Constraints on methane oxidation in ice-covered boreal lakes
Abstract
Boreal lakes can be ice covered for a substantial portion of the year at which time methane (CH4) can accumulate below ice. The amount of CH4 emitted at ice melt is partially determined by the interplay between CH4 production and CH4 oxidation, performed by methane-oxidizing bacteria (MOB). Yet the balance between oxidation and emission and the potential for CH4 oxidation in various lakes during winter is largely unknown. To address this, we performed incubations at 2°C to screen for wintertime CH4 oxidation potential in seven lakes. Results showed that CH4 oxidation was restricted to three lakes, where the phosphate concentrations were highest. Molecular analyses revealed that MOB were initially detected in all lakes, although an increase in type I MOB only occurred in the three lake water incubations where oxidation could be observed. Accordingly, the increase in CO2 was on average 5 times higher in these three lake water incubations. For one lake where no oxidation was measured, we tested if temperature and CH4 availability could trigger CH4 oxidation. However, regardless of incubation temperatures and CH4 concentrations, ranging from 2 to 20°C and 1–500 μM, respectively, no oxidation was observed. Our study indicates that some lakes with active wintertime CH4 oxidation may have low emissions during ice melt, while other and particularly nutrient poor lakes may accumulate large amounts of CH4 below ice that, in the absence of CH4 oxidation, will be emitted following ice melt. This variability in CH4 oxidation rates between lakes needs to be accounted for in large-scale CH4 emission estimates.
Key Points
- CH4 oxidation in lake water from three out of seven ice-covered lakes occurred after one month of incubation at 2°C
- Factors other than CH4 availability, temperature, and presence of methanotrophs, limited CH4 oxidation
- CH4 oxidation limitation was partially explained by phosphate availability and bacterial community interactions
1 Introduction
Lakes are significant contributors to the global methane (CH4) budget, contributing up to 16% of the total CH4 emissions into the atmosphere [Bastviken et al., 2011]. High-latitude lakes represent a large fraction of all lakes on Earth [Verpoorter et al., 2014] and can be ice covered for many months of the year [Prowse et al., 2012]. During this time, CH4 produced in anoxic sediments and exported to the water column by ebullition or diffusion accumulates in the water column below ice [Michmerhuizen et al., 1996; Phelps et al., 1998]. At the same time, in aerobic waters below ice, biological CH4 oxidation has the potential to decrease concentrations of CH4. Since CH4 emissions from lakes can be substantial during the ice melt period, accounting for 3–84% of the annual CH4 released from lakes [Karlsson et al., 2013], understanding the constraints of CH4 oxidation prior to ice melt is important. However, current knowledge on CH4 oxidation during winter is sparse with no consensus on the rate of CH4 oxidation [Michmerhuizen et al., 1996; Kankaala et al., 2006; Martinez-Cruz et al., 2015], making it difficult to account for the amount of CH4 that is oxidized below ice prior to ice melt.
CH4 oxidation is carried out by methane-oxidizing bacteria (MOB), which use CH4 as a carbon and energy source. Aerobic CH4 oxidation typically takes place where both CH4 and oxygen (O2) are present in large enough quantities and has been suggested to be primarily limited by CH4 substrate supply [cf. Duc et al., 2010]. The highest rates of CH4 oxidation in lakes are often found in transition zones between anoxic and oxic environments, where MOB profit from high CH4 concentrations while also having access to O2, the thermodynamically most favorable electron acceptor [Carini et al., 2005; Bastviken, 2009]. Conditions at the water-ice interface can also be favorable for CH4 oxidation when CH4 ebullition from anoxic sediments elevates CH4 concentrations in oxic surface waters [Ricão Canelhas et al., 2016]. However, studies investigating CH4 oxidation in cold surface waters below ice are limited, while the direct effect of temperature on CH4 oxidation has been a matter of debate. Some studies suggest that CH4 oxidation is indirectly temperature dependent, since CH4 production is temperature related and CH4 oxidation is driven by CH4 availability [Segers, 1998; Kankaala et al., 2006; Duc et al., 2010; Borrel et al., 2011]. Other studies conversely suggest that CH4 oxidation is directly related to temperature since cold temperatures may restrain overall microbial activity [Sundh et al., 2005; Martinez-Cruz et al., 2015].
