Interrogating the methane paradox in freshwater wetland soils: A combined multi-omics

and geochemical approach

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy

in the Graduate School of The Ohio State University

By

Jordan Angle

Graduate Program in Microbiology

The Ohio State University

2018

Dissertation Committee

Dr. Kelly Wrighton, Advisor

Dr. Joseph Krzycki

Dr. Virginia Rich

Dr. Michael Wilkins

Dr. Gil Bohrer

Copyrighted by

Jordan Angle

2018

2

Abstract

The methane paradox - the phenomenon of unexpected biological methane production in oxygenated habitats – has been long-documented in marine waters and more recently inferred in freshwater lakes and soil habitats. In Chapter 1, the two primary mechanisms underpinning this phenomenon are described. I then identify the biomarkers and organisms implicated thus far in methane production from oxygenated habitats. Lastly, the contributions of the methane paradox to site-wide emission estimates and the implications of this process for global methane predictions are discussed.

Chapter 2 presents the first genome enabled understanding of organisms performing the methane paradox in well-oxygenated soils. Oxygenated soils from a freshwater wetland located adjacent to Lake Erie contained significantly higher in situ methane concentrations and nine times greater methanogenic activity than corresponding deeper soils. Metagenomic and metatranscriptomic sequencing resulted in the discovery of

Candidatus Methanothrix paradoxum, a novel acetoclastic methanogen species which accounted for nearly all of the inferred methanogenic activity in oxygenated soils. This oxic surface activity was estimated to contribute up to 80% of site-wide methane fluxes.

Chapter 3 extends the genomic analyses to the larger methanogenic community in the Lake Erie freshwater wetland. Here I first compare methane production potential rates across surface and deep soils and show surface soils typically have greater methane ii production rates than deeper, anoxic soils. Metagenomic analyses demonstrated distinct clades of mcrA sequences from surface and deep soils and metatranscriptomic analyses were used to profile the activity of these sequences in each depth. This chapter also provides a more detailed genome-resolved investigation of the dominant and active methanogen across the site, Candidatus Methanothrix paradoxum. Lastly, the metabolic potential and activity of other methanogen genomes from both surface and deep habitats are initially described.

Chapter 4 is a summary and discussion-oriented chapter focused primarily on how higher water levels at the wetland are affecting dissolved oxygen concentrations and the subsequent effect upon methanogenesis in these soils. First, I leverage publicly-available water level data and preliminary geochemical and methanogenic activity measurements to assess the role of hydrology as a driver of methanogenesis across the site. This prior Fall

(2017), water levels up to 1.4 meters higher than my earlier dissertation sampling events

(Chapters 2 and 3) resulted in the first reported anoxic conditions in surface soils at this site. Methane production potential data collected from these deep soils was higher than in years prior, while surface soils was similar to previous years. This finding, while preliminary, has important ramifications as it suggests that aerobic carbon decomposition in the surface soils may be not be required to “unlatch” more recalcitrant carbon leading to the generation of methanogenic substrates and sustaining methane production in these soils

– a concept expanded upon in Chapter 4. This chapter ends with summarizing unanswered questions throughout the dissertation, including the status of isolation attempts for

iii

Candidatus Methanothrix paradoxum and introducing future efforts to characterize dissolved organic carbon and study anaerobic microsites in the bulk oxic soils.

Cumulatively, this work provides an alternative view the to the paradigm that methanogenesis is only relegated to anoxic portions of the soil column. Findings generated here revealed the identity, activity, and distribution of methanogens along freshwater wetland gradients - insights vital to predicting greenhouse gas flux from these climatically relevant ecosystems. A greater understanding of the microbiology facilitating methane emissions in terrestrial saturated soils could improve the accuracy of methane emissions modeling efforts currently underway.

iv

Dedication

With tremendous gratitude I dedicate this work to my family, especially my mother and father Teresa and Jeff Angle, who continuously encouraged me to explore my scientific interests, pursue my passions, and have supported me beyond measure.

v

Acknowledgments

I would first like to thank Dr. Kelly Wrighton for seeing something in me and giving me the opportunity to join her newly-started lab. She created a challenging, stimulating, and extremely rewarding graduate education experience and it was an honor to grow scientifically and personally under her tutelage. It has been obvious that Kelly’s motivation is the best interest of her students and I’m profoundly appreciative of this. I’m truly proud of what the lab and wetlands project have become, and humbled to have served a role in helping to build them.

I would also like to thank the members of my dissertation committee – Dr. Joe

Krzycki, Dr. Virginia Rich, Dr. Gil Bohrer, and Dr. Michael Wilkins for providing crucial feedback and offering their advice and encouragement throughout my graduate career.

The work I have done would not be possible without the contributions of the entire roster of the Wrighton Lab. I would like to thank Rebecca Daly for being a tremendous source of knowledge and advice; we are all spoiled by having access to a scientist like Reb in the lab. The other graduate students – Lindsey Solden, Garrett Smith, and Mikayla

Borton – have all not only contributed to my scientific body of work immeasurably but have been tremendous friends throughout my graduate career as well.

Outside of our lab, I have had the opportunity to work with and learn from some amazing scientists to tackle challenges on the wetland project, including Dr. Christopher vi

Miller, Dr. Adrienne Narrowe, Dr. Timothy Morin, and Camilo Rey-Sanchez. I would also like to thank Mike Johnston, Kay Stefanik, and Lennel Camuy-Velez for their efforts in executing field research which yielded data found within my dissertation. Additionally, I would like to thank Dr. Kristi Arend and the entire staff of Old Woman Creek for being tremendously accommodating hosts for our research endeavors. During my time at OSU,

I’ve been fortunate to have made many life-long friends who helped me navigate the last five years including Jon Lamb, Chris Phelps, Bob Danzak, and Alan Kessler.

I have been tremendously fortunate to be supported by an amazing group of family and friends. My parents Jeff and Teresa Angle have been nothing but perfect in fostering my educational endeavors. Their constant love, care, and support has given me the strength

I needed to climb this mountain. I would like thank the entire rest of my family, with a special recognition to my Great Aunt Wanda Martin who was critical in encouraging my education and fostering my creativity. I would also like to give special thanks to my close friends Sean, Brandon, Ryan, Kye, and Travis for being the very best friends a guy could ask for as well as Kailey and Keiko for constant companionship. I would also like to thank

Charles, Justin, and Kadie for encouraging me to push myself and go back for the Ph.D.

To all of you, my sincere and heartfelt thanks. I wouldn’t be who I am or have been able to achieve this without you all.

vii

Vita

Personal Information

Current-2013 Graduate Student Researcher and teaching assistant, The Ohio State University Current-2013 Visiting Lecturer, McNeese State University 2013-2010 Instructor of Biology, McNeese State University 2010 Instructor of Biology, Frontier Community College 2009 M.S. Biological Sciences, Eastern Illinois University 2009-2007 Graduate Student Researcher and teaching assistant, Eastern Illinois University 2008 Instructor of Biology, Olney Central College 2007 B.S. Biological Sciences, Eastern Illinois University 2003 Fairfield Community High School

Publications

1. A.B. Narrowe, J.C. Angle, R.A. Daly, K.C. Stefanik, K.C. Wrighton, and C.S. Miller. High- resolution sequencing reveals unexplored archaeal diversity in freshwater wetland soils. Environmental Microbiology, 19: 2192–2209. (2017)

2. J.C. Angle, T.H. Morin, L.M. Solden, A.B. Narrowe, G.J. Smith, M.A. Borton, C. Rey-Sanchez, R.A. Daly, G. Mirfenderesgi, D.W. Hoyt, W.J. Riley, C.S. Miller, G. Bohrer, and K.C. Wrighton. Methanogenesis in oxygenated soils is a substantial fraction of wetland methane emissions. Nature Communications, 8: 1567. (2017).

Fields of Study

Major Field: Microbiology

viii

Table of Contents

Abstract ...... ii Dedication ...... v Acknowledgments ...... vi Vita ...... viii List of Tables ...... xii List of Figures ...... xiii Chapter 1. The methane paradox: methanogenesis in oxic habitats challenges a paradigm ...... 1 1.1 An introduction to the methane paradox ...... 1 1.2 Mechanism 1 - Archaeal methanogenesis: Metabolic pathways and organisms implicated in the methane paradox ...... 6 1.3 Mechanism 1 - Archaeal methanogenesis: Investigating methanogen oxygen tolerance ...... 10 1.4 Mechanism 2 - Bacterial heterotrophy: Genetic pathways and organisms implicated in oxic methane production ...... 16 1.5 Global distribution and ecosystem context for the methane paradox ...... 19 1.5.1 Marine ...... 20 1.5.2 Freshwater ...... 21 1.5.3 Soils ...... 22 1.6 Global contribution to methane flux ...... 24 1.7 Concluding comments ...... 28 Chapter 2. Methanogenesis in oxygenated soils is a substantial fraction of wetland methane emissions ...... 30 2.1 Introduction ...... 30 2.2 Results ...... 31 2.2.1 Methanogens are most active in oxic, surface soils ...... 31 2.2.2 Candidatus Methanothrix paradoxum is present and active in oxic soils ...... 39 2.2.3 Candidatus Methanothrix paradoxum is the dominant methanogen across the wetland and globally distributed across other hydric soils ...... 51

ix

2.2.4 Oxic soil methanogenesis contributes substantially to methane flux ...... 55 2.3 Supplementary Material ...... 57 2.3.1 Supplementary Note 1 – Greenhouse gas emissions and estimates ...... 57 2.3.2 Supplementary Note 2 – Metagenomic and metatranscriptomic analyses ...... 58 2.3.3 Supplementary Note 3 – Comparative Methanothrix genomic analyses ...... 62 2.3.4 Supplementary Note 4 – Candidatus Methanothrix paradoxum biogeography ...... 63 2.3.5 Supplementary Note 5 – Site level scaling analyses ...... 65 2.3.6 Supplementary Discussion ...... 66 2.4 Chapter 2 Methods ...... 70 2.4.1 Field sampling ...... 70 2.4.2 Soil and porewater geochemical analyses ...... 71 2.4.3 Collection of dissolved gasses and greenhouse gas emission ...... 72 2.4.4 Transport and production model ...... 73 2.4.5 Eddy covariance collection and data processing ...... 74 2.4.6 Site level methane budget ...... 76 2.4.7 Metagenomic analyses ...... 77 2.4.8 Meta-Transcriptomic analyses ...... 79 2.4.9 Phylogenetic analyses ...... 81 Chapter 3. Methanogen diversity and function at Old Woman Creek: A genome resolved view spanning spatial and temporal gradients ...... 83 3.1 Introduction ...... 83 3.2 Results ...... 86 3.2.1 Methane production potentials are typically greater in oxic surface soils than corresponding deeper soils ...... 86 3.2.2 An expanded view of the wetland mcrA activity, including deep soils ...... 89 3.2.3 Expanding the known genetic potential and activity of the dominant wetland methanogen – Candidatus Methanothrix paradoxum ...... 96 3.2.4 Examining the established and potential oxygen tolerance mechanisms of Candidatus Methanothrix paradoxum ...... 99 3.2.5 Examining the unannotated yet highly active genes of Candidatus Methanothrix paradoxum ...... 102 3.2.6 Exploring additional potential mechanisms for the survival of Candidatus Methanothrix paradoxum in wetland soils ...... 103

x

3.2.7 The taxonomy, novelty, and genetic potential of other methanogen genomes recovered from the wetland ...... 105 3.2.8 The activity of other methanogen genomes recovered from the wetland ...... 109 3.2.9 Conclusions ...... 113 3.3 Chapter 3 Methods ...... 115 3.3.1 Methane production potential determination ...... 115 3.3.2 Assessing potential function of unknown Candidatus Methanothrix paradoxum transcripts ...... 116 3.3.3 Determining mcrA transcript relative abundance ...... 116 3.3.4 Methanogen genome recovery and quality determination ...... 118 3.3.5 Wetland methanogen genome activity assessments ...... 119 Chapter 4. Future directions in methanogen research at Old Woman Creek ...... 120 4.1 Rising water level at Old Woman Creek and the impact on soil dissolved oxygen concentrations ...... 120 4.2 Preliminary data on the impact of decreased soil dissolved oxygen on in situ methane concentration and methane production potentials ...... 125 4.3 Towards the future: Isolating and fully characterizing Candidatus Methanothrix paradoxum ...... 130 4.4 Towards the future: Unraveling the effects of dissolved organic carbon upon wetland methanogenesis ...... 131 4.5 Summary and concluding comments ...... 133 References ...... 135 Appendix A: porewater collection and concentration protocol ...... 152

xi

List of Tables

Table 1 – Metagenomic and metatranscriptomic sample sequencing data ...... 61 Table 2 – Candidatus Methanothrix paradoxum genome bin characteristics ...... 61 Table 3 – Unique genes detected as transcribed in wetland methanogen genomes ...... 111

xii

List of Figures

Figure 1 – Global methane paradox sites ...... 5 Figure 2 - Pathways to methane production in both traditional archaeal methanogenesis and in bacterial phosphate-scavenging routes ...... 7 Figure 3 - Potential mechanisms for methanogenesis in the presence of oxygen ...... 12 Figure 4 - The species distribution of phnJ transcripts by ecosystem from environmental metatranscriptomes ...... 18 Figure 5 - A pictorial representation of the primary ecosystems where the methane paradox has been implicated ...... 19 Figure 6 – OWC Schematic and Sampling Guide Overview ...... 32 Figure 7 – Methane emission rates and correlation of methanogenic activity to geochemical parameters ...... 34 Figure 8 – Methane concentrations and production rates across soil depths ...... 36 Figure 9 – The relationship between soil and dissolved oxygen concentration and methanogenic activity with depth and ecosites from Summer ...... 38 Figure 10 – Genome recovery and average nucleotide identity reveal a new species of Methanothrix termed Candidatus Methanothrix paradoxum ...... 40 Figure 11 - A concatenated ribosomal tree depicting the phylogenetic placement of the 6 surface soil-acquired Candidatus Methanothrix paradoxum genomes ...... 42 Figure 12 – Evidence that Candidatus Methanothrix paradoxum are similar to genotypes in other environmental metagenomes and metatranscriptomes (S3 tree) ...... 44 Figure 13 – Evidence that Candidatus Methanothrix paradoxum are similar to genotypes in other environmental metagenomes and metatranscriptomes (mcrA tree) ...... 45 Figure 14 – Candidatus Methanothrix paradoxum genes transcribed in oxic soils ...... 46 Figure 15 – Mapping of metatranscript reads to methanogen diversity sampled in the metagenomic dataset shows Candidatus Methanothrix paradoxum are responsible for a majority of mcrA transcripts in oxic soils ...... 47 Figure 16 – Candidatus Methanothrix paradoxum (genome M1) transcript abundance patterns shared across seasons and ecosites ...... 50 Figure 17 – Candidatus Methanothrix paradoxum are dominant methanogens in the OWC surface soil communities based on metagenomic relative abundance analyses ..... 53 Figure 18 – Candidatus Methanothrix paradoxum is globally distributed in a variety of ecosystems ...... 54 Figure 19 – Percent methane generated in ecosites over the season as determined from the diffusion/generation model ...... 56 Figure 20 – Fall and Summer methane production potential (MPP) rates ...... 88 xiii

Figure 21 – Metagenome-recovered mcrA sequence transcription from surface and deep soils, visualized by sequence placement on a RAxML nucleotide tree...... 93 Figure 22 Seasonal, ecosite, and depth-resolved view of metagenome-recovered mcrA sequences ...... 95 Figure 23 – Visualization of Candidatus Methanothrix paradoxum genetic potential and activity in oxic surface soils...... 97 Figure 24 – Methanogen genome transcript activity by season and depth ...... 112 Figure 25 Water depth and dissolved oxygen concentrations at the lower estuary monitoring station, Old Woman Creek, 2014 – 2017 ...... 122 Figure 26 – A comparison of water level and dissolved oxygen concentrations in “open water” soils – Summer 2015 and Fall 2017 data ...... 123 Figure 27 – in situ methane concentrations following increased water levels and decreased dissolved oxygen concentrations ...... 127 Figure 28 – 10-day methane production rates following increased water levels and decreased dissolved oxygen concentrations ...... 128 Figure 29 – Sample schematic of proposed bioreactor design ...... 132

xiv

Chapter 1. The methane paradox: methanogenesis in oxic habitats challenges a paradigm

1.1 An introduction to the methane paradox

This chapter was reproduced verbatim from “J.C. Angle, G. Bohrer, and K.C. Wrighton.

Unraveling the methane paradox: Insights into methane production in oxic waters and soils.

FEMS Microbiology Reviews (in review)”. The text benefited from the writing and editing contributions from co-authors G. Bohrer and K.C. Wrighton.

Methane, a potent greenhouse gas, is generated in numerous habitats globally1.

While both abiotic and biotic methane generation exists, biotic generation by microorganisms accounts for a large percentage of the total methane emitted to the atmosphere2. Historically, biotic methane production was believed to be performed strictly by anaerobic archaea know as methanogens. Based on physiological data demonstrating oxygen toxicity to methanogens, microbial methane production in nature is constrained to reduced, anoxic soils with limited external electron acceptor availability. Assumptions about methanogen oxygen sensitivity extend to global methane emission modeling efforts, which dramatically limit or fail to account for methane production in the presence of oxygen3. More recent landmark research efforts have presented an alternative view, reporting biological methane production from oxygenated fresh4 and marine waters5, and even more recently oxic soils6. More importantly, several studies have shown methane generation in oxygenated environments is a significant contributor to site-wide methane emissions7,8. The term methane paradox is used to describe the process where methane is microbially produced in oxygenated habitats.

1

Methanogens - defined as obligatory anaerobic archaea that produce methane gas as a metabolic byproduct - were thought to be extremely oxygen sensitive9. Based on their substrate usage these organisms are classified into three broad metabolic groups: (i) acetoclastic (acetate), (ii) hydrogenotrophic (hydrogen, carbon dioxide), or (iii) methylotrophic (methanol, methylamines). Some genera of methanogens, e.g.

Methanosarcina, are generalists that can use all substrate types, while many other methanogen genera are specialized for a single substrate10. Methanogens are active across a range of anoxic habitats including marine11,12 and estuarine sediments13,14, freshwater habitats including lakes15,16 and rivers17, and terrestrial soils including rice paddies18, peatlands19, and wetlands20. In these systems, when assuming the same substrate (e.g. hydrogen, acetate), the free energy available from methanogenesis is less than other respiratory processes like oxygen, nitrate, ferric iron reduction, and sulfate21. Moreover, reports show methanogens have lower substrate affinities, further reducing their competiveness for substrates with other respiratory organisms (e.g. sulfate-reducing bacteria)22. Collectively, these thermodynamic explanations are used to justify why methanogenesis is less competitive in scenarios where electron acceptors are present. It is however possible that these constraints are minimized in the case of electron donor excess and microsite variability23–25.

In addition to thermodynamic exclusion inhibiting methanogens in oxygenated habitats, methanogens contain essential enzymes that are not functional in the presence of oxygen. Oxygen sensitive genes include iron-sulfur proteins, flavoproteins, and catecholamines. When oxidized, these enzymes form reactive oxygen species such as

2 hydrogen peroxide, superoxide, and hydroxyl radicals26. These reactive oxygen species can depolymerize nucleic acids and oxidize both polysaccharides and fatty acids to nonfunctional states27. Methanogens also contain enzymes that require reduced iron and nickel cofactors for enzyme catalytic activity and in the presence of oxygen these compounds can also be oxidized, rendering enzymes non-catatlyic28. These metal-cofactor containing enzymes are vital to methane production and include key genes in the methanogenesis pathway including hydrogenases, acetyl-CoA cleaving CO dehydrogenase, and methyl coenzyme M reductase29. Alternatively, genome based analyses of methanogens have suggested these organisms widely encode enzymes to combat this oxidative damage30, potentially allowing for greater oxygen tolerance than was previously accounted for. In the section entitled, “Investigating methanogen oxygen tolerance” we discuss this alternative view in more detail.

In addition to the traditional anaerobic archaeal methanogens, recent research has uncovered an alternative route for methane production in oxygenated marine and fresh waters via aerobic heterotrophic bacteria. Here, bacteria generate methane gas not directly from an energetic pathway but instead as a waste product during phosphate acquisition31,32.

This C-P lyase pathway, which is described in detail in the “Bacterial Heterotrophy:

Genetic pathways and organisms” section, involves the utilization of multiple enzymes to harvest the phosphate group from organic phosphate compounds. These compounds are transported into the cell from the external environment via a transporter complex, where they are cleaved to release diphosphate for cellular use. Following the diphosphate harvest, the organic group is released and diffuses out of the organism, which in the case of released

3 methyl groups provides a source of methane32. Although the genes necessary for the C-P lyase pathway are widespread in many different bacterial lineages, the activity of this pathway is thought to be limited to phosphate-limited conditions or environments. In this chapter, metatranscriptomes from a variety of environments were mined to infer the global distribution and activity of these methane-producing heterotrophic bacteria at the ecosystem scale.

Here, to discern the differences in these two genetic pathways in this review, we refer to canonical methane production by methanogens as methanogenesis and methane production via the C-P lyase pathway as bacterial methane release. Here we summarize the literature where the methane paradox has been suggested to occur, and map these studies by ecosystem category onto a global framework (Figure 1, Supplementary Dissertation

Table 1). From the literature mining we recovered 38 studies dating from 197133 and spanning unique geographical locations that implicate microbial methane production in the presence of oxygen. Of these studies, 37% are reported from marine waters, 39% from freshwaters, and 24% from soils. In both marine and freshwater systems, methane was found to be generated in supersaturated waters34,35, while in soils methane production was reported from incubation studies in the presence of dissolved oxygen concentrations as high as 19%36.

