Methane, a significant greenhouse gas, has up to 34 times the global warming potential over 100 years compared to carbon dioxide (Myhre et al., 2013). Paddy fields contribute substantially to atmospheric methane concentrations and release 25–300 Tg of CH₄ per annum (Bridgham et al., 2013), representing 10–20% of global methane emissions (Conrad, 2009; Bodelier, 2011). In the next decades, the land area designated for rice cultivation is predicted to increase even further. Without mitigation measures, this will result in elevated methane emission to the atmosphere.

The microbial community structure of paddy fields is influenced by several environmental and anthropogenic factors. Alteration in microbial community composition in paddy fields have been studied with respect to flooding (Rui et al., 2009), fertilization and straw application (Bao et al., 2016), temperature (Conrad et al., 2009; Noll et al., 2010), rice cultivar and soil type (Conrad et al., 2008), and plant growth stage (Breidenbach and Conrad, 2015). Paddy fields provide a habitat for both aerobic and anaerobic methanotrophs. Aerobic methanotrophs are found in the oxic layers of the soil and in oxic microhabitats of the rhizosphere. Methanogenic archaea, anaerobic methanotrophic archaea and/or bacteria thrive preferentially in the anoxic compartments of the waterlogged soil. The flux of methane to the atmosphere is the net result of production and consumption by methanogenic and methanotrophic microorganisms.

Since the discovery of “Bacillus methanicus” (Söhngen, 1906), aerobic methane-oxidizing bacteria (MOB) have been extensively studied. MOBs were long considered the only microbes capable of oxidizing methane. Currently, MOB belong to the phyla Proteobacteria and Verrucomicrobia (Op den Camp et al., 2009; Semrau et al., 2010). Proteobacterial aerobic methanotrophs inhabit a wide variety of environments, ranging from tundra soil (Dedysh et al., 2004) and arctic permafrost (Liebner et al., 2009) to sewage treatment sludge (Ho et al., 2013). Phylogenetic analyses of both 16S rRNA and the particulate methane mono-oxygenase subunit A (pmoA) gene have classified Proteobacteria into Gammaproteobacteria (Type I methanotrophs) and Alphaproteobacteria (Type II methanotrophs) (Trotsenko and Murrell, 2008; Semrau et al., 2010). Type I methanotrophs belong to the genera Methylosarcina, Methylobacter, Methylomonas, Methylomicrobium, Methylosoma, Methylosphera, and Methylovulum (Type Ia) and Methylococcus, Methylocaldum, Methylogaea, Methylohalobius, and Methylothermus (Type Ib). Alphaproteobacterial MOB belong to the genera Methylocystis and Methylosinus (Type IIa) and the genera Methylocella, Methylocapsa, and Methyloferula (Type IIb) (Dumont et al., 2014; Zheng et al., 2014; Knief, 2015). Aerobic methanotrophs have been detected in several paddy field soils (Ho et al., 2011; Lüke and Frenzel, 2011; Lee et al., 2014), and furthermore, it has been suggested that Type I methanotrophs can likely outcompete Type II methanotrophs for substrates in these nitrogen-loaded environments (Zheng et al., 2014). Compared to the proteobacterial aerobic methanotrophs, the more recently discovered Verrucomicrobia often inhabit more extreme environments with low pH values and/or high temperatures (Dunfield et al., 2007; Op den Camp et al., 2009; Sharp et al., 2014; van Teeseling et al., 2014).

Rice cultivation under waterlogged conditions creates anoxia in the majority of soil compartments and, consequently, provides a suitable habitat for methanogenic microorganisms. Rice maturation with the developed and decaying rhizosphere, releases root exudates that, together with dead roots, provide organic matter for an anaerobic food chain. Oxygen influx to soil occurs through diffusional transport via the aerenchyma and radial oxygen loss of the rice roots (Armstrong, 1971; Li and Wang, 2013). Although traditionally considered strict anaerobes, methanogens have been detected in the rhizosphere and on rice roots in several studies (Chin et al., 2004; Xu et al., 2012; Edwards et al., 2015; Lee et al., 2015). Lee et al. (2015) observed a higher abundance of methanogens in the rhizosphere than in bulk soil (Lee et al., 2015). The methanogens in the rhizosphere may live in non-active roots where no oxygen is released or, alternatively, may be oxygen tolerant and have mechanisms to counteract reactive oxygen radical species, as investigated for Rice Cluster I (RC I) (now known as Methanocella) methanogens (Erkel et al., 2006). The genomes of RC I harbor genes encoding catalase, three different superoxide anion scavengers, superoxide dismutase and two different super oxide reductase genes for oxygen detoxification (Erkel et al., 2006). The up-regulation of catalase genes in response to oxygen exposure has been observed in both Methanosarcina and Methanocella (Angel et al., 2011).