The aim of this study was to test the CH4 oxidation potential in surface water collected during winter below lake ice. We performed a comparative lake screening for wintertime CH4 oxidation potential in seven lakes characterized by varying water chemistry. In addition, for one of the seven lakes, a more detailed incubation at different temperatures and CH4 concentrations was carried out to test if temperature and CH4 availability could trigger CH4 oxidation in a single system. In all incubations CH4 oxidation was determined by tracking changes in CH4 concentration over time. In addition, microbial community composition analysis was conducted to identify the presence of MOB at the start and the end of the incubation.
2 Materials and Methods
2.1 Lake Survey Sampling
For the CH4 oxidation incubation experiment, water was collected from seven ice-covered lakes in early 2015: Björklinge-Långsjön, Erken, Fyrsjön, Lumpen, Lötsjön, Malstasjön, and Plåten (Figure 1). The lakes were chosen based on differences in morphometry and water chemistry.
Water for the incubations was collected in polypropylene bottles from a drilled hole in the ice at a depth of 0.5 m below the bottom of the ice. The depth of 0.5 m was chosen to avoid sampling water that had been disturbed by the drilling of the ice. In situ water temperature, O2, and conductivity were measured using a HQ40d Portable Multiparameter sonde (HACH). pH, CH4 concentration and dissolved organic carbon (DOC) of the lake water were measured upon returning from the field. CH4 measurements were made using the headspace technique, where a 60 mL polypropylene syringe (Plastipak- Becton-Dickson) was first filled with bubble free lake water. A headspace of ambient air was immediately introduced and the syringe was shaken for at least 1 min to equilibrate dissolved CH4 from the water with the headspace. The resulting headspace was transferred to a closed serum vial filled with a saturated NaCl solution (sample injection replaced NaCl solution in the vial as described in detail in Bastviken et al. [2010]) to preserve the sample until it was analyzed. Analyses were performed on a gas chromatograph (GC) (Agilent Technologies 7890A GC Systems) equipped with a flame ionization detector. CH4 concentrations were calculated according to Henry's law, correcting for temperature according to Lide and Frederikse [1995] and corrected for CH4 in ambient air (also measured on the GC). For DOC analyses water was filtered through a precombusted 0.7 µm glass fiber filter (Whatman GF/F) and then measured on a total organic carbon (TOC-V) analyzer (Shimadzu scientific Instruments).
2.2 CH4 Oxidation Potential in Multiple Lakes
Incubations were performed in 120 mL glass bottles that had been acid washed overnight in 10% HCl, subsequently rinsed repeatedly with Q-grade water (Milli-Q Advantage A 10 ultrapure water purification system) and finally muffled for 4 h at 450°C. The clean bottles were filled with 60 mL of lake water and capped with 10 mm thick butyl rubber stoppers and aluminum crimp seals. The headspace in the vials was then evacuated and pressure equilibrated with outside air by inserting a needle through the stopper for 2 min. CH4 concentration in the ambient air was measured to determine the exact amount of CH4 in the headspace at the start of the experiment. To obtain the desired CH4 concentration of 100 μM CH4 in the water of each incubation bottle, 16 mL of a predetermined CH4 gas stock was first injected, using a syringe connected to a needle, and mixed with the headspace. Subsequently, 16 mL of headspace gas was taken out with the same syringe to maintain the atmospheric pressure in the bottles and used as sample for initial stable isotope analysis (detailed below).The CH4 concentration of the gas stock was determined by the partitioning of CH4 between the headspace and the water according to Henry's Law.
For the lake survey water samples, three replicate bottles and one autoclaved lake water control bottle were set up for each lake. Bottles were incubated at 2°C and placed in a water bath equipped with water pumps that created constant whirl movement and mixing of the vials for 15 min every second hour. Initial CH4 and carbon dioxide (CO2) were measured in the headspace after 1 day (start) and after 26 days (end) for most lakes (Björklinge-Långsjön after 28 days). In addition, gas for stable isotope analysis and water for the analysis of nitrate (NO3−), phosphate (PO43−), and sulfate (SO42−), bacterial abundance, and bacterial community composition were sampled at the start and end of the incubation (described below).
2.3 Temperature and CH4 Concentration Effects in Björklinge-Långsjön
To test if CH4 concentration and temperature trigger the oxidation of CH4, incubations were set up with water collected below ice from Björklinge-Långsjön. Since Björklinge-Långsjön is a mesotrophic lake that had intermediate levels of NO3−, PO43−, and CH4 but did not show any sign of CH4 oxidation in the survey (see section 9, it was a suitable candidate for the detailed incubation. Björklinge-Långsjön is situated in central Sweden (Figure 1), has clear water, and is surrounded mostly by forest and agriculture (~40% each). The lake is groundwater fed, draining catchment soils with a marine origin, which causes it to have higher conductivity and a greater buffering capacity. Incubations were set up in the same manner as detailed above. However, in this case, bottles were amended with four different CH4 concentrations: 1, 10, 100, and 500 μM, with five replicate bottles and one control bottle with autoclaved lake water for each concentration. Bottles were placed in four temperature control chambers, set to 2, 4, 10, and 20°C, equipped with water baths (i.e. 4 × 4 treatments and in total 96 bottles including controls).
