The sediments and the extracted porewater used in this study were from the methane-generating depth (below 20 cm from the water–sediment interface from station A). They are mostly carbonatic-clay and contained 7% iron oxides (Vigderovich et al., 2022) and about 3% total organic carbon (TOC). Dissolved sulfate concentrations decrease from about 0.5 mM at the surface sediment to depletion around 10 cm depth, where dissolved Fe(II) appears and increases with depth up to 80 μM at 30 cm depth. Dissolved methane concentrations increase with sediment depth, reaching a maximum of 2 mM at 10–15 cm depth. The concentrations decrease then to 0.5 mM at 30 cm depth. The dissolved organic carbon (DOC) concentrations in the porewater increase with sediment depth, from ~6 mg C L⁻¹ at the sediment–water interface to 17 mg C L⁻¹ at 25 cm depth (Adler et al., 2011; Sivan et al., 2011; Bar-Or et al., 2015).

Sediment cores were collected using a gravity corer on four day-long sampling campaigns (Table 1), on the research vessel Lillian, between 2017 and 2021 from station A in the center of the lake (water depth 42 m). In each campaign, 1–2 Perspex cores of 50 cm long were collected for the incubation experiment, and another 10 cores were collected for porewater extractions. For the porewater extraction, sediment from the methane-generating zone (sediment depth > 20 cm) of each core was transferred to a 5 L plastic container onboard. The cores and the container were brought to the lab, and the cores were kept at 4°C, while porewater was extracted on the same day by centrifugation as described in Vigderovich et al. (2022).

Five experiments are presented in this study, one of them (experiment C) is a slurry experiment set up with freshly collected sediment and is described below. The other four experiments (A, B, D, and E) are long-term two-stage slurry experiments (details in Vigderovich et al., 2022). In short, sediments from the methane generating zone of the collected cores were transferred within 48 h of their collection, under anaerobic conditions, to a 250 mL pre-autoclaved glass bottle and diluted with porewater extracted from the methane generating sediments to reach a 1:1 sediment-to-porewater ratio (pre-incubated slurry). The incubations were flushed with N₂ (99.999%, MAXIMA, Israel) and methane (¹²CH₄ + ¹³CH₄, 99.99%, MAXIMA, Israel and 99%, Sigma-Aldrich, respectively) was injected into the incubations to reach 20% of the headspace. After at least 3 months of incubation, sub-samples (18–20 g each) from each pre-incubated slurry were transferred under a laminar hood, with continuous flushing of N₂ to 60 mL pre-autoclaved glass bottles. The slurry was then diluted with filtered (0.22 μm) fresh anoxic porewater from the same depth as the sediments to reach a 1:3 sediment-to-porewater ratio. All experiments were kept in the dark at 20°C. The bottles were shaken before every porewater sampling point, before oxygen measurements, and after every oxygen injection. This was to ensure a homogeneous distribution of the oxygen and the dissolved constituents. We describe below each experiment; details summary of the experiments can be found in Table 1 and Figure 1.

In experiment A, seven sub-samples of pre-incubated (set-up in August 2017) slurry were transferred to seven 60 mL experiment glass bottles. Hematite (Sigma-Aldrich, <5 μm, 99%) was added to six of the bottles to reach final concentration of 10 mM, as was done previously in Vigderovich et al. (2022). Each slurry was further diluted with fresh, filter-sterilized, and anoxic porewater and was crimped-sealed. The final headspace volume in the experiment bottles was 20 mL. The bottles were flushed with N₂ for 5 min, shaken vigorously, and flushed again thrice (Sivan et al., 2014) to confirm anoxic starting conditions. This was verified with an optical oxygen sensor (details in the analytical methods below). The killed control bottle was autoclaved twice, cooled and only then hematite was added to the killed control bottle. Finally, 1.5 mL of methane was injected into all the experiment bottles (1 mL ¹²CH₄ + 0.5 mL ¹³CH₄) to reach final concentration of 7.5% methane in the headspace. The experiment consisted of four treatments, 0% O₂ + hematite, 1% O₂ + hematite, 0.1% O₂ + hematite, and killed control with 1% O₂ + hematite (“% O₂” refers to the oxygen concentrations in the headspace). The treatments were set up in duplicates. The duration of this experiment was 70 days. During that time porewater samples were taken for dissolved Fe(II) concentrations and δ¹³C-DIC analyses. Oxygen gas (99.999%, MAXIMA, Israel) was injected into the specific bottles once a week and the oxygen levels in the headspace were monitored. A sample was taken from the 1% O₂ treatment at the start of the experiment and after 52 days for metagenome analysis. The experiment bottles were kept after the experiment ended, and a sample for lipid analysis was taken after 558 days.