We then go on to summarize the genetic pathways and resistance strategies underpinning the two mechanisms for microbial methane production in oxygenated habitats. Next, we summarize some of the key methane paradox research discoveries organized by ecosystem type. Lastly, we quantify the impacts the methane paradox has on

4 methane emissions across ecosystems. It is our hope that by highlighting the prevalence of methane paradox research across ecosystems, this process can be better quantified at the ecosystem scale, and potentially, if warranted, accounted for in global methane biogeochemical models in the future.

Lake Marine Soil

Figure 1 – Global methane paradox sites Sites where the methane paradox has been implicated worldwide, updated from Tang et al37. Sites where methane production or oxic methanogenesis has been observed are depicted, with color representing the habitat type of the site. Summary of the publications shown on this map are provided in Supplementary Dissertation Table 1.

5

1.2 Mechanism 1 - Archaeal methanogenesis: Metabolic pathways and organisms implicated in the methane paradox

Several marker genes can identify methanogens and their substrate preferences, with the key pathways summarized in this review (Figure 2, Supplementary Dissertation

Table 2). In all archaeal methanogens sampled to date, the heterodimeric methyl coenzyme

38 M reductase (Mcr) enzyme performs the final methane-generating step . The gene for the

A subunit (mcrA) of this enzyme is often used as a biomarker for characterizing the diversity and activity of methanogens in ecosystems39.

6

CO2 CH4 acetate Methyl-CoM acetyl-CoA 5-Methyl-THMPT 1 4 7 10 14 18 21 24 5 8 11 15 19 22 25 2 6 6 9 12 16 20 23 3 13 17

CH4 Methyl-CoM formate Formyl-MFR 5-Methyl-THMPT 10 14 18 21 38 26 28 30 11 15 19 22 39 34 35 36 37 27 29 31 12 16 20 23 40 13 17

32 CO

33 2

CH4 Methyl-CoM compounds methylated 41 42 43 18 21 38

Archaeal Methanogenesis 44 45 46 19 22 39 44 47 48 20 23 40 44 49 50

CH4 α-D-ribose-1-Mpn α-D-ribose-1-Mpn 5-phospho-α-D-ribose- -5-triphosphate 1,2-cyclic-phosphate -5-phosphate

MPn 51 Bacterial 52 54 55 56 57 58 59 60 53 C-P lyase pathway

Figure 2 - Pathways to methane production in both traditional archaeal methanogenesis and in bacterial phosphate-scavenging routes

7

Genes indicative of specific methane production routes are highlighted in color. Acetoclastic methanogenesis-specific genes (highlighted in orange) are cdhABCDE/cooS (#4-9) and hdrDE (#24,25). Hydrogenotrophic methanogenesis genes utilized in carbon dioxide-reduction genes (highlighted in crimson) are ftr, mch, mtd, and mer (#34-37 respectively). Methylotrophic methanogenesis-specific genes (highlighted in purple) are mtaABC, mtbA, mttBC, mtbBC, and mtmBC (#41-50 respectively). Bacterial C-P lyase pathway genes, necessary for phosphate-scavenging and leading to subsequent methane release (highlighted in blue) are phnDECIGHLKMJ (#51-60 respectively). A full list of the genes depicted on this chart is located within Supplementary Dissertation Table 2.

Until recently, knowledge of methanogen diversity was confined to representatives within the Euryarchaeota, however recent metagenomic and single cell studies have indicated that methanogens may extend to other archaeal phyla, including members of the

Bathyarchaeota and Verstraetearchaeota40–42. Notably, many of these newly sampled methanogens also have potentially divergent mcrA genes41, and may be missed in gene surveys using historical mcrA primer sets. This newly sampled diversity may inspire new primers43 or non-amplification based methods for characterizing archaea8 in ecosystems, so that diversity, abundance, and activity of methanogens can be accurately captured across ecosystems.

Here we highlight genes indicative of possible methanogen metabolic strategies.

For instance, acetoclastic methanogens encode a carbon-monoxide dehydrogenase/acetyl-

CoA synthase complex (cdhABCDE/cooS) and the hdrDE complex, necessary for the breakdown of acetate and reduction of the final disulfide compound; respectively10. Genes specific for methylotrophic methanogenesis often include coenzyme M methyltransferases specific for the C1-carbon substrates, including those for methanol (mtaABC), monomethylamine (mtmBC), dimethylamine (mtbBC), or trimethylamine (mtbA)44,45. The 8 hallmarks for hydrogenotrophic methanogens are less distinctive, but some hydrogenotrophic methanogens can uniquely utilize formate, which requires a multi- subunit formylmethanofuran dehydrogenase complex (fwd or fmd genes). Additional caution should be used when inferring substrate use from genomes as some methanogens, e.g. Methanosarcina, can use a range of substrates 46–48. We note that expression data may be required to differentiate active pathways49. Additionally, many methanogenic genes are also present in non-methanogenic archaea and bacteria, so the presence of some of these genes, rather than reconstructed pathways, may not be indicative of methanogens 46.

In scaling from genes to organisms, we surveyed the methane paradox literature for specific methanogen taxa that have been associated with oxygenated habitats. These methanogens are largely implicated on the field scale based on correlative analyses, where methanogens that are in high relative abundance are identified and were inferred to correspond to increased methane production from these samples. Five main genera are often identified in methane paradox studies and include: (i) Methanosarcina6,50,51, (ii)

Methanocella6,50,51, Methanobrevibacter50, (iii) Methanoregula8,4, and (iv) Methanothrix

(formerly Methanosaeta)41,8,51,4. We note, however, that association of these taxa with methane in oxic habitats may be an artifact of their distribution or capacity for detection with current primer sets. Today, studies moving beyond 16S rRNA gene presence and towards deciphering the genome content of active methanogens in oxic habitats are fairly limited at the ecosystem scale. However, with the growing evidence of the methane paradox in the past five years, we anticipate future investigations will more directly link organisms to methane production activity in these unexpected methanogenic habitats.

9

1.3 Mechanism 1 - Archaeal methanogenesis: Investigating methanogen oxygen tolerance

Today all surveyed methanogens are considered strictly anaerobic and oxygen- sensitive organisms. Consequently, in this section we reviewed the literature for enzymes produced by methanogens that may counteract this oxygen toxicity, protecting methanogenic enzymes in oxic environments. We synthesized data from laboratory and field studies where these anti-oxidant strategies were shown to be important to methane production. We also surveyed other methods methanogens may employ to minimize oxidative damage in bulk oxygenated environments.

The capacity for anaerobic organisms to survive in the presence of oxidative stress is not a new concept, with research dating back over 50 years52,53. Early work determined that the most important enzymes for combating oxygen toxicity were superoxide dismutase

54,55 - (Sod), catalase (Kat), and various peroxidases . Sod disproportionates O2 to O2 and

H2O2 while Kat catalyzes the decomposition of H2O2 to O2 and H2O. Work by Kirby et al demonstrated the presence of sod in methanogens56, findings that were extended to other methanogens soon after57–61. Since then, it has been extensively documented that methanogens encode a diversity of oxygen detoxification strategies, research that is best summarized in recent comparative genomic analysis of methanogens by Jasso-Chavez et al30. Across genome-sequenced methanogens, the most common oxygen detoxification mechanisms (ordered by prevalence) are: thioredoxin / thioredoxin reductase (trx and trxr), rubrerythrin (rbr), peroxiredoxin (prx), desulfoferrodoxin (dfx), rubredoxin (rbx), glutaredoxin (grx), peroxidase (px), catalase (kat), superoxide dismutase (sod), F420H2 oxidase (fprA), and cytochrome D oxidase (cyd). Building on these findings, another recent 10 study examined antioxidant features from representative genomes from six well- established methanogen Orders. This study concluded that methanogens belonging to the

Methanocellales, Methanomicrobiales, and Methanosarcinales Orders (e.g.

Methanosarcina, Methanocella, and Methanothrix genera discussed above) have an enrichment of anti-oxidant genes. It is interesting to note that the enrichment of oxygen tolerance genes in these orders relative to other methanogenic Orders corresponds to the more frequent sampling of these taxa in microaerophilic and even oxic environments62.

Physiological studies with a handful of cultivated isolates have demonstrated that specific methanogens can remain metabolically active in the presence of oxygen or oxidative stress (Figure 3). Methanobrevibacter arboriphilus cell suspensions still utilized

FprA to reduce mM concentrations of oxygen63, while Methanobrevibacter cuticularis was capable of simultaneous methane production while reducing local levels of toxic molecular oxygen64. Consistent with these findings, some Methanobrevibacter spp. tolerate aeration and water stress65. In another model methanogen used in oxygen tolerance tests,

Methanosarcina acetivorans cultures were subjected to pulses of up to 1% oxygen for up to 6 months. This resulted in the generation of air-adapted methanogen populations, which were equivalent in methane production and protein content to anoxic control tubes, except for the oxygen exposed cells had greater number of transcripts for oxygen detoxification genes (e.g. sod, kat, and px) and activity of subsequent enzymes30. Additional oxidative stress-testing with Methanosarcina revealed that cell growth was only reduced by 25% in the presence of hydrogen peroxide (up to 1 mM) via the activation of thioredoxin/thioredoxin reductase system66. While it is intriguing that Methanobrevibacter

11 and Methanosarcina used as model methanogens in laboratory investigations are often identified as enriched in oxic environments, the extent to which responses of pure-culture methanogens under laboratory conditions translates to their behavior in natural conditions is still largely understudied.

Figure 3 - Potential mechanisms for methanogenesis in the presence of oxygen A. Laboratory studies have provided evidence for the upregulation of superoxide dismutase (sod), catalase (kat), peroxiredoxin (px), and thioredoxin (trx) in methanogens (depicted as black ovals) under oxidative conditions. Sod and Kat produce the following reactive species O2 and H2O2, and O2 and H2O, respectively, depicted here. B. The presence of these oxygen detoxification mechanisms in numerous methanogen genomes and their activity in laboratory oxidative stress tests has been previously reviewed30, as well as observed in situ

12 activity6,8, all of which are demonstrated by black boxes here. C. A proposed model for avoiding oxic conditions is the localization of methanogens within anoxic microsites in heterogeneous soil, where aerobic heterotrophic organisms coupled carbon oxidation to consumption of oxygen, generating regions with below detectable oxygen concentrations. Additionally, methanogens may be part of biofilms, further minimizing oxygen exposure and subsequent oxidative damage.

Recent investigations from soil reactors under laboratory conditions provided new evidence for methanogen oxygen tolerance in complex communities. In one of the most compelling laboratory examples, despite the presence of up to 21% oxygen, microcosms with desert biological soil crusts produced methane (100-1000 nmol per gram soil over 42- day incubation), which was two log-fold lower than anoxic control soils in the same experiment6. Notably, under oxic conditions methane isotopic signatures determined that methanogenesis shifted from acetoclastic to hydrogenotrophic pathways, concomitant with a decrease in Methanosarcina and an increase of Methanocella. The quantification of kat genes was undisguisable between oxic and anoxic conditions. This is in agreement with the results by Zhang et al67 who failed to measure increased expression of catalase in Methanosarcina barkeri in response to air exposure, but in contrast to those of

Brioukhanov et al68 who reported an increase in catalase transcription in response to oxidative stress. Thus, it seems from comparative genomics, physiology, and enrichment culture experiments, there is no “one size fits all” strategy, as uniform oxygen or oxidative stress responses within or between methanogen species are lacking.

In one of the only field studies we are aware of linking methanogen identity and geochemical measurements of methane production to methanogen activity measurement8 13

(discussed in detail in Chapter 2), mcrA transcription was nine times greater in oxic surface than deep anoxic pore waters. Moreover, one species, Candidatus Methanothrix paradoxum was inferred to be responsible for 84% of the mcrA transcripts in oxic soils over two seasons. Notably, metatranscripts from these same soils failed to identify activity of oxygen tolerance genes in methanogens or other organisms, suggesting oxygen detoxification may not be a requirement for sustaining the methane paradox under natural soil conditions. It was noted that many genes of unknown function and also genes for protein repair were highly transcribed, suggesting yet to be defined oxygen tolerance strategies may be present (discussed in detail in Chapter 3). Based on the failure to detect oxygen tolerance genes, it was posited that methanogenesis may confined to anoxic sites occurring in bulk oxic soils (e.g. biofilms, aggregates, or locally depleted oxygen zones), and the methanogens were not exposed to the bulk oxygen measured in the habitat (Figure

3).

In reviewing the literature, anoxic microsites have been evoked across ecosystems as a plausible explanation for methane production by methanogens in oxic habitats.

Examples of anoxic microsites that could sustain methanogenesis include protection within cells flocs, biofilms, exploiting the natural heterogeneity of soil particles, and living intracellularly inside host symbionts. Research with Methanosarcina barkeri demonstrated that survival during oxygen exposure was aided by the formation of cell flocs69. In granular sludge, methanogen-containing communities were capable of tolerating up to 41% headspace oxygen with only 50% inhibition, with bacterial oxygen consumption presumed to facilitate local anoxic microenvironments, enabling a redox-protected habitat for

14 anaerobic methanogenesis to occur70. More recent work showed that methanogen isolates could tolerate oxygen exposure and drying conditions significantly better when mixed with fresh or sterile soil, suggesting that the methanogens may reside in soil regions protected from oxic conditions65.

The idea that anoxic microsites in oxic soils can support anaerobic metabolisms

(e.g. denitrification and iron reduction) is not novel71,72. For methanogenesis, it was suggested local regions of active heterotrophic respiration form anoxic microsites that sustain anaerobic metabolisms73–75. Additionally, several works outside soil aggregates have demonstrated that anoxic microsites can form within plant detritus particles21,76, as well as in excrement of zooplankton77–82 and fish83,84 in marine systems. Future research directions and challenges will involve the development of more resolved methods for quantifying anoxic microsites and linking these data to measurements of microbial methanogenesis activity.

Thinking beyond free-living methanogens, symbiotic methanogens living intracellularly are also implicated as a source of methane in oxic environments. Protozoa, like ciliates, have long been known to contain endosymbiotic methanogens85 and this host- methanogen relationship has been reviewed thoroughly86,87. Interestingly, a genus of ciliate

Metopus, which supports methanogenesis intracellularly by Methanothrix, produce oxidative protection enzymes (e.g. sod), potentially for their endosymbiont88. There is also evidence to suggest that methanogens are active within earthworm species89,90, however data also exists suggesting that earthworms lower net methane production via aeration of soils91. Extending beyond intracellular relationships, several citations suggest

15 methanogenesis occurs in the presence of algae92,93, with suggestions that marine algae

Emiliania huxleyi produce methane under oxic conditions independent of associations with archaeal methanogens by metabolizing bicarbonate and methionine, as determined by 13C isotope work94. Lastly, research conducted in the Arctic ocean concluded that respiration from local bacteria maintained anaerobic conditions conducive for methane production inside bacterial cells despite enhanced oxygen concentrations in the environment95.

1.4 Mechanism 2 - Bacterial heterotrophy: Genetic pathways and organisms implicated in oxic methane production

Aerobic bacterial methane release via the C-P lyase pathway utilizes a fundamentally different set of proteins to scavenge phosphorus-containing compounds from phosphorus-limited environments, resulting in methane release. In this process, methylphosphonate (or other organic phosphonates) are transported into the cell from the external environment via a transporter complex (PhnCDE) before removal of an adenine

(via PhnIGHL) and cleavage of the C-P bond (PhnM), releasing a diphosphate group for cellular use. Following the phosphate harvest, the methyl group (or another organic group) is released from the intermediate (PhnJ). The methane is now free to diffuse out of the organism32,96. Like mcrA for methanogens, the phnJ gene can be used as a marker gene for the C-P lyase pathway97,98 (Figure 2). It has been demonstrated that 31P NMR can also be utilized to track phosphorylated substrates and subsequent degradation to aid in identifying active C-P lyase pathway31. For aerobic bacterial methane production to occur via the phosphate-scavenging pathway, data from laboratory isolates signifies that the

16 concentration of environmental inorganic phosphate must be relatively low, as the pathway is completely repressed at ~30 uM inorganic phosphate32.

The environmental distribution of the C-P lyase pathway remains relatively undefined. To address this knowledge gap, here we mined environmental metatranscriptomes using reported phnJ reference sequences32,99 [BLAST e-value of 1e-20 to Joint Genome Institute Integrated Microbial Genomes and Metagenomes, data mined on

10/10/17]. Sequences were assessed for similarity to known organisms, site origin, and source habitat (Figure 4). In total, our analysis recovered 202 putative phnJ transcripts

(Supplementary Dissertation Table 3, Supplementary Dissertation File 1). These 202 phnJ sequences were found in 19 metatranscriptomes from various environments: 9 from marine systems (ocean waters – 6, estuary waters – 3), 9 from freshwaters (lakes -8, hot springs waters -1), and one soil (peatland). These results show that transcripts most similar to those from Rhodospirillales and Rhodobacterales are present across a range of habitats, while sequences most similar to Burkholderiales references were found almost exclusively in marine systems. Overall, lake and marine transcripts predominate the 202 transcripts, accounting for 144 of total phnJ transcripts. This increased detection in lake and marine systems may be impacted by the increased metatranscript sampling in these systems.

Linking these sites back to our known methane paradox sites (Figure 1), revealed that these metatranscript locations were not previously implicated in methane paradox publications.

More extensive analyses defining the environmental conditions and organisms that facilitate bacterial methane production, combined with studies determining the

17 contribution of this process to methane emissions, are needed to quantify the significance of this process on ecosystem and global scales.

phnJ environmental transcripts by ecosystem Lake Marine Estuary Hot Spring Peatland

Burkholderiales Rhodobacterales Enterobacteriales Rhodospirillales Oceanospirillales Others Rhizobiales Unknown

Figure 4 - The species distribution of phnJ transcripts by ecosystem from environmental metatranscriptomes The gene phnJ encodes the protein of the C-P lyase pathway necessary for methane gas release. Demonstrated here are the environmentally-active phnJ transcripts recovered from IMG environmental metatranscriptomes via BLAST of phnJ reference sequences32,99. The pie charts depict the relative contribution of transcripts from each Order (distinguished by color) towards the total detected transcripts from a given ecosystem. The complete JGI reference information for each recovered sequence and the sequences themselves are located within Supplementary Data Table 3 and Supplementary Data File 1, respectively.

18

1.5 Global distribution and ecosystem context for the methane paradox

Numerous publications from distinct, globally-distributed habitats have noted and hypothesized biological explanations for methane production in oxygenated environments.

Here we classify these sites as marine, freshwater, and terrestrial to provide a global summary of the methane paradox studies published to date (Figure 1, Supplementary

Dissertation Table 1). A recap of the pathways for methane production by ecosystem is shown in Figure 5.

Soils Freshwater Lakes Marine Waters O O O O 2 2 2 2 O2 O2 MPn CH4 CH4 Archaeal C-P lyase pathway intracellular anoxic methanogens MPn CH4

O X O X Archaeal C-P lyase pathway 2 2 Archaeal methanogens methanogens

Figure 5 - A pictorial representation of the primary ecosystems where the methane paradox has been implicated Traditional archaeal methanogenesis has been reported to occur in both in anoxic regions as well as in bulk oxic soils. Archaeal methanogens are also suggested to be the source of methane in surface, oxygenated lakes. More recently, however, in freshwater lakes with extremely low total phosphorus concentrations (<1 µM), bacterial phosphorus scavenging via the C-P lyase pathway has also been shown to result in methane release100. In oxic marine waters, several mechanisms yield methane production, including the C-P lyase pathway, the creation of anoxic conditions (not pictured), and subsequent methanogenesis within bacterial cells.

19

1.5.1 Marine

The term methane paradox was originally coined in marine systems101, thus a large amount of work has explained methane supersaturation in oxic waters34,35,102. In these systems, it was thought that the lack of other available terminal electron acceptor facilitated methanogenesis in these environments. There are currently competing hypotheses for the source of this methane. Karl and colleagues, in one of the landmark methane paradox papers, demonstrated a microbial metabolism that generated methane in marine oxic zone with oxygen concentrations ~210 uM5. In this study, surface seawater was collected from

Station ALOHA in the Pacific Ocean and used in amendment studies to demonstrate that aerobic bacteria could utilize methylphosphonate resulting in release of methane gas. They also noted the C-P lyase pathway for phosphonate utilization and release of methane had been described in E. coli previously, and screened their marine metagenomes for the presence of phosphonate-utilization gene homologs, revealing gene homologs present in surface marine waters. The C-P lyase pathway has been suggested to occur more broadly in marine systems, as laboratory studies have demonstrated organisms acquired from these marine systems perform this pathway under extreme phosphorus starvation98,103.

Other marine sites have proposed anaerobic methanogenesis is responsible for methane production in oxygenated waters. For instance, respiration from unicellular organisms was shown capable of creating anoxic conditions where methanogens could feasibly operate, allowing for the production of methylotrophic methanogenesis from

DMSP in this methane and oxygen rich zone (ranging from 290-400 uM)95. Additionally, van der Maarel et al discovered 16S rRNA genes for the methanogen Methanococcoides 20 methylutens in both sea water particles as well as in the digestive tract and feces of the flouder fish Platichthys flesus in the North Sea with a dissolved oxygen concentration of

231 uM84. In conclusion, it appears both anaerobic methanogenesis and C-P lyase pathways are active sources of methane production in oxic marine waters and future work will be needed to assign emissions contributions.

1.5.2 Freshwater

Of the many systems in which the methane paradox has been observed, there is substantial research from freshwater lakes. While detectable methane in oxic surface waters can be attributed to transportation from its source in anoxic sediments104,105 many exceptions have shown oxic methane production in lakes. Some of the initial research demonstrated that in deep lakes the water depth prevented methane gas from being driven solely via diffusion from littoral or benthic methane in sediments35,4,102,106–110. Additional works have since shown the methane paradox is active for lakes of variable sizes and depths111–113.