Both acetoclastic and hydrogenotrophic methanogens have been identified in paddy fields. Methanogenic archaea of the order Methanosarcinales derive methane from the methyl group of compounds such as methanol and methylamine, and until now, only Methanosarcina and Methanosaeta are known to use acetate for methane production (Jetten et al., 1992; Costa and Leigh, 2014; Welte and Deppenmeier, 2014). Hydrogenotrophic methanogens belonging to the orders Methanomicrobiales, Methanobacteriales and Methanocellales have been commonly found in paddy fields, with the exception of Methanococcales, which barely have been detected (Watanabe et al., 2010; Lee et al., 2015). Many of these hydrogenotrophic methanogens can use formate as a substrate but are unable to utilize acetate. Archaea belonging to RC I (Methanocella) (Kögel-Knabner et al., 2010), which forms a separate phylogenetic lineage branching between the orders Methanosarcinales and Methanomicrobiales, are considered key methanogens in rice fields. The reaction stoichiometry of methanogenesis (Conrad and Klose, 1999) indicates that acetoclastic methanogens could contribute approximately two-thirds to methane production, consistent with the dominance of acetoclastic over hydrogenotrophic methanogenesis in paddy fields (Krüger et al., 2001).

Previous theories suggesting a decrease in methane flux as a result of direct stimulation of methanotrophs after amendment with nitrogen fertilizers were unable to link observations to the activity of the denitrifying anaerobic methanotrophic bacteria and archaea as these microorganisms, were discovered only recently compared to the aerobic methanotrophs. Nitrite- and nitrate-dependent anaerobic oxidation of methane (AOM) were first described in 2006 in an enrichment culture consisting of archaea distantly related to ANME-2d and of bacteria that consume nitrite as an electron acceptor to oxidize methane anaerobically (Raghoebarsing et al., 2006). This novel denitrifying, methanotrophic bacterium of the candidate division NC10 was named ‘Candidatus Methylomirabilis oxyfera’ (Ettwig et al., 2010). Despite its preference for an anoxic habitat, it is postulated to have an intra-aerobic metabolism. The genome of the bacterium contains all genes of the aerobic methanotrophic pathway and encodes a particulate methane mono-oxygenase complex that can use the O₂ released from nitric oxide for methane oxidation, similar to aerobic methanotrophs (Ettwig et al., 2010).

The genome of ANME-2d archaea was sequenced in 2013 and responsible organism named ‘Candidatus Methanoperedens nitroreducens’ (Haroon et al., 2013). This nitrate-reducing archaeon employs a reverse methanogenesis pathway to oxidize methane. The genomes of three different strains of ‘Candidatus Methanoperedens nitroreducens’ have been published, and the necessary genes for nitrate reduction and the methanogenic pathway have been identified (Haroon et al., 2013; Arshad et al., 2015; Vaksmaa et al., 2017a,b). Nitrite- and nitrate-dependent AOM microorganisms and/or activity have been detected in several freshwater environments, including paddy fields (Vaksmaa et al., 2016; Welte et al., 2016). Recently it was demonstrated that ‘Candidatus Methanoperedens nitroreducens’ can also oxidize methane using iron as electron acceptor (Ettwig et al., 2016).