Experiment B was set-up from pre-incubated slurry (set-up in September 2018), similarly to experiment A. It consisted with the seven following treatments ¹³CH₄, 10 mM hematite, hematite + BES, 0.3% O₂ + hematite, 1% O₂ + hematite, 20% O₂ + hematite, 1% O₂ + hematite + BES, and killed control + 1% O₂ + hematite + BES. 1 mL of BES stock solution (0.8 M) was injected into the specific treatment bottles to reach a final concentration of 20 mM, as was previously shown to inhibit the AOM in these sediments (Bar-Or et al., 2017). The final headspace volume of the bottles was 24 mL. The treatments were set up in duplicates. Due to a mistake, ¹³C-labeled methane was injected into all experiment bottles in two pulses, 200 μL in the beginning of the experiment and another 1 mL after 21 days, reaching 5% methane in the headspace. It should be noted that in this experiment (and in experiment D) only ¹³C-labeled methane was injected into the bottles, in a lower volume than experiment A. The methane concentrations were enough to sustain methane oxidation, and the higher labeling resulted in a faster ¹³C-labeling of the DIC in the bottles without oxygen. During the experiment time of 111 days, water samples were taken for dissolved Fe(II) concentrations and δ¹³C_DIC. In addition, the oxygen levels were monitored in the headspace, and oxygen gas was injected into the specific bottles once a week.

Fresh sediments for experiment C were collected from the methane generating zone of a core collected in March 2021, were cut under anaerobic conditions into a zip lock bag, and 7 g of sediment sample was transferred to eight pre-autoclaved 60 mL glass bottles under a laminar hood. Hematite was added to all bottles (final concentration of 10 mM). Fresh, filter-sterilized, and anoxic porewater from the same depth as the sediments was added to reach a 1:3 sediment-to-porewater ratio. The final headspace volume was 32 mL. The bottles were crimped-sealed and flushed with N₂ for 1 h and stored in the dark at 4°C for 5 days. To the headspace of each bottle, 1.5 mL of air was injected to reach 1% oxygen in the headspace. 1.5 mL of methane was added to five of the bottles, and 1.5 mL of N₂ to the rest. Methane concentrations in the bottles headspace were 4.2%. The experiment consisted of two treatments: hematite + O₂ + CH₄ in a N₂ headspace and hematite + O₂ in a N₂ headspace. Oxygen concentrations were monitored in the headspace daily and when the levels depleted, the bottles were flushed with N₂, subsequently, air was reinjected to all bottles, and CH₄ to the relevant treatment bottles. The experiment bottles were sampled for dissolved Fe(II) concentrations and metagenome analysis (not presented here). The experiment’s duration was 37 days.

Experiment D was set-up from a pre-incubated slurry (set-up in December 2018) in the same fashion as experiment B and consisted of four treatments, ¹³CH₃OH, ¹³CH₄, killed control +¹³CH₃OH, and live control (no additions). For the methanol labeling, a stock of 101 mM concentration of ¹³C-labeled methanol was prepared. Then, 0.2 mL of the stock was injected into the relevant experiment bottles. The final ¹³CH₃OH concentration in the bottles was 0.56 mM. The killed control bottles were autoclaved twice and cooled, only then ¹³CH₃OH was injected into them. The final head space volume in the experiment bottles was 24 mL. One milliter of ¹³C-labeled methane was injected into the relevant treatment bottles (4.2% methane in the headspace). All the treatments were set up in triplicates except for the killed control, which was set up in duplicates due to the limited amount of the original 1:1 slurry that was used for this experiment. The duration of this experiment was 147 days, in which porewater was sampled for dissolved Fe(II) concentrations and δ¹³C_DIC, and the headspace was sampled for methane concentrations and δ¹³C_CH4. Two samples for 16S rRNA amplicon-based sequencing were taken from the unamended slurry at the beginning of the experiment (as t₀), and another sample, after the experiment ended, at day 459 of incubation from the ¹³CH₃OH treatment. Samples for lipid analysis from the ¹³CH₄ and ¹³CH₃OH treatments were taken after 462 days.

