δ13C-CH4 reveals CH4 variations over oceans from mid-latitudes to the Arctic

The biogeochemical cycles of CH4 over oceans are poorly understood, especially over the Arctic Ocean. Here we report atmospheric CH4 levels together with δ13C-CH4 from offshore China (31°N) to the central Arctic Ocean (up to 87°N) from July to September 2012. CH4 concentrations and δ13C-CH4 displayed temporal and spatial variation ranging from 1.65 to 2.63 ppm, and from −50.34% to −44.94% (mean value: −48.55 ± 0.84%), respectively. Changes in CH4 with latitude were linked to the decreasing input of enriched δ13C and chemical oxidation by both OH and Cl radicals as indicated by variation of δ13C. There were complex mixing sources outside and inside the Arctic Ocean. A keeling plot showed the dominant influence by hydrate gas in the Nordic Sea region, while the long range transport of wetland emissions were one of potentially important sources in the central Arctic Ocean. Experiments comparing sunlight and darkness indicate that microbes may also play an important role in regional variations.

Pacific Ocean and revealed three distinct latitudinal bands of δ 13 C-CH 4 14 . The latitudinal variations of CH 4 and δ 13 C-CH 4 are important for understanding the chemical and dynamic processes that control their distributions. Although there were reports on CH 4 distributions from 85°N to 67°S , the sources and influencing factors are still poorly understood, especially in the Arctic region 15 . Since 2007, the CH 4 concentrations had stabilized but increased again 16,17 . Systematic observations of CH 4 and δ 13 C-CH 4 over oceans remain limited. The present knowledge of atmospheric CH 4 is insufficient for describing all the variations affected by regional influencing. At high latitudes, methane is supersaturated in the surface waters of the Arctic Ocean 18,19 and European coastal areas 20 . Spatial and temporal observation of CH 4 is essential to identify and quantify the CH 4 sources. However, the direct atmospheric CH 4 concentration data and δ 13 C-CH 4 measurements are meagre over oceans from the mid-to high latitudes of the North Hemisphere, especially over the Arctic Ocean where the physical and chemical characteristics of the oceans waters have changed in response to climatic warming.
In this study, we describe new shipboard determinations of atmospheric CH 4 concentrations and δ 13 C-CH 4 measurements conducted from offshore China to the central Arctic Ocean, covering the latitudes and longitudes of 31.1°N-87.4°N and 22.8°W-90°E-166.4°W, during the 5 th Chinese National Arctic Research Expedition (CHINARE 2012). This study is the first to report the temporal and spatial distributions of atmospheric CH 4 concentrations, combined with δ 13 C-CH 4 measurements, over an extensive spatial scale. The δ 13 C-CH 4 measurements revealed factors that affect CH 4 variation.

Results
Trends of atmospheric CH 4 concentrations and δ 13 C-CH 4 . The spatial and latitudinal distributions of CH 4 concentrations determined during CHINARE 2012 are shown in Figs 1a and 2a, respectively. The CH 4 concentrations varied between 1.65 and 2.63 ppm. By the statistic analysis approximately 79% of the data ranged from 1.80ppm to 2.00ppm, with a median concentration of 1.88 ppm (mean: 1.88 ± 0.12 ppm), indicating that local episode influences were minimal (Fig. 2b). Based on evaluation using the Kolmogorov-Smirnov test, the CH 4 concentrations were revealed to be distributed inhomogeneously along the cruise track (p < 0.05), even when the four highest values were excluded. The distribution of CH 4 concentrations showed no obvious relationship with latitude outside the Arctic Ocean, which was consistent with observations from the South Atlantic 13 . However, the CH 4 concentrations in the Arctic Ocean (> 66.5°N) fluctuated in a more consistent manner, especially in the central Arctic Ocean (> 80°N), where concentrations increased with latitude (r = 0.44, p < 0.01).