Besides so far know methanotrophs and methanogens, recent investigations of microbial “dark matter” discovered key genes of the methane pathway to be present in phyla, which previously were not linked to the ability to produce or consume methane. Phylum Bathyarchaeota, renamed from Miscellaneous Crenarchaeotic Group is a deeply branching phylum consisting of 17 sub-groups (Kubo et al., 2012). It is abundant in marine environments but is also found in extreme habitats like hot springs, cold sulfur springs, Polar Regions and in mesophilic habitats like sewage waste, fresh water lakes and paddy fields. Though there are no pure isolates, based on culture independent methods, their function was speculated to be important in the global cycle of carbon (Parkes et al., 2005). To date there are eight different genomes annotated, out of which two BA1 and BA2 are hypothesized to be methane metabolizers (Evans et al., 2015; He et al., 2016).

Majority of previous studies of paddy field microbial communities have focused on either a specific group of microorganisms or environmental or anthropogenic effect on methane emissions or sampling had been carried out at a single depth, hindering direct comparison. The aim of the present study was to explore how the microbial communities in a paddy field are influenced by spatial factors along a depth gradient. The objectives of this study were (i) to characterize the bacterial and archaeal communities in a paddy field soil core by 16S rRNA gene amplicon sequencing with a focus on methane cycle-related organisms; (ii) determine the abundances of total bacteria, total archaea, ‘Candidatus Methanoperedens nitroreducens,’ NC10 phylum bacteria and Bathyarchaeota; and (iii) estimate the anaerobic methane oxidation potential using nitrate, nitrite and iron as electron acceptors at different soil depths.

Paddy fields are a major source of methane emitted to the atmosphere. The flux of methane is controlled by the microbial community present in the soil, particularly by methanogens and methanotrophs.

The vertical profile of the methane gradient included a higher methane concentration in the top 15 cm of the soil core, followed by a drastic drop. At a depth of approximately 28 cm, methane was nearly undetectable. This depth correlates with the interface of annual plowing and undisturbed soil as well as the rice root penetration depth.

The profile of the total abundance of microorganisms along the depth gradient followed the same trend as methane. The highest copy numbers of both bacteria and archaea were detected at a depth of 10 cm, followed by a decrease to 25 cm, after which the microbial population size was a few orders of magnitude smaller. The total bacterial and archaeal population sizes correlate well with previous reports. The total bacterial and archaeal 16S rRNA gene copy numbers in a Chinese paddy field ranged from 1.4^∗10¹⁰ to 2.9^∗10¹⁰ per g_dw and 5.4^∗10⁸-1.7^∗10⁹ per g_dw, respectively (Ahn et al., 2012). In paddy fields in the Philippines, the total bacterial copy numbers and archaeal copy numbers were on the order of 10¹⁰ and 10⁸ per g_dw, respectively (Breidenbach and Conrad, 2015). In paddy fields in India, 9.6^∗10⁹-1.4^∗10¹⁰ bacterial 16S rRNA copies per g_dw and 7.13^∗10⁷-3.02^∗10⁸ archaeal 16S rRNA copies per g_dw were reported (Singh et al., 2012). We recorded maximum bacterial and archaeal abundances of 5.6 ± 1.4^∗10⁹ and 1.0 ± 0.3^∗10⁹ 16S rRNA gene copies per g_dw, respectively.

The rice root system has been described as the key determinant in shaping the microbial community via release of root exudates, decaying roots and organic matter as well as oxygen (Kuzyakov and Blagodatskaya, 2015). Diffusion of oxygen to the soil creates micro-oxic niches for oxygen-dependent microorganisms. We detected sequences belonging to aerobic methanotrophs throughout the soil core. The relative abundance of aerobic methanotrophs was highest at a depth of 10 cm and was twice as high as of that in the surface layer. Along the entire depth gradient, MOB were dominated by Type I Methylococcaceae, followed by Type II Methylocystaceae. Methylococcaceae have been detected in several environments with low oxygen concentration, even tolerating periods of hypoxia (Hernandez et al., 2015). The presence of these aerobic methanotrophs in low oxygen environments, such as the investigated paddy field, could possibly be explained by their denitrifying ability as has been demonstrated for Methylomonas denitrificans, which during hypoxia carries out nitrate reduction and methane oxidation (Kits et al., 2015). Other Methylococcaceae, such as Methylobacter contain in their genome besides respiratory nitrate and nitrite reductases as well genes necessary for dinitrogen fixation (Kalyuzhnaya et al., 2015).