Experiment E was set-up from a pre-incubated slurry (set-up in December 2018) similarly to experiment D and consisted of five treatments, live control (no additions), ¹³CH₃OH, ¹³CH₃OH + BES, ¹³CH₃OH + hematite, and hematite. Each treatment was set up in duplicates. Hematite (Sigma-Aldrich, <5 μm, 99%) was added to reach a final concentration of 10 mM. BES was added to the bottles as in experiment B, and ¹³CH₃OH was added as in experiment D. The final head space volume in the experiment bottles was 24 mL. All treatments were set up in duplicates. The duration of this experiment was 129 days, in which porewater was sampled for dissolved Fe(II) concentrations and δ¹³C_DIC, and the headspace was sampled for methane concentrations. Samples for 16S rRNA amplicon-based sequencing were taken from the unamended slurry at the beginning of the experiment and after the experiment ended, at day 284 of incubation, from the ¹³CH₃OH + BES and ¹³CH₃OH + hematite treatments. Samples for lipid analysis from the same treatments were taken after 287 days.

We measured the change in Fe(II) concentrations after the addition of methane or nitrogen in experiment C in four time points (after 0, 11, 23, and 37 days). To assess the change over time, we performed a separate statistical analysis for both treatments, via one-way repeated measures ANOVA. To analyze the differences between treatments, Two-way repeated measures ANOVA was performed. Post hoc tests were performed by pairwise t-tests, with Bonferroni correction for multiple testing. To achieve normality Fe(II) concentrations were Log10 transformed. Analysis was performed in R using the rstatix package.

Three incubation experiments (A, B, and C) tested how exposure to limited amounts of oxygen affects the methanotrophy and iron reduction in methane-generating sediments of Lake Kinneret. In experiment A, oxygen was injected repeatedly to reach two final concentrations (1 and 0.1%) in the headspace. The δ¹³C_DIC values of the 1% O₂ treatment increased intensively and reached up to 2,500‰ (Figure 2A; Supplementary Table S1). In the 0.1% treatment, the values increased by 37‰ during the experiment. The values of the bottles without the addition of oxygen also increased during the experiment, but only by 19‰. The average initial dissolved Fe(II) concentrations were 20 ± 8 μM. The highest Fe(II) concentrations change was in the 1% O₂ treatment (Figure 2B; Supplementary Table S2). The change in Fe(II) concentrations of the control and the 0.1% O₂ treatments were 16 and 11 μM, respectively. Using the more frequent O₂ measurements in the beginning of the experiment, the oxygen consumption rate was calculated to be 0.03% O₂ per g sediment per day (Figure 2C).

To corroborate the findings obtained in experiment A and to extend our comprehension of the impact of oxygen on iron reduction, experiment B (Figure 3; Supplementary Figure S1) was set up. Oxygen was injected to reach three different concentrations of 0.3, 1, and 20% in the headspace. The δ¹³C_DIC values of the 20% O₂ treatment were the highest (>10,000‰), and the 1% O₂ treatments, with and without BES, reached approximately 2,000‰ after 60 days (Figure 3A; Supplementary Table S3). Both treatments were not measured for δ¹³C_DIC after 60 days because of a memory effect of the IRMS (due to the very high isotopic values). The δ¹³C_DIC values of the 0.3% oxygen treatment reached 2,381‰ at the end of the experiment. The isotopic values of the ¹³CH₄-only and hematite treatments increased as well and reached 261 and 210‰ (respectively) by the end of the experiment. The δ¹³C_DIC values of the hematite and BES treatment did not change throughout the experiment (Supplementary Figure S1). The average initial dissolved Fe(II) concentrations were 40 ± 6 μM. The change in the Fe(II) concentrations increased in all treatments, except for the 20% O₂ treatment which remained the same with low concentrations (Figure 3B; Supplementary Table S4). The highest change was in the 1% O₂ treatment with the addition of BES, then the treatment with 1% O₂ without BES, then 0.3% O₂ treatment. Dissolved Fe(II) concentrations of the treatments without oxygen increased as well, however much lower than the treatments to which oxygen was injected.

The third experiment (experiment C) used fresh sediments and consisted of two treatments, one with a headspace of N₂ and methane and the other with a headspace of only N₂. Air was injected into both treatments to reach a 1% O₂ concentration in the headspace and reinjected each time the oxygen depleted. The average initial dissolved Fe(II) concentrations were 27 ± 3 μM. Fe(II) concentrations measured in the treatment with methane addition did not alter significantly throughout the experiment. A slight non-significant average decrease (of 8 μM) was noted between day 0 and day 11. Final concentrations after 37 days were 24.3 ± 5 μM. In the treatment without methane, concentrations decreased significantly (F = 122.1, p = 9.1×10⁻⁶). Fe concentrations were lower at the end of the experiment after 37 days by 22.7 μM (t = 11.1, p = 0.032) and stood at 4.5 ± 1.2 μM. Similarly, post hoc pairwise comparisons indicate concentrations were higher at day 0, compared to the other time points (Figure 4; Supplementary Data). Two-way repeated measured ANOVA showed a significant correlation between time and treatment (F = 73.7, p = 4 × 10⁻⁵). Post hoc analysis revealed Fe(II) concentrations did not differ at day 0. Subsequently, at the following time points (11, 23, and 37 days), concentrations were significantly and consistently lower at the treatment without methane (Figure 4; Supplementary Data).

