Geographic and seasonal variation of dissolved methane and aerobic methane oxidation in Alaskan lakes

Introduction Conclusions References Tables Figures


Introduction
Freshwater ecosystems are an important source of atmospheric CH 4 , responsible for 20 6-16 % of global emission to the atmosphere (Bastviken et al., 2011). Northern lakes are responsible for as much as 6 % of these global CH 4 emissions (Walter et al., 2007). Methane emission from aquatic ecosystems is significantly mitigated by CH 4 oxidation (MO) by aerobic methanotrophs, a group of gram-negative bacteria that use CH 4 as a carbon and energy source (Murrell et al., 1993;Trotsenko and Murrell, Several environmental factors directly affect aerobic MO in freshwater ecosystems. First, methanotrophy depends on the availability of both CH 4 and O 2 . Higher MO rates are usually found at the oxic/anoxic interface, where both CH 4 and O 2 are present (Utsumi et al., 1998a, b;Bastviken et al., 2002;Liikanen et al., 2002;Carini et al., 2005;Schubert et al., 2010). In turn, CH 4 and O 2 concentrations depend on numerous other 15 processes involved in biogeochemical carbon cycling (Fig. 1). Among these, the most important are methanogenesis producing CH 4 , primary production and atmospheric diffusion supplying O 2 , and several aerobic metabolic processes that compete with MO for available O 2 (Dzyuban, 2010).
In addition to autochthonous and allochthonous carbon inputs to lakes, permafrost 20 thaw can provide an additional source of labile organic carbon to fuel methanogenesis and carbon mineralization in thermokarst (thaw) lakes (Zimov et al., 1997;Walter et al., 2006). MO in northern regions is therefore directly and indirectly linked to permafrost type and landscape processes that are highly variable. Permafrost ranges from sporadic to continuous and is also variable in composition (Jorgenson et al., 2008). Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | non-yedoma permafrost and permafrost-free soils are typically thinner than those of yedoma deposits (Ping et al., 2008;Tarnocai et al., 2009;Hugelius et al., 2014). Many northern lakes are located in continental climate zones, subject to contrasting seasonal conditions with long, cold winters followed by relatively short, but warm summers. Although psychrotolerant and psychrophilic methanotrophs have been 5 reported (Omelchenko et al., 1993(Omelchenko et al., , 1996Bowman et al., 1997;Trotsenko and Khmelenina, 2002), MO occurs more efficiently at mesophilic temperatures, from 20 to 35 • C (Semrau et al., 2010). During winter, northern lakes are covered by a thick ice layer for seven to nine months. Surface lake ice impedes oxygen transfer from the atmosphere to the lake and, when snow-covered, substantially reduces light 10 penetration and oxygen production by photosynthesis (White et al., 2008;Clilverd et al., 2009). Thus, the combination of low temperature and limited oxygen availability suggests lower MO rates in northern lakes in winter than in summer. Given the number of parameters having a potential effect on MO, as well as the patchwork of seasonal and geographic conditions found among northern lakes, MO 15 is expected to exhibit large geographic and seasonal variations that still remain to be characterized. The goal of our study was to determine these variations through measurement of dissolved CH 4 and O 2 as well as MO rates in the winter and summer in 30 lakes along a south-north transect in Alaska.

Site description
We sampled 30 Alaskan lakes during two field campaigns, one in late winter (March-April 2011) and one in summer (June-July 2011). To evaluate the effects of latitudinal variation and permafrost type on MO, lakes were selected along a transect from the southcentral Alaskan coast on the Kenai Peninsula to the Arctic Ocean near Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | discontinuous, sporadic, or no permafrost in south-central Alaska; discontinuous to isolated yedoma permafrost in the interior of Alaska; and continuous permafrost in northern Alaska. In this work, for simplicity, lakes located in yedoma-type permafrost areas will be referred to as "yedoma lakes" and all others as "non-yedoma lakes". Geographic variability along the north-south Alaska transect has been previously 5 described for ecosystems, climate, geology, and permafrost type (Gregory-Eaves et al., 2000;Jorgenson et al., 2008;Smith et al., 2010). Additionally, Sepulveda-Jauregui et al. (2014) quantified the surface area of the selected lakes (0.002-1.45 km 2 ), their trophic states (ultraoligotrophic to eutrophic), and their annual CH 4 fluxes (0.5-317 g CH 4 m −2 yr −1 ). Table 1 shows the location and permafrost type of the selected 10 lakes.