Atmospheric δ 13 C-CH 4 varied from − 50.34% to − 44.94% with a median value of − 48.63% (meane: − 48.55 ± 0.84%) (Fig. 2c). The mean value was lower than the mean δ 13 C-CH 4 (− 47.44%) reported for the northern hemisphere 9 . The higher frequencies ranged from − 49.5% to − 48% (Fig. 2d). A Kolmogorov-Smirnov test indicated heterogeneous distribution of δ 13 C-CH 4 along the cruise track (p < 0.05). From mid-to high latitudes, the δ 13 C-CH 4 measurements showed a slight decreasing trend with latitude (r = − 0.23, p < 0.01). A similar decreasing trend with northern latitude was also observed between 65°S and 50°N 9 during the Pacific Ocean cruise.  Table 1). For the sample regions outside the Arctic Ocean, the mean (± SD) CH 4 leves in the OC, JS, SO and BS areas were 1.90 ± 0.05 ppm, 2.00 ± 0.21 ppm, 1.90 ± 0.05 ppm and   Although the averages over CS, CAO, and NS were similar, some relatively higher values were found in the CAO, potentially indicating the possible presence of a CH 4 source. The mean measurement of δ 13 C-CH 4 in the OC region was − 47.49 ± 1.24%, which was similar to the mean δ 13 C-CH 4 (− 47.44%) for the Northern Hemisphere 8  Temporal variations of atmospheric CH 4 concentrations and δ 13 C-CH 4 . The CH 4 concentrations in July and September for the same sampling regions are shown in Fig. 3a. Non-parametric tests indicated that the values in July and September were not significantly different. Figure 3b also shows no significant differences between the measurements of δ 13 C-CH 4 in July and September in the OC, JS, and BS areas. However, the δ 13 C-CH 4 values in NPO and CS regions were greater in July than in September.

Regional variations of atmospheric CH 4 concentrations and
The diurnal and nocturnal CH 4 concentrations and δ 13 C-CH 4 measurements for the same sampling regions are shown in Fig. 4. Non-parametric tests revealed no significant differences between day and hight CH 4 concentrations and δ 13 C-CH 4 measurements.

Discussion
Atmospheric CH 4 concentrations and δ 13 C-CH 4 over the oceans might be influenced by sources and sinks, e.g., long-range transport of anthropogenic emissions or natural sources emitted from the ocean, and by oxidation by Cl and OH radicals or microbes. Although oceanic CH 4 was not measured simultaneously in this study, environmental parameters were recorded for further analysis.
The role of oxidation. The phase of δ 13 C-CH 4 in the seasonal cycle is consistent with the kinetic isotope effect (KIE), which is due to OH and/or Cl radicals oxidizing δ 12 C-CH 4 faster than δ 13 C-CH 4 , resulting in atmospheric methane enriched in δ 13 C-CH 4 9 . From mid-to high latitudes, the δ 13 C-CH 4 showed a slight decreasing trend with latitude in sunlight. The latitudinal loss of δ 13 C-CH 4 may be due to decreasing enriched input of δ 13 C or chemical oxidation. Considering the potential fuel source influences, the variation of CO with latitude and the air masses of all the samples are shown in Figure  S1. The CO concentrations showed a decreasing trend up to about 50°N, indicating a decrease in local contribution by anthropogenic sources, such as fossil fuels, north of 50°N. The back-trajectories of the air masses further confirmed the influence of continental sources of CO in the OC and JS areas (< 50°N). However, the latitudinal decreasing trend of δ 13 C-CH 4 remained under background air, suggesting the potential role of oxidation. It has been reported that OH radicals in the troposphere are the primary sink for global atmospheric CH 4 21 . Cl radicals may also contribute to CH 4 loss over oceans. To determine the potential reactive process for CH 4 , the contributions of the OH and Cl radicals were calculated as follows. The average concentrations of the OH and Cl radicals in the marine boundary layer are about 7 × 10 5 -2.9 × 10 6 molecules·cm −3 and 1.8 × 10 4 molecules·cm −3 , respectively [22][23][24][25] ; mean CH 4 concentration is 1.88 ppm; and the rate constant at 8 °C based on the mean sampling temperature for OH and Cl radicals is 4.42 × 10 −15 and 7.59 × 10 −14 cm −3 ·molecules −1 ·S −1 , respectively 26 . Assuming the reactive height is 25 m, based on the sampling height, the CH 4 consumption for OH and Cl radicals is 8.72 × 10 −3 -3.61 × 10 −2 mg·m −2 ·d −1 and 3.85 × 10 −3 mg·m −2 ·d −1 , respectively. The effect of Cl radicals on CH 4 is similar to that of the OH radicals. It is unclear if the level of Cl radicals varies with latitude. However, OH radicals can decrease from low to high latitudes 23,27 . In addition, sunlight intensity at the high latitudes is lower than in the mid-latitudes. Thus, reduced oxidation in the high-latitude region might result in depleted δ 13 C-CH 4 . A similar principle could explain higher values of δ 13 C-CH 4 in July compared to September over the CS area. The significantly higher sunlight intensity over the CS area in July compared with September indicated stronger oxidation potential (Fig. 5). However, the variations between July and September over the OC, JS, BS and NPO regions were not consistent with oxidation results. Most regions outside the Arctic Ocean were influenced by continental sources, and complex sources inputs may influence the oxidation results ( Figure S1).