Metagenomic analysis was performed on the 1% O₂ + hematite treatment at t₀ and after 52 days. The results show an increase in the relative abundance of the methanotrophic bacteria Methylomonas (5.4%), Methylobacter (2%), and the methylotrophic bacterium Methylotenera (1.5%) during the experiment (Figure 5A). A small increase was also observed in the relative abundance of Desulfuromonas (0.02%) and Geobacter (0.04%). In this treatment, a decrease of 0.2 and 2.7% in the relative abundance of archaea Methanothrix and the Methanofastidiosales order (respectively) was observed.

Isotopes and concentrations analyses of PLFAs during the oxygen experiment provide a base for calculating the production of newly formed fatty acids. Our analysis shows a pattern indicative of aerobic methanotrophs being highly stimulated (Figure 5B), with new ¹³C production ranging between 4 and 20 ng ¹³C/g dw observed for dominating C_14:0, C_16:1ω7, C_16:1ω5, and C_16:0. This PLFA pattern matches the one provided by Bar-Or et al. (2017) (Supplementary material), but the new production is four times higher during active addition of oxygen_, as performed here.

We explored the involvement of methanol in methanotrophy and iron reduction, due to our former suggestion of potential methane activation by archaea and the release of available intermediates to the methanotrophs (Bar-Or et al., 2017). Of those, methanol would be the most probable candidate. During those anoxic slurry experiments, we observed an inhibition of the overall process by BES addition. Using ¹³C-labeled methanol, two incubation experiments (D and E) tested the involvement of methanol. In experiment D (Figure 6; Supplementary Figure S2), the methane concentrations increased in all treatments except for the killed control. The highest change in methane concentrations was observed in the methanol treatment (6.2 μmol/g dw). The change in methane concentrations was the same in the no-addition and the methane treatments (2.6 μmol/g dw; Figure 6A). The δ¹³C_CH4 values in the ¹³C-labeled methanol treatment reached the highest value (67,000‰) compared to the no-addition value (−18‰) and the killed control (300‰; Figure 6B). Throughout the experiment δ¹³C_DIC values of the ¹³C-labeled methanol treatment increased to ~2,500‰ in, and to 260‰ in the ¹³C-labeled methane treatment (Figure 6D). The average initial dissolved Fe(II) concentrations were 28 ± 6 μM. The dissolved Fe(II) concentrations increased in all treatments except for the killed control treatment, with the highest increase noted in the ¹³C-labeled methane treatment (13.8 μM; Figure 6C).

In experiment E, BES and hematite were added to the slurries in addition to ¹³C-labeled methanol to test the involvement of methanogens. Methane concentrations increased in all treatments except for the methanol + BES treatment. The methanol and the methanol + hematite treatments show the highest change of 5.7 μmol/g dw, while the no-addition and hematite treatments increased by 3 and 2 μmol/g, respectively (Figure 7A). The δ¹³C_DIC values in the methanol and the methanol + hematite treatments increased during the experiment by 2,389 and 2,409‰, respectively (Figure 7B). The δ¹³C_DIC of the methanol + BES increased by the end of the experiment by 2,809‰; however, the slope during the first 14 days was relatively low compared to the slope between day 14 and the end of the experiment (Supplementary Figure S3). The average initial dissolved Fe(II) concentrations were 50 ± 7 μM. Dissolved Fe(II) concentrations increased by about 7 μM in the hematite and methanol + hematite treatments (Figure 7C). In the methanol + BES treatment, the concentrations increased by 12 μM throughout the experiment. The concentrations of the no-addition and methanol treatments remained the same.