A landmark freshwater methane paradox study by Grossart et al4 demonstrated that, in a temperate oligotrophic lake located in Germany, methane was produced in waters oversaturated with oxygen (313-625 uM) and was not impacted by the presence or absence of methylphosphonate. Furthermore, archaeal acetoclastic methanogens closely related to uncultured Methanoregula and Methanothrix, along with mcrA gene transcripts, were detected in the oxygenated lake waters, clearly demonstrating methanogen populations were active in bulk oxygenated waters. The attachment of methanogens to photoautotrophs was confirmed by visualization methods in laboratory studies, and the presence of 21 phototrophs resulted in increased methane production relative to controls. This study was one of the first in the field to demonstrate the activity of methanogens in oxygenated habitats.

More recent studies have suggested that acetoclastic methanogenesis (rather than hydrogenotrophic or methylotrophic) may contribute to methane production in oxic lake waters. A study by Bogard et al7 with data collected from a small, shallow oligomesotrophic lake in Canada used isotopic insights to verify that acetoclastic methanogenesis occurred in oxic surface lake waters ranging from 45-129% saturation.

Here they also estimated soil methane flux and measured ebullition. Using these data, they scaled the average areal estimates of diffusion at specific sites to the entire lake using surface area data and determined that methane production in oxic lake waters contributed

20% to overall methane flux from the lake in the summertime. Similar to these findings, more recent research from a mesotrophic lake in Switzerland also determined methane production in saturated and supersaturated oxic surface layers, with isotopic data demonstrating acetoclastic methanogenesis likely accounted for up to 90% of methane emitted to the atmosphere114. Alternative to the activity of methanogens, other laboratory enrichments have highlighted the capacity for C-P lyase activity from a phosphorus-limited lake located in the United States100. Analogously, Yao et al32, also showed that phn genes were more abundant in areas of the lake with significantly lower phosphorus concentration.

1.5.3 Soils

In addition to water sources, there is a wealth of data demonstrating the presence of methanogens in bulk oxic soils which, at the least, can survive these conditions69,115,116 22 and become active again under anoxic conditions. The works of Conrad and collaborators demonstrated this presumed activity in environments including rice paddies and forest soils117 as well as savanna and desert soils118 more than 20 years ago. Later work demonstrated that desert soils, while generating higher mcrA copy numbers under anoxic

3 7 conditions, still produced 10 -10 mcrA gene copies in oxic microcosms (21% O2 headspace), dominated by Methanocella (hydrogenotrophic methanogens)6. In addition to arid desert soils, unexpected oxygenated soil methane production has been observed in other terrestrial soils such as pastures119, rice paddies120, and tropical soils36. Work in pasture soil from Scotland evaluated the presence of methanogenic archaea under different treatment regimens and found that, in well-drained soils, RC1 and Methanosarcina methanogen genera were still detected and presumably responsible for detected methane production119. Soil from a Japanese rice paddy was inoculated into oxic mesocosms, where the methanogens (both hydrogenotrophic and acetoclastic) Methanothrix, Methanosarcina, and Methanobacterium persisted for up to 7 days, while Methanocella actually increased in relative abundance120.

To understand the importance of the methane paradox activity to soil emissions, we further elaborate on previous work8, discussed briefly above under “Mechanism 1” and covered in more detail in Chapter 2. Here a single organism - Candidatus Methanothrix paradoxum - was inferred to be a primary driver of methane production in oxic soils. To quantify the implications for this methanogenesis activity to the wetland methane budget, a multi-stage processed was applied. This process decomposed eddy covariance flux data into individual land cover type contributions, then used a diffusion model for dissolved

23 methane concentrations was applied to determine the depth-resolved location of methanogenic activity within the soil that incorporated dissolved oxygen concentration data to calculate the proportion of methane generated in oxic soils. Finally, this data was scaled to the site observations over the entire wetland, adapting a scaling method from

Bogard et al7 and the data was integrated over the summer month time-course for which experimentation was conducted. The combined efforts of this work demonstrated that, at the wetland-wide level, 40-90% of the methane being emitted from the wetland originated in oxic soils. This study provided direct biological and geochemical evidence for oxic soil methane production and clearly demonstrated that the methane paradox may be critically important to methane emissions in some soil systems.

1.6 Global contribution to methane flux

Despite their relatively small land coverage, wetlands represent the largest source of atmospheric methane (20-40%). However, variations in these wetland emission budgets are high, with over 25% uncertainty. Accurately predicting net methane fluxes from these systems depends on multiple interacting geochemical, ecological, and metabolic constraints that are poorly understood, oversimplified, or missing in global biogeochemical models121. Currently, based on historical assumptions, methanogenesis in oxygenated environments is not factored into global methane biogeochemical models3. However, a growing body of field data suggests a large portion of methane emitted from lakes (20-

7,114 8 90% ) and soils (up to 80% ) may be occurring in oxic layers. To precisely forecast methane emissions today and in the future, the physiological constraints on methanogen activity in these environments must be identified. Moreover, if microbial processes are 24 important to predictions then it becomes imperative to improve accuracy in scaling from site-level biological and geochemical data to habitat-wide estimations.

All global ecosystem models focus on resolving the rates of carbon uptake by the ecosystem. Hydrological modules within these models characterize soil moisture and identify the locations and grid-cell fractions that are inundated or saturated. One of the major difficulties for global models stems from the lack of accurate observations and understanding of the geographic distribution of wetlands and inundated area121, which is the first step at simulating methane fluxes. Furthermore, the typical coarse resolution of global models (10-100 km2) make most wetlands and flooded soil areas represented as a sub-grid-scale processes and not directly resolved122,123. Wetland biogeochemical modules within ecosystem models assume that methane production turns some of the modeled carbon pool into methane, and then further diffuse this produced methane to the atmosphere or consume it by oxidation. All models today assume that methane generation is limited to the anoxic soil layers. The way by which this limitation is enforced in each model is dependent of the complexity of the representation of hydrological and biogeochemical processes in the model. Methanogenesis is either indirectly limited to the anoxic soil horizon by being restricted to a certain depth below the water table (e.g., VISIT model124,

JSBACH model125, explicitly limited to the anoxic function (e.g. LPJ model126), directly limited by redox potential (e.g., CLM4Me model3, TEM model127, or limited through oxidative downregulation of the production of methane substrates by anaerobic fermentation processes (e.g., Ecosys model128). A key consequence of representing methanogenesis in oxic soil layers is the fraction of the carbon pool that is available for

25 methane production. As soil-carbon concentrations and particularly dissolved organic matter can have a strong vertical profile, a large fraction of the carbon pool that is stored above the anoxic horizon, is currently assumed by models as unavailable for methane production. A second large consequence is related to temperature dependency. All models assume that methane production rates are positively related to temperature. Soil temperature vary with depth, and during the summer season when most methane production occurs, the shallower soil layer are typically warmer. Shifting a significant fraction of methane production upward in the soil column towards the more oxic layers also means that production occurs at higher temperatures and thus may occur at faster rates than currently assumed by models.

We have identified several areas for experimental advancement and data integration relevant to understanding methane production across ecosystems. First, model representation of anoxic microsites needs to be validated. represent the effects of small- scale heterogeneity within the soil, such that the complex physical structure of soils results in anoxic zones with associated redox gradients within oxic soils. Most models assume spatially uniform O2 concentration within a soil layer (if resolved to layers at all) at a given pixel or inundated sub-pixel patch, despite the reality of a highly structured soil environment. Recent process-based reactive transport models have incorporated microsites in an aggregate model. While these models predict anoxic soil aggregate zones where other anaerobic processes can occur (e.g. nitrous oxide129–131) they need to include methanogenesis and the capacity to predict anoxic microsites needs to be validated in future modeling and experimental work, respectively.

26

Second, biological based processes are being incorporated into models, but lack process-based experimental validation. Biological methane processes simulated in some models include the interrelated activities of anaerobic fermenters, hydrogen producing acetogens, acetoclastic methanogens, and hydrogenotrophic methanogens8,128. The simulation of these activities is based on the stoichiometries and energetics of the transformations mediated by each microbial process. Ongoing field work that combines isotopic and metatranscriptomics data on the activity of these microbial processes and their linkages to physical and hydrological conditions, substrate concentrations, and methane production and emission rates within different redox zones, ecological sub-patches types, and site-wide levels will help constrain specific parameterization of these processes.

Comparisons of model predictions to experimental data over multiple seasons and depths will establish confidence in the predictive capabilities for soil-atmosphere methane exchanges under spatially dynamic conditions. A reasonable grand challenge question, is to answer to what extent and level of resolution does the microbial ecology of the system matter?

Here we summarize clear evidence for the extent of methanogenesis in oxic environments and show this process can be significant contributor to overall site wide methane emissions. These findings have important ramifications for global biogeochemical models, as current simulations down-regulate methanogenesis in surface soil layers due to oxygen concentrations, potentially greatly underestimating methane emissions. It is therefore critical that future global biogeochemical model research be aware of, and

27 potentially account for, methanogenesis in bulk oxygenated environments, more accurately predicting net wetland methane emissions and their effects on climate.

1.7 Concluding comments

The methane paradox – the generation of methane gas by microorganisms in oxic habitats

– is a global phenomenon with potentially drastic implications for accurately predicting global methane budgets. Here we outline methane paradox sites, which is assuredly a gross underestimation of these sites. We then addressed the two current mechanisms for this methane production, either from anaerobic methanogens or via bacterial methane release.

Lastly, we summarize some of the seminal work across ecosystems and note specific systems where the methane paradox has been shown to be a contributor to a significant portion of site wide emissions. This is, to our knowledge, the first inventory of methane paradox sites that also includes instances from terrestrial habitats, and the first examination of phnJ activity globally. Future research challenges include assessing the microbial mechanisms for methane production in these diverse habitats - especially measuring the extent of C-P lyase pathway across methane paradox sites - and attempting to quantify its contribution to methane flux from these environments. Also, research will be needed to more clearly link microorganism identity and metabolic activity to better understand biological controllers on the methane paradox. Data collected on finely resolved spatial and temporal scales, with tightly coupled chemical data, is also warranted to provide a framework for the physical and chemical constraints on methane production. Lastly, laboratory experiments conducted both in pure cultures and enrichment cultures will

28 provide valuable physiological knowledge that can be translated to the field and ecosystem scale.

29

Chapter 2. Methanogenesis in oxygenated soils is a substantial fraction of wetland

methane emissions

This chapter was reproduced verbatim from “Angle and Morin et al (Nature

Communications, 20178)”. The text benefited from the writing and editing contributions of all authors. Supplemental Data Tables for this chapter, since already published, are available online at https://www.nature.com/articles/s41467-017-01753-4 and thus are not included in Supplementary Dissertation Tables. The numbering of these materials in the dissertation is consistent with the content found online. The main text and extended data figures in the publication have been have been integrated into the dissertation and the numbering of all the Chapter 2 figures reflects this incorporation.

2.1 Introduction

The Modelling and biological studies investigating methane flux from wetlands discount microbial methane production in surface, oxic soils121,132. The basis of this assumption is that critical methanogen enzymes are inactivated by oxygen and methanogens are poor competitors with other microorganisms for shared substrates133,134.

Because of the assumed physiological constraint that oxygen has on methanogens, global terrestrial biogeochemical models limit soil methane production in the presence of

3 dissolved oxygen (DO) .

Recent reports present an alternative view that in some ecosystems methanogenesis also occurs in oxic environments, known as the methane paradox. In freshwater lakes, isotopic and molecular biology techniques provided evidence for the presence and activity 30 of methanogens in well-oxygenated portions of the water column37,4,7. Similarly, isotopic signatures in oxygenated soils and activity measurements from soil laboratory enrichments have provided intriguing evidence for methanogenesis in soils with up to 19% oxygen36,6.

Despite this mounting, indirect evidence, comprehensive genomic investigations that link methanogens to methane production in any oxic habitat in situ are lacking.

Here we analyze observations from the Old Woman Creek (OWC) National

Estuarine Research Reserve, a freshwater wetland at the shore of Lake Erie in Ohio. In this study, we experimentally assess biological methane production and emission in freshwater wetland soils across multiple spatial and temporal gradients. The results presented here provide the first ecosystem-scale demonstration of methane production in bulk-oxic soils, its microbial drivers, and the global significance of this currently under appreciated process.

2.2 Results

2.2.1 Methanogens are most active in oxic, surface soils

To account for differences associated with distinct ecological sites in the wetland

(ecosites), we sampled soils beneath three land coverage types: emergent vegetation (plant) periodically-exposed mud flats (mud); and continuously submerged under open water

(water) (Figure 6). Seasonal variability, especially the effects from photosynthesis and climate, was accounted for by sampling the three ecosites in summer (peak primary production) and late fall before freezing (senescence), while differences in vertical oxygen distributions were examined in 5 cm intervals up to 35 cm deep (Supplementary Data 1).

31

Figure 6 – OWC Schematic and Sampling Guide Overview A. The National Oceanic and Atmospheric Administration (NOAA) field site, Old Woman Creek (OWC), is a 573-acre wetland located adjacent to Lake Erie. Orange boxes designate the location of the sampling transect, comprised of soils beneath three ecosites (plant, mud, and water). Here we monitored the biogeochemistry from this transect over two seasons, Fall (November, 2014) and Summer (August, 2015). Previously, we monitored with 16S rRNA gene and geochemistry multiple transects across the site and showed strong replication between cores from the same ecosite43. B. Inset cartoon depicts the sampling transect in greater detail, showing the 3 ecosites (2 m2) as well as the meteorological station and eddy covariance tower (indicated by the red star). C. Inset of the replicate core sampling within a given transect, showing ~4 soil cores being collected and the corresponding dialysis peeper always located < 1 m proximity to sample cores. Chamber measurements were collected in duplicate over each ecosite as discussed in methods. D. A sampling guide including the number of the spatial and temporal sampling events, the total number of samples collected. Note, due to increased mcrA qPCR transcript abundance (Figure 9), metatranscript data collection was performed on plant and mud ecosite samples. Abbreviations used include: ecosite as plant (P, green), mud (M, orange), and open water (W, blue) and soil sample depth as surface (S, 0-5 cm) and deep (D, 25-35 cm). 32

All ecosites were net methane emitting during both summer and fall sampling seasons (Figure 7). In summer, regardless of ecosite (plant, mud, water), the porewater

DO profiles were similar; for instance, depths shallower than 10 cm were always oxic while soils deeper than 25 cm were always anoxic (Figure 8, Supplementary Data 1). In situ porewater dialysis samplers (peepers) measured the greatest methane concentrations in oxic, surface porewaters in the four summer months sampled (June-Sept). For mud and water ecosites, we paired these concentration measurements with direct surface flux measurements from static chambers, and used a dynamic diffusion model to calculate the net methane source (production and destruction) rate at each layer within the soil column

(Supplementary Note 1). Compared to non-oxygenated soil layers, methane was frequently produced in larger amounts in oxygenated layers, in some instances up to an order of magnitude more, but the proportion varied with season and ecosite (Figure 8). These findings demonstrate that the methane paradox occurs in wetland soils and provides the first evidence for the extent to which it operates over spatial and temporal gradients. These

33 findings demonstrate that the methane paradox occurs in wetland soils and provides the first evidence for the extent to which it operates over spatial and temporal gradients.

a 105 ) Plant -1

s Mud -2 Water 104

103

102

1 Methane Flux Rate (nmol m 10 N A J J A S O 2014 2015

b ript

ransc ed Oxygen t soil methanev rmate o Acetate F MethanolDepth R (Pearson) mcrA in situ Dissol Total Carbon 1 mcrA transcript

in situ soil methane 0.5 + correlation Dissolved Oxygen Total Carbon 0 Acetate correlation no significant Formate −0.5 Methanol Depth −1.0 - correlation Figure 7 – Methane emission rates and correlation of methanogenic activity to geochemical parameters

34

A. Methane flux was measured via non-steady state chamber method, while paired emission, biological and geochemical samples were collected in November 2014 (N, red) and August 2015 (A, red). The x-axis depicts the chamber samplings across time, with each time point consisting of monthly flux data from the ecosites represented by color. Positive methane flux rate is depicted on the y-axis. B. Summer soil methanogenic activity (from qPCR of mcrA) and corresponding soil geochemistry measurements were assessed for significant correlations. Surface and deep soil sample data from triplicate cores (3 each from plant, mud, and open ecosites) is included in the analysis (Supplementary Data 1). The heatmap depicts Pearson correlation where statistically significantly (p<0.05) positive correlations (orange/red), statistically significantly negative correlations (green/blue), and a lack of a significant correlation (black). The correlation between mcrA transcript number and acetate concentrations, suggests an important role for acetoclastic methanogenesis in these wetland soils, findings consistent with other reports from soils and lakes135. Data regarding the ecosite-level differences in methane emissions have been discussed previously136.

35

Mud Rate a 1.4 b c oxic

(mol CH 200 12.6 0 transition 4 m 23.8 -3

day -200 Depth (cm) -1

35.0 anoxic ) 0 0.2 0.4 CH (mM) 4 Jul Aug Sep

Water Rate 1.4 d eD oxic

(mol CH 100 12.6 0 transition 4 23.8 m

-3 -100 day Depth (cm) -1

35.0 anoxic ) 0 0.2 0.4 CH (mM)

4 Jul

Aug Sep

Figure 8 – Methane concentrations and production rates across soil depths A. Pore-water dialysis peepers provide 2.8 cm resolved depth methane measurements. B and D. Monthly in situ porewater dissolved methane concentrations in mud and water- covered soils with data collected from June (blue), July (yellow), August (red), and September (purple). Black dashed lines depict the 95% confidence interval for location of the oxic to anoxic transition. C and E. The calculated net methane volumetric fluxes in soils columns from mud and water ecosites show seasonal methane production (orange and red) in oxic soils (Supplementary Note 1).

36

In order to measure methanogenesis activity from these surface and deep soils, we quantified methyl-coenzyme reductase subunit A (mcrA) gene transcripts, a key functional gene for inferring methanogenesis activity38. On average, across all ecosites and seasons, oxic soils contained nine times more mcrA transcripts and twice the methane concentration per gram of wet soil than anoxic soils (Figure 9, Supplementary Data 1). Methanogen activity was positively correlated to porewater dissolved organic carbon and acetate concentrations, but not to other soluble methanogenic substrates like formate, methylamines, and methanol (Figure 7B, Supplementary Data 1). Taken together, these findings suggest that methanogens utilizing acetate may be responsible for sustaining the methane paradox in these soils.

37 a methane emissions (nmol m-2 s-1) 1294 711 512 CH CH 4 4 CH 4 CH4 CH4 CH4

Plant Mud Water

in situ methane (μmol g-1 ) b 0 0.6 0 0.6 0 0.6

0 200 0dissolved oxygen (μM)200 0 200 0

5 oxic

10 depth (cm)

20 anoxic 30 0 5.0 0 5.0 0 5.0 mcrA transcript log10 copy # (g )

Figure 9 – The relationship between soil and dissolved oxygen concentration and methanogenic activity with depth and ecosites from Summer A. Schematic of the three ecosites examined in this study with methane emissions shown in colored boxes and depicted by red lines (Figure 7A). B. Dissolved oxygen concentrations (black boxes), transcripts for mcrA (colored bars), and porewater methane concentrations (red triangles) in soils. Error bars reflect SE (mcrA) and SD (oxygen), n =3.

38

2.2.2 Candidatus Methanothrix paradoxum is present and active in oxic soils

Paired metagenomic and metatranscriptomic sequencing provided the first holistic insight into the methanogens active in oxic environments. From metagenomics sequencing we reconstructed six (two estimated to be >90% complete) genomes from oxic soils that represent a new species of methanogenic archaea. Based on whole genome comparisons and phylogenetic analyses (e.g. 16S rRNA, concatenated ribosomal protein, and mcrA)

(Figure 10) theses genomes clearly represent a new species within the genus Methanothrix

(formerly Methanosaeta).

Based on these analyses this new species was phylogenetically most closely related to M. concili, a methanogen species widely distributed in anoxic terrestrial methanogenic environments, such as flooded rice paddy soils and lake sediments93,137. Comparative genomic analyses between these wetland genomes to four genomes from cultivated

Methanothrix, demonstrated the Candidatus Methanothrix paradoxum genomes expanded the Methanothrix pangenome by 27%, with 467 genes uniquely encoded in our wetland genotypes. Of these unique genes, the majority (55%) lacked any functional annotation information (Figure 12C). Here we propose the name Candidatus Methanothrix paradoxum, after the implied role for this organism in the soil methane paradox (Figures

11-13, Supplementary Note 2).

From our metatranscriptomic analyses, we conclude methanogenesis in oxic soils is conducted primarily via a canonical acetoclastic pathway (Supplementary Note 3,

Supplementary Data 2). Transcripts from these genomes were in the top 3% of all community-wide metatranscripts and accounted for on average 84% of the mcrA transcripts 39 in surface soils (Figures 14 and 15, Supplementary Data 3). In addition to the methanogenesis pathway, genes for protein synthesis and energy generation were consistently and highly expressed in both seasons and ecosites (Figures 14 and 16), signifying active methanogenesis by this organism stably occurs in these oxic wetland soils.