Over the first 2 weeks, headspace CH4 and CO2 were regularly monitored every second to third day. Subsequently, the interval between sampling events was increased to approximately 1 week, with a total length of the incubation time of 53, 52, 44, and 43 days for 2, 4, 10, and 20°C, respectively. In order to avoid underpressure in the vials when withdrawing samples repeatedly, 1 mL of atmospheric air stock was injected into the vial and mixed with the headspace before collecting 1 mL of headspace gas to be analyzed for CH4 on the GC. Prior to injection into the GC, all samples were manually diluted with N2 (Air Liquide). The ambient atmospheric pressure and the dilutions from air stock additions, as well as dilution with N2, were accounted for when calculating CH4 and CO2 concentration in the headspace.
2.4 Chemical and Biological Analyses of the Incubations
As an additional measure of CH4 oxidation, gas samples for isotope analysis of δ 13C-CH4 (i.e., 13C/12C) were taken at the start and end of the incubation experiment. A change in the isotope composition with an enrichment of the heavier 13C isotope indicates methanotrophic bacterial activity. Briefly, gas samples of 16 mL were injected into 12 mL Exetainer® vials (Labco Ltd, High Wycombe, UK). Stable isotope analysis was carried out at the Stable Isotope Facility at UC Davis following standard procedures (http://stableisotopefacility.ucdavis.edu) using an isotope ratio mass spectrometer (ThermoScientific DELTA V Plus, Bremen, Germany).
The water chemistry of incubation water was characterized at the start and end of the experiment. Water for analyses of NO3−, PO43−, and SO42− was first prefiltered through rinsed 0.2 µm membrane filters (Pall Corporation) and then analyzed by ion chromatography (IC) on a Metrohm IC system (883 Basic IC plus and 919 Autosampler Plus) fitted with a Metrosep A Supp 4/5 guard column and a Metrosep A Supp 5 analytical column (150 × 4.0 mm).
Further, water for analysis of bacterial abundance was collected at the start and end of all incubations. Briefly, cells were fixed with 0.2 µm filtered formaldehyde (2% final concentration) and stored at 4°C prior to analyses. Cells were stained with the fluorescent nucleic acid stain Syto13® (Molecular probes, Invitrogen, Carlsbad, CA, USA) according to the protocol of del Giorgio et al. [1996] and were then counted with a flow cytometer equipped with a 488 nm blue solid state laser (Cyflow Space, Partec, Görlitz, Germany) using green fluorescence for triggered particle scoring. Cell counts were analyzed using Flowing Software version 2.5 (Perttu Terho, Centre for Biotechnology, Turku Finland).
Cells for DNA extractions and analyses of bacterial community composition were captured by filtering 100 mL of water at the start of the incubation and 40 mL at the end through 0.2 µm Supor 200 filters (Pall Corporation, Port Washington, NY, USA). The filters were stored at −80°C. The Power Soil DNA isolation kit was used to extract DNA as recommended by the manufacturer (MoBio Laboratories, Carlsbad, CA, USA). Bacterial 16S rRNA genes were amplified by PCR and analyzed as described in detail in Sinclair et al. [2015]. Barcoded amplicons from individual samples were pooled and sequenced with 2 × 300 cycles using the Illumina MiSeq platform at the SciLifeLab SNP/SEQ sequencing facility hosted by Uppsala University. Raw sequence data were quality filtered using an in-house pipeline [Sinclair et al., 2015], and sequences were grouped in operational taxonomic units (OTUs) using UPARSE with a 3% dissimilarity threshold [Edgar, 2013] and taxonomically annotated using CREST with SilvaMod [Lanzén et al., 2012]. Estimates of relative abundances (%) of methanotrophs in the bacterial community were performed after random resampling of the OTU table to the smallest sample size using R (version 3.1.1) and the Vegan package [Oksanen et al., 2013]. Raw sequence data have been deposited at the Sequence Read Archive (SRA) in GenBank® under accession number SRP063993.