Samples from experiments D and E were analyzed for the isotopic composition of bacterial fatty acids and archaeal-derived isoprenoid hydrocarbons (Table 2; Supplementary Table S5). The δ¹³C values of the archaeol-derived phytane in the ¹³C-methanol treatments with and without hematite addition were 1,600 and 2,300‰, respectively. In contrast, in treatments with methane and ¹³C-methanol plus BES δ¹³C values were − 3 and − 5‰, respectively. Fatty acids mostly indicative of heterotrophic bacteria (iC_15:0, Aepfler et al., 2019) from the ¹³C-labeled methanol treatment were slightly enriched in ¹³C. The enrichment was more pronounced in the treatments with the addition of BES and hematite. In addition, there was a strong ¹³C-enrichment in the fatty acid C_16:1ω7, which can be affiliated with methylotrophic bacteria (Guckert et al., 1991), specifically considering the conditions applied here. The highest value was found in the treatment with BES (up to 850‰), then with hematite, and a small enrichment in the treatment with methanol only. No change in δ¹³C values of fatty acids was observed in the methane treatment. A combination of fatty acid carbon isotope values and their corresponding concentrations were used to calculate the portion of newly produced fatty acids (in ng ¹³C/g dw; Supplementary Figure S7). These patterns are different from that observed for the oxygen experiment which is highly specific for aerobic methanotrophs and shows much higher ¹³C incorporation. Nonetheless, new production is still fairly high in the methanol treatments with new production ranging from 0.5 to 4.5 ng ¹³C/g dw for the most diagnostic fatty acids iC_15:0 and C_16:1ω7. Other fatty acids such as aiC_15:0, C_16:0, and C_18:1ω7 are also showing new production values higher than 1.0 ng ¹³C/g dw, especially when methanogens were inhibited or when iron reduction was stimulated by hematite additions. This trend is also visible using the sum of all newly produced bacterial fatty acids as an indicator of the overall turnover capacity of bacteria during the different treatments (Table 2). The highest amount of new production was found when oxygen is introduced (57.9 ng ¹³C/g dw), whereas the lowest amount was detected during addition of ¹³CH₄ (1.2 ng ¹³C/g dw). Methanol additions resulted in new productions of 7.8 ng ¹³C/g dw, and 20.3 and 16.1 ng ¹³C/g dw, when BES and hematite was used, respectively.

To study the microbial community associated with methanotrophy related to methanol addition, samples supplemented with methanol (experiment D), methanol and BES, and methanol and hematite (experiment E) were sent for 16S rRNA amplicon-based sequencing. Following taxonomic classification, our analysis indicated archaea accounted for a significant percentage of the microbial population of all treatments (between 28.3 and 37.5%, Supplementary File 1), and the remaining reads were assigned to bacteria. All ASVs of both experiments are also presented in Supplementary File 1. A substantially larger number of Bacterial Classes and Orders was noted (experiments D: 129 classes and 228 Orders; experiment E: 112 classes and 199 Orders), compared to Archaea (experiment D: 17 classes and 24 Orders; experiment E: 18 classes and 28 Orders). Supplementary Figure S4 presents the most abundant (>1%) Bacterial Classes and Archaeal Orders in experiments D and E. Six prominent bacterial Classes accounted for about 50% of the bacterial reads, with slight alteration observed from T0 to the methanol, methanol + BES and methanol + hematite additions: Anaerolineae, Dehalococcoidia, Gammaproteobacteria, Sva0485, Thermodesulfovibrionia, and Aminicenantia (Supplementary Figure S4B). Similarly, the six prominent Archaea orders, accounted for over 80% of the archaeal reads, with minor variations: Bathyarchaeia, Methanomicrobiales, Marine Benthic Group D and DHVEG-1, Woesearchaeales, Methanosarciniales and Methanofastidiosales (Supplementary Figure S4A).

We identified methanogens of the orders Methanosarciniales, Methanomassiliicoccales, Methanomicrobiales, Methanocellales, Ca. Methanomethyliales and Ca. Methanofastidiosales. Methanogens of the genus Methanomethylovorans (0.6% of total reads), were only detected after methanol addition to the slurries, compared to t₀ of experiment D (Supplementary Figures S5A, S6A). Similarly, the genus Methanosarcina (2.4% of total reads) was predominantly found after methanol addition. An increase in the relative abundance of Ca. Methanomethylicus was noted (from 0.16 to 0.32% of total reads). In contrast, the relative abundance of the Methanomassiliicoccaceae family was reduced (from 0.38 to 0.28% of total reads) in the presence of methanol. Slight decreases were observed after methanol addition in the genera Methanosaeta (2.6–1.9% of total reads), Methanoregula (1.36–1.5% of total reads), and Methanolinea (3.2–2.75% of total reads). As expected, after BES addition, the relative abundance of all the methanogens decreased compared to t₀ of experiment E, excluding Methanosarcina which was only observed after BES addition (Supplementary File 1).