Figure 10 – Genome recovery and average nucleotide identity reveal a new species of Methanothrix termed Candidatus Methanothrix paradoxum

40

A. Similarity matrix of average nucleotide identity (ANI) between reconstructed Candidatus M. paradoxum genomes greater than 50% complete (M1-M4, M6) and other available Methanotrix genomes. B. Pie-chart representation of recovered Candidatus Methanothrix paradoxum genome completeness, based on single copy gene analyses, coloring denotes ecosite source with orange (mud), green (plant), blue (water). C. Comparative genome analyses revealed flexible and core Methanothrix genomes, with 467 genes unique to Candidatus Methanothrix paradoxum genome. D. Pan-genome analyses demonstrated the contribution of each Methanothrix genomes (>50% complete) to the total pan-genome of Methanothrix genus.

41

Figure 11 - A concatenated ribosomal tree depicting the phylogenetic placement of the 6 surface soil-acquired Candidatus Methanothrix paradoxum genomes 42

15 single copy ribosomal gene proteins (RpL2, 3, 4, 5, 6, 8, 14, 15, 18, 22, and 24 and RpS 3, 10, 17, and 19) were extracted from the genomes of all Methanosarcinales isolate genomes on the Joint Genome Institute-Integrated Microbial Genomes and Microbiomes JGI-IMG/M database (accessed 12/01/16). Bootstrap percentages ≥75 (black) or 100 (red) are denoted by node circle color. All of the Methanothrix genomes reconstructed in this wetland (colored by ecosite) are closely related to each other and phylogenetically distinct from previously sampled Methanothrix isolate genomes from a thermophilic anaerobic bioreactor, sewage sludge, and anaerobic sludge blanket reactor (Data from IMG). Our wetland genomes (labeled Methanothrix 1-6, or M1-M6) represent the first Methanothrix genomes reconstructed environmental shotgun sequencing data. These genomes share > 98% average nucleotide identity with each other and < 80% average nucleotide identity with any prior Methanothrix isolate genomes, supporting our conclusion that these genomes represent a new Methanothrix species, here proposed as Candidatus Methanothrix paradoxum. The aligned concatenated FASTA input file used to generate this figure is provided (Supplementary Data 6).

43

Methanothrix harundinacea 6Ac (CP003117) % similarity

Methanothrix thermophila PT (NC008553) to outgroup

Methanothrix concilii GP6 (NC015416)

OWC Plant

OWC Mud

OWC Water >70%

Barrow, AK metagenome M4-ASP1-1 Twitchell Island, CA M1-NSP1 Surface metatranscript Twitchell Island, CA >99% Surface metagenome M3-ASO1 M6-ASM2 Activated Sludge M2-NSM2 metatranscriptome (IL) Delaware, NJ >84% metatranscriptome

Figure 12 – Evidence that Candidatus Methanothrix paradoxum are similar to genotypes in other environmental metagenomes and metatranscriptomes (S3 tree) We selected the S3 ribosomal proteins (rpsC gene) as a marker as it was consistently transcribed at a high level between summer and fall seasons (Figure 8C) and in both ecosites. All genes were on unbinned scaffolds, not from reconstructed genomes from environmental metagenomic studies. Ribosomal protein S3 sequences from Candidatus Methanothrix paradoxum genome bins are denoted in bold (ecosite denoted by color), while sequences (>70% amino acid identity) recruited from other publically available metagenomic datasets are colored according to the legend. Similar to our findings, genotypes of Candidatus Methanothrix paradoxum are active in surface soils from a temperate wetland on Twitchell Island138 (red and black) and also in activated sludge (brown). Black circles indicate bootstrap values ≥75. The input FASTA file used to generate this figure is provided (Supplementary Data 7).

44

OWC Plant

OWC Mud

8 OWC Water

Barrow, AK metagenome Twitchell Island, CA Surface metatranscript Twitchell Island, CA Surface metagenome to outgroup >88% Activated Sludge metatranscriptome (IL) Delaware, NJ metatranscriptome

Methanothrix concilii GP6 (NC015416)

M3-ASO1 M1-NSP1 >96%

Figure 13 – Evidence that Candidatus Methanothrix paradoxum are similar to genotypes in other environmental metagenomes and metatranscriptomes (mcrA tree) Phylogenetic analysis constructed using the methanogenesis functional marker protein mcrA amino acid sequences >88% similar to Candidatus Methanothrix paradoxum. Binned wetland Candidatus Methanothrix paradoxum sequences (bold, M1, M3) are highly similar to transcripts from other surface wetland soils from Twitchell island138 (red). Sequences included in this analysis had an amino acid similarity ≥88% to sequences from metagenomic datasets on JGI IMG (12/01/16). Black circles indicate bootstrap values ≥75. The input FASTA file used to generate this figure is provided (Supplementary Data 8).

45 a 100 b 4 Ca. Methanothrix paradoxum mcrA genes highly transcribed mtrE in both seasons acs 3 cdhC

50 2 Methanogenesis ATP synthase CH Ribosomal 4 1 Repair ACS CDH MTR MCR transcript abundance (%) Transcription/Translation Summer Gene Expression Other acetate

mcrA 0 Unannotated 0 P M P M 0 1 2 3 4 Fall Summer Fall Gene Expression

c mcrA hsp20 4 thsA Ukn2 cdhC atpA atpB rpsC eif2a acs mtrE tuf Ukn1 3

2

1 Oxygen detoxification Gene Expression rubr1 rubr2 0

Figure 14 – Candidatus Methanothrix paradoxum genes transcribed in oxic soils A. Taxonomic assignment and relative abundance of mcrA transcripts in surface soils assigned to Candidatus Methanothrix paradoxum (black), Methanoregula (dark grey), and other methanogens (light grey). B. The relationship between the 100 most transcribed genes (by log10 FPKM) in each season, with gene functional categories denoted in color and key steps of the methanogenesis pathway highlighted. C. Gene expression levels for selected genes from B, across all samples with color legend used from B – black line, boxes, and whiskers represent the median, quartiles, and minimum/maximum of the log10 FPKM values) For comparison, oxygen detoxification genes are not consistently transcribed at detectable levels.

46

Figure 15 – Mapping of metatranscript reads to methanogen diversity sampled in the metagenomic dataset shows Candidatus Methanothrix paradoxum are responsible for a majority of mcrA transcripts in oxic soils

47

On the left, a phylogenetic tree with mcrA reference nucleotide sequences from isolated methanogens (bold) and these wetland metagenome-derived sequences denoted by ecosite (black, not bold). Candidatus Methanothrix paradoxum mcrA sequences are shown in the grey box, two of which were recovered in genome bins (denoted M1, M4, colored). Metatranscripts from plant (P) and mud (M) ecosites in fall and summer were mapped to this mcrA sequence database. The bar chart on the right summarizes the normalized transcript abundance (scale from 0-150,000 fragments per kilobase per million mapped (FPKM)). Data are the average from triplicate cores collected in each ecosite and season. Methanothrix account for 84% of the recruited mcrA metatranscript reads. Bootstrap values ≥75 (black), or 100 (red) are denoted by circle color on the node. Collapsed node “other methanogens” contains nucleotide sequences from the genera Methanobacterium, Methanobrevibacter, Methanocaldococcus, Methanococcus, Methanosphaera, Methanothermobacter, Methanothermococcus, Methanothermus, and Methanotorris. The input FASTA file used to generate this phylogenetic analysis and the mapping results are found in Supplementary Data 9 and Supplementary Data 3 respectively.

Prior laboratory investigations have shown that methanogens in pure culture or from soil mesocosms upregulate antioxidant mechanisms to attenuate oxygen toxicity64,65,68. Consistent with those reports, Candidatus Methanothrix paradoxum genomes encode known oxygen detoxification genes including those for stabilizing free radicals, reducing toxic reactive oxygen species, and for repairing oxidative disulfide damage (Supplementary Note 3, Supplementary Data 2). However, these genes were not unique to our wetland genomes, present in similar abundances across all other

Methanothrix spp. More notably, oxygen tolerance genes were not consistently transcribed in our oxic wetland soil samples by Candidatus Methanothrix paradoxum or any other methanogen. To illustrate the minimal transcript detection, we reported the two most abundant oxygen detoxification genes (rubrerythrin) alongside other more highly transcribed genes (Figures 14C and 16). Additionally, we cannot rule out that some of the highly and consistently transcribed genes lacking a known annotation in our surface 48 methanogens (Figure 14) could play roles in oxygen detoxification, however the use of remote homology detection via structural modelling139 and HMM searches for PFAM domains failed to identify putative oxygen detoxifying genes in these highly expressed but unannotated genes. These metatranscriptomic findings demonstrate for the first time that oxygen detoxification is not a requirement for sustained anaerobic methanogen activity in oxic habitats.

Accounting for the black queen hypothesis140, we did consider that oxygen tolerance could be provided to Candidatus Methanothrix paradoxum by other members in the soil community. In our metatranscriptomes, we recovered transcripts for a catalase gene and several superoxide dismutase genes belonging to non-methanotroph

Gammaproteobacteria; however only one of these transcripts was detected in as many as

5 of the 12 samples, and at very low abundances. Importantly, none of these transcripts were highly abundant in our dataset, nor correlated to methanogen activity, suggesting that the ability for methanogens to compensate for oxygen toxicity is not likely to originate from other community members. Together, these findings demonstrate that known oxygen detoxification mechanisms used by other methanogens in the laboratory are not a requirement to sustain methanogenesis in these oxic wetland soils.

49

Fall Summer Plant Mud Plant Mud acsA cdhA cdhC cdhD mtrE mcrA Methanogenesis mcrB mcrD mcrG atpA atpB atpC atpD atpE ATP synthase atpI atpL rpsC rplA Ribosomal rplJ thsA thsB hsp20 dnaK Protein Repair panA psmB rpoA1 rpoD infB Transcription/ eif2a tif2 tuf Translation fusA Lipoprotein ATPase S.layer Other rpa Stress Unn1 Unn2 Unn3 Unn4 Unn5 Unannotated Unn6 Unn7 Unn8 rubr1 rubr2 O2 detoxification 4 3 2 1 0

Gene expression (log10 FPKM)

Figure 16 – Candidatus Methanothrix paradoxum (genome M1) transcript abundance patterns shared across seasons and ecosites

Log10 FPKM values are shown for each replicate transcriptome for a subset of genes. Gene abbreviations are shown with assignment to functional categories performed manually (Supplementary Data 3). Beyond genes in the pathway for methanogenesis pathway (red) other genes with high transcript relative abundance include those for energy generation (yellow), protein production and repair (green and blue), cell surface (other), and unannotated genes (grey). Despite encoding multiple oxygen detoxification mechanisms (Supplementary Data 2), we show the two highest expressed oxygen detoxification genes (pink) were not comparatively highly or consistently transcribed by ecosite or season (Supplementary Note 3). The mapping results used to construct this figure are provided (Supplementary Data 3), as well as a complete list of M1 genes (Supplementary Data 5).

50

Analysis of the methane paradox literature revealed several possible explanations for methane production in oxic habitats. In our data, we failed to find metabolite or molecular evidence supporting methane production from microbial decomposition of methylated compounds31 or by protozoan endosymbionts141 (Supplementary Discussion).

Instead, the data convincingly demonstrates methane production by canonical methanogenic archaea as the drivers of methane in these oxic soils. Together the absence of transcripts for known oxygen tolerance genes and the activity from multiple methanogen genera (e.g. Methanoregula accounted for 16% of mcrA transcripts in summer (Figure 15), suggests a more general explanation for the methane paradox in these soils. We suggest rather than having special adaptations, surface methanogen activity may be confined to anoxic subfactions (e.g. microsites, soil aggregates, or particles) with locally depleted soil oxygen concentrations relative to otherwise overall oxic surrounding soils. This hypothesis is not without warrant, as anoxic microsites were shown to facilitate anaerobic metabolisms in bulk oxygenated soils (e.g. for denitrification, iron reduction) and particle-associated models explain methanogenesis in oxic lake waters4,7.

2.2.3 Candidatus Methanothrix paradoxum is the dominant methanogen across the

wetland and globally distributed across other hydric soils

To assess the contribution of Candidatus Methanothrix paradoxum to methanogenesis in this wetland and beyond, we mined our data and public databases for highly similar (>99%) 16S rRNA gene sequences. In this wetland, this methanogen species is cosmopolitan, recovered from 97% of soil samples collected from various depths, ecosites, and time points over three years. Moreover, as we previously reported43, these 51 methanogens are dominant members of the oxic soil community and unlike other methanogens show a strong enrichment in the top 5 cm of soil (Figure 17, Supplementary

Note 4). Candidatus Methanothrix paradoxum are also globally distributed, detected in 102 locations across four continents spanning a range of habitats including rice paddy, wetland, and peatland soils (Supplementary Data 4). From these analyses, we infer Candidatus

Methanothrix paradoxum is well adapted to diverse hydric soils and sediments.

Our analysis of ribosomal 16S rRNA genes from previous methane paradox publications revealed that Candidatus Methanothrix paradoxum was detected and often acknowledged as a critical member in ten studies where the methane paradox was previously reported (Supplementary Note 4, Figure 18). For instance, many of the reported

Methanosaeta sequences in oxic lake waters share greater 99% 16S rRNA gene identity with Candidatus Methanothrix paradoxum7 (Figure 18). We posit that perhaps the increased activity of Candidatus Methanothrix paradoxum over other acetoclastic methanogens in oxic soils may be due to its competitive substrate acquisition under low acetate concentrations142 (<1 mM) found in these wetland surface soils (Supplementary

Data 1) and others soils143. Similarly, a recent report on the importance of acetoclastic methanogenesis to the methane paradox in lakes also alluded to low acetate concentrations in oxic surface waters as a potential contributer7. Our findings demonstrate that the

Candidatus Methanothrix paradoxum genotypes reconstructed here are widespread and active, potentially contributing to methanogenesis in a wide variety of oxic, yet high- methane habitats globally.

52

Plant Mud Open 4 4 4 M1 M2 not detected Rank: 24/54 Rank: 12/111 Fall

2 2 2

0 0 0 0 20 40 0 30 60 90 0 20 40 60 80

4 4 4 M4 M6 M3 Abundance (%) Rank: 4/218 Rank: 22/203 Rank: 19/181 Summer

2 2 2 Relative

0 0 0 0 50 100 150 200 0 50 100 150 200 0 50 100 150

Ca. Methanothrix paradoxum Methanolinea

Other Methanothrix All other methanogens

Methanoregula Methanotroph

Methanomassiliicoccus Other

Figure 17 – Candidatus Methanothrix paradoxum are dominant methanogens in the OWC surface soil communities based on metagenomic relative abundance analyses Rank abundance curves for the microbial community from surface (0-5 cm) soil metagenomes. The y-axis depicts the percent relative abundance of the rpsC (30S ribosomal protein S3) gene in the assembled metagenome. The rank and relative abundance of Candidatus Methanothrix paradoxum rpsC genes in each sample is denoted in red color and summarized in the left corner. The relative abundance of rpsC genes assigned to methane cycling organisms are also denoted: other Methanothrix (grey), Methanoregula (green), Methanomassiliicoccus (purple), Methanolinea (orange), all other methanogens (crimson), and methanotrophs (blue) are also shown.

53

102 studies with sequences >99% identity to a Candidatus Methanothrix paradoxum Environment Wetland; Peatland or Marsh Sludge; Wastewater Freshwater sources Rice Paddy Other Permafrost; Fen; Arctic Marine; Estuary

AY817738 Methanothrix harundinacea b CP000477 Methanothrix thermophila PT AB679168 Methanothrix soehngenii CP002565 Methanothrix concilli GP6 ParadoxEnvironmentLocation AJ608188 Kemnitz et al. 2004 KC604427 JN649116 Yavitt et al. 2012 EU155908 Cadillo-Quiroz et al. 2008 KC875584 AB655115 Itoh et al.2013 AY531696 Chan et al. 2005 JF727934 AB653609 Itoh et al. 2013 GQ340374 Location AY454761 JF510103 Grossart et al. 2011 AB288619 North America GQ906620 AB479409 South America CU917341 FJ705109 KF198609 Europe CU916658 JN397696 Middle East KF198608 JQ781241 Reis et al. 2013 HQ407441 Asia HQ404318 DQ260367 KC604424 JX426845 KF186065 CU916809 AB650673 Itoh et al. 2013 AB550818 AM085540 JF431892 FJ887767 KT323105 EU255751 KF431940 JQ792597 Stoeva et al. 2014 KJ619666 KC895404 KU297843 KT265196 KC923115 KU297835 JN397643 JN397916 M1-Candidatus Methanothrix paradoxum AF424767 JF304112 AB266904 AB479394 KC676307 KF198526 CU916001 EU591661 KF198726 JQ866636 JF313795 GU388766 LN796000 KP101356 JF712540 AB653986 Itoh et al. 2013 JN596399 AB653606 Itoh et al. 2013 KC437280

Figure 18 – Candidatus Methanothrix paradoxum is globally distributed in a variety of ecosystems 54

A. 868 16S rRNA genes from 102 studies were recovered from public databases that were >99% similar to the Candidatus Methanothrix paradoxum 16S rRNA gene from genome bin M1. Pie chart represents studies where Candidatus Methanothrix paradoxum was detected, shown by environment type (Supplementary Data 4, Supplementary Note 4). B. Maximum likelihood phylogenetic tree constructed with representative 16S rRNA sequences identified in A, with black node circles indicating bootstrap values >70%. The environment where the sequence was recovered from is denoted by color. Geographic location is indicated in greyscale. Importantly, several Methanothrix sequences detected under oxic conditions or where the methane paradox was cited are noted in black under paradox. References for these papers are provided by first authors last name and year of publication4,143–149. Taken together with our other data (e.g. Figures 14 and 15) the new species of Methanothrix we propose here is globally distributed and active in high methane- flux environments, suggesting this lineage may be a predominant contributor to global methane production in anoxic and oxic environments. The input FASTA file used to generate this figure (Supplementary Data 10) and the metadata (Supplementary Data 4) are provided. These results were obtained in conjunction with Adrienne Narrowe (laboratory of Dr. Christopher Miller, University of Colorado Denver).

2.2.4 Oxic soil methanogenesis contributes substantially to methane flux

To understand the importance of methanogenesis in oxic soils, we estimated the contribution of this process to the total methane budget in this wetland using simplifying assumptions (Supplementary Note 5). We first decomposed the eddy covariance flux signal into its ecosite level contributors150. We then applied a diffusion model of pore water dissolved methane concentrations of to determine the location of microbial methane activity within soil columns. We overlaid the microbial activity profile with the dissolved oxygen concentration profile to determine microbial activity in the oxic layers. Previously,

Bogard et al7 used a scaling method to demonstrate methanogenesis in the oxic portion of the water column contributed to 20% of lake-wide emissions. Using a similar approach,

55 when integrating over the course of this study, we estimated that between 40 and 90% of methane emitted originated in oxic soil layers (Figure 19).

Figure 19 – Percent methane generated in ecosites over the season as determined from the diffusion/generation model These data represent a synthesis of the 10% best performing realizations of the microbial activity terms (R(t,z)) as determined by the Markov Chain. Red lines show the integrated production/consumption of methane over the oxic zone, interpolated over time. The heavy red dashed line indicates the net neutral methane generation point. Black lines represent the fraction of methane production that can attributed to the strictly oxic layer (i.e. production above the 97.5th percentile line of the oxic horizon). The shaded areas of both lines represent one standard deviation of the 4000 R(t,z) realizations. Oxic layers were almost always a net source, with the exception of August in the mud ecosite. The percent contribution depended on the total production within the soil column as well as the level of production in the oxic layers. As the footprint of the site is primarily open water (97% when accounting for only open water and mud, as we do here), the percent generation curve of open water dominated the site level budget when calculating the total percent generation in the oxic layers. These results were obtained and figure created by Dr. Tim Morin (previously a graduate student in the laboratory of Dr. Gil Bohrer, The Ohio State University), and provided with their consent.

56

This study provides the first genome-resolved view of the methane paradox in any ecosystem and identifies the important contribution of a newly defined and globally distributed methanogen species, Candidatus Methanothrix paradoxum. We provide clear evidence for the extent of methanogenesis in oxic soils from multiple seasons and ecosites, and show this process is a significant contributor to overall wetland methane emissions.

These findings have important ramifications for global biogeochemical models, as current simulations down-regulate methanogenesis in surface soil layers due to oxygen concentrations, potentially greatly underestimating methane emissions. It is therefore critical to refine global biogeochemical models to account for methanogenesis in bulk oxic surface soils, more accurately predicting net wetland methane emissions and their effects on climate.

2.3 Supplementary Material

2.3.1 Supplementary Note 1 – Greenhouse gas emissions and estimates

The concentration profile of methane is not directly indicative of the vertical distribution of microbial activity in the soil because changes in methane concentration profiles between any two points time are due to several processes: 1) microbial activity

(i.e., methanogenesis and methanotrophy), 2) transport between soil layers, and 3) flux between the soil layers and the atmosphere. To isolate the effect of microbial activity, it is therefore necessary to account for the transport of methane, both between soil layers and leaving the system. Transport within layers can be caused by molecular diffusion, bulk flow of porewater, ebullition, and via plants. In permanently flooded soils, bulk flow can

57 be assumed to be negligible, reducing the problem in complexity. Transport out of the soil has been documented to occur in one of three ways: 1) molecular diffusion, 2) plant transport, and 3) ebullition3.

Chamber measurements of fluxes were used as the upper boundary condition for the diffusion model whose results are show in Figure 8. Chamber flux measurements are filtered for ebullition and so are representative only of diffusion as an egress mechanism.

By focusing on mud and open water, we were thus able to disregard plant transport. We disregarded ebullition both between soil layers and out of the soil column from this analysis. This makes our estimates a lower boundary on how much methane could be produced in oxic soils. This is because ebullition is effectively a numerically unaccounted- for sink term in the upper layers. If this sink term were to be quantified and accounted for in the analysis, the microbial activity term (R) in the affected layers would necessarily have to be increased to compensate for the methane lost. Methane moved from lower layers as bubbles also may not be directly emitted as gases in bubbles are often reabsorbed as dissolved gases during transport, limiting the impact of this simplifying assumption.

