sites were established in the eastern tundra of Ardley Island: (1) the control site AE1 had transparent chambers; (2) site AE2 had transparent chambers covered by a 0.03-mm filter membrane; and (3) site AE3 had transparent chambers covered by a 0.06-mm filter membrane (Fig. 1c). During summer 2014/2015, N 2 O and CH 4 fluxes were also measured at three observation sites in the upland tundra on the Fildes Peninsula: (1) the control site GW1 had transparent chambers; (2) site GW2 had transparent chambers covered by a 0.03-mm filter membrane; (3) site GW3 had transparent chambers covered by a 0.06-mm filter membrane (Fig. 1b). There were no differences in the dominant vegetation species and phytomass among the three sites in each study area 15 . These observation sites were characteristic of the typical surface and vegetation within the tundra ecosystems in maritime Antarctica.

UV radiation measurement.
To test whether the UV radiation polyester filter membrane with different thicknesses could decrease solar ultraviolet radiation, we used an UV radiation instrument (Photoelectric Instrument Factory, Beijing Normal University, Beijing, China) with UV radiation sensors and data loggers (model UV-II) to measure the UV intensity. The sensors, which were manually mounted under the chambers with different thickness polyester filter membrane, collected UV data at 5-min intervals and the measured data displayed by the instrument was the radiant exposure (mW cm −2 ). The instrument was calibrated by the manufacturer and was used within the one-year interval of the validity for this calibration. The order of measurements was randomized to ensure that the measuring sequence did not bias the results and each site had three replicate measurements. During the period from Dec 24, 2011 to Feb 5, 2012, the UV radiation intensity was measured eight times at sites AW1, AW2 and AW3. These data indicated that the filter membrane significantly (ANOVA and LSD test, P < 0.05) decreased the UV radiation transmitted to the chamber (Fig. 2a). The UV radiation through site AW1 plots was not affected, the UV-A and UV-B decreased by 20% through the site AW2 plots and by 50% through the AW3 plots (Fig. 2b). were immediately transferred to 17.8 ml glass vials, which had been evacuated in advance 14,15 . More information on the in situ N 2 O and CH 4

Analysis of N 2 O and CH 4 concentrations and calculation of flux. The methods of analyzing N 2 O
and CH 4 concentrations and flux calculation were described in detail in our previous papers 12, 15 . In brief, gas samples were analyzed using gas chromatography (GC-HP5890 II, USA; Shimadzu GC-14B, Japan; Shimadzu GC-12A, Japan) to measure N 2 O and CH 4 concentrations. Their emission fluxes were calculated by fitting the experimental data to a linear least squares plot (N 2 O and CH 4 concentrations vs. time). More information is given in Supplementary Materials S2.
Measurements of environmental variables and soil properties. Soil temperatures (ST 0 , ST 5 and ST 10 ) were measured in situ using a ground thermometer inserted into the corresponding depth at the sampling sites. Meteorological data, e.g. air temperature (AT), daily sunlight time (ST), precipitation and total daily radiation (TDR) were acquired at the weather station of Great Wall Station. Soil samples were collected in the chamber plots after the fieldwork was completed in the summers of 2011/2012 and 2014/2015. The soils were sampled using a PVC tube (height: 15 cm; diameter: 6 cm), which was sealed and stored at 4 °C until analysis. Soil moisture was determined by oven drying at 105 °C to a constant weight. Each soil sample was homogenized manually and a subsample (fresh weight: 10 g) was extracted with 100 mL of 1 M KCl for 1 h and then filtered and analyzed for NH 4 + -N and NO 3 − -N, which were determined using a colorimetric method based on Berthelot's reaction and ion chromatography 14,15 . The TOC content in the soils was determined by the chemical volumetric method 12 and and TN was analyzed using automatic elemental analysis (Elementar Vario EL, Hanau, Germany). The pH was determined after a 1:3 (soil:solution) dilution of soil with distilled water 15 .
Statistical analysis. The standard error (SE) was used to estimate the uncertainty of the mean of individual fluxes. All the data for N 2 O and CH 4 fluxes were expressed as the mean ± SE. Differences in N 2 O fluxes or CH 4 fluxes under different UV radiation intensities were examined using one-way repeated ANOVA and LSD tests at the P = 0.05 level. The relationships between soil parameters and N 2 O and CH 4 fluxes were addressed using Pearson correlation analysis (P = 0.05 level). The contribution of the reduction in UV radiation to tundra N 2 O or CH 4 fluxes was calculated as: CT 0.03 = MF 0.03 -MF con and CT 0.06 = MF 0.06 -MF con . CT 0.03 and CT 0.06 indicate the contribution of the 20% and 50% reduction in UV radiation to tundra N 2 O or CH 4 fluxes, respectively. MF 0.03 , MF 0.06 and MF con indicate the mean N 2 O or CH 4 fluxes under the 20% and 50% reduction in UV radiation and under the control at the ambient UV level, respectively. All statistical analyses were performed using SPSS 20.0 (http://www.spss.com.cn/) and Microsoft Excel 2016 (https://products.office.com/zh-cn/excel) for Windows 10.