Aerobic methanotrophs were also observed in both experiments (Supplementary Figures S5B, S6B). The Gammaproteobacteria family Methylophilaceae, comprising type I aerobic methanotrophs (Deng et al., 2019), was detected at low relative abundance (0.02 and 0.05% of total reads) after the addition of methanol. The family Methylococcaceae (class Gammaproteobacteria), which is also comprised of type I methanotrophs (Taubert et al., 2019), was found in all samples of both experiments. After methanol addition, Methylococcaceae relative abundance decreased (from 0.28 to 0.21% of total reads). An increase (from 0.2 to 0.3%) was noted after the addition of hematite. Methylocystis (Alphaproteobacteria), a type II strictly aerobic methanotroph (Bowman, 2006; Belova et al., 2011), was observed in both experiments at low relative abundance (between 0.04 and 0.08% of total reads). Although Methylocystis relative abundance did not change significantly after methanol addition, when methanol and hematite were added, it increased compared to t₀ (from 0.05 to 0.08% of total reads). Methanol addition also increased the relative abundance of bacteria capable of iron-reduction: ca. Omnitrophus (from 1.1 to 1.64% of total reads), Anaeromyxobacter (from 0.34 to 0.47% of total reads), Desulfuromonas (from 0.07 to 0.15% of total reads) and Thermoanaerobaculum (from 1.44 to 1.57% of total reads). Even more profound increases were observed for most of the aforementioned iron reducers in the presence of hematite (Supplementary Figures S5C, S6C, Supplementary File 1).

Methane oxidation in Lake Kinneret sediments has been observed in pore-water profiles, models, on-top core, and incubation experiments; however, the observed oxidation was considered anaerobic due to the anoxic nature of the sediments (Adler et al., 2011; Sivan et al., 2011; Bar-Or et al., 2017; Vigderovich et al., 2022). Nevertheless, evidence for aerobic methanotrophy was presented in different microbial profiles (Bar-Or et al., 2015) and incubation experiments (Bar-Or et al., 2017; Elul et al., 2021) of Lake Kinneret sediments and, similarly, in other highly reduced freshwater environments around the world (Blees et al., 2014; Milucka et al., 2015; Oswald et al., 2016b; Martinez-Cruz et al., 2017; Van Grinsven et al., 2020, 2021; Su et al., 2022). This may be due to the slow release of remnant oxygen from clay-containing sediments, as proposed in Wang et al. (2018), or due to the continuous production of low oxygen levels in the anoxic environment, which is immediately used by methanotrophs and thus does not poison the anaerobes (Ettwig et al., 2010; Dershwitz et al., 2021; Kraft et al., 2022). An alternative explanation may be the potential survival of methanotrophs performing anaerobic metabolism under hypoxia conditions (Kits et al., 2015; Orata et al., 2018; Zheng et al., 2020; Cheng et al., 2022). In Lake Kinneret, aerobic methanotrophs and Fe-AOM co-occur at the same depth, raising the possibility that iron reduction is somehow associated with aerobic activity. Here we explored the potential aerobic methanotrophy in hypoxic methane-generating sediments of Lake Kinneret and assessed its influence on iron reduction.

The link between aerobic methanotrophy and iron reduction was investigated by repeated oxygen injections into three sets of initially anoxic slurry incubation experiments with/out of methane in the headspace and with/out inhibition of methanogenesis. Our results indicate first that the aerobic methanotrophs in these sediments can be activated by even small oxygen levels, which means that they are in a dormant-like state or possess the ability to survive under anoxic conditions by using other electron acceptors as was previously shown with related methanotrophs (Kits et al., 2015; Orata et al., 2018; Zheng et al., 2020; Cheng et al., 2021).

Second, interestingly, the net iron reduction increased along with the oxygen levels, up to 1% O₂ treatment. This increase was unexpected since dissolved Fe(II) was thought to be oxidized quickly when oxygen is introduced to the system, as can be seen in the 20% O₂ treatment (Figure 3). Generally, Fe(III) reduction is considered to occur mainly under anoxic conditions, where it acts as an electron acceptor instead of oxygen, and the produced Fe(II) is stable (Straub et al., 2001; Lovley et al., 2004). Nevertheless, iron reduction under aerobic conditions has recently been demonstrated in pure culture of iron reducers (Zhang et al., 2019). Zhang et al. (2019) also noted a delay in Fe(II) oxidation under oxic conditions due to metabolites (citric and gluconic acids) and self-produced-siderophores secreted by the specific iron reducers (Actinobacteria) that were tested. These metabolites can bind Fe(II) to form stable complexes. In our slurries, a net increase in Fe(II) was observed in all the treatments where oxygen was injected. Higher levels of injected oxygen (up to 1%) resulted in higher Fe(II) levels. In experiment C, where fresh sediments were incubated (Figure 4), it seems that there is no Fe(II) build-up without the presence of methane, suggesting that methanotrophy is required to explain the increase in net iron reduction. This implies that the methanotrophs/methylotrophs contribute to the observed iron reduction, perhaps by secreting metabolites that are actively used by iron reducers (as a carbon source, for instance) or by functioning as Fe(II)-binding ligands.