Sampling and field measurements
We sampled lake water offshore and usually near the center of each lake. In the winter, the ice cover was drilled through with a motorized auger (0.3 m in diameter). Using a Hydrolab DataSonde (Hach Hydromet, Loveland, CO, USA), we measured 15 temperature, pH, chlorophyll a, and dissolved oxygen (DO). The Hydrolab was calibrated regularly, before and after each section of the latitudinal lake transect (four sections per transect, approximately one calibration per week). All parameters were measured at 0.5 or 1 m depth intervals throughout the water column, except in Dolly Varden L. where measurement intervals were increased to every 5 m from 15 to 20 25 m depth. In lakes shallower than 1 m, we measured Hydrolab parameters at three distributed depths throughout the lake water column.
Water samples for MO rates and dissolved CH 4 concentration were taken at a depth of within 1 m of the ice-water interface in winter and usually at 0.75-1 m water depth in summer. Due to differences in lake depths and thicknesses of the ice sheets, in some 25 lakes, lake-bottom water was sampled, while in other lakes near-surface or mid water column water was sampled. Samples were taken with a horizontal Van  Water density derived from surface and bottom water temperatures were used to determine the relative water column stability (RWCS; Padisak et al., 2003). Lakes with RWCS > 56.5 were considered fully stratified, lakes with RWCS < 16.3 were considered fully mixed, and lakes with intermediate RWCS were considered partially stratified (Branco et al., 2009). Similarly, we determined whether an oxycline was 5 present in each lake based on a sharp DO gradient or presence of an oxic/anoxic interface. Detailed temperature and DO profiles are available as a Supplement in Sepulveda-Jauregui et al. (2014). We report all results in mean ± standard deviation (SD). 10 To avoid long delays in sample transfer from remote locations to the laboratory, we used a previously described method for determination of dissolved CH 4 concentrations (HE-TDLAS; Sepulveda-Jauregui et al., 2012). This method consisted of determining the CH 4 concentration in the headspace of an equilibration vial containing a known volume of lake water and in which gas/liquid equilibrium has been reached by 10 s 15 of vigorous shaking. The CH 4 concentration in the headspace was determined using a laser beam crossing the headspace of the equilibration vial. This measurement was conducted with a modified open-field CH 4 analyzer (GasFinder 2; Boreal Laser, Edmonton, Canada). The CH 4 concentration in the water sample was calculated from the measured headspace concentration according to Henry's law (see  Jauregui et al., 2012 for details).

Dissolved CH 4 concentration and MO rate
We determined duplicate MO rates in one water sample from each lake as described above, using a modified HE-TDLAS method to allow for measurement of MO in the field. This new method was based on a previous development using the HE-TDLAS method for the determination of methanogenic activity (Martinez-Cruz et al., 25 2012). Two 60 mL lake water subsamples from a single Van Dorn bottle sample were gently transferred to two 100 mL equilibration vials (duplicates). Equilibration vials were immediately closed with rubber stoppers and vigorously shaken for 10 s to transfer 4218 Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | most of the dissolved CH 4 contained in the water sample to the headspace. Next, the headspace was vented, the vial was closed, and the sample was shaken again to evacuate the residual CH 4 content of the water sample. Using this procedure, more than 99.5 % of the original CH 4 content of the sample was evacuated. The equilibration vials were then closed with rubber stoppers and aluminum crimp caps, 5 spiked with 0.6 mL CH 4 (99.0 % purity; Air Liquide, Houston, TX, USA) injected with a disposable syringe, and vigorously shaken for 10 s. This approach allowed MO tests to be conducted with an initial standard CH 4 concentration in the liquid phase (∼ 0.6 mg L −1 ). It also provided an initial CH 4 to O 2 molar ratio of 0.062, significantly below the stoichiometric ratio (0.5), ensuring no O 2 limitation.