In addition to detecting sequences of well-known proteobacterial aerobic methane oxidizers, sequences belonging to methanotrophic Verrucomicrobia were identified in this study. Detailed phylogenetic analysis revealed the presence of aerobic methanotrophs distantly related (85% nucleotide identity of the 16S rRNA gene) to cultured members of Candidatus Methylacidiphilum. Only a very small number of verrucomicrobial methanotrophs have been detected in ecosystems other than acidic volcanic areas, including paddy field soil (GenBank JF984005.1), forest soil (GenBank JF420089), lake sediment (GenBank GU305773) arctic soil (GenBank HQ462506) and a few other environments. The reported cultivated strains originate exclusively from extreme hot or acidic environments in Italy, Kamchatka or New Zealand. Further studies are needed to determine if the microbes found in less extreme environments also contain pmoA genes in their genome and have the capability to oxidize methane. We hypothesize that there is a niche for these aerobic verrucomicrobial methane oxidizers in less acidic methane-rich environments such as paddy fields.

The translation of 16S rRNA gene sequencing read numbers to copy numbers indicated that the methanogen population abundance was highest at a depth of 10 cm, with 4.5^∗10⁸ copies per g_dw, followed by a decline in abundance to 60 cm, with 2.7^∗10⁴ copies per g_dw. The methanogenic population size determined previously in the same Italian paddy field was 10⁷–10⁸ copies per g_dw (Conrad and Klose, 2006). Compared to other sampling sites, our observed abundances are slightly higher than the previously reported methanogen abundances of 1.1^∗10⁷ or 1.4^∗10⁷ copies per g_dw (Singh et al., 2012) or 10⁴–10⁵ copies per g_dw (Hou et al., 2000). A previous vertical profile study of methanogens identified the highest abundance based on mcrA gene copy numbers at a depth of approximately 20 cm in three Japanese paddy fields, peaking at 10⁷ (Watanabe et al., 2010). Together, these results suggest that the methanogenic zone is located approximately 10–20 cm below the soil surface and co-occurs with the end of the main root system in soil.

The community analysis of methanogens revealed a diverse composition throughout the soil core. The methanogenic community was dominated by Methanosaeta, Methanosarcina, Methanobacterium, Methanoregulaceae, and the RC I cluster (Methanocella), which have also been found previously in temperate climate paddy fields (Conrad and Klose, 2006; Watanabe et al., 2010). The community throughout the core was dominated by the strictly acetoclastic Methanosaeta, followed by more versatile Methanosarcina spp. The sampling time of the soil at the end of the growing season, when most root exudates are released (Aulakh et al., 2001) and the ammonia concentration is highest, may explain the methanogen community structure (Singh et al., 2012). Methanosarcina spp. have been shown to be present during the rice-growing season, whereas during pre-planting, tilling or post-harvest, Methanosaeta were present in lower numbers (Singh et al., 2012), correlating with the lower concentrations of acetate available in the soil (Kruger et al., 2002). In paddy field soil, acetate-dependent methanogenesis (acetoclastic) generally dominates over hydrogen-dependent methanogenesis (hydrogenotrophic), as demonstrated by ¹³C-labeling experiments (Conrad, 1999, 2005; Conrad et al., 2002; Zhang et al., 2016).