2.5 CH4 Oxidation
(1)
For the lake survey (seven lakes incubated with 100 μM at 2°C, three bottles, and one control per lake) an analysis of variance (ANOVA) was run to test for a significant difference between ΔCH4 (dependent variable) of the lake to the grouped controls, indicating CH4 oxidation. Lake was set as the independent variable (eight groups: Björklinge-Långsjön (n = 5), Erken (n = 3), Fyrsjön (n = 3), Lumpen (n = 3), Lötsjön (n = 3), Malstasjön (n = 3), Plåten (n = 3), and control (n = 7)). A post hoc Tukey-Kramer test was used to determine which lakes differed from the control. We assumed that significant CH4 oxidation occurred with p value < 0.05. The residuals of the ANOVA analysis were tested for normal distribution (Shapiro-Wilk, p value > 0.05) and homogeneity of variance (Levene's test, p value > 0.05).
In the extended incubation with Björklinge-Långsjön water, it was assessed whether CH4 oxidation had occurred in any of the different treatments, using separate ANOVA tests for each temperature room (2, 4, 10, and 20°C). For all four ANOVA tests, ΔCH4 was set as the dependent variable and treatment (five groups: 1 μM (n = 5), 10 μM (n = 5), 100 μM (n = 5), and 500 μM (n = 5) and control (n = 4)) was set as the independent variable. Again, post hoc Tukey-Kramer tests were used to determine which treatment had significant CH4 oxidation.
Since CH4 oxidation causes isotopic fractionation, δ13C-CH4 was used as an independent confirmation of CH4 oxidation in lakes Malstasjön, Erken and Lötsjön. Based on the incubations, we could also empirically determine the isotope fractionation factor (α value) at 2°C according to Coleman et al. [1981]. For each bottle the fraction (%) of the CH4 oxidized under closed system conditions (fclosed) was calculated from the isotope fractionation factor (α) using the Rayleigh model for a closed system [Liptay et al., 1998].
2.6 Bacterial Community Composition
Differences in bacterial community composition at the start and the end of the incubations were assessed using Bray-Curtis dissimilarity measures on rarefied OTU tables and were visualized in a nonmetric multidimensional scaled plot. Further, differences in start and end bacterial community composition were compared between lakes where CH4 oxidation was observed with lakes where CH4 oxidation was not observed. The difference in end bacterial community composition comparison between lakes was run an additional time with MOB excluded from the test. The compositional difference, from the Bray-Curtis dissimilarity, between lakes was tested with permutational multivariate analysis of variance (PERMANOVA) analysis (R software), in order to reveal significant differences in community composition. To determine if the relative abundance in MOB at the end of the incubation was significantly higher in lakes with observed oxidation compared to lakes where oxidation was not observed, we used a nonparametric Wilcoxon test (nonnormally distributed data, Shapiro-Wilk's test result: p < 0.01). An indicator species test was also performed, using the indicspecies package with the R software (version 3.1.1), to compare the co-occurring communities of the lakes that showed oxidation with the emerging communities in lakes where no CH4 oxidation was observed. The indicator species test classifies an OTU as exclusive when it is consistently found in either group of lakes. The identified OTUs indicate high association with the sample group and can be used to characterize the sample [Dufrêne and Legendre, 1997]. Prior to analysis, OTUs classified as known methanotrophs were removed, in order not to bias the lakes that had clearly shown oxidation from the ones that had not, and the OTU table was subsampled in order to perform the indicator species test. The test results are given as indicator value (indval) ranging from 0 to 1, with higher values representing a better prediction indication. The significance of the indicator value was then tested using a permutation test (permutations = 999) and the p value was adjusted for repeated measurements.
3 Results
For all seven lakes, the surface waters sampled below the ice were supersaturated in CH4 (relative to atmospheric levels) and contained MOB (Tables 1 and 2). The water was also oxic in the six lakes where O2 was measured (Table 1). Despite these properties, relative change in CH4 concentration at the end of the incubations was only significantly different from the control in the lakes Malstasjön, Erken, and Lötsjön (ANOVA, Tukey-Kramer adjusted p value < 0.01), indicating CH4 oxidation occurred in three of the seven lakes (Table 2). In lake Björklinge-Långsjön, where no oxidation was observed at 2°C, 100 μM, the restriction of CH4 oxidation was confirmed at different temperatures (2, 4, 10, and 20°C) and CH4 concentrations (1, 10, 100, and 500 μM). None of these treatments were statistically different from the control (ANOVA, Tukey-Kramer adjusted p value > 0.05).