10
Equilibration vials were incubated for 10-12 days in a water bath inside insulated boxes placed in our vehicle. In the winter, the vials were maintained at 2 ± 2 • C in a water bath with ice supplements; in the summer, the vials were maintained at 15 ± 2 • C. The temperature of the water bath was measured daily. We measured the CH 4 concentration in the equilibration vials daily using the HE-TDLAS method 15 described in detail by Sepulveda-Jauregui et al. (2012). Briefly, dry control MO test vials containing only CH 4 standards were read by the TDLAS for calibration. Each experimental equilibration vial was vigorously shaken for 10 s to reach phase equilibrium and then immediately placed in the laser beam path, after which a stable HE-TDLAS reading was typically observed within 5 s. Five readings were taken for each 20 MO test vial and recalibration was conducted after measuring each set of test vials to ensure instrument stability. The field HE-TDLAS method allowed measurement of dissolved CH 4 and MO rates. This technique was simple, rapid (about 60 s per sample measurement), non-invasive, and avoided complications and long delays in sample transfer from remote locations to the laboratory. 25 We calculated the total CH 4 concentration (C CH 4 = total CH 4 mass present in the gas and liquid phases divided by the sample liquid volume) in each vial during the MO tests. MO rates were determined from the decrease in C CH 4 , 1996;Segers, 1998) in which C CH 4 and C O 2 represent the actual dissolved CH 4 and DO concentrations measured in the lake, respectively, and K S-CH 4 and K S-O 2 are the apparent affinity constants of the methanotrophic community, for CH 4 and DO, respectively: Average K S-CH 4 and K S-O 2 values for lakes have been determined by previous studies: ;Liikanen et al., 2002;Lofton et al., 2013) and K S-O 2 = 0.624 ± 0.064 mg L −1 (mean ± SD; Lidstrom and Somers, 1984;Frenzel et al., 1990). To the best of our knowledge, the highest K S-CH 4 reported in lakes is 0.704 mg L −1 (Liikanen et al., 2002). It should be noted that these reported K S 15 values refer to the apparent affinity constants for the methanotrophic community, rather than the half-saturation constant for the CH 4 monooxygenase enzyme that catalyzes CH 4 oxidation. The potential error caused by using previously reported K S , instead of experimentally determined values will be considered in the discussion section.
To establish the extent of potential MO limitation by CH 4 or DO, two limitation factors BGD 12,2015 Geographic and seasonal variation of dissolved CH 4 and aerobic CH 4 oxidation K. Martinez-Cruz et al.
A limitation factor of 100 % means that 100 % of a process ceases to occur due to the absence of the limiting substrate, while a limitation factor of 0 % indicates a process 5 occurring at maximum rate (r = r max ). When β > γ, CH 4 was considered to be the limiting factor; conversely, when γ > β, DO was considered to be the limiting factor.

Statistical analyses
Significant differences between all parameters except r and r max were determined using the Tukey-Kramer (TK) multiple comparison test performed after analyses of 10 variance (ANOVA; p < 0.05). Because of the non-normal distributions of r and r max , significant differences were determined using the Kruskal-Wallis (KW) test for nonparametric data (p < 0.05). Statistical analyses were conducted using the NCSS 2000 Statistical Analysis System software (Number Cruncher Statistical Systems, Kaysville, UT, USA). Linear regressions were also conducted to determinate MO rates using 15 Wolfram Mathematica 7.0 (Wolfram, Minneapolis, MN, USA).