2.3.2 Supplementary Note 2 – Metagenomic and metatranscriptomic analyses

Metagenomic assembly and binning of these surface soil samples yielded 58 bins, with eight identified as methanogens, and six of these as Candidatus Methanothrix paradoxum. The other sampled methanogen bins (48% and 61% complete) were most closely related to Methanoregula spp., which were less abundant and less active methanogens of the surface soil community sampled here. Candidatus Methanothrix paradoxum genomic bin quality and completion are summarized below (Table 2), but 58 ranged from 31-91% estimated completeness, with low numbers of overages (<3% in the most complete genomes) (Figure 10). The most metabolically complete genome, M1, was used as a population representative. We recovered one 16S rRNA gene fragment (1472 bp) in the most metabolically complete genome M1. Metabolic profiling of the Candidatus

Methanothrix paradoxum genomes was performed manually, and, to account for any misbinning, we confirmed that any gene included in the metabolic summary was supported by other genes on the contig annotated as Methanothrix and having similar GC and coverage to the overall bin. A summary of the metabolic capabilities and the gene transcripts detected is included (Supplementary Data 2). All of the pathways required for acetoclastic methanogenesis were present and highly transcribed across both ecosites and seasons, a finding consistent with our porewater substrate data that showed a significant positive correlation between acetate concentrations and mcrA transcripts (Figure 7B). On the other hand, essential genes for methanol and methylamine activation or utilization were not present, while genes for hydrogen utilization were present but not transcribed. These findings reflect a lack of significant correlation between mcrA transcripts and these other methanogenic substrates in our porewater (Figure 7B). Based on this data, and other reports from previously characterized Methanothrix spp.151,152 we consider it likely that

Candidatus Methanothrix paradoxum is also an obligate acetoclastic methanogen.

To identify Candidatus Methanothrix paradoxum genes that were highly expressed across both seasons, we identified the top 100 transcribed genes in each season, resulting in 140 unique genes from summer and fall. To show gene transcription patterns shared across both seasons, the log10 FPKM for each season was plotted, and 73% (102) of these

59 genes were found to share high levels of transcription in both summer and fall surface soils

(Figure 14B, white oval). These findings clearly show that key genes in the methanogenesis pathway are highly transcribed in surface soils. Additional indicators of

Candidatus Methanothrix paradoxum activity in these oxic surface soils include the high relative abundance of transcripts for protein synthesis (transcription and translation) and energy generation (ATP synthesis). Notably, genes for protein repair (e.g. chaperones) were also highly transcribed, suggesting protein turnover and repair may be a mechanism of oxidative protection of proteins used by methanogens in oxic soils.

To gain insight into the environmental distribution of Candidatus Methanothrix paradoxum genomes, both the phylogenetic marker gene 30S small subunit ribosomal protein 3 (rps3) and the functional marker gene mcrA found within the genome M1 were used to query environmental metagenomes available on the Joint Genome Institute

Integrated Microbial Genomes/Microbiome (JGI-IMG/M December 2016). These analyses clearly show that genotypes similar to the reconstructed genomes here are present in other hydric soils from Barrow, Alaska and Twitchell Island, California. Also of interest, similar to our reports that Candidatus Methanothrix paradoxum mcrA and rpsC genes were highly transcribed in both seasons and ecosites (Figure 14C), these genes were also highly transcribed in surface soils from Twitchell Island, clearly demonstrating that Candidatus

Methanothrix paradoxum can contribute to methane cycling across geographically distinct wetlands (Figures 12 and 13).

60

Table 1 – Metagenomic and metatranscriptomic sample sequencing data

Data type - Same name Read Length (bp) Read count Total Sequencing (Gbp) Metagenome - Fall Plant 100 98581091 19.72 Metagenome - Fall Mud 100 208578991 41.72 Metagenome - Fall Open 100 96172864 19.23 Metagenome - Summer Plant 151 276581518 83.53 Metagenome - Summer Mud 151 231981210 70.06 Metagenome - Summer Open 151 231999210 70.06

Metatranscriptome - Fall Plant 1 151 121148486 36.59 Metatranscriptome - Fall Plant 2 151 101678848 30.71 Metatranscriptome - Fall Plant 3 151 135941546 41.05 Metatranscriptome - Fall Mud 1 151 138053478 41.69 Metatranscriptome - Fall Mud 2 151 128188682 38.71 Metatranscriptome - Fall Mud 3 151 122537220 37.01 Metatranscriptome - Summer Plant 1 151 115335708 34.83 Metatranscriptome - Summer Plant 2 151 138225486 41.74 Metatranscriptome - Summer Plant 3 151 129084508 38.98 Metatranscriptome - Summer Mud 1 151 134446440 40.60 Metatranscriptome - Summer Mud 2 151 118650014 35.83 Metatranscriptome - Summer Mud 3 151 143765256 43.42

Table 2 – Candidatus Methanothrix paradoxum genome bin characteristics

Genome Genome Land Single copy name name Cover gene Largest Assembled Number of (Short) (Full) Season Type Completion overages contig length in bin contigs in bin M1 M1-NSP1 Fall Plant 90% 2.9% 23,510 1,471,312 252 M2 M2-NSM2 Fall Mud 91% 1.9% 38,789 1,751,596 238 M3 M3-ASO1 Summer Water 76% 4.8% 14,739 1,170,295 263 M4 M4-ASP1-1 Summer Plant 57% 3.8% 10,710 688,846 179 M5 M5-ASP-2 Summer Plant 31% 5.7% 5,525 208,261 68 M6 M6-ASM2 Summer Mud 79% 3.8% 21,054 1,171,139 249

61

2.3.3 Supplementary Note 3 – Comparative Methanothrix genomic analyses

Here we expand on the main text and provide an inventory of the oxygen tolerance genes reported in other methanogens and mined from Methanothrix genomes: thioredoxin

/ thioredoxin reductase (trx and trxr), rubrerythrin (rbr), peroxiredoxin (prx), desulfoferrodoxin (dfx), rubredoxin (rbx), glutaredoxin (grx), peroxidase (px), catalase

(kat), superoxide dismutase (sod), F420H2 oxidase (fprA), and cytochrome D oxidase

(cyd)30. From our wetland genomes, we recovered known oxygen detoxification genes, including those for stabilizing free radicals (sod), reducing toxic reactive oxygen species

(kat, px, rbr, rbx), and for repairing oxidative disulfide damage (prx and trx). These data are summarized in Supplementary Data 2. Also, in contrast to prior reports of methanogens in oxic laboratory experiments, no oxygen tolerance genes were expressed abundantly

(single sample >1000 FPKM) or consistently (present >3 samples) in our field data.

It is possible that Methanothrix confinement to anoxic microenvironments with the bulk oxic layer may be a possible explanation for the activity of methanogens in our wetland surface soils. One anoxic microhabitat would be the formation of biofilms. While numerous publications have reported on the presence and contributions of Methanothrix in biofilms from diverse systems, including pipes153, wastewater digestors154,155, and even carbonate chimneys156, very little work has examined the specific genes involved in biofilm formation. A collection of genes implicated in methanogen biofilm formation or conditions157–159, as well as more general bacterial biofilm genes160–162, were queried to the two most complete genomes of Candidatus Methanothrix paradoxum and to the community metatranscriptome. None of these biofilm-associated genes (1e-5 identity) 62 were recovered in Candidatus Methanothrix paradoxum genomes, nor were these biofilm genes detected transcribed in the community metatranscriptomes.

Of course, we cannot rule out contributions from yet-annotated and highly expressed genes in Candidatus Methanothrix paradoxum in oxygen adaptation, especially since some of these were potential s-layer or extracellular proteins. Lastly, metatranscript recruitment plots between our most complete genome and the nearest Methanothrix neighbor (M. concilii) demonstrated recruitment of a log-fold more genes using our wetland genotypes. This finding showcases the value of reconstructed genomes when functionally profiling in situ metabolisms from soils or other habitats containing high numbers of uncultivated or genomically undersampled strains.

2.3.4 Supplementary Note 4 – Candidatus Methanothrix paradoxum biogeography

Based on the data collected here and our prior study43, we sampled the microbial community in these wetland soils by 16S rRNA sequencing for two years (Nov 2013, Nov

2014, Feb 2015, March 2015, Aug 2015) and from multiple ecosites (n=126). These samples spanned a range of depths, with 23.8% being collected from the first 12 cm surface soils. Across these samples we recovered 1560 (Nov 2013) and 5663 (remaining dates) bacterial and archaeal OTUs (defined at 97% nucleotide identity). Seven of these OTUs were classified as belonging to the genus Methanothrix, but due to short amplicon size further taxonomic resolution was not possible (V4 16S rRNA region, ~300 bp).

The 16S rRNA data from mud and open water samples (n=60), published previously, used a larger amplicon size (V3-V6 region of the 16S rRNA gene) and targeted the Archaeal 16S rRNA with archaeal-specific primers, allowing for greater phylogenetic 63 resolution and deeper sampling of Methanothrix strains43. Consistent with our metagenomic findings from surface soils (Figure 17), OTUs representing Methanothrix spp. (max relative abundance 47%, mean 21% +/- 8%) and hydrogenotrophic

Methanoregula spp. (max relative abundance 10%, mean 4% +/- 2%) were the two most abundant methanogens across the wetland. Additionally, we observed OTU-level differences in abundance along soil depth gradients. One Methanothrix OTU (OWC_a1), which was 100% identical to the 16S rRNA gene recovered from our Candidatus

Methanothrix paradoxum genome, was enriched in the surface soils relative to the deep

(mean 7% +/-6% greater within-core relative abundance in shallow samples vs. deep samples) and represented the most abundant archaea in surface soils. These 16S rRNA gene results from a prior year, and from more ecosites, support our metagenomic rank abundance curves (Figure 17) and suggests findings generated here may extend much more broadly across larger spatial and temporal time scales in these wetland soils.

The surface-enriched Candidatus Methanothrix paradoxum 16S rRNA gene from the M1 genome has 99% or greater nucleotide similarity to sequences found globally in

102 studies representing a variety of ecosystems (Figure 18). The distribution of these studies includes 28% wastewater, 34% freshwater (dominated by lake waters), 7% estuary or marine, 8% permafrost (with equal distribution across mountain and arctic/boreal), and

10% wetlands (including rice paddy, peatland, marsh) (Supplementary Data 4). We acknowledge that this distribution is largely affected by the sampling of these ecosystems, but the data highlight the broad relevance of Candidatus Methanothrix paradoxum across

64 ecosystems and geographic regions. A subset of representative sequences from this survey is included in a phylogenetic analysis (Supplementary Figure 9, Supplementary Data 4).

From this meta-analysis, notable was the prevalence Candidatus Methanothrix paradoxum in surface soils (often oxygenated), including tropical streams148, arctic wetlands149, and temperate peatlands144,145. Moreover, we show representatives similar to

Candidatus Methanothrix paradoxum were present in prior soil and lake studies where the methane paradox was suggested (Figure 18). Of particular interest, particle-associated

Methanothrix (some of which were highly similar to Candidatus Methanothrix paradoxum) were inferred to be responsible for methane production in oxic lake waters, one of the first methane paradox publications4. In terrestrial systems, Candidatus Methanothrix paradoxum were also enriched and inferred to be active in the top 5 cm of soils7,120,149,163,164, some of which were shown to be oxic and have high numbers of transcripts from mcrA belonging to Methanothrix163. Collectively these findings, in light of our results, suggest

Candidatus Methanothrix paradoxum may be a critical driver of methanogenesis in oxic habitats from both aquatic and terrestrial systems.

2.3.5 Supplementary Note 5 – Site level scaling analyses

Here we based our site level scaling on a method similarly used by Bogard et al7, which determined the contribution of oxic methanogenesis to overall lake methane flux.

We estimate that between 40 and 90% of methane emissions across the site is driven by oxic soil production. Quantification of the proportion of emitted methane due to generation in the oxic soil zone is a non-trivial problem and we acknowledge that there are some coarse assumptions made in our estimate, which we discuss below. The rates of methane 65 emission observed from the 3 ecosites were at the high end of the rates reported in similar wetlands136.

The net methane activity values generated in this study (Figure 8) result from methanogenesis and methanotrophy at each soil layer. The individual contributions of these processes to our predicted net activity values are unknown and to partition one must make assumptions on how oxygen and alternate electron acceptors affect these rates. Net negative activity layers in the soil almost certainly still have methanogenesis, but the rate is lower than co-located methanotrophy. However, some of the generated methane may be mobilized towards the soil/water interface before the full amount is consumed (displaced by incoming methane from other layers). In order to be emitted, methane generated in the deep layers must pass through the oxic zone, which may well decrease its effective transmission to the atmosphere as large portions of it may be consumed in methanotrophic reactions as it passes through.

Furthermore, positive net methane activity values in the shallow layers are here treated purely as methanogenesis. In reality, there must be methanotrophy in these layers but methanogenesis must be high enough to offset this sink in order to produce the activity levels we observe. Future quantification should include detailed modeling of observed tracers or in depth isotopic analysis165 to provide more comprehensive accounting of the origin of emitted methane.

2.3.6 Supplementary Discussion

Several mechanisms have been proposed to explain methane production in oxic habitats, here we discuss these hypothesis in light of our data. First, based on the increased 66 methanogenesis activity, organismal abundance, and methane production in oxic soils, we conclude that diffusion from methanogen activity in deeper anoxic layers112,166 is not a major contributor in our system. Similarly, our biological and modeling evidence does not support methane produced from UV-irradiated plants167 or as a by-product of heterotrophic decomposition31,5. As additional evidence that this process is driven by methanogens and not via microbial decomposition of methylated compounds5, we failed to detect methylphosphonate and its derivatives in our NMR porewater metabolite data, and we failed to detect phnJ transcripts (or any phn subunits involved in this pathway) in our community metatranscriptomics data31,32,96. Moreover, while it has been suggested in other ecosystems that methanogens may find oxygen shelter inside protozoans, we failed to detect 18S rRNA sequences from any known methanogen ciliate hosts in our EMIRGE reconstructions88,168, nor did we find consistent eukaryotic signal correlating to methanogen activity in our metagenomic data. This signifies that it is unlikely that endosymbiont methanogens are the primary source of methane in these soils. Here we show methanogenesis activity in oxic soils is driven by canonical, likely free-living methanogens.

Our data is the first methane paradox study to show which methanogens are transcriptionally active in bulk oxic habitats. Based on our metatranscript data that demonstrates i) multiple methanogens can be active in oxic soils and ii) that failed to identify a known genetic mechanism explaining increased activity of Methanothrix (e.g. oxygen tolerance, oxygen detoxification), we consider it plausible that surface soil methanogens may not be encountering the high levels of oxygen measured in porewaters.

67

Quantification of anoxic microsites in soils and the mechanisms sustaining these zones represents areas for future research. When considering explanations for the increased activity of methanogens in bulk oxygenated soils, it could be possible that analogously to wastewater digesters and microbial fuel cells, active carbon decomposition by other members of the community produce local anoxic conditions favorable for methanogens169,170. Analogously, Bogard et al7 suggested that fermentative bacteria create the conditions for anoxic methanogenesis in oxic surface lake waters. In wetland soils, it is also recognized that the combination of labile dissolved organic matter (DOM) and reduced rates of gas diffusion through saturated soil pores can facilitate the formation of anoxic microsites that fuel other anaerobic metabolisms (e.g. nitrate and iron reduction) in bulk oxic soils71,72,171.

Along these lines, we consider it likely that increased lability and input of dissolved

DOM from overlaying plants or surface waters contributed to the increased methanogenic activity in surface (0-5 cm) relative to deeper (>20 cm) soils. In support of this possibility, prior reports of DOM from porewaters in these wetland soils demonstrated structural differentiation with depth, with DOM molecular weight and aromaticity increasing with a depth below 5 cm172. In surface soils, this possible increased carbon input and decomposition (due to increased electron accepting capacity of these soils) may have contributed to the significantly greater (~on average twice as much) acetate we detected in surface compared to deep soils. We note that acetate is a non-conservative substrate, with a presumed high rate of turnover in oxygenated soils, thus the absolute concentrations in the soil may be an underestimation of methanogen substrate availability. Thus, DOM input

68 and composition in surface soils could fuel local regions of heterotrophy, leading to oxygen consumption and the generation of acetoclastic methanogen substrates, together facilitating methane production in surface soils.

Increased concentrations of methanol and formate detected in deeper soils

(Supplementary Data 1, worksheet 1) indicates methanogenesis in these soils is not likely facing substrate limitation. However, we and others172 have shown that deeper soils have increased Fe(II) concentrations (Supplementary Data 1, worksheet 1), which could directly or indirectly impact methanogen activity. Directly, using pure cultures and soil measurements, methanogenesis was shown to be inhibited by addition of amorphous

173,174 Fe(OH)3 and humic acids . Indirectly, increased metal ion chelation or absorption to the soil matrix in deeper soils could limit availability of required methanogen co-factors

(e.g F430)175. Biological competition for substrates176,177, vitamins, or cofactors178 with other microbial taxa or viral predation179,180 could further constrain methanogens in deep soils. Ongoing research coupling integrated biogeochemical, molecular DOM characterization, and omics technologies is required to better understand the factors facilitating methanogenesis in these soil horizons.

Regardless of the mechanism, our findings that methanogenesis occurs in oxygenated soils and contributes significantly to wetland wide methane flux has important ramifications for modeling efforts. Models that simulate methane production assume down-regulation of methanogenesis in these soil layers due to oxygen concentrations, underestimating methane emissions. As a consequence, soil conditions are diagnosed as appropriate for methane production at greater depths and after longer flooding periods than

69 are actually observed. Further understanding of microsite evolution may alter the perceived sensitivity of methane emissions to air and water temperatures, as shallower sites show higher temperature fluctuations that correlate more strongly with air temperature than deeper soil layers. It may therefore be critical to account for these processes in biogeochemical models to improve predictability of net wetland methane emissions and their effects on climate.

2.4 Chapter 2 Methods

2.4.1 Field sampling

The field location, Old Woman Creek National Estuarine Research Reserve

(41°22’N 82°30’W), is a 573-acre freshwater wetland and reserve located on the southern point of Lake Erie near Huron, Ohio. This site is co-operated by the National Oceanic and

Atmospheric Administration (NOAA) and Ohio Department of Natural Resources. This is one of 28 coastal (only two are in the Great Lakes region) NOAA designated sentinel research sites. The site consists of a permanently flooded channel surrounded by marsh, mud flats, and forested upland habitat. We collected soil cores from three (~2m2) ecologically differentiated sites (ecosites): plant, mud, and open. Four or more water saturated soil cores were collected per ecosite to a depth of 35 cm (width 7 cm) using a modified Mooring System soil corer. Cores were kept on ice in the field until processing in the laboratory (no more than 2 hours), where soils were immediately hydraulically extruded25, sub-sectioned into surface (0-5cm) and deep (23-35 cm), and then transferred

70 into separate sterile Whirl-pak bags for RNA extraction (stored -80°C), DNA extraction

(stored -20°C), and geochemical analysis (stored 4°C).

2.4.2 Soil and porewater geochemical analyses

Soil total carbon (TC) and porewater dissolved organic carbon (DOC) were analyzed via Shimadzu TOC-L with SSM-5000A solid sample combustion unit attachment using methods described181. Concentrations of soil and porewater acetate, nitrate, nitrite, and sulfate were determined via ion chromatography. For soils 5 grams of soil was mixed with 5 ml of MilliQ water (1:1 v/v), filtered with a 0.2 um filter, and quantified using a

Dionex ICS-2100 Ion Chromatography System with an AS18 column with standard curves performed for each anion. To more directly pair soil methane concentrations to microbiological soil data, in situ methane concentrations were calculated as described previously182 following immediate transfer to 4°C for transport and measurement on a

Shimadzu GC-2014 gas chromatograph.

Soil porewater was extracted using methods and infrastructure previously described in detail from this wetland150,172. Porewaters were then sent to the Pacific Northwest

National Laboratory and metabolites were identified by 1H NMR as described previously183. Metabolomic responses were characterized using the EMSL 800 MHz and

600 MHz NMR spectrometers equipped with cryogenically-cooled triple resonance probes for their high sensitivity and quantitation determined via 1H NMR metabolite libraries

(presently ~1,000 metabolites). 2-D NMR metabolomics methods including 1H-13C correlation experiments (HSQC’s), and connectivity experiments TOCSY, and COSY on

71 a subset of samples (<8) to enhance metabolite identification. Geochemical and metabolite data was analyzed in relationship to methanogenesis activity by linear correlations determined via Pearson correlation (p < 0.05).

2.4.3 Collection of dissolved gasses and greenhouse gas emission

Surface emissions were measured by non-steady-state chambers, with floating chambers used for measurements over open water. Chambers were measured in duplicate in each ecosite and season150 and sampling was coordinated to peeper measurement times.

Additional greenhouse gas emissions were collected with an eddy co-variance and meteorological station (3m tall tower, site-wide footprint). We have previously shown that both chambers and eddy co-variance measurements provide congruent measurements150.

Porewater dialysis samplers (peepers) were used to sample for dissolved methane, carbon dioxide, and nitrous oxide below ground monthly, with a vertical resolution of

2.8cm, throughout the upper 56cm of soils with minimal disturbance to the soil184–186.

Hydrogen was not measured in porewater from the dialysis peepers. The peepers feature

20 vertically stacked windows covered with a 0.1 µm dialysis membrane that allows the water inside the windows to equilibrate with dissolved gas concentrations outside. Gas concentrations in the peeper samples were quantified using a Shimadzu GC-2014 gas analyzer. The design and sample collection with both chambers and peepers followed protocols previously described150. Both chamber and peeper measurements were taken simultaneously, once a month during the 2015 growing season.