| Lake | ID | Coordinates (N, E) | LA (km2) | Zavg (m) | Temp (°C) | O2 (mg L−1) | Cond (μS cm−1) | pH | CH4 (μM) | DOC (mg L−1) |
|---|---|---|---|---|---|---|---|---|---|---|
| Malstasjön | M | 59.77, 18.64 | 0.24 | 3.4 | 0.2 | 9.6 | 307 | 7.9 | 0.97 | 15.4 |
| Erken | E | 59.84, 18.63 | 23.7 | 9.0 | 0.3 | 13.4 | 239 | 8.1 | 1.11 | 10.7 |
| Lötsjön | Ls | 59.87, 17.95 | 0.58 | 6.0 | 0.4 | 12.7 | 201 | 7.9 | 0.04 | 9.7 |
| Björklinge | BL | 60.03, 17.57 | 2.46 | 6.1 | nd | nd | nd | 8.0b | 0.43 | 6.8 |
| Plåten | P | 59.86, 18.54 | 0.02 | 2.9 | 0.7 | 5.7 | 93 | 6.7 | 0.04 | 28.0 |
| Fyrsjön | F | 59.79, 18.51 | 0.16 | 2.6 | 0.7 | 9.9 | 203 | 7.8 | 0.61 | 14.3 |
| Lumpen | L | 59.96, 17.28 | 0.24 | 1.3 | 1.1 | 7.2 | 62 | 6.5 | 0.44 | 27.1 |
- a Lake water chemistry was measured from water collected at 0.5 m below the bottom of the ice. Lakes with detectable CH4 oxidation are in bold.
- b Measured below ice in winter of 2012; nd = not determined.
| ID | ΔCH4 | ΔCO2 | Start MOB (%) | End MOB (%) | CH4 oxidation (%) |
|---|---|---|---|---|---|
| M | −0.37 | 1.70 | 2.1 | 60.6 | 41 |
| E | −0.17 | 2.75 | 11.5 | 51.2 | 21 |
| Ls | −0.08 | 0.90 | 3.5 | 36.8 | 8 |
| BL | 0.01 | 0.52 | 1.6 | 0.1 | nd |
| P | −0.01 | 0.30 | 1.1 | 0.4 | nd |
| F | −0.02 | 0.29 | 1.2 | 0.7 | nd |
| L | 0.02 | 0.34 | 4.1 | 1.2 | nd |
- a Lakes with detectable CH4 oxidation are in bold and the corresponding average fractions of CH4 oxidation is reported; nd = not determined. For lake abbreviations refer to Table 1.
The enrichment in 13C, i.e., increasing δ 13C-CH4 values, further confirmed methanotrophic activity in lakes Malstasjön, Erken, and Lötsjön. In these three lakes the fraction of CH4 oxidized during the 26 day incubation, determined by 13C fractionation for a closed system, ranged from 7 to 42%, with an average of 41, 21, and 8%, respectively, for lakes Malstasjön, Erken, and Lötsjön (Table 2). The isotope fractionation factor (α) at 2°C, used to determine the fraction of CH4 oxidized, ranged between 1.017 and 1.038.
The CO2 concentration in the headspace of the bottles in all incubations increased from the start to the end. However, the ΔCO2 was on average 5 times higher in lakes where CH4 was oxidized (mean ± SD, ΔCO2: 1.78 ± 0.93) compared to lakes where no oxidation was observed (mean ± SD, ΔCO2: 0.36 ± 0.11) (Table 2).
The composition of bacterial communities differed significantly between the start and end of the incubation for all seven lakes (PERMANOVA, F = 3.85, R2 = 0.12, p = 0.003; Figure 2). Comparing bacterial communities in lakes where CH4 oxidation was observed with bacterial communities in lakes where CH4 oxidation was not observed, we found that start communities were not significantly different (PERMANOVA, F = 0.71, R2 = 0.12, p = 0.75), whereas end communities were significantly different (PERMANOVA, F = 15.64, R2 = 0.49, p =0.002). In order to investigate if the difference in bacterial community at the end of the incubation could be related to taxa other than MOB, we also investigated differences in community composition with MOB excluded from the test. Also here, bacterial communities at the end of the experimental incubation remained significantly different when comparing lakes with and without observed CH4 oxidation (PERMANOVA, F = 5.7, R2 = 0.26, p = 0.001).
MOB were represented by the families Methylocystaceae, Methylococcaceae, and Methylacidiphilaceae in all lakes at the start of the incubation. The relative abundance of MOB only increased in the three lakes, where significant CH4 oxidation was observed (Figure 3), which was explained by a significant increase in Methylococcaceae (nonparametric Wilcoxon test: p < 0.0001).