Physicochemical parameters
The sampled lakes were shallow; other than four atypical lakes with a maximum known depth > 20 m (lakes #4, #24, #26 and #30), the average lake depth in summer was BGD 12,2015 Geographic and seasonal variation of dissolved CH 4 and aerobic CH 4 oxidation K. Martinez-Cruz et al. 4.5 ± 2.6 m (mean ± SD). During winter, none of the lakes was completely frozen at the sampling stations. Liquid water was always present underneath the ice cover, which ranged in thickness from 0.60 to 1.25 m (mean ± SD, 0.81 ± 0.14 m). The mean temperature throughout the lake water columns was 2.4 ± 0.6 • C (mean ± SD, n = 103) in the winter and 13.9 ± 2.4 • C (mean ± SD, n = 235) in the summer. According to 5 RWCS, during the summer, 15 lakes of the 28 for which a complete temperature profile was determined were fully thermally stratified. Six lakes were partially stratified and seven lakes were mixed. During the winter, 16 of 18 lakes were fully mixed, while two lakes were partially stratified and none was fully stratified. Overall, only one third of the temperature profiles indicated clear stratification.
Lake water pH ranged from 5.9 to 8.2 in winter and from 6.3 to 9.2 in summer among the study lakes. Chlorophyll a was only detected during the summer, ranging from 1.0 to 45.9 µg L −1 (detection limit, 0.03 µg L −1 ). The concentration of dissolved CH 4 in the 30 lakes ranged from 0.01 to 14.77 mg L −1 during the winter and from 0.02 to 1.51 mg L −1 during the summer ( Table 2). The DO concentration at the same depths 15 ranged from 0.10 to 13.63 mg L −1 during the winter and from 0.22 to 11.07 mg L −1 during the summer (Table 2). During summer, a clear oxycline was observed in all yedoma lakes, but only in six of 20 non-yedoma lakes. In contrast, during winter, an oxycline was not observed in any of the yedoma lakes, which were largely anaerobic throughout the whole water column. We observed an oxycline in winter in four of 13 20 non-yedoma lakes. Overall, an oxycline was observed in 30 % of the DO profiles. Temperature-oxygen profiles for all 30 studied lakes are shown in Sepulveda-Jauregui et al. (2014). Figure 3 shows the statistical distributions of the dissolved CH 4 and DO concentrations, as well as the TK comparisons. Significant differences were observed 25 between yedoma and non-yedoma lakes (p < 0.05). In yedoma lakes, the CH 4 and DO concentrations were significantly higher and lower, respectively, than in nonyedoma lakes during both seasons (TK test, p < 0.05). In addition to differences related to permafrost type, higher CH 4 concentrations and lower DO concentrations were BGD 12,2015 Geographic and seasonal variation of dissolved CH 4 and aerobic CH 4 oxidation K. Martinez-Cruz et al. observed during the winter than in the summer (Fig. 3) and an apparent geographic trend was observed. Higher dissolved CH 4 and lower DO concentrations were found in lakes from central Alaska than in those from southern and northern Alaska (Sepulveda-Jauregui et al., 2014).

Methane oxidation rates 5
The HE-TDLAS method allowed us to determine the potential MO potential in the field in all studied lakes. Figure 4 shows three representative C CH 4 trends observed in the MO vials. In some cases, MO began on the first day of incubation (Fig. 4a) and the initial slope of the change in C CH 4 was taken into account in determining the MO rate. In about 60 % of the cases during the summer and 80 % during the winter, a lag phase was 10 observed (Fig. 4b) before C CH 4 began to decrease. This behavior, termed "induction of MO", has previously been reported for various soils (Bender and Conrad, 1995;Dunfield et al., 1999). In lakes in which this pattern was observed, the lag phase was not taken into account and the MO rate was instead determined from the slope of C CH 4 after the lag phase. When no significant decrease in C CH 4 was observed during the first 15 seven days ( Fig. 4c; TK test, p < 0.05), we assumed an MO rate of zero, consistent with previous reports for various soils (Whalen et al., 1990;Bender and Conrad, 1995;Dunfield et al., 1999). We observed MO rates of zero in only three non-yedoma lakes during winter. Otherwise, no correlation with lake morphology, season, or permafrost type was observed in regard to the existence of a lag phase or its duration.