The iron reduction in the methane-generating sediments could be performed by iron-reducing bacteria, but also by methanogens (Sivan et al., 2016) and even by aerobic methanotrophs (Zheng et al., 2020); all three have been found in this depth of Lake Kinneret sediments (Bar-Or et al., 2015). It seems that before oxygen injection, natural iron reduction occurs in anoxic sediments by different iron reducers and perhaps by methanogens (Elul et al., 2021). When oxygen is injected, it is used for aerobic methanotrophy and biotic/abiotic Fe(II) oxidation. As a result, new and highly reactive (less crystalline) iron oxides precipitate. The increase in reactive iron oxides encourages iron reduction when oxygen levels are low and net iron reduction increases. In addition, some of the aerobic methanotrophs (i.e., Methylomonas and Methylosinus species) are known to be able to switch to iron reduction metabolism when oxygen levels are low (Zheng et al., 2020), this contributes to the increase in net Fe(II) concentrations. Experiment C indicates that when methanotrophy is scarce, there is no increase in the net iron reduction. This suggests that dissolved Fe(II) does not only increase due to the presence of oxygen but that methanotrophic bacteria activity is necessary for Fe(II) to accumulate, as was shown by Zhang et al. (2019). This potentially encourages iron reduction due to metabolites or an intermediate release during methanotrophy.

Hematite is considered a less reactive and more stable iron oxide compared to non-crystalline Fe-oxides, such as ferrihydrite, goethite or other iron(hydr)oxide (Poulton et al., 2004). Here, hematite was chosen to be the Fe(III)-oxide added to the slurries due to its stable nature, so that it will not disturb the system and shift it toward iron cycling. In addition, it is found naturally in the sediments and was shown to be the most available iron oxide for Fe-coupled AOM in incubations with Lake Kinneret sediments (Bar-Or et al., 2017). Here, in the hypoxic incubations, where oxygen was injected regularly, an aerobic metabolic pathway for methanotrophy was observed that promoted the iron reduction. It could be that addition of more reactive iron oxide to the experiments would cause more intense iron reduction.

From the results of the metagenome analysis on the 1% O₂ treatment of experiment A, it is evident that the methane oxidation is most likely performed by the type I methanotrophs Methylomonas and Methylobacter (Figure 4A). This is in line with previous observations from Lake Kinneret sediments, where the Methylococales order was detected in a microbial sediment profile and a slurry incubation experiment (Bar-Or et al., 2015, 2017; Elul et al., 2021). These methanotrophs were also noted in other anoxic/hypoxic freshwater environments (Martinez-Cruz et al., 2017; Cabrol et al., 2020; Van Grinsven et al., 2020, 2021; Su et al., 2022). Another aerobic bacterium detected in experiment A is the non-methane-oxidizing methylotroph Methylotenera. This methylotroph is known to co-occur with both Methylomonas and Methylobacter and oxidizes methanol excreted by the latter two as an intermediate during the methanotrophy process (Beck et al., 2013; Oshkin et al., 2015; Cao et al., 2019). The iron reducers Geobacter and Desulfuromonas abundance increased as well. Both are well-documented anaerobic iron-reducing bacteria in aquatic sediments and soils (Lovley et al., 2011; An and Picardal, 2015). As expected, the abundance of most methanogens and ANME-1 did not change during the experiment. The decrease in abundance of Methanothrix and the order Methanofastidiosales during the experiment suggests that ANME and methanogens are not involved in the observed methanotrophy, as indicated by the geochemical results. The results of the lipid analysis of the 1% O₂ treatment are in line with the metagenomic results (Figure 5B). There is a similar pattern of the PLFAs to the one previously observed in an anaerobic incubation experiment (Bar-Or et al., 2017) but given the apparent amount of new production in iC_15:0 (1.3 ng ¹³C/g dw) we can speculate that this is the result of concomitant heterotrophic activity (Aepfler et al., 2019).