An indicator species test revealed 11 indicator OTUs that were typical for the lakes featuring CH4 oxidation, whereas three OTUs were indicative for lakes where CH4 oxidation was not observed (Table 3). In the group of lakes with CH4 oxidation, 4 out of the 11 indicator OTUs were methylotrophs from the facultative methylotroph group Methylophilaceae. Accordingly, an increase in relative abundance of both Methylococcaceae and Methylophilaceae was observed in lakes where CH4 oxidation occurred (Figure 3).
| Taxonomical Assignment | Indval | p Value | Freq |
|---|---|---|---|
| β-proteobacteria-Methylophilaceae | 0.99 | 0.03 | 14 |
| β-proteobacteria-Methylophilaceae | 0.99 | 0.03 | 10 |
| Flavobacteria-Cryomorphaceae | 0.87 | 0.03 | 11 |
| β-proteobacteria-Methylophilaceae | 0.84 | 0.03 | 19 |
| Sphingobacteria-Sphingobacteriaceae | 0.83 | 0.03 | 15 |
| β-proteobacteria-Oxalobacteraceae | 0.79 | 0.03 | 16 |
| Actinobacteria-Sporichthyaceae | 0.78 | 0.04 | 20 |
| β-proteobacteria-Methylophilaceae | 0.77 | 0.03 | 10 |
| γ-proteobacteria-Moraxellaceae | 0.77 | 0.03 | 9 |
| Actinobacteria-Sporichthyaceae | 0.68 | 0.03 | 23 |
| Actinobacteria-Sporichthyaceae | 0.67 | 0.03 | 6 |
| β-proteobacteria-Comamonadaceae | 0.78 | 0.03 | 23 |
| Phycisphaerae-Phycisphaeraceae | 0.74 | 0.03 | 20 |
| β-proteobacteria-Comamonadaceae | 0.66 | 0.03 | 23 |
- a The test results are given as an indicator value for each taxa (indval), the adjusted p value depicts the significance level of the permutational test (significance level < 0.05), and the relative frequency (freq) is reported as the number of times the taxa appeared in the sample.
Among the lake water characteristics measured, initial PO43− concentrations stood out as being higher in the three lakes with significant CH4 oxidation compared to the lakes where CH4 oxidation was not observed (Table 4). In the mentioned three lakes, most of the measured inorganic nutrients were depleted during the month-long incubation except for SO42− which remained at a stable level. In all three lakes PO43− decreased to below detection limit (<10 µg L−1), while NO3− decreased by 23, 99, and 16%, for Malstasjön, Erken, and Lötsjön, respectively (Table 4).
| ID | Start NO3− (mg L−1) | End NO3− (mg L−1) | Start PO43− (µg L−1) | End PO43− (µg L−1) | Start SO42− (mg L−1) | End SO42− (mg L−1) | Start Abund (105 cells mL−1) | End Abund (105 cells mL−1) |
|---|---|---|---|---|---|---|---|---|
| M | 10.40 | 8.03 | 74.1 | <10 | 42.7 | 42.0 | 2.78 | 1.20 |
| E | 0.83 | 0.01 | 96.1 | <10 | 29.3 | 29.3 | 1.63 | 7.80 |
| Ls | 1.39 | 1.17 | 25 | <10 | 5.7 | 5.6 | 4.77 | 5.85 |
| BL | 1.13 | 0.89 | 16 | <10 | 69.4 | 70.5 | 4.43 | 8.95 |
| P | 0.90 | 0.88 | <10 | <10 | 11.9 | 11.8 | 8.59 | 7.77 |
| F | 1.00 | 1.04 | <10 | <10 | 13.2 | 13.5 | 2.83 | 4.28 |
| L | 1.27 | 1.32 | <10 | <10 | 6.1 | 6.0 | 4.66 | 2.60 |
- a Lakes (ID) with detectable CH4 oxidation are in bold. For lake abbreviations refer to Table 1.