20
The potential MO rate r max ranged from 0.000 to 0.488 mg L −1 d −1 during the winter and from 0.073 to 1.339 mg L −1 d −1 during the summer (Fig. 5a). Seasonal variation of r max was significant, with summer r max up to 47 times higher than winter rates. Permafrost type was also an important determining factor, because during the summer, yedoma lakes had higher r max than non-yedoma lakes (KW test, p < 0.05); specifically, 25 r max was 0.71 ± 0.36 and 0.29 ± 0.16 mg L −1 d −1 (mean ± SD) for yedoma and nonyedoma lakes, respectively. However, during the winter, no significant differences were observed between yedoma and non-yedoma lakes. In addition to differences related to permafrost type, an apparent latitudinal pattern was also observed, with higher r max for lakes from central Alaska compared to those from southern and northern Alaska (Fig. 5a). 5

Geographic and seasonal variations in physicochemical parameters
In yedoma lakes, the CH 4 and DO concentrations were significantly higher and lower, respectively, than in non-yedoma lakes during both seasons. This observation is most likely due to higher organic carbon and nutrient inputs associated with thawing permafrost in yedoma-type lakes. Walter Anthony et al. (2014) and Sepulveda-Jauregui 10 et al. (2014) showed that thawing yedoma permafrost not only provides ancient (Pleistocene-aged) organic carbon stimulating CH 4 production but also phosphate and nitrogen (ammonium), which promotes contemporary plant growth in and around lakes. Since terrestrial plant matter surrounding lakes gets deposited in thermokarst-lake sediments as lakes laterally expand, both enhanced allochthonous and autochthonous 15 productivity of yedoma-type lake ecosystems results in higher rates of contemporary organic matter loading to sediments of yedoma-type lakes compared to non-yedoma lakes (Walter Anthony et al., 2014). Contemporary plant matter decomposes in part to form CH 4 in surface lake sediments, whereas ancient yedoma carbon is largely introduced to lakes at greater depths in the thaw bulb sediments beneath 20 lakes. Thus higher organic carbon and nutrient inputs in yedoma-type lakes promote higher anaerobic and aerobic metabolism and accordingly, lower DO concentrations.
Conversely, higher organic carbon inputs promote higher rates of methanogenesis in the sediments (Huttunen et al., 2003), leading to higher dissolved CH 4 concentrations in the lake water column. Introduction In both yedoma and non-yedoma lakes, higher CH 4 concentrations and lower DO concentrations were observed during the winter than in the summer (Fig. 3). This seasonal variation can be attributed to thick ice covering the lakes in winter. Ice cover impedes gas exchange between the water and the atmosphere, promoting CH 4 buildup in the water column (Phelps et al., 1998;Bastviken et al., 2004;Juutinen et al., 2009) 5 and hindering oxygen transfer from the atmosphere, except in some locations where high-flux ebullition seeps allow gas exchange through local holes in lake ice (Greene et al., 2014). Ice and snow also reduce light penetration and oxygen production by photosynthesis beneath the ice (White et al., 2008;Clilverd et al., 2009). The absence of detectable levels of chlorophyll a in ice-covered lakes during March and April (see 10 results section) despite the longer springtime photoperiod was supportive evidence of reduced photosynthesis under the ice. In summer, although CH 4 production was higher due to warmer sediments, ice was not a physical barrier to CH 4 exchange between the lake water and the atmosphere (Fig. 1).