The role of sources and atmospheric transport. To determinthe potential role of sources or sinks, the variations of δ 13 C-CH 4 versus mixing ratio changes were calculated using the approach proposed by Allen et al. (2001) 25 . Assuming that the removal of δ 13 C-CH 4 is by OH radicals in a closed well-mixed box, the effective rate coefficient for δ 12 C and δ 13 C removal by OH radicals are referred to as k 12 and k 13 . We adopted the value k 13 /k 12 = 0.9946 28 , for which ∈ = k 13 /k 12 − 1 is defined as the KIE. The relationship between changes in δ 13 C-CH 4 and changes in mixing ratio can be expressed as Δδ ≈ ∈ ( + δ )Δ / C 1 0 C0, which relates the δ 13 C-CH 4 variations (Δ δ ) around the mean δ 13 C-CH 4 value (δ 0) to relative mixing ratio variations (Δ C/C0), where C0 is the mean mixing ratio over the cruise track. If we plot Δ δ versus Δ C/ C0, the KIE line of slope can be obtained. The details about the expression were presented in a previous report 25 . We can also obtain the variations of δ 13 C-CH 4 and mixing ratio changes in different regions (Fig. 6a). The values over the OC, JS, SO, NS, and BS areas were above the KIE line (OH oxidation line), representing enriched δ 13 C-CH 4 . Enriched δ 13 C-CH 4 may mean more oxidation by OH radicals. If the enriched δ 13 C-CH 4 was influenced by other forms of oxidation such as by Cl radicals, the CH 4 concentration should be lower. The higher values signified an anthropogenic source, especially for the samples over the OC and JS regions that were far from the KIE line, suggesting that anthropogenic sources might play a major role. The values from the CAO, CS, and NPO areas were below the KIE line and the depleted δ 13 C-CH 4 might indicate a natural source or less oxidation. If the lower values over the CAO, CS, and NPO areas were a reflection of reduced oxidation only, the concentrations would be higher and thus, the low values signify natural sources.