To explain the results of the previous anaerobic incubation experiment (Bar-Or et al., 2017), the production of potential intermediates that can be channeled from archaea to aerobic methanotrophs was suggested. As the most probable candidate is methanol, its role was tested in anoxic methane-generating sediments in two slurry incubation experiments. Our results indicate that under anaerobic conditions, methanol additions fueled methylotrophic methanogenesis rather than methanol oxidation, as opposed to the fatty acids analyses of the 1% O₂ experiment (Figure 5B; Supplementary Figure S8) and previous observations (Bar-Or et al., 2017). In general, methanol addition increased methane concentrations and higher δ¹³C_DIC values, originating from methylotrophic methanogenesis. This aligns with the strong ¹³C-enrichment of methylotrophic methanogen lipids (i.e., phytane and phytenes; Table 2; Supplementary Table S5). These isoprenoid hydrocarbons are indirect indicators of archaeol and hydroxyarchaeols produced by methylotrophic methanogens of the order Methanosarcinales (Sprott et al., 1993). Accordingly, the 16S rRNA sequencing results fit the biogeochemical observations and show changes in the relative abundance of various methanogens during the experiment. These ranged from acetoclastic and hydrogenotrophic methanogens, i.e., Methanosaeta, Methanoregula, and Methanolinea (Jetten et al., 1992; Oren, 2014), to methylotrophic methanogens of the genera Methanosarcina, Methanomethylovorans, and Methanomethylicus (Smith, 1978; Ranalli, 1986; Lomans, 1999; Vanwonterghem, 2016; Supplementary Figures S5A, S6A). In the methanol treatment (experiment D), there was just a slight ¹³C-enrichment and hence an incorporation of methanol into bacterial fatty acids (Table 2). Nonetheless, this suggests that methanol addition also stimulated the activity of bacteria, either directly as a substrate or by consuming ¹³CO₂ derived from methylotrophic methanogenesis, as expected in this kind of environment (Dijkhuizen and Harder, 1984).

In experiment E, some slurries were amended with BES and hematite, in addition to ¹³C-methanol, to test the potential of methanol turnover when methanogenesis is inhibited or additional electron acceptors available. Similarly, high δ¹³C_DIC values in these treatments (Figure 7) suggest the direct oxidation of methanol by bacteria, which is also reflected in higher ¹³C-incorporations compared to those observed during the methanol-only experiment (Table 2). Even though the FA C_16:1ω7 is associated with methylotrophs (Guckert et al., 1991), they were not detected in the 16S rRNA sequencing results (Supplementary File 1). Alternatively, this FA could indicate the involvement of aerobic methanotrophs; however, their relative abundance decreased compared to the t₀ treatment (Supplementary Figure S6). Considering the increase in Fe(II) concentrations, these new production patterns of bacterial fatty acids may indicate the activity of iron reducers (Teece et al., 1999; Zhang et al., 2003). When methanogenesis was inhibited, there was a marginal ¹³C-enrichment in phytane and phytenes compared to the original values in the sediment (−32‰, Vigderovich et al., 2022). These results indicate that methylotrophic methanogens very likely outcompete heterotrophic bacteria for methanol in a natural system.

Hematite addition with and without methanol mostly encouraged net iron reduction. According to the δ¹³C_DIC results, hematite does not appear to influence methanol oxidation. However, the isotopic composition of bacterial fatty acids revealed a more complex picture. It seems that the addition of hematite doubled the activity of the heterotrophic bacteria (Table 2), indicating a shortage of electron acceptors in the slurry. The sequencing analysis showed a similar picture, where the relative abundance of several bacteria capable of iron reduction (i.e., ca. Omnitrophus, Anaeromyxobacter, and Thermoanaerobaculum) increased during the incubation time. Similar results were noted for the phylums Zixibacteria and Sva0485. Prior reports identified iron-reducing genes in both members (Treude, 2003; Kerin, 2006; Losey, 2013; Tan, 2019; Garber, 2020; Casar, 2021; Williams, 2021). Interestingly, the type I aerobic methanotroph family Methylococcaceae was found in all treatments; however, it only increased in the methanol and hematite treatment. Members of this family have been previously found in suboxic and anoxic environments, such as lake sediments (Bar-Or et al., 2015; Martinez-Cruz et al., 2017; Elul et al., 2021; Su et al., 2022) and in the anoxic hypolimnion of freshwater lakes (Blees et al., 2014; Oswald et al., 2016a; Rissanen et al., 2021). This increase could indicate a potential metabolism that can sustain these methanotrophs in natural anaerobic environments.