The total methane concentration in soil over the course of rice maturation peaks at the flowering and ripening stage (Singh et al., 2012). Previous studies in Italian paddy fields have demonstrated that methane emission rates reach approximately 400 nmol CH₄ per g_dw d^-1 70–80 days after flooding (Kruger et al., 2005) or even approximately 600 nmol per g_dw d^-1 (Conrad and Klose, 2006). We previously observed methanogenic activity of the same paddy field soil of 432 nmol and 358 nmol per g_dw d^-1 without and with the addition of methane, respectively, in incubation assays (Vaksmaa et al., 2016). In the current soil core, the highest methanogenic activity was recorded at a depth of 5–10 cm, with rates of 369 and 444 nmol per g_dw d^-1 with and without the addition of methane. In control incubations in which methane was added, methanotrophic activity was initially observed. After 3 weeks, methanogenesis became the dominant process, with methane oxidation rates identical to those observed in the control treatment without the addition of methane. Furthermore, ferrihydrite added to slurry incubations seemed to stimulate methanogens in the top 20 cm. In those treatments, no initial methanotrophic activity was observed; only stimulation of methanogenic activity was recorded, even when both methane and possible electron donors were supplied. Previous studies in wetlands and paddy fields have demonstrated that the addition of poorly crystalline iron, such as ferrihydrite, has an inhibitory effect on methanogenic activity (Achtnich et al., 1995; Lueders and Friedrich, 2002). In contrast, the addition of highly crystalline iron oxide species of hematite or magnetite stimulates methanogens enriched from paddy field soil via a positive effect on either direct interspecies electron transfer or the availability of diffusive carriers such as hydrogen or formic acid (Kato et al., 2012; Holmes et al., 2017).

Methanotrophic bacteria of the NC10 phylum have been previously detected in paddy field soil based on the 16S rRNA gene or pmoA genes and activity assays with nitrite and methane. Conflicting results regarding the vertical distribution of NC10 phylum bacteria in soil have been reported. Zhou et al. (2014) indicated that the highest abundance of 1.0^∗10⁸ copies per g_dw occurred at a depth of 100–120 cm (Zhou et al., 2014). This finding was supported by a study by Hu et al. (2014) of a Chinese paddy field, in which the highest copy number abundance of 1.5 ± 0.2^∗10⁶ to 4.5 ± 0.3^∗10⁶ copies per g_dw was observed at a depth of 50–60 cm. However, the methane-oxidizing potential of soil slurries amended with nitrite was highest at a depth of 90–100 cm, with values of 1.68 ± 0.03 to 2.04 ± 0.06 nmol of CO₂ per g_dw (Hu et al., 2014). By contrast, in a subtropical paddy field soil core sampled to 100 cm, the abundance of NC10 phylum bacteria was highest at the 0–10 cm depth, with 1.0 ± 0.1^∗10⁵ copies per g_dw, followed by 7.5 ± 0.4^∗10⁴ at 30–40 cm and a subsequent gradual decrease, with no detection at depths of 70 cm and beyond (Wang et al., 2012).

The phylogenetic comparison of the 16S rRNA gene reads obtained from amplicon sequencing revealed that, in the top layers, the 16S rRNA gene reads were assigned exclusively to group B (Ettwig et al., 2009; Welte et al., 2016), with nucleotide identities of 95.6–96.7% to ‘Candidatus Methylomirabilis oxyfera.’ Sequences belonging to Group A of NC10 phylum bacteria were found only at depths of 40 cm and below. This distribution is consistent with previous reports in which 16S rRNA gene sequencing and relative read abundance indicated that these nitrite-dependent AOM bacteria formed the largest subset of sequencing reads among total bacterial reads at depths of 50 cm and 100 cm (Ding et al., 2015). In our activity assays with nitrite, we observed the highest methane oxidation potential of 55 nmol per g_dw h^-1 in samples from 10 to 20 cm, which correlates with the first peak of high abundance of NC10 phylum bacteria. However, all the sequences at 10–20 cm all belonged to group B, for which no methane-oxidizing ability has been demonstrated thus far and needs further investigation.