However, this box model only provided potential sources due to it requiring knowledge of how the out of the box values changing. We therefore applied Keeling plot approach to further investigate the regional variations. If the CH 4 is emitted into the atmosphere from a single source, the isotopic ratio of the source can be inferred as an end-member for the baseline on the Keeling plot 29,30 . It was reported that biogenic sources were dominant at Spitsbergen 29 . Figure 6b shows examples of δ 13 C-CH 4 plotted against the reciprocal of CH 4 for the regions over NS and CAO regions, both of them close to Spitsbergen, and the OC region influenced by anthropogenic sources. NS was divided into east NS (NSE) close to Spitsbergen and west NS (NSW) close to Iceland, based on the geographical location. The Keeling plot can be used to understand the processes controlling isotope discrimination and to estimate the isotopic ratio of a source 31 . The CH 4 collected in NSE gave a source with − 52.46% (r = 0.33, p < 0.05), which is similar to the observations at Zeppelin station during Arctic springtime 29 . The main source in the NSE was gas field emissions as few air masses over this were from known emission areas. In contrast, the δ 13 C-CH 4 of − 44.75% (r = 0.35, p < 0.05) signature in NSW indicated enriched δ 13 C inputs. Iceland is a geothermal country. Air masses close to Iceland with heavier δ 13 C may have contributed to the increment in NSW. However, in CAO we collected CH 4 data in a scale from 1.8 to 2.0 ppm, the highest frequency range. This indicated a source with − 62.45% (r = 0.50, p < 0.01) suggesting that the wetlands dominated. The isotope data was consistent with the Siberian railroad and the Ob river with − 62.9% 32 . The main air masses in CAO move across the west Siberian coast, also confirming the wetlands emission ( Figure S2). But there were some enriched and depleted sources inputs, indicating CAO was influenced by complex mixing sources. It has been demonstrated that extra sources may change seasonal variation 25 . One example outside the Arctic Ocean is the OC region which has complex mixing sources (Fig. 5b). Complex mixing sources may influence the results of seasonal variation outside the Arctic Ocean.
The role of microbes. The δ 13 C-CH 4 should be enriched in sunlight and depleted in darkness in the same region because of the higher photochemical oxidation rate in sunlight. However, we found no significant differences in the CH 4 concentrations and δ 13 C-CH 4 measurements between sunlight and darkness in this study. This might be due to microbes producing more CH 4 , which can be deduced based on the case study of CH 4 variation in sunlight and darkness over the central Arctic Ocean. The CH 4 fluxes on sea ice in sunlight and darkness are shown in Fig. 7. CH 4 fluxes on sea ice had positive (emission) or negative (absorption) values 19 . Methanogenic bacteria and methanotrophic bacteria can occur in cold marine waters and in sea ice 33,34 . Thus, the CH 4 emissions might come from the CH 4 in the water 18 and from CH 4 production by microorganisms in the sea ice 35 . During CHINARE 2010, we suggested that negative fluxes could be associated with both photochemical and biochemical oxidation 19 . However, photochemical oxidation cannot explain why lower CH 4 fluxes were observed in the dark than in sunlight. The negative fluxes could be attributed to the role of methanotrophs. Llight inhibits the growth and activity of methanotrophic bacteria 36 , which could result in the reduced loss of CH 4 in sunlight compared with darkness. Additionally, temperatures are higher in sunlight than in darkness and methanogenic bacteria can increase CH 4 production at higher temperature 37 . Archaeal populations of methanogenic and methanotrophic bacteria can be abundant in cold and temperate environments 37 . In temperate environments, the depleted δ 13 C-CH 4 produced by microbes in the sunlight might offset the sink of chemical oxidation.

Experimental Methods
Sampling gas. During CHINARE 2012 (July-September 2012) air samples were collected from the marine boundary layer using 17.5-ml vacuum vials (manufactured by the Institute of Japanese Agricultural Environment) and 0.5-l Tedlar gas bags to determine the CH 4 concentrations and values of δ 13 C-CH 4 , respectively. The samples in the gas vacuum vials were sealed with a butyl-rubber septum and a plastic cap, following the same sampling method used in the research on Antarctica 38 . The cruise covered the eight geographical areas shown in Fig. 1a. The sampling location was the fifth deck of the icebreaker Xuelong, which was about 25 m above sea level. To avoid contamination by ship emissions and anthropogenic factors the samples were collected upwind. The gas vacuum vial was equilibrated in the air for about 1 min using a two-way needle above the head. Samples were collected two or three times each day. The sampling times after 06:00 and 18:00 (local time) were considered as day (sunlight) and night (darkness), respectively. Ancillary data including sunlight intensity and CO concentrations analysed using an EC9830 monitor were also recorded along the cruise track 39 . CH 4 fluxes on sea ice in sunlight and darkness (simulating day and night) were determined using a static chamber technique less than 2 h at five short-term ice stations (sites shown in Fig. 1b). The procedure was based upon a previous report 14 . The inner diameter of the cylindrical chamber was 0.4 × 0.3 m. The open-bottomed acrylic resin chambers were placed on collars installed at the measurement sites. The use of the collars allowed the same spot to be measured repetitively, ensures that the chamber is well sealed, and minimizes site disturbance. One chamber allowed sunlight transmission and the other did not permit sunlight. Once the chamber was set up, the head-space samples were immediately transferred into the vacuum vial using a two-way needle 19 . The sampling procedures were repeated at 20-or 30-min intervals for about 2 h at each site. All collected samples were analysed in the laboratory of the Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China.