4 Discussion
The lake screening incubations revealed that wintertime CH4 oxidation potential among lakes varied greatly; some lakes had a high potential for CH4 oxidation, while in other lakes CH4 oxidation was not observed even after 1 month of incubation. These patterns were confirmed by multiple independent measurements; CH4 concentration changes, δ13C-CH4 values, and increase in abundance of MOB. Previous studies have found CH4 oxidation in lakes to be limited by CH4 availability [e.g., Kankaala et al., 2006; Duc et al., 2010] with greatest wintertime CH4 oxidation found in relatively deep waters where CH4 concentrations are highest [e.g., Bastviken et al., 2002]. Since the CH4 concentration chosen for the lake survey incubation (100 μM) was higher than CH4 concentrations in most ice-covered boreal lakes (mean CH4 concentration of 3.39 and 53.62 μM for surface and bottom waters, respectively [Juutinen et al., 2009]), the potential CH4 concentration limitation threshold should have been exceeded, and therefore, CH4 oxidation was expected. However, our study revealed that elevated CH4 concentrations were not sufficient to activate CH4 oxidation in more than half of the studied lakes, in spite of ubiquitous presence of MOB (Figure 3). Similarly, lake water from Björklinge-Långsjön incubated at varying CH4 concentrations, spanning from concentrations (1 μM) similar to the in situ conditions (Table 1) to well above a potential CH4 limitation threshold (500 μM), did not show CH4 oxidation, further confirming that CH4 oxidation was not dependent on CH4 availability. Additionally, it has been argued that CH4 oxidation potential can vary with temperature [e.g., Martinez-Cruz et al., 2015]. For example, He et al. [2012] found that CH4 oxidation was highest at incubation temperatures most similar to in situ conditions for that study (21°C), while CH4 oxidation at temperatures of 4 and 10°C was very low. Despite incubating lake water at near in situ temperature in the present study (~2°C) and also at 4, 10, and 20°C, in the case of Björklinge-Långsjön, CH4 oxidation could still not be observed for any of the tested temperatures or CH4 substrate levels. Hence, our results indicate that factors other than temperature, CH4 availability, and MOB presence may limit CH4 oxidation during winter below ice.
Other variables shown to limit CH4 oxidation have been previously reviewed [Hanson and Hanson, 1996; Bastviken, 2009] and include O2 availability, pH, salinity, light intensity, inorganic nitrogen concentrations (NH4+ or NO3−), and zooplankton predation on bacteria, including the rather large methanotrophs. Since our study focused on CH4 oxidation during winter, when the lakes were covered by ice and snow, the effects of light and zooplankton predation on CH4 oxidation may have been reduced. The pH and salinity (conductivity used as a proxy) varied between lakes, but no clear pattern was observed between the lakes with CH4 oxidation and the lakes where this process was not detected. The effects of NH4+ and NO3− on CH4 oxidation have been contradictory since both inhibition and stimulation of oxidation has been reported previously (reviewed in Bodelier and Laanbroek [2004]). Furthermore, a clear relationship between NO3− and CH4 oxidation was not found in this study, and it therefore appears that CH4 oxidation was limited by some other factor than previously proposed.
Phosphorus availability played an important role in CH4 oxidation and heterotrophic respiration since the three lakes with the highest PO43− concentrations were the only lakes where CH4 oxidation was observed and had highest ΔCO2 (Table 2). Furthermore, even though MOB were present in all lakes at the start of the incubation, MOB abundance only increased in the three lakes with highest PO43− concentrations (Figure 3). MOB are considered slow growing bacteria with low growth rates [van Bodegom et al., 2001] and often need several weeks to replicate [Horz et al., 2002]. Since all organisms require phosphorus for cell division, energy transformations, and cell maintenance, MOB may be out competed for nutrients by faster growing heterotrophic bacteria. Accordingly, the PO43− available in the three lakes (Table 4) where CH4 oxidation was detected could potentially have provided sufficient phosphorus to sustain both fast growing heterotrophs and slow growing MOB. Higher productivity and organic carbon turnover rates, as shown by higher CO2 production in lakes with observable CH4 oxidation (Table 2), supports this view. In addition, CH4 oxidation likely contributed substantially to the increase in total CO2 production over the incubation time in CH4 oxidizing incubations.
A methanotroph-methylotroph relationship was observed in our incubations, as methylotrophs from the Methylophilaceae family were identified as an indicator in lakes with CH4 oxidation (Table 3) and since their relative abundance increased toward the end of the incubation in these lakes (Figure 3). The relationship between methanotrophs and methylotrophs may be beneficial, since it has been suggested that high methanol concentrations may inhibit methanotrophs and co-occurrence with methylotrophs able to assimilate methanol would thus reduce this compound to levels that would enable the methanotrophs to thrive [Anthony, 1982]. To what extent this happens in the natural environment, with nutrient patches and micro niches, is difficult to predict and requires further investigation.