In addition to detecting nitrite-dependent AOM bacteria of the NC10 phylum, we observed high numbers nitrate-dependent AOM archaea ‘Candidatus Methanoperedens nitroreducens’ throughout the soil core. However compared to NC10 phylum bacteria, these archaea were more abundant at all depths, except 60 cm, where NC10 phylum bacteria outnumbered ‘Candidatus Methanoperedens nitroreducens.’ Sequences classified as ‘Candidatus Methanoperedens nitroreducens’ have been detected previously in paddy fields, including fields in Vercelli, Italy (Lueders et al., 2001; Conrad et al., 2008), Chinese paddy fields (Xu et al., 2012), and Korean paddy fields (Lee et al., 2015) as well as in natural wetlands (Narrowe et al., 2017). We previously quantified and detected ‘Candidatus Methanoperedens nitroreducens’ in an Italian paddy field based on 16S rRNA gene (Vaksmaa et al., 2016) and mcrA gene sequences in high abundance (Vaksmaa et al., 2017a). High relative sequence abundance has also been observed in other paddy fields based on the 16S rRNA gene, with 60% of all archaeal reads classified as ‘Candidatus Methanoperedens nitroreducens’ (GOM Arc I) at a depth of 60 cm in bulk soil (Lee et al., 2015). In a study by Lee et al. (2015), the soil core depth profile exhibited the same trend observed in the current study (Lee et al., 2015). The abundance of ‘Candidatus Methanoperedens nitroreducens’ increased with depth, peaking at 20 cm with 1.8 ± 0.3^∗10⁷copies per g_dw. The activity assays performed with nitrate and methane indicated that the activity was highest at a depth of 5–10 cm (57 nmol per g_dw d^-1), followed by a depth of 10–20 cm (33 nmol per g_dw d^-1). We previously observed a methane-oxidizing potential of 80 nmol methane per g_dw d^-1 in mixed and sieved soil slurry from a depth of 10–20 cm (Vaksmaa et al., 2016). The present study is the first to evaluate potential methane oxidation rates utilizing nitrate as an electron acceptor in different depths of a paddy soil core. Recent research has revealed that ‘Candidatus Methanoperedens nitroreducens’ not only can couple nitrate reduction to methane oxidation but is also able to reduce oxidized metals (Ettwig et al., 2016) and may play in important role in both methane and iron cycling in natural and man-made wetlands (Narrowe et al., 2017).

Finally the large number of Bathyarchaeota observed in this paddy field and other wetland systems is intriguing, and their potential role in methane cycling needs further investigation (Evans et al., 2015; Narrowe et al., 2017). For the microbial community members, which the function is still unknown, we detected Bathyarchaeota to be present throughout soil core with highest abundance at 5 cm with 2.1 ± 1.1^∗10⁸ 16S rRNA gene copies per g_dw. From the total archaeal community, these account for almost 50%. Albeit their unknown function, their high abundance and wide distribution indicates that though the function is unknown they might be relevant microorganisms. Up to date, the members of phylum Bathyarchaeota have been detected in a wide range of habitats from terrestrial to marine, cold and hot temperatures or surface and subsurface environments. Generally they are known to be abundant in marine environments (Teske et al., 2002; Lipp et al., 2008; Kubo et al., 2012). Similarly, other studies showed Bathyarchaeota to be present in freshwater environments (Porat et al., 2010; Li et al., 2012). Previous studies have detected Bathyarchaeota in paddy fields as well (Lee et al., 2014, 2015) with abundance of Bathyarchaeota increasing from 17 to 23% in three different phases of rice cultivation (Breidenbach and Conrad, 2015) and with relative abundance up to 42% in paddy field sub-soils (Bai et al., 2017).

In summary, we observed high diversity of the archaeal and bacterial microbial communities throughout the soil core and determined the methane-oxidation potential with various electron acceptors at several soil depths. This study highlights the usage of various electron acceptors for the AOM process. Our findings provide support for the significant role of ‘Candidatus Methanoperedens nitroreducens’ carrying out nitrate-dependent and/or iron-dependent AOM in paddy fields. NC10 phylum bacteria seem to play a less significant role in AOM in paddy fields. The as-yet unknown functions of members of the Candidatus Methylacidiphilum genus and Bathyarchaeota in paddy field soil will hopefully be explained in studies in the near future. We acknowledge that the small sample size of our study does have its limitations, and future studies should include more samples in order to more accurately estimate the contribution of AOM in paddy fields on a larger scale.

AV and CL designed the research and carried out the fieldwork. EL and GV provided the access to the sampling station, and supported the design and fieldwork. AV carried out the experiments in the laboratory. TvA and AV collected and interpreted the sequencing data. AV, CL, KE, and MJ drafted and finalized the manuscript with input from all authors. The manuscript was checked by a professional peerwith.com editor.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.