Determination of CH 4 concentrations and fluxes. An Agilent 7890 A gas chromatograph (GC)
with a flame ionization detector (FID) was used to determine the CH 4 concentrations. The GC-FID was equipped with an auto-injection system controlled by a computer program and a back-flushed system of 10-port valves. The chromatographic column was a 2-m stainless steel column filled with high-performance Molecular Sieve 13X. The column and detector temperatures were 85 and 250 °C, respectively. The flow rates of N 2 , H 2 , and air were 25, 60, and 380 ml/min, respectively. CH 4 standard gas at 10ppm was produced by the National Institute of Metrology, China (NIMC). The GC analysis and calibration were according to GB/T 8984-2008 (NIMC). The calibration scale had a range of 0.95~49.8 ppm. The GC instrument was calibrated using CH 4 standard gas every twelve samples. The variance coefficient (CV) for each measured sample and each time was < 1%. CH 4 fluxes were calculated using the following equation: P(CH 4   Determination of δ 13 C-CH 4 . The δ 13 C-CH 4 value was measured using a Thermo Finnigan Mat-253 Isotopic Mass Spectrometer. The Mat-253 mass spectrometer has a fully automated interface for the pre-GC and pre-concentration of trace gases. Full details of the method were described by Cao 41 and a brief description is given here. In this study, 100-ml gas samples were injected into vacuum glass bottles. If the gas sample was < 100 ml, inert gases without CH 4 was added to the bottle and the bottle was adjusted to normal pressure. The air-sampling bottles were installed into pre-concentration. After their thresholds were blown by He gas, the valves at either end of the sampling bottles were opened and the samples blown into the cold trap by He. At the temperature of − 196 °C, only the volatile components (N 2 , O 2 , Ar, and CH 4 ) can enter the 1000 °C burning furnace via the cold trap and an aluminous oxidative pipe filled with three 0.13-mm nickel wires. During the test period, CH 4 is oxidized into CO 2 and H 2 O. The CO 2 produced from the CH 4 combustion was collected by another cold trap and transported into a third cold trap. Then, CO 2 was passed into the GC for further separation. The calibrated standard CO 2 was injected into the ionic source three times continuously every 30  were accepted by cup2, cup3, and cup4, respectively. Adjusting the flow rate of the reference gas controlled the peak intensity of m/z 44 to within 2v-3v. The No. 2 peak was set as the standard sample peak. The CH 4 peak occurred at about 870 s and the ratio line was positive. According to the ratios of the No. 2 CO 2 peak and the sample peak, the δ 13 C PDB for the CO 2 from the CH 4 was obtained. The 2.02 μ l/l compressed CH 4 was from the same source. Then, 25 ml of compressed CH 4 was injected into a 100-ml glass bottle with an inert gas filling under normal pressure nine times. The standard deviation for δ 13 C-CH 4 in the compressed air was 0.196% based on the nine repeated measurements. The different CH 4 concentrations showed good relationships with ionic flows of m/z 44, and the correlation coefficient was 0.983. GBW04407 (carbon black, national standard substance produced by NIMC with − 23.73% for δ 13 C VPDB ) were used to calibrate the carbon isotope 40 . Isotope ratios were defined as δ 13 C = [(R sample /R standard )− 1] × 1000[%], where δ 13 C is the δ value of the carbon isotope and R is the ratio of the heavy isotope to the light isotope.