In addition to phosphorus availability and microbial community interactions, other factors may have further enhanced MOB growth and subsequent CH4 oxidation, as e.g., Lake Erken had the highest PO43− at the start of the incubation but did not have the highest rate of CH4 oxidation. Rather, Lake Malstasjön had the highest rate of CH4 oxidation and this was also manifested in higher relative MOB abundance (Figure 3). Among the lakes studied by Bastviken et al. [2002] biomass of MOB has previously been shown to correlate positively with CH4 oxidation [Sundh et al., 2005], a result that is similar to the observations in the present study. Besides the nutrients measured in this study (PO43−, NO3−, SO42−), there are several other elements that may play a role in the enzyme activity of MOB. For example, the methane monooxygenase enzyme is limited by copper availability [Semrau et al., 1995] and in the carbon assimilation pathway the hexulose phosphate synthase enzyme is enhanced by magnesium and manganese [Ferenci et al., 1974].
As well as nutrient effects, different groups of MOB are shown to be active in different environments and such biogeographical patterns may also affect CH4 oxidation. MOB can be divided into three functional groups depending on the formaldehyde fixation pathway (type I, type II, and type X—a subgroup of type I [Bowman, 2006]). Metabolic specialization in MOB to different temperature optima between population subgroups, e.g., type I and type II [Hanson and Hanson, 1996] has been proposed, which could make them particularly sensitive to sudden temperature and environmental changes. In the three CH4 oxidizing lakes we found that the same type I MOB, Methylococcaceae, dominated at the end of the incubation. This agrees with previous studies that have detected type I MOB more often in lakes with low temperature [Sundh et al., 2005], while type II MOB have been reported in tropical lakes [Dumestre et al., 2001].
A multitude of factors can influence wintertime CH4 oxidation, and since it is so far not known how these factors interact, further studies are needed. Specifically, multifactorial experimental studies would be beneficial to resolve the constraints of nutrient limitation on wintertime CH4 oxidation across different lakes by enriching lake water incubations with different amounts of nutrients to determine if there is a nutrient limitation threshold. In vitro studies have shown that cultures of methanotrophs can either be enhanced or inhibited by the addition of micronutrients [Boiesen et al., 1993]; however, the extent to which these dependencies translates to environments where complex bacterial communities also will react to changes in nutrient additions needs further investigation. Complex interactions between different factors such as CH4 and nutrient availability as well as temperature should then also be considered.
To conclude, factors other than CH4 availability and temperature limited wintertime CH4 oxidation in our lake water incubations. Based on our results, we propose that phosphorus availability, at least partially, limits MOB growth and thus also CH4 oxidation. We also suggest that interactions between different bacterial groups may play a role as indicated by observed community changes. Overall, this suggests that the potential for CH4 oxidation below ice varies between different lakes and likely depends on nutrient availability and bacterial community interactions. In a broader context, the presence or absence of wintertime CH4 oxidation is a critical factor for accurately estimating CH4 emissions during ice melt, which have been suggested to be important. Our study, therefore, indicates that CH4 emissions could vary between lakes, as nutrient poor lakes may have lower wintertime CH4 oxidation compared to nutrient rich lakes. Future changes in climate and land use may alter the availability of both nutrients and other microbial substrates in lakes. For this reason, it is important to have a solid understanding of linkages between abiotic drivers, the microbial systems, and CH4 oxidation, not only for understanding present day CH4 dynamics below ice but also for predicting future CH4 emissions from seasonally ice-covered boreal lakes.
Acknowledgments
We thank Jan Johansson for the help with field sampling. We thank Henrik Reyier and Lena Lundman for laboratory support at Linköping University, Sivakiruthika Natchimuthu for GC support, Omneya Ahmed for amplification and barcoding work, and Christoffer Bergvall for Ion Cromatography expertise. This work was funded by the Malméns foundation (grants to M.R.C. and B.D.), the Swedish Research Council, and the Swedish Research Council FORMAS (grants to S.B. and G.W. for the overall project and to D.B. regarding analyses and laboratory support at Linköping University), the NordForsk approved Nordic Centre of Excellence “CRAICC” (grant to G.W.), and Carl Tryggers stiftelse (grant to A.E.). DNA sequencing was performed at the SciLifeLab SNP/SEQ facility hosted by Uppsala University, and bioinformatic analyses were carried out using high-performance computing resources made available via Uppsala Multidisciplinary Center for Advanced Computational Science (UPPMAX). Raw sequence data have been deposited at the Sequence Read Archive (SRA) in GenBank® under accession number SRP063993. All other data referred to in this article can be obtained by contacting the corresponding author at [email protected].