Conclusions References
Geographic variations were also observed with higher dissolved CH 4 and lower DO 15 concentrations being found in lakes from central Alaska than in those from southern and northern Alaska. However, this apparent latitudinal pattern was related to the higher concentration of yedoma lakes in central Alaska. No significant latitudinal trend was observed when yedoma and non-yedoma lakes were analyzed separately (TK test, p < 0.05).
20 Figure 3 shows that when relatively high CH 4 concentrations were found, relatively low DO concentrations were observed and conversely, when low dissolved CH 4 concentrations were found, higher DO concentrations were observed. This pattern was particularly clear in yedoma lakes: in winter, a CH 4 concentration of 7.32 ± 5.86 mg L −1 (mean ± SD) was found, while the DO concentration was 25 0.13 ± 0.03 mg L −1 (mean ± SD). In the same yedoma lakes, the summer CH 4 concentration was 0.49 ± 0.52 mg L −1 (mean ± SD), while the DO concentration was 3.19 ± 3.24 mg L −1 (mean ± SD The trend toward higher CH 4 concentrations and lower DO concentrations in winter than in summer was not as strong in non-yedoma lakes as in yedoma lakes (Fig. 3). These results provide additional evidence that organic carbon inputs to yedoma lakes fuel methanogenesis and MO more strongly than in non-yedoma lakes. Another reason is that yedoma lakes have a significantly higher ebullition year round (Walter et al.,  2014) found that 80 % of CH 4 in ebullition bubbles trapped by lake ice dissolves into the lake water column, leading to elevated dissolved CH 4 beneath the ice. Another possible explanation for higher MO in yedoma lakes compared to non-yedoma lakes may be related to microbial community composition, but this was beyond the scope of our study.

Limiting elements of MO rates
The actual MO rates r estimated from r max , reduced the magnitude of the MO, with r ranging from 0.000 to 0.124 mg L −1 d −1 during the winter and from 0.017 to 0.538 mg L −1 d −1 during the summer (Fig. 5b). . Similarly, r values were 1 to 50-fold higher in the summer than in the winter. We attribute this finding to the temperature dependence of methanotrophy (Semrau et al., 2008;Borrel et al., 2011), but also to the limited DO concentration under the ice cover during the winter. In addition to seasonal variations, permafrost type was also a determining factor 20 of r. As observed with r max , during summer yedoma lakes showed 2-3 times higher r than non-yedoma lakes (KW test, p < 0.05; r = 0.28 ± 0.17 mg L −1 d −1 , mean ± SD, yedoma lakes; r = 0.09 ± 0.08 mg L −1 d −1 , mean ± SD, non-yedoma lakes). Higher r values for yedoma lakes in summer is most likely linked to the higher dissolved CH 4 concentration in presence of a relatively high DO concentration above the oxycline 25 (Fig. 3). An apparent latitudinal trend was observed, with higher r and r max for lakes from central Alaska compared to those from southern and northern Alaska (Fig. 5). This apparent trend was associated with a higher number of yedoma lakes in central Alaska. No significant latitudinal trend in MO was observed when yedoma and nonyedoma lakes were analyzed separately. The actual MO rates; r, were determined from r max and CH 4 and DO concentrations using two affinity constants, K S-CH 4 and K S-O 2 . These affinity constants are highly 5 variable, because their determination is challenging and subject to relatively high determination error (Segers et al., 1998) and because the methanotrophic community is sensitive to numerous factors and changes over time and space (Carini et al., 2005;He et al., 2012). For instance, Lofton et al. (2014) reported a variation of 150 % in K S-CH 4 within the hypolimnetic water column of two lakes with similar 10 characteristics. The determination of MO rates may, therefore, be subject to large error if reported values are used instead of experimental parameters or if an error occurs in experimental K S determinations. To quantify these potential errors, a sensitivity analysis was conducted. We arbitrarily modified K S-CH 4 and K S-O 2 and calculated the resulting r (Eq. 1) using the experimental r max , C CH 4 , and C O 2 measured in the 30 15 lakes. Figure 6 shows the impact of varying K S-O 2 (Fig. 6a) and K S-CH 4 (Fig. 6b) on the sensitivity coefficient (∆r/∆K S ), where ∆r is the variation in r (%) for a given variation in K S (%). In Fig. 6, ∆r/∆K S is expressed as a function of K S , which is K S normalized over the range reported in the literature, with 1.0 being the average K S reported (e.g. 20 0.110 mg L −1 for K S-CH 4 ). Figure 6a shows that ∆r/∆K S-O 2 ranged from 6 to 12 % for all lakes and seasons, except in yedoma lakes during winter, where an error of up to 200 % was estimated. This relatively high sensitivity to error in K S-O 2 in yedoma lakes during winter is likely due to DO concentrations close to K S-O 2 . Likewise, Fig. 6b shows that ∆r/∆K S-CH 4 ranged from 2 to 9 % for all lakes and seasons, except in 25 non-yedoma lakes during summer where an error of up to 100 % was estimated. As above, the latter is likely due to CH 4 concentrations close to K S-CH 4 . This sensitivity analysis shows that, other than for K S-O 2 in yedoma lakes during winter and K S-CH 4 in non-yedoma lakes during summer, errors or variations in K S would have relatively little impact on determination of methanotrophic rates. Overall, under the conditions of the present study, taking into account yedoma and non-yedoma lakes in each season, the weighted average of ∆r/∆K S-O 2 and ∆r/∆K S-CH 4 ranged from 8 to 40 %. This error is significantly lower than the standard deviation of r, which averaged 119 %. From Eqs. (2) and (3), we estimated that, during the summer CH 4 was the main limiting element in 25 out of 30 lakes. In contrast, during winter, CH 4 was the main limiting element in 10 of 26 lakes (Table 2). Notably, during the winter, DO was the limiting element for all seven yedoma lakes, while during the summer, MO was limited by CH 4 for all non-yedoma lakes. A similar sensitivity analysis was done on β and γ, as done with r, to estimate if the 10 estimated limiting element would change for variable K S . The results showed no impact on the limiting element in the 30 lakes and for both seasons. These results confirm that MO was mainly controlled by DO and CH 4 availability, which in turn, depended on the season and landscape processes. A potential bias in our r estimates may have arisen from taking duplicate samples at 15 a single depth in each lake. The literature has clearly shown that a higher MO rate is often found at the oxic/anoxic interface in stratified lakes (Utsumi et al., 1998a(Utsumi et al., , 1998bBastviken et al., 2002;Carini et al., 2005;Pimenov et al., 2010;Schubert et al., 2010). Estimation of MO rates consistently measured at a single depth that was not necessarily located at the oxic/anoxic interface may have neglected potentially 20 higher rates occurring at the oxic/anoxic interface in stratified lakes. However, in the present study, the sampled lakes were in many cases shallow, relatively well mixed, and without a clear oxycline (see Results section), suggesting a relatively homogeneous water column. Utsumi et al. (1998b) observed homogeneous MO rates at all depths of a shallow and mixed temperate lake, while Rudd and Hamilton (1978) also reported 25 homogeneous MO rates during overturn of a dimictic lake. Determination of MO rates at the oxic/anoxic interface, in the few cases in which such an interface was observed, would likely have indicated higher MO rates. Thus, the results of r presented here, may be underestimated to an unknown extent.