72

Temperature probes and co-located oxygen level measurements (via a PreSens

Fibox 4 handheld oxygen meter) provided vertical detail near each peeper-measurement location. The oxic horizon was determined by fitting a reverse exponential curve to all dissolved oxygen data collected per patch per month. The horizon was taken as where the

187 curve crossed 20 (micromol O2 / kgH2O) . To determine the upper and lower bounds of the horizon, we identified the soil depths at which the 2.5th and 97.5th% confidence intervals of the exponential fit crossed the same 20 (micromol O2 / kgH2O) threshold. The oxic horizon and confidence bounds were interpolated linearly between measurement periods.

2.4.4 Transport and production model

A numerical model was used to combine chamber and peeper measurements to determine the rates of methane production/oxidation in different layers of the wetland soil.

A diffusion model was separately created for mud and open ecosites (not plants), due to the complexity of including plant transport and roots. We discretized Fick’s 2nd Law

(equation 1) in 1-D using an implicit backwards Euler method to account for diffusive transport within the soil column. A production/oxidation term was included to account for the implied biological activity.

Equation 1 �� �, � � �� �, � = � �, � + � �, � �� �� ��

Here, C is the soil pore water concentration of methane, z is the vertical position in the soil column, D is the temperature dependent diffusion coefficient, t is the time in days, and R is a methane sink/source (generation/oxidation) term. The temperature profile was

73 determined through measurements made with nearby soil temperature probes. A Neumann no-flux boundary condition was prescribed at the bottom of the soil column. We used a known flux top boundary condition (implemented by discretizing Fick’s 1st Law) which was prescribed based on time interpolated chamber measurements. Each month’s measured concentration profile was used to model the next month’s first using an ignorant guess of

R (determined by solving the above with a month-long time step). We then refined the time step to 0.1 days and used a Markov-Chain Monte Carlo Metropolis Hastings (MCMC-

MH)188,189 approach with 40,000 repetitions to alter the value of R along the vertical column in order to minimize the error between the modeled future methane concentration profile and its measured value. We took the average of the 10% best performing MCMC runs as the microbial activity. Uncertainty was quantified as 1 standard deviation of the 4,000 selected runs. This simplistic model interprets observational concentration data as production/oxidation with no assumptions about the oxic conditions of the soil, providing a unique way of observing the data.

2.4.5 Eddy covariance collection and data processing

Eddy covariance data were collected from July to October 2015. The flux calculation approach was fully outlined previously190,191. Briefly, a 3D rotation was applied to wind observations to force the vertical and cross wind components gathered from the sonic anemometer (CSAT3, Campbell Scientific, Logan UT) to average to 0 for each half- hour192. To correct for the separation of the sensors, the time series of concentration

193 measurements were shifted in time using the maximal-covariance approach . Carbon

74 dioxide (net ecosystem exchange, NEE), methane, and water vapor flux (latent heat flux,

LE) were corrected as previously described194 to account for the effects of changes in the densities of dry air and water vapor. Frequency response corrections for LE and methane fluxes, which are based on concentration measurements from open-path gas analyzers (LI-

COR Bioscience, LI-7500 for water vapor and carbon dioxide, and LI-7700 for methane)

193,195 were calculated and validated as previously described . The absorbance spectrum of methane is temperature dependent. We therefore combined an absorbance spectrum correction with the WPL correction as detailed in the LI-7700 manual (LI-COR, 2010).

Day-night transition was calculated using shortwave radiation observations from the nearby meteorological station. Night was defined as when shortwave radiation dropped below 10 W/m2. The standard empirical approach of defining a seasonal thresh-old value of friction velocity (u*) that indicates an insufficient level of turbulent mixing was used to reject invalid data41. The minimum value allowed for a u* threshold was 0.2 m/s. The u* filter was used for both carbon dioxide and methane fluxes on the assumption that if the turbulence is sufficient to provide adequate mixing conditions for carbon flux measurements it will be sufficient to do the same for methane. Eddy covariance flux data was gap-filled to Morin et al.191 using the automated neural network (ANN) approach, an expanded version of the method commonly used in flux sites191,196,197, introduced by Papale and Valentini198.

75

2.4.6 Site level methane budget

To determine the site level methane budget and the oxic production contribution, we used an expanded version of the fixed frame eddy covariance scaling methodology developed for mud and open ecosites150. Briefly, this method combines eddy covariance data with a footprint method and ecosite flux measurements collected using the chamber method to decompose the eddy covariance flux signal into its contributing parts by ecosite.

We used the Detto footprint method199, which is a 2D expanded version of the Hsieh model200. We used monthly varying displacement height and roughness lengths to represent the Typha spp. growing around the tower. There were 4 relevant ecosites with footprint contributions to the eddy covariance tower: 1) open water, 2) Typha spp. 3)

Nelumbo spp., and 4) mud flat.

Equation 2

� = �� + �� + �� + �� where F’ of each ecosite indicates the relative flux strength of that ecosite at the landscape level (different than that provided by the chamber measurements), and e is the footprint contribution of that patch per day.

Chamber flux data determined the relative flux strengths of each ecosite compared to open water fluxes. This provided a solvable system of equations for the relative flux strength contributing to the eddy covariance tower (F’) with a daily temporal resolution.

Equation 3 � � � = = � � 76

� � � = = � � � � � = = � �

We interpolated each ecosites percent of production in the oxic layers to daily values over the course of the study. We scaled these values to the site level by integrating spatially over the site level footprint and temporally over all times we were able to model methane production.

Equation 4 , � � � + � � � , , , � = , � � + � � , where f is the site level percent area contribution of each ecosites and p is the percent generation in the oxic zone (determined by the diffusion model) of each ecosite.

2.4.7 Metagenomic analyses

Genomic DNA was extracted from triplicate 0.5g of soil using a MoBio PowerSoil

DNA Isolation Kit following manufacturers protocol. DNA from three representative Fall and Summer surface soil samples (plant, mud, and water) were sequenced at The Ohio

State University and The Joint Genome Institute using an Illumina Library creation kit

(KAPA Biosystems) with solid-phase reversible immobilization size selection. The quantified libraries were then prepared for sequencing on the Illumina HiSeq 2500 sequencing platform utilizing a TruSeq Rapid paired-end cluster kit, v4. We obtained 304

Gbp of metagenomic sequencing (Table 1). Sequence assembly generated ~3.8 Gbp of 77 contiguous sequences >5 kbp from the six surface soil metagenome samples. Fastq files were generated using CASSAVA 1.8.2. and were trimmed from both the 5’ and 3’ ends using Sickle, then each sample was assembled individually using IDBA-UD with default parameters as previously described183,201. Scaffold coverage was calculated by mapping reads back to the assemblies using Bowtie2202.

Genes on scaffolds ≥5 kb were annotated as described previously203,204 by predicting open reading frames using MetaProdigal205. Called genes were compared using

USEARCH206 to KEGG, UniRef90, InterproScan207 with single and reverse best hit (RBH) matches greater than 60 bits reported. The collection of annotations for a protein were ranked: Reciprocal best BLAST hits (RBH) with a bit score >350 given the highest (A) rank, followed by reciprocal best blast hit to Uniref with a bit score >350 (B rank), blast hits to KEGG with a bit score >60 (C rank), and UniRef90 with a bit score greater than 60

(C rank). The next rank represents proteins that only had InterproScan matches (D rank). The lowest (E) rank comprises the hypothetical proteins, with only a prediction from

Prodigal but a bit score <60.

Assembled scaffolds were binned into genomes based on GC, coverage, and taxonomic affiliation across samples using ESOM and Metabat208. For each bin, genome completion was estimated based on the presence of core gene sets (highly conserved genes that occur in single copy) for bacteria (31 genes) and archaea (104 genes) using

Amphora2209 using a method previously reported183. Taxonomic placement of the genome bins was based on phylogenies of 16S rRNA genes recruited from the bin and/or ribosomal protein analyses. Overages (gene copies >1 per bin) indicating potential misbins, and

78 discrepant GC and phylogenetic signal were used to manually remove potential contaminating scaffolds from the bins.

Genome-wide average nucleotide identity (ANI) values from our five >50% complete reconstructed Methanothrix genomes (Table 2) and comparisons to three existing

Methanothrix genomes were calculated from the Kostas lab calculator (http://enve- omics.ce.gatech.edu). Metabolic analyses were performed largely using the M1 genome

(Supplementary Data 5) as a model and are included in the Supplementary Note 1. To identify unique gene features in reconstructed genomes that differ from isolate

Methanothrix genomes, we created an ITEP210 database and compared all genes via all versus all blast, orthoMCL clustering and sqlite database generation. A cluster is defined as having bidirectional-best-hits based on a percent identity cut off.

2.4.8 Meta-Transcriptomic analyses

Metatranscripts were performed on three cores sampled in two ecosites (plant and mud) during two seasons (Fall and Summer) (n=12), as these soils demonstrated the highest surface methanogenesis activity. Total RNA was extracted from selected soils previously analyzed by metagenomics using ~2g. of soil via MoBio PowerSoil Total RNA Isolation

Kit and DNA Elution Kit following manufacturer instructions. For metatranscriptomics,

RNA was processed according to JGI established protocols. Briefly, rRNA was removed using the Ribo-Zero rRNA Removal Kit. Stranded cDNA libraries were generated using the Illumina Truseq Stranded RNA LT kit. The rRNA depleted RNA was fragmented and reversed transcribed using random hexamers and SSII (Invitrogen) followed by second

79 strand synthesis. The fragmented cDNA was treated with end-pair, A-tailing, adapter ligation, and 8 cycles of PCR.

We obtained 462 Gbp of metatranscriptomic sequencing (Table 1). The resulting

~150 bp nucleotide sequences were trimmed as described above (see metagenomic section) and separately mapped via Bowtie2202 to three databases (the metagenomic scaffolds >5 kb, mcrA gene database from pure cultures and metagenome scaffolds >1 kb, the

Candidatus Methanothrix paradoxum genomes) allowing for up to three mismatches. The normalized transcript abundance for each gene was calculated via Cufflinks211 using the rescue method for multi-reads and the mapped data was reported as fragments per kilobase per million mapped (FPKM). FPKM values for Methanothrix genes (from transcripts mapped to genes on Methanothrix scaffolds) were averaged across the 12 samples and ranked, with genes assigned to functional categories manually.

The top 100 most abundant Candidatus Methanothrix paradoxum transcripts in each season (Fall and Spring) were individually determined from the six samples in each season (triplicate samples in plant and mud ecosites). The resulting mean Fall and Summer log10 FPKM gene transcript relative abundance were plotted, and gene functional category was manually assigned and depicted by color. For the most expressed genes in each functional category across both seasons, we also plotted the log10 FPKM values from all

12 samples using boxplots, with mean transcript relative abundance shown by black line.

The responses of genes highly transcribed in both seasons and present in 50% of samples per season were also reported in a heatmap constructed using R pheatmap function.

80

RT-qPCR using random-primed cDNA synthesis followed by cDNA quantification was performed to quantify mcrA transcripts. Briefly, RNA was synthesized to cDNA via manufactures protocol with SuperScript III Reverse Transcriptase kit, RNase OUT recombinant ribonuclease inhibitor, and random primers. Primers for mcrA39 were previously reported. The quantification protocol and conditions followed from

Franchini163, with a slight deviation in reaction mix including 1 µl of cDNA, 10 µL 2X

SsoAdvanced Universal SYBR Green Supermix reaction buffer, 0.5 µM of each mls and mcrA-rev primer, balanced with 20 µl total volume with PCR grade H20. Serial diluted mcrA genes amplified from M. acetovorans DNA were used as standards. Statistical differences in mcrA transcript copy number between depths were evaluated via ANOVA

(df = 17, p < 0.05).

2.4.9 Phylogenetic analyses

Single gene and concatenated genes analyses were performed as previously described201. Reference datasets for the 15 ribosomal proteins chosen as single-copy phylogenetic markers (RpL2, 3, 4, 5, 6, 14, 15, 18, 22 and 24 and RpS 3,8, 10, 17, and 19), small subunit ribosomal protein 3 (rps3), and mcrA were created using sequences mined from the NCBI and Joint Genome Institute Integrated Microbial Genomes/Microbiome

(JGI-IMG/M) databases (December 2016). Each individual protein dataset was aligned using MUSCLE 3.8.31 and then manually curated to remove end gaps212. For amino acid phylogenetic analyses (S3 and concatenated ribosomal trees), we used ProtPipeliner, a python script developed in-house for generation of phylogenetic trees

81

(https://github.com/lmsolden/protpipeliner). The pipeline runs as follows: alignments are curated with minimal editing by GBLOCKS213, and model selection conducted via

ProtTest 3.4214. A maximum likelihood phylogeny for the concatenated alignment was conducted using RAxML version 8.3.1 under the LG model of evolution with 100 bootstrap replicates215 and visualized in iTOL216. For the mcrA nucleotide tree, a similar analysis was used except under the GTRCAT model.

The V4 portion of 16S rRNA genes was sequenced at Argonne National Laboratory at the Next Generation Sequencing facility with a single lane of Illumina MiSeq using 2 x

251 bp paired end reads and analyzed as described previously201. The full-length 16S rRNA sequence recovered from the Ca. Methanothrix paradoxum M1 genome was used to recover a single Methanothrix OTU from the V4 dataset that was >99% similar to the metagenomic recovered sequence. Additionally, the 16S rRNA fragment sequence from the M1 genome was searched against GenBank using BLASTN (evalue 1e-10, 100000 alignments, 100000 descriptions). Hits (>500 bp) of at least 99% identity were retained.

Genbank records for these sequences were parsed to find study title, and sequence source environment was assigned from the title and summarized (Supplementary Data 4). A maximum likelihood phylogenetic tree of Candidatus Methanothrix paradoxum 16S rRNA gene sequences was built using RAxML 8 in ARB217 using all Silva SSURef NR99

(v128) Methanothrix sequences ≥ 900 bp, the M1 16S rRNA fragment sequence from this study, with Methanosarcina reference sequences as the outgroup. Additional reference sequences from the BLAST search above were aligned using SINA218 and added to the tree using the parsimony add function in ARB.

82

Chapter 3. Methanogen diversity and function at Old Woman Creek: A genome resolved

view spanning spatial and temporal gradients

3.1 Introduction

Many studies have examined soil methanogen communities, however a majority of this research is based on marker gene analyses, such as the 16S rRNA gene. These studies have identified genera such as Methanothrix, Methanoregula, Methanobacterium, and

Methanomassiliicoccus while correlating the abundance of specific methanogens to measured geochemical factors or substrate distributions along wetland gradients43,219–221.

Only recently has the in situ metabolic potential of methanogens been explored via the application of metagenomic sequencing and assembly in soil systems222–225.

The application of metagenomics to ecosystem science has uncovered new insights into the ecophysiology of methane-cycling organisms, several of which have predictive consequences on an ecosystem scale. For instance, recent work in this area has expanded our knowledge of methanogenic archaeal diversity and physiology. Methanogens have been inferred to occur in the Bathyarchaeota, which contained the mcrA gene and metabolic machinery for methane production222, representing the first time methanogen diversity extended beyond the phylum Euryarchaeota226. Similarly, one year later, Vanwonterghem et al characterized the archaeal phylum Verstraetearchaeota, demonstrating some of these genomes also contained novel mcr genes and the pathway for methylotrophic methanogenesis42. Conversely, the genomic potential for methane oxidation through a reverse methanogenesis pathway was also found in anoxic marine sediments227, termed

83 anaerobic methane-oxidizing archaea (ANME). More recent genomic work on these organisms has detailed their potential utilization of electron donors sulfate, iron, and nitrate228. These advances clearly highlight the important insight metagenomics has contributed to our knowledge of archaeal methane cycling in the past five years.

Consistent with our research in Chapter 2, there is growing body of metagenomic- based evidence that much of the methane production in northern latitude methanogenic ecosystems, e.g. peat and permafrost, may be contributed by a single dominant methanogen species in each ecosystem223,229. This finding may have important ramifications for modeling, as despite the large diversity in soils, methanogenesis may be confined at least under narrowly defined timescales (yearly) to a few select taxa. It is clear today, especially highlighted by the recent paper by Parks et al (which recovered 8000 new genomes)230 that recovering high quality genomes from metagenomes is not an insurmountable barrier even in complex ecosystems like soils. This rapid evolution of genome reconstruction from complex environments should increase our ability to accurately identify the important methane cycling organisms in vital soil ecosystems such as wetlands. Levering these genomic in accordance with corresponding geochemical data can aid model development by rectifying the accuracy of processes underlying model predictions.

Beyond sampling the genomic potential, a handful of studies have linked expression data to reconstructed genomic data from complex soil systems as we have done here. A study by Mondav229 et al used metaproteomics and metagenomics to discover a novel methanogen – Candidatus Methanoflorens stordalenmirensis – and with isotopic data showed this methanogen was highly active in thawing permafrost. In addition to

84 metaproteomics, metatranscriptomes can also be mapped back to reconstructed genomes, enabling insight into which pathways are transcribed under different environmental regimes. The few cases in soil systems where these methods have been applied include examining active methanogenic pathways in Arctic peat231,232 and European temperate wetlands233. Work by Tveit et al232 combined metagenomics and metatranscriptomics to demonstrate a shift in the identity of active methanogens across a temperature gradient.

Hultman et al223 combined metagenomic, metatranscriptomic, and metaproteomic techniques to detail the identities of the active methanogenic community in response to a variety of simulated warming conditions. Work on the applications of ‘omics approaches to peatland methane cycling was summated by Mackleprang et al224 in 2016. Very recently, work by Tas et al225 utilized these methods to recover novel genomes and link microbial processes to landscape microtopography and greenhouse gas emissions from tundra soil.

These relatively few studies highlight the value of assessing the dynamic activity of methanogens under in situ conditions across numerous spatial and temporal scales, and how that activity ultimately affects methane emissions from these sites. While Chapter 2 shed light on the contribution of the methane paradox to overall methane emission in Old

Woman Creek, Chapter 3 more holistically interrogates the methanogen genomes recovered across the surface and deep soils.

In chapter 2 we observed Candidatus Methanothrix paradoxum dominates the oxic surface soil mcrA transcript abundance and actively performs a wide variety of metabolic activities. In this chapter, we extend the activity of this organism to deeper soils and provide a more detailed metabolic model for this genome. In addition to the further investigation

85 of Candidatus Methanothrix paradoxum, we also determined the methane production potentials of the surface and deep soils under standardized anoxic conditions using a cultivation-based approach. We have updated the metagenome-recovered mcrA gene tree

(independent of binning), including deep-soil recovered mcrA sequences, and determined the transcriptional activity of these sequences in surface and deep soils as an alternative to genome-binned view of the soils. To expand our knowledge on the roles of the other methanogens which inhabit the wetlands, we have included additional taxonomic, metabolic, and activity analyses of 16 methanogen genomes reconstructed from both surface and deep soils of Old Woman Creek.

3.2 Results

3.2.1 Methane production potentials are typically greater in oxic surface soils than

corresponding deeper soils

In chapter 2, we demonstrated that surface soils at Old Woman Creek had ~9 times more mcrA transcripts compared to deep soils over the Fall and Summer sampling seasons.

Using the same soils as examined in Chapter 2 (Fall 2014, Summer 2015) the methane production potential (MPP) from these soil communities was examined under redox- standardized conditions. Here, uninoculated soil was incubated under anoxic conditions

(N2 headspace) at 20°C. By not adding additional substrate we hoped to discern the impacts of native carbon on methane production and sustainability in these soils. The headspace from triplicate reactors for each sample was sampled for methane at Day 0, Day 10, and

Day 40, and methane production rate over that time was calculated.

86

Almost uniformly, MPP from surface soil communities were greater than there corresponding deep soil counterparts (Figure 20), with the only exception being Fall plants, in which the deep soils had a significantly greater 10-day production rate than their corresponding surface soils. Across the ecosites, the highest methane production rates were observed in Fall mud and Fall open water surface soils, which were significantly greater than the deep counterparts (Figure 20, ANOVA p < 0.05). Summer plant surface soils also produced significantly more methane than their deep counterparts. MPPs from OWC were more than two orders of magnitude greater on average than the highest reported rates from coastal salt marsh soils234 over comparable time frames (7 days in marsh soils, 10 days in

OWC soils, both were incubated at similar temperatures). Similarly, the highest previously reported 50-day MPP was from a Chinese wetland soil, which was approximately two orders of magnitude less than that of OWC soils over a 40-day incubation219, again with similar temperatures of incubation (25°C and 20°C, respectively). These data provide more additional evidence on top of previously published data8,136 to demonstrate the high methane production and emission capacity of Old Woman Creek soils.

Over both seasons, there was a consistent trend that methane production occurred on average four times faster rate during the Day 0 to Day 10 window than over the full 40- day testing window. These rates may have hinted at differences in carbon availability between the surface and deep soils. The MPP surface soils continue to produce methane up to and past the Day 40 measurement point, where most deep soils stopped producing methane between Day 10 and Day 40 measurement points. While total carbon

87

20 Fall Summer ) -1 *

day * -1 15 -C g 4

10 *

* 5 Day 10 Methane Production Potential (ug CH

Day 10 Day 10 Day 40 Day 40 0 Day 40 Day 10 Day 40 Day 10 Day 40 Day 10 Day 40

Figure 20 – Fall and Summer methane production potential (MPP) rates MPP experiments quantified methane production rates over a 40 day period. Green, orange, and blue represent plant, mud, and open water ecosites, respectively, while lighter and darker shades represent surface and deep depths within each ecosite, respectively. Initial production rates (Day 0 to Day 10) and total production rates (Day 0 to Day 40) were calculated for all ecosites in the Fall and Summer seasons. The surface soils (denoted by a lighter color) with significantly greater methane production rates than corresponding deep soils are designated by * (ANOVA p < 0.05). These data are also found in Supplementary Dissertation Table 4.

88 concentration did not differ statistically within an ecosite for each season (p>0.05

ANOVA), across all ecosites the surface soils had 37% more total carbon than deeper soils.

We think it is unlikely that this statistically-insignificant difference in total carbon between surface and deep soils was the sole contributing factor to the significantly greater MPP rates in surface soils. Rather, we propose that the molecular composition of this surface carbon ultimately resulted in the significantly greater acetate concentrations in surface soils

(p<0.05 ANOVA) and sustained methanogenesis in our laboratory MPP. In summary, our

MPP results suggest that the molecular formula, rather than concentration, may be an important yet underappreciated factor sustaining methanogenesis in these wetland surface soils.

3.2.2 An expanded view of the wetland mcrA activity, including deep soils

It has been previously demonstrated that surface soils at OWC contribute up to 80% of the methane to the total wetland methane budget8 and the organism contributing ~84% of the total mcrA activity in these surface soils is Candidatus Methanothrix paradoxum.

These findings however also suggested that other methanogens were active and prevalent in surface soils. For instance, in Chapter 2 (Figure 15), we showed on average 16% of the surface mcrA transcripts could be attributed to other methanogen genera, with sequences representative of Methanoregula, Methanolinea, and Methanomassiliicoccus comprising the majority of that 16%. We also demonstrated in Chapter 2 (Figure 17) that rps3 genes representative of Methanoregula, Methanolinea, and Methanomassiliicoccus were detectable members of the soil surface community. In this chapter, I wanted to focus more extensively on these methanogens. First, I compiled mcrA sequences obtained from our 89 surface soils with sequences obtained from metagenomes from deep (~25 cm) soils collected in 2013 (kindly provided by my collaborator, Dr. Miller). In addition, since the data in Chapter 2 was published, the Wrighton laboratory has also obtained the first metatranscript data from deep soils (Summer plants, Aug 2015). Together these new data offered an opportunity to assess methanogen diversity and activity in both surface and deep soils.

Previous metagenomic analyses recovered 54 mcrA sequences reconstructed from our prior surface metagenomic assemblies. The addition of deep metagenomes acquired from Fall 2013 samples added 147 additional mcrA sequences to the collection, including

46 partial (>50%) and 18 near-complete to complete (>80%). Newly acquired deep soil mcrA sequences were most similar to prior sequences previously assigned to

Methanomassiliicoccus, Methanolinea, Methanoregula, Methanothrix, and ANME-2D.

To assess the activity of the mcrA sequences in our revised database, metatranscriptomic reads from 5 season, ecosite, and depth combinations (n=15 in total) were once again mapped to the updated mcrA sequence pool. Specifically, these transcripts now included metatranscripts from triplicate deep samples (Summer 2015 plant), as well as the original 12 surface metatranscript samples described in Chapter 2. Of the 201 mcrA gene sequences meeting the length criteria (>33.3%), 97 recruited transcripts with default coverage parameters in at least 1 of the 15 samples. Interestingly, discreet mcrA clades showed depth-specific activity. Of the 15 Methanomassiliicoccus sequences recovered from deep metagenomes, 3 distinct clades emerged, each respectively demonstrated deep- specific activity, surface-specific activity, and sequences that were active at both depths

90

(Figure 21). Additionally, there was another Methanothrix species (distinct from

Candidatus Methanothrix paradoxum) recovered from deep soils. These 11 new

Methanothrix sp. sequences were primarily active in deep soils, but did show some minor activity in surface soils as well. These findings of surface and deep differentiated

Methanothrix strains corroborates our team’s earlier 16S rRNA based publication, which observed two abundant strains of Methanothrix that were depth-resolved across the wetland43.

Interestingly, the 33 mcrA sequences most closely related to an ANME-2D draft genome (JGI IMG Genome ID 2505380168) were transcribed almost exclusively in deep soils. Additionally, a collection of 22 mcrA sequences which are most closely related to but divergent from Methanolinea sequences show transcription primarily in deep soils, as was also the case with an additional clade of 15 sequences most closely related to

Methanolinea tarda. Similar to the findings of Methanomassiliicoccus, 3 distinct clades of sequences closely related to Methanoregula boonei were active in either surface, deep, or both depths. To visualize differences in these transcript abundances, normalized deep soil metatranscript (n=3) and surface soil metatranscript (n=12) mean FPKM values for the distinct mcrA clades are shown on Figure 21 in blue (deep) and red (surface) soils

(Supplementary Dissertation Table 5).

Remarkably similar to previous analyses with the prior mcrA sequence database8, sequences from the Candidatus Methanothrix paradoxum clade represented 84% of all mcrA surface soil activity. Only 6 methanogen clades constituted greater than 1% of the surface soil transcript abundance – two Methanoregula clades (5% and 2%), two

91

Methanomassiliicoccus clades (3% and 1%), and the divergent Methanothrix clade (2%).

It is worth noting that our view of surface transcript abundance and the organisms that contribute to this activity was not significantly altered by the addition of new mcrA sequences.

Transcripts from the deep soil samples were more diverse than transcripts from the corresponding surface soils, although the genus Methanothrix collectively dominated mcrA transcription accounting for just over 50% of the total mean FPKM in the deep soils. The divergent, non-pardoxum Methanothrix species recovered from the deep metagenomes contributed 27% of the total mean FPKM, followed closely by Candidatus Methanothrix paradoxum sequences with 24% total mean FPKM. The next most abundantly transcribed clades where the Methanoregula branch clade (16%), the deep-active Methanolinea clade, the Methanomassiliicoccus clade active in both surface and deep soils (8%), and the

ANME-2D sequences (8%). It is important to note that, while expressing mcrA here, the overall metabolism of these ANME-2D genomes cannot be concluded from this analysis alone. While some have shown that some ANME can produce methane235, recent work has demonstrated organisms from this OWC clade are responsible for methane oxidation coupled to nitrate236–238 or iron239 as electron acceptors. The transcript abundance of each mcrA sequence clade is also visualized by season and ecosite discretely (n=5) in Figure

22.

92

sequence sequence transcription from surface and deep soils, visualized by sequence placement on a

mcrA recovered recovered -

Metagenome

–

93 21

Figure RAxML nucleotide tree. On the left, a phylogenetic tree with mcrA reference nucleotide sequences from isolated methanogens and these wetland metagenome-derived sequences denoted by metagenome of acquisition. Sequences are grouped if many sequences exist within a localized branch, denoted as “OWC group name”. OWC sequence clusters are furthermore colored by whether they were found active exclusively in deep soil (blue), exclusively in surface soil (red), or in both surface and deep soils (purple). Bootstrap values are reported on the branches. The bar chart on the right summarizes the normalized transcript abundance (scale from 0-800,000 fragments per kilobase per million mapped (FPKM)). Meta-transcript data are the mean from cores collected in each ecosite and season (surface soils n=12, deep soils n=3). The FPKM table used to generate this figure is available in Supplementary Dissertation Table 5.

94

ANME_SUMS

Methanoflorens_Deep

Methanolinea_branch_Deep

Methanolinea_tarda

Methanolinea_tarda_Deep FPKM Methanolinea_tarda_Surface 0 Methanomassiliicoccales_both 2e+05 Methanomassiliicoccales_deep 4e+05 Methanomassiliicoccales_Surface 6e+05 Methanoregula_boonei 8e+05 Methanoregula_branch_Deep

Methanoregula_formiculum

Methanoregula_formiculum_Deep

Methanoregula_formiculum_Surf

Methanothrix_deeper

Methanothrix_paradoxum

vg a vg vg vg vg a a a a

all_P_ all_M_ F F Sumr_P_ Sumr_M_ Sumr_P_D_

Figure 22 Seasonal, ecosite, and depth-resolved view of metagenome-recovered mcrA sequences

Metagenome-recovered mcrA sequence transcription from surface and deep soils, grouped by phylogenetic clade demonstrated on the mcrA tree (Figure 21) and quantified as mean (n=3) FPKM for a given season, ecosite (P = plant, M = mud), and depth (surface = red, deep = blue) combination. Candidatus Methanothrix paradoxum dominated the abundance of mcrA transcripts in surface soils, consistent with previous analyses. These data are also available in Supplementary Dissertation Table 5. 95

3.2.3 Expanding the known genetic potential and activity of the dominant wetland

methanogen – Candidatus Methanothrix paradoxum

Previous work demonstrated the contribution of Candidatus Methanothrix paradoxum to the overall methanogenic activity of the wetland, which has been reinforced through the expanded and updated data found in Chapter 3. Here, I examine in detail the genetic potential and activity of this critical methanogen based off our genome reconstruction and metatranscriptomic mapping efforts, the cumulative efforts of which are highlighted in Figure 23.

To evaluate the genetic potential of this surface methanogen, the genes specific to methanogenesis were summarized. All physiologically characterized Methanothrix species are considered obligatory acetoclastic methanogens. Consistent with this, our OWC genome contained genes for the incorporation of acetate, including acetyl-CoA synthetase

(acs) gene. The genome does not contain the alternative, 2-stage acetate incorporation mechanism mediated by the products of the acetate kinase and phosphate acetyltransferase

(ackA and pta genes, respectively). The genome also contains the genes for the 5 subunits of the acetyl-CoA decarbonylase/synthase complex (cdhABCDE), necessary to cleave the

240 acetyl-CoA molecule and transfer the methyl group to H4SPT . Furthermore, the Acs and

Cdh complex was shown to be actively transcribed via transcript mapping to the genome.

While several genes required for formate utilization were retained in the genome, only 2 of the genes necessary for this process were found active in an inconsistent and unabundant fashion. None of the genes required for utilization of methylated compounds (mtaABC, mtbA, mttBC, or mtmBC) were detected in the genome. 96

Methanothrix paradoxum genetic potential and activity in oxic surface soils. activity genetic oxic potential Methanothrix in paradoxum and

Candidatus Visualization of of Visualization

–

23

Figure 97

Individual genes for each category of biological function are numbered 1-208. Genes not identified in the genome are shown as white boxes, while genes found in the genome are denoted in black boxes. Genes determined to be transcriptionally active in surface soils are colored in red. The following complexes required for acetoclastic methanogenesis are represented as such - acetyl-CoA synthetase and CHD complex: genes 1 and 4-9, the Mtr complex: 10-17, the Mcr complex: 18-23, the Hdr complex: 24-28, the Fpo complex: 29- 40, and the V/A type ATP synthase complex: 41-49. Oxygen detoxification genes 111- 121 represent superoxide dismutase, catalase, rubredoxin, rubrerythrin, thioredoxin, thioredoxin reductase, peroxidase, glutatredoxin, peroxiredoxin, desulfoferrodoxin, and ferredoxin respectively. Protein repair genes 122-135 represent thermosome subunits A, B and D, heat shock protein 20, chaperone proteins dnaJ and dnaK, cochaperone GrpE, prefoldin subunits alpha and beta, proteasome subunits alpha and beta, proteasome activating nucleotidases alpha and beta, ATP dependent protease lonB, and HtpX protease respectively. A full list of genes represented here are available in Supplementary Dissertation Table 6.

Continuation of the methanogenic metabolism proceeds through the tetrahydromethanopterin S-methyltransferase (Mtr) enzyme, encoded by the mtrA-H subunits. The genome of Candidatus Methanothrix paradoxum contains the first 5 of these subunits (mtrABCDE), with 2 of the subunits showing activity. While activity of subunits

A-H have previously been detected in Methanothrix49, only subunits A-E were shown to be conserved in all methanogenic archaea46. It has been demonstrated that subunits BDGA are typically found within their own operon, perhaps meaning that the slight incompleteness of the genome may have contained the genes for the remaining undetected subunits.

The methyl-coenzyme M reductase complex (Mcr) is the canonical methanogenic enzyme, as it results in the direct generation of methane. The genes encoding this complex and an affiliated associate enzyme (mcrABGCD and atwA) were all present and active consistently across the samples. A downstream complex required to reduce the 98 heterodisulfide formed by the activity of the Mcr complex (Hdr) was present and active as well. In this case, the hdrDE-encoded complex was solely active, not the hdrABC-encoded complex which is also present in the genome. The hdrDE complex is the one traditionally

241 used in Methanothrix , consistent with our data. Furthermore, subunits A-O of the F420H2 dehydrogenase complex (fpo) were present within the genome, with 6 shown active. As expected based on previous literature, the “F” subunit was not found present, staying consistent with other Methanothrix which are expected to utilize a “headless” version of the Fpo complex47. Also consistent with prior work, the genomes did not contain additional hydrogenases such as Ech, Rnf, or Vho/t47. Finally, the V/A-type H+/Na+ transporting

ATPase (atpA-L) were present and active.

To summate the findings of central carbon metabolism, as expected, numerous genes were present but the TCA cycle was incomplete and transcription of some of these

TCA cycle genes was occurring. While genes for the pentose phosphate pathway were present, they were not found to be active. Seven genes required for pyruvate metabolism were present, however only 4 were detected active. Additionally, genes required for the

TCA cycle were incomplete and uniformly inactive. Several genes for tRNA and amino acid biosynthesis and transport mechanisms were detected, fully detailed in Supplementary

Dissertation Table 6.

3.2.4 Examining the established and potential oxygen tolerance mechanisms of

Candidatus Methanothrix paradoxum

To cope with potential exposure to oxidative stress agents, methanogen genomes contain numerous oxygen detoxification mechanisms capable of negating oxidative stress 99 of many forms30. Since an overwhelming majority of data related to oxygen detoxification activity in methanogens has been performed under lab conditions, there is little data about the usage of these genes under in situ, field-relevant conditions. As briefly discussed in

Chapter 2, our M1 genome did not contain abnormally high copy numbers of oxygen detoxification genes compared to M. concilii, and even lacked a catalase gene. Genome

M2, retrieved from mud soil, contained a catalase gene and additional copies of peroxidase, rubredoxin, thioredoxin, and peroxiredoxin compared to M1. While tempting to speculate on the usage of additional detoxification gene copies in a seasonally-exposed mudflat, we do not see consistent or highly-transcribed copies of these genes in methanogens across surfaces soils from any seasons or land coverage type tested (including genes found on unbinned scaffolds). While there is precedence of Methanothrix expressing energetic and acetoclastic methanogenesis genes in concert with genes for oxygen detoxification and protein repair157, we did not observe such oxygen detoxification mechanisms highly or consistently active under field conditions. Both M1-binned rubrerythrin genes showed low- abundant mapping transcripts, but only in 3 or less samples. Genomes M2, M3, and M6 also had sparing thioredoxin reductase transcripts. Genomes M2 and M3 had a peroxiredoxin that had sparing transcript mapping.

In addition to traditional oxygen detoxification mechanisms, other mechanisms of surviving bulk-oxic zones have been proposed. Numerous general signaling pathways have been demonstrated to be response to oxidative stress in archaea, including a suite under control of the RosR transcription factor242. Genome M1 contains 5 genes found via blast

(evalue 1e-5) similar to genes under control of RosR with 2 detected in the transcriptome

100

– a cold shock protein (CspA) and a response regulator most similar to Hlx2. These genes were only detected in 1 and 3 metatranscriptome samples respectively however, so if they do confer some measure of oxygen tolerance, the conditions were not frequently encountered.

Several highly transcribed Candidatus Methanothrix paradoxum genes documented here serve to repair misfolded proteins in response to stress conditions, including oxidative stress. Heat shock protein 20 (hsp20) is active under cold temperature stress, performing chaperone-like protein folding243 as well as some evidence to suggest that heat shock proteins are capable of repairing mistranslated and improperly folded proteins due to severe oxidative stress244. Universal stress protein A (uspA) has been described as a small cytoplasmic protein capable of increasing cell survival under multiple long-term stresses245 which has been previously demonstrated to be active in industrial methanogenic reactors246. Although not thoroughly studied in methanogens, mutation experiments of UspA have disrupted the cells ability to survive starvation and stress from a number of conditions, including oxidative stress247. Thermosome gene subunits (thsA,B, and D), encode subunits of archaeal-specific chaperone proteins which also function in helping rescue proper protein folding. Proteasome subunit beta (psmB) are used to create a proteasome complex with hydrolase activity broadly throughout archaea248, with a primary function for targeted protein breakdown249. The chaperone family hsp70 and a corresponding co-chaperone (dnaK and dnaJ) help to ensure that improperly folded proteins with erroneously-exposed hydrophobic zones are properly refolded. These proteins have been previously documented as up-regulated in methanogens during times

101 of cold stress250. Cold shock protein subunit A (cspA) respond to changes in temperature and contain DNA-binding domains, thought to help influence transcript stability.

Proteosome-activating nucleotide genes (panA and panB) are though to function similarly in regulating the change from growth to stationary phase but may differ in protein substrate or binding affinity251. The ATP-dependent protease (lonB) is a membrane protease known to regulate membrane proteins and regulators of the stress response in H. volcanii252. While speculative, it is not without precedent that these Candidatus Methanotrhix paradoxum genes highly transcribed in oxic surface soils could be helping to limit the exposure to or damage resulting from oxygen exposure in some yet undiscovered way.

3.2.5 Examining the unannotated yet highly active genes of Candidatus

Methanothrix paradoxum

In addition to genes speculatively contributing to tolerating oxygenated conditions, many completely unknown or hypothetical genes were some of the most highly transcribed genes of Candidatus Methanothrix paradoxum in oxic surface soils. Each unannotated amino acid sequence representing the genes of unknown function from the top 100 transcripts were submitted to Phyre2139, a program which identifies likely domains in the predicted protein structure and reports functional domains. Each amino acid sequence submitted to Phyre2 yielded the 10 most likely domain results, categorized by potential function (Supplementary Dissertation Table 7). Phyre2 predicted annotations of the amino acid sequence of unknown transcripts covered an average of 23% of each sequence, ranging from 12 to 44%. The best hits to these unannotated transcripts were to hydrolases, cell signaling molecules, transcription regulators, replication proteins, type ii secretion 102 systems, methyltransferases, and amino acid synthase proteins. Although highly speculative, these domain level predictions offer new insight to the role of these unannotated yet highly transcribed methanogen genes.

3.2.6 Exploring additional potential mechanisms for the survival of Candidatus

Methanothrix paradoxum in wetland soils

Based on the absence of transcripts for oxygen reduction or tolerance mechanisms, we hypothesized that methanogens may subsist in bulk-oxic environments by localizing within anoxic microsites, an idea explored to some extent previously in Chapter 1. This idea for these OWC soil was inspired in part because of work with granular sludge, where studies indicated up to a 41% oxygen headspace only caused a 50% inhibition of methanogenic activity70. This hypothesis of anoxic microsites to sustain methanogenesis has been widened to include environmental habitats like soils, where these pockets can provide protection from drying events65. While difficult to prescribe specific gene activity to facilitating life in microsites, we examined our OWC methanogen genomes for a number of previously identified physiological features or adaptations in other characterized

Methanothrix genomes. In sludge, Methanothrix play a vital role in the physical structure of the granules by forming long filamentous cells, and these could help in the formation of sheaths limiting penetration of diffusible oxygen or providing structural support for communities in microaggregates. The sole protein responsible for filament physical structure is MspA253. Surprisingly, neither genome M1 or M2 contained any predicted amino acids with a blast result to MspA (evalue >1.0E-5). Furthermore, the transition from short cells to longer, filamentous cells is carried out via quorum sensing, where 103 carboxylated acyl homoserine lactones act as the signaling agent159; a process known to dramatically increase methane production in reactor environments254. A filI-filR quorum sensing system regulates filament formation in this strain, with the product of the filI gene being an AHL synthase enzyme159. While genome M1 had 5 potential candidate genes returned via blast result >bit score 60 to filI, none were detected in the metatranscriptome.

In fact, M1 only had 1 sensor gene found to be transcriptionally active (scaffold 185_20) which is annotated as a response regulator domain with a PAS sensor, although the true annotation is not well resolved. Interestingly, this gene was consistently, abundantly, and solely active in the Fall plant samples. Taken together, this data indicates that M1 either utilizes drastically different means to produce classical Methanothrix filament structures or it does not create long filamentous cells at all in this natural habitat, potentially having a different cell morphology than wastewater Methanothrix spp. Regardless, we currently lack evidence to suggest that filamentous structure is a means for M1 to cope with potential oxidation stress. Another function of Methanothrix which may be helpful for creating or sustaining anoxic microsite habitats is adhesion mechanisms, however no genes for

Methanothrix cellular adhesion mechanisms have yet to be reported, hindering efforts to understand potential adhesion principles in our natural system. The collective lack of understanding about how the dominant methanogen is capable of activity in natural conditions reinforces the need to improve means for isolation so that the numerous unknown capabilities may be physiologically and genetically identified.

104

3.2.7 The taxonomy, novelty, and genetic potential of other methanogen genomes

recovered from the wetland

Given the insights gleaned from a Candidatus Methanothrix genome recovery, we then attempted to recover genomes from other methanogen species in surface and deep soils. A total of 16 unique methanogen genomes were reconstructed either directly from field samples or from laboratory methanogen enrichments which were determined to be >

50% complete (Supplementary Dissertation Table 8). All field-derived genomes

(excluding enrichment-derived genomes with higher contamination values) had < 30% contamination. All but two genomes contained S3 ribosomal proteins, and four genomes contained mcrA sequences. For most of the genomes, taxonomy was able to be clearly established using phylogenetic marker genes for methanogens (e.g. S3, mcrA, or 16S rRNA), which included four Methanolinea (Lin-34, Lin-89, Lin-124, and Lin-44), two

Methanomassiliicoccus (Mx-87 and Mx-72), two Methanoregula (Reg-114 and Reg-En), and five Methanothrix including 3 Candidatus Methanothrix paradoxum (Thrix-135,

Thrix-5, M1, M4, and M7). The remaining three genomes were confidently able to be assigned to Order Methanomicrobiales, however further taxonomic assignment was not confidently possible based as the predicted taxonomy of their single copy genes or examined marker genes. These genomes – representing unknown members of the Order

Methanomicrobiales, were designated as Umm-69, Umm-113, and Umm-71. Genomes

M1, Reg-114, and Umm-113 contained 16S rRNA gene sequences (1,402 bp, 302 bp, and

1,072 bp respectively).