Conclusions
We developed a new method based on a TDLAS for the determination of MO rates together with dissolved CH 4 concentration in lakes in the field. This method was successfully applied to 30 lakes along a north-south transect and allowed for the determination of MO potentials ranging from 0.000 to 1.339 mg L −1 d −1 in winter and 5 summer. MO rates showed high seasonal and geographic variability. In addition to temperature effects, the main factors controlling MO were: (1) CH 4 availability during the summer, limited both by exchange with the atmosphere and by MO itself, (2) DO availability during the winter, mainly due to ice cover impeding gas exchange with the atmosphere and primary production; and (3) inputs of organic substrates to lakes, mainly related to the presence or absence of yedoma permafrost as an additional source of carbon and nutrients. These results indicate that MO may substantially mitigate the increase in CH 4 emission predicted by permafrost thawing (Khvorostyanov et al., 2008;Walter Anthony et al., 2014). Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Methanogenesis Methanogenesis
Carbon release Thaw bulb Figure 1. Carbon cycling in northern high-latitude lakes during the summer and winter. Carbon (C org ) release from primary production and landscape processes promotes CH 4 production and competes with MO for O 2 .   Figure 6. Sensitivity analysis of the impact of an error or variation in K S-O 2 (a) and K S-CH 4 (b) on r; yedoma lakes in winter (-), yedoma lakes in summer (-· -·), non-yedoma lakes in winter (-·· -··) and non-yedoma lakes in summer (----).