105

Of the genomes recovered by these analyses, 2 were the previously-examined M1 and M4 genomes, representatives of Candidatus Methanothrix paradoxum8. Another

Candidatus Methanothrix paradoxum genome was reconstructed in the new analyses, here labeled as M7. We also recovered more partial Methanothrix genomes Thrix-5, from soils following an anoxic laboratory enrichment, and Thrix-135, from deep soils.

Methanolinea is a genus of hydrogenotrophic methanogens inferred from mcrA transcript data to be active in wetland surface and deep soils. Of the 4 reconstructed

Methanolinea genomes, 2 came from surface soils (Lin-124, Lin-44) and 2 came from deep soils (Lin-34, Lin-89). All Methanolinea genomes contained genes for utilization of carbon dioxide and formate to drive hydrogenotrophic methanogenesis and did not show evidence of utilization for methanogenic substrates acetate or various methylated compounds, consistent with the physiology of the isolated representative255. These data are consistent with concentrations of formate in the Summer and Fall soils ranging from undetectable to

55 mM. As also expected, the genome also contains pathways for the production of methanogenic accessory compounds methanofuran and coenzyme F420. Lin-124 contained an ATP-utilizing potassium transport system noted previously to potentially help combat low inorganic nutrient availability in peatlands256. Lin-124 did not contain any detectable energy-conserving hydrogenase (ech) subunits as expected to be found in a

Methanolinea genome, however these were detectable in Lin-34 (recovered from deep soils), suggesting the genome was likely incompletely sampled. Like previous surface soil recovered genomes, these genomes also contained numerous oxygen detoxification mechanisms such as superoxide dismutase, catalase, and rubredoxin.

106

Two other hydrogenotrophic methanogens had reconstructed genomes, and were most similar to the genus Methanoregula (Reg-114 from deep soils and Reg-En obtained from an enrichment culture). The 322 base pair 16S rRNA gene fragment found within the genome shared 99.3% similarity with the overlapping portion of a Methanoregula boonei

6A8 reference sequence acquired from Silva, however this is a small fragment. Their genomes contained many of the required genes for hydrogenotrophic methanogenesis, but only partially complete, probably due to the incompleteness of the genomes (~70% complete).

Of the 3 methanogen genomes grouped in the “Umm” group (Unknown

Methanomicrobiales), genome Umm-113 contained a 1,072 base pair 16S rRNA gene fragment with a most similar Silva database match (only 92.6% identity) to an unknown environmental Methanomicrobiales sequence, reinforcing the novelty of this genome. We could not accurately assess what genus or species genomes Umm-69 and Umm-71 would be, presumably because of phylogenetic novelty. As expected for being members of Order

Methanomicrobiales257, these genomes contained genes consistent with hydrogenotrophic methanogenesis including mch, mtd, mer, and frh. Although all three genomes contained

S3 ribosomal proteins, the predicted most accurate phylotypes for these genes differed between genera Methanoculleus (98% similarity), Methanoculleus (52% similarity), and

Methanoregula (88% similarity) for Umm-69, Umm-71, Umm-113 respectively. The mcrA gene sequence contained within Umm-113 most closely resembled that of a Methanolinea.

Two Methanomassiliicoccus genomes were reconstructed, both from deep soils. As expected based on taxonomy, these genomes do not contain the required genes for

107 hydrogenotrophic or acetoclastic methanogenesis, but rather methylotrophic methanogenesis. Specifically, the Mx-87 genome included genes annotated as a trimethylamine methyltransferase (mttB) as well as single putative copies of monomethylamine and dimethylamine corrinoid methyltransferases, however no evidence of methanol utilization through MtaABC. On the contrary, Mx-72 contained a gene annotated as mtaA for methanol utilization, but did not contain detectable genes for monomethylamine, dimethylamine, or trimethylamine utilization. In accordance with prior reports, Mx-87 contained subunits for the methyl viologen-dependant hydrogenase (Mvh),

Hdr complex, and homologs to Fpo complex subunits233.

Overall, the genetic potential of the additional methanogens genomes recovered from wetland soils were as expected based upon taxonomy from a methanogenic pathway perspective. However, for the novel genomes, there is clearly more research to be done in elucidating the metabolic function. Additionally, more detailed metabolic analyses and metatranscriptomic data will be needed to understand the in situ metabolism of these organisms. It is worth nothing that many of the methanogen genera for which genomes were recovered are genera we may expect to find in methane paradox sites, based off previous literature discussed in Chapter 1. To expand upon work started here, comparative assessments of related genera could be conducted to determine unique genes present in the genomes of interest as was performed with Candidatus Methanothrix paradoxum via iTEP210 or by newly developed, multi-parameter assessment methods such as ezTree258.

Additionally, more robust forms of phylogenetic determination efforts, such as

108 concatenated ribosomal trees, could be performed on the three “Umm” genomes to determine a most likely taxonomy.

3.2.8 The activity of other methanogen genomes recovered from the wetland

All 16 reconstructed genomes were pooled for transcript mapping and 11 of them had at least 1 reported successful transcript mapped. The three Candidatus Methanothrix paradoxum genomes recovered the greatest number of unique transcripts with M1, M7, and M4 genomes recruiting 182, 123, and 95 transcripts detected respectively. Of the non-

Methanothrix genomes, Lin-34 had the most genes detectably transcribed, with 48 unique genes (Table 3). To simplify downstream analyses, a transcript had to be detected in at least 3 of 15 metatranscript samples to be considered present.

The expression of key methanogenesis genes is visualized in Figure 24. Transcripts for the acetoclastic marker gene cdhA were only detected in the M1 genome. Six total genomes, expected to be hydrogenotrophic based on their presumed taxonomy, demonstrated detectable transcription of hydrogenotrophic marker genes mtd, mer, or both in the case of Lin-124 and Umm-69. There was no consistent pattern of surface or deep exclusivity based solely off these data. Additionally, key marker genes mcrA and mcrB were also examined, revealing evidence for deep methanogenesis in the Umm-113 genome but at least limited surface activity by genomes Lin-124, Lin-34, and Umm-71 in addition to the already known activity of the M1 genome. Of special note, Methanolinea genome

Lin-124 contained an mcrA sequence depicted in Figure 21. Furthermore, evidence of active cell division via genes minD or ftsZ was observable for genomes Lin-34, Umm-71, and Umm-69 exclusively in deep soils whereas the three Candidatus Methanothrix 109 paradoxum genomes and Lin-124 showed transcription of these same genes in surface soils, signifying active cell division. It should be noted that although two high quality

Methanomassiliicoccus genomes could be reconstructed in deep soils, there was almost no detected transcript that mapped to these genomes, suggesting these organisms were not active under the in situ conditions at the time of sampling events.

Aside from methanogenesis and cell division genes, there were no additional methanogen oxygen tolerance genes – including superoxide dismutase, catalase, rubreythrin, etc. detected in transcript data. While this is the case, much like observed in

Candidatus Methanothrix paradoxum genomes previously discussed, there were a number of active protein repair and recovery genes active in newly-reconstructed genomes.

Genomes Lin-124, Lin-44, Umm-69, Umm-71, and Lin-34 also transcribed protein repair genes, although none of these were transcribed in significantly greater abundances in surface soils. Aside from methanogenesis genes and some protein repair genes, there were numerous transcription and translation factors transcribed, as well as ATP synthases, DNA polymerases, and ribosomal proteins, indicating these methanogens were likely overall metabolically active. Also of note, transcripts for archaeal flagellin genes were recovered in four genomes (Lin-34, Lin-44, Umm-69, and Umm-113), suggesting motility may be important at least in part for some methanogens. A complete list of genes detected transcribed in the methanogen genomes under in situ conditions is found in Supplementary

Dissertation Table 9.

110

Table 3 – Unique genes detected as transcribed in wetland methanogen genomes

# of Genome Genome detected code transcripts M1 7_Metabat_PLANT_2014_11.fa 182 M7 21_Metabat_MUD_2014_2015_coassembly.fa 123 M4 4_Metabat_PLANT_2015_08.fa 95 Lin-34 M3D3-D4_reassembled_scaffolds_ge1kb.fasta 48 Umm-69 69_Metabat_MUD_2014_2015_coassembly.fa 42 Umm-71 C4D4_metabatSS.71.fa 35 Lin-124 124_Metabat_PLANT_2015_08.fa 12 Lin-44 44_Metabat_PLANT_2015_08.fa 12 Umm-113 O3D3_metabatSS.113.fa 6 Mx-87 O3D3_metabatSS.87.fa 1 Mx-72 O3D4_metabatSS.72.fa 1

111

Genome Gene Function Plant Mud Plant Mud Plant acetoclastic M1 cdhA methanogenesis

Lin-34

Umm-71 mtd Umm-69

Lin-124 hydrogenotrophic methanogenesis Umm-113

Lin-34 mer Umm-69

Lin-124

Umm-113 M1 mcrA Lin-124 general methanogenesis Umm-113 Lin-34 Umm-71 mcrB M1

Lin-34 Umm-71 cell divison minD Umm-69 FPKM Lin-124 10000 M1 20000 40000 Umm-69 ftsZ 60000 M4 100000 M7

Surface Surface Deep

Fall Summer Summer

Figure 24 – Methanogen genome transcript activity by season and depth 112

On the left is a list of methanogen genomes for which a gene of interest had a reported transcript activity in at least one of the five ecosite/depth/season combinations. Bubble plots represent the abundance of a transcript for a corresponding gene of the listed genomes across 4 surfaces (Fall plant, Fall mud, Summer plant, Summer mud – red bubbles) as well as Summer plant deep soils (blue bubbles). Transcripts highlighted here include genes for all methanogens (mcrA and mcrB), genes for acetoclastic methanogenesis (cdhA), genes for hydrogenotrophic methanogenesis (mtd and mer), as well as cell division and reproduction markers (minD and ftsZ). Transcripts indicative of active methylotrophic methanogenesis were not observed in reconstructed genomes. Data represented here is available in Supplementary Dissertation Table 9.

3.2.9 Conclusions

Regardless of genome, consistent evidence for active methanogenesis are transcribed across wetland gradients (ecosites, seasons, depths). This includes genes for methanogenesis, ribosome synthesis, protein repair, DNA replication, transcription, translation, and cell division. Consistent with prior reports in Chapter 2 that Candidatus

Methanothrix paradoxum does not transcribe oxygen detoxification genes, we extend these findings more globally to all methanogens across oxic surface soils. This provides even more evidence that alternative means, such as anoxic microsites for example, may be prominently contributing to oxic-soil methanogenesis. Here, we supplement our prior knowledge of one methanogen with a broader knowledge of methanogen diversity and metabolisms active along wetland gradients.

Obviously the more complete the genome the greater the chance one has to robustly understand the activity of organisms in the wetland, and it should be noted that it would be ideal to expand our knowledge of transcription in deep soils to additional sites moving forward to more easily capture differences in depth-driven activities of organisms who are 113 found and active in across the oxygenated vs. anoxic soil profile. Interestingly, while

Candidatus Methanothrix paradoxum still dominates the activity of surface soils, a divergent clade of Methanothrix mcrA sequences exist that were recovered from deep soils and are uniquely more active in these deep environments. While genome Thrix-135 may in fact be the organism responsible for this activity, the lack of mcrA sequence in the genome makes this currently impossible to determine. Also of interest, methylotrophic methanogenesis was a player in wetland soil activity based off of unbinned mcrA activity work, however the Mx-87 and Mx-72 genomes successfully reconstructed did not appear to be the organisms who represent that community, based upon the lack of detected activity within their genomes. Additional efforts need to be taken to better link near-complete genomes to mcrA gene diversity we have sampled across the wetland.

Cumulatively, these data revealed for the first time the depth-localized transcriptional differences of mcrA sequences, and new genomes for genera implicated by this analysis were successfully recovered from both surface and deep soils. These data, combined with the first examination of genome activity in surface and deep habitats, present even more reason to continue studying the deep soils through metagenomics and metatranscriptomics. Additionally, the methane production potential experiments highlight that numerous factors can influence the ability of soil communities from different ecosite, season, and depth combinations to produce methane, a metric comparable easily between different ecosystems. It is vital to continue the exploration of these soils across the scales using a combination of laboratory and computational techniques to more accurately

114 understand and predict the physiological responses of these organisms under various natural conditions they face currently and are likely to face in the future.

3.3 Chapter 3 Methods

3.3.1 Methane production potential determination

To evaluate the methane concentrations in wetland soil incubation vials, methane production potential (MPP) experiments219,234 were conducted. Five grams of soil from each depth/site combination was transferred into triplicate sterile, sealed 30 mL amber vials. Sealed vials (18 for each season, Fall and Summer) were driven anoxic for 20 minutes under 99% N2 gas and immediately transferred to 4°C for transport to inhibit methane production. Vials were individually removed from the 4°C conditions, vortexed for 20 seconds to release methane bound in the pores, and 5 mL headspace methane was removed and injected into a Shimadzu GC-2014 (Serial No. C114850) gas chromatograph to get a baseline for “Day 0” methane concentrations. Currents were 150 mV and 80 mV and carrier gasses of argon and helium set to flow rates of 25 mL/min and 26 mL/min respectively. Peak areas were converted via known standards ran in conjunction to yield

µg methane (CH4-C) per gram sediment in each vial. To determine the potential methane production for each site/depth community given identical anaerobic conditions, vials were incubated immediately following initial headspace removal and allowed to incubate for 10 days and 40 days before re-sampling 5 mL of headspace gas. Once again, the previously- described Shimadzu GC-2014 was used for gas analysis and peak areas were converted via a 99% methane dilution series of standards ran in conjunction to yield µg methane per gram

115 sediment in each vial. Rates of methane production were then determined based off these production values. Statistical differences between treatment groups were evaluated using

ANOVA tests, with significance being established as p < 0.05. Settings for this GC were identical to previously described.

3.3.2 Assessing potential function of unknown Candidatus Methanothrix

paradoxum transcripts

Open reading frames that hit to hypothetical proteins in databases used

(Tigerfam259, Uniprot260, KEGG261, InterProScan207) or lacking a significant hit to a protein in the databases (at bit score >60) were labeled as “unannotated.” In order to gain further insight into some of these unannotated genes, we selected the highly-transcribed genes in our Methanothrix bins to search for functional catalytic domains. These genes were uploaded to Phyre2139 and processed with default settings

(http://www.sbg.bio.ic.ac.uk/~phyre2/html/page.cgi?id=index). The resulting 10 best hits for each unknown transcript was collected into Supplementary Dissertation Table 8.

3.3.3 Determining mcrA transcript relative abundance

For the sake of both metagenome-acquired mcrA gene and binned genome analyses, metagenomes from deep soils were utilized as provided courtesy of Christopher Miller and

Adrienne Narrowe at the University of Colorado Denver. These soils represent the “D3” and “D4” depth zones from the open water and mud ecosites of Transect 3, as previously described43. For these 4 samples, 54-60 GB of sequencing was acquired per sample. These metagenome reads were quality trimmed via BBDuk (sourceforge.net/projects/bbmap) and

116 assembled by JGI using megahit assembly262 following sequencing and mud ecosite metagenomes were reassembled via IDBA-UD (parameters mink 55 –maxk 124 –step 10) to improve results. To aid in binning, each metagenome had a coverage profile generated by mapping each of the metagenomic read sets to each assembly using BBmap v 36.99.

Each assembly was binned via Metabat208 v 0.32.4 using superspecific parameters with manual curation of genome bins as required.

To expand upon the prior investigation into mcrA transcript abundance (Figure 16), the same Fall (November 2014) and Summer (August 2015) surface soil metagenome assemblies (6 in total) were investigated for the presence of mcrA genes using both BLAST and HMM263 searches. Additionally, the same techniques were performed on metagenomes acquired from deep soils in October 2013. A pool of all mcrA sequences recovered from these analyses were collected in Geneious and merged with a fasta file of full-length nucleotide sequences collected from JGI. This sequence pool was first length-trimmed to contain only mcrA sequences which reaching 34% of the total nucleotide sequence length or longer. The resulting sequences were then put through a quality control step of aligning them to a reference sequence via the MAFFT alignment feature in Geneious264. Sequences which failed to properly align to any portion of the mcrA reference sequences were discarded. The resulting sequences were aligned via MUSCLE alignment using default parameters and this alignment was visualized in Geneious for anomalous sequences. From this alignment, a maximum likelihood phylogeny tree was created using RAxML version

8.3.1 under the GTRCAT model of evolution with 100 bootstrap replicates215 and visualized in iTOL216. The mcrA sequences which made it on to the tree had transcripts

117 mapped onto them and corresponding transcript abundaces (FPKM) calculated, as previously described in Chapter 2 methods above for both surface metatranscripts (Fall plant, Fall mud, Summer plant, Summer mud, n = 12), and deep metatranscripts (Summer plant, n=3). For each transcriptionally-active mcrA sequence, the cumulative mean transcript abundance for each depths was calculated.

3.3.4 Methanogen genome recovery and quality determination

For all surface and deep metagenomes, assembly and binning was performed as previously reported (Chapter 2 methods), with only minor modification. In brief, each metagenome assembly was binned via Metabat208. Coassemblies were also generated for the mud and plant ecosites which combine metagenome sequences from these ecosites regardless of sampling season prior to assembly of the metagenomic data in an attempt in improve downstream binning quality. Thus, genomes discussed here represent genomes from traditionally-assembled metagenomes as well as coassembled metagenomes, and are labeled as such when discussed. Also included in these analyses are two metagenomes recovered from sequencing efforts performed on methanogen enrichments which had been previously developed in the lab, following the conclusion of methane production potential experiments. Genomes recovered from the various metagenomes were qualified for completion and contamination via two different methods: 1) the presence of core gene sets

(highly conserved genes that occur in single copy) for Bacteria (31 genes) and Archaea

(104 genes) using a method previously reported183, and 2) CheckM analyses265. Taxonomic placement of the genome bins was based on phylogenies of 16S rRNA genes recruited from the bin and/or ribosomal protein analyses. Once genomes were determined to be 118 methanogen genomes, BLAST analyses were conducted to locate 16S rRNA genes and mcrA gene sequences found within. Data reported here are methanogen genomes recovered with a completion greater than 55% and contamination less than 30%, with the exception of the enrichment-recovered genomes.

3.3.5 Wetland methanogen genome activity assessments

The genomes depicted in Table 5 had all surface and deep metatranscripts (n=15) mapped to them collectively as previously described (Chapter 2 methods), so that comparisons could be made between transcript abundance of genes from different genomes. The resulting transcriptional activity profiles were visualized in multiple ways.

Firstly, a genome cartoon was illustrated depicted the genes absent from (white), found within (black), and found actively transcribed in (red) Candidatus Methanothrix paradoxum M1 genome (aka 7_Metabat_Plant_2014_11). Secondly, difference in a gene mean transcript abundance between 4 surface and 1 deep soil ecosite/season combinations were visualized via the bubble function of ggplot2266 in R Studio.

119

Chapter 4. Future directions in methanogen research at Old Woman Creek

4.1 Rising water level at Old Woman Creek and the impact on soil dissolved oxygen concentrations

Like most natural field sites, conditions are constantly in flux at Old Woman Creek.

Sampling efforts spanning multiple seasons, ecosites, and depths are inherently variable.

Recently in Summer-Fall 2017, the water level at the site has risen by roughly a meter in a few months, resulting in submersion of previously exposed vegetation and mud-flats across the site. While some previous works have examined how rapid water level changes could affect anaerobic process such as nitrogen utilization by wetland soil microbial communities267,268, very few studies have examined the effects of similar conditions on soil methanogens, especially under in situ conditions. Although preliminary, here we present geochemical and methanogenic activity data from OWC soils, sampled during these unusually high water levels. I will conclude this chapter and the dissertation as a whole by looking to the future of the wetland methanogen project with avenues that are currently under exploration, and those which may be beneficial as the project continues to move forward and grow.

To demonstrate the recent changes in water level at Old Woman Creek, I have utilized the water depth and dissolved oxygen concentration measured at the lower estuary monitoring station, near transects from which sample data was collected. To visualize this,

I have utilized the System Wide Monitoring Program (SWMP) Graphing Application available through NOAA and generated Figure 25. This figure shows the relationship

120 between water level and the dissolved oxygen concentration in the water column. In the two most recent years, the “mouth” of the wetland – the seasonally open connection between the wetland and Lake Erie – has not opened until later in the year compared to previous years. The opening of the mouth results in a rapid draining of the wetland

(dropping water level) and consequent increase in water dissolved oxygen concentration, observable as the steep drop and raise in the blue and red lines, respectively (Figure 25).

While this dissolved oxygen concentration comes from the water column and not the soil, there is preliminary soil dissolved oxygen data collected near the peak of water levels in

Fall 2017 (November). The water level on the date of sampling in Fall 2017 was ~1.1 meters higher than that of sampling in Summer 2015 (August), which probably contributed greatly to the reduction in soil dissolved oxygen concentrations (Figure 26). There is also a difference in water dissolved oxygen concentrations as measured at the monitoring station between these sampling events – with means of 8.9 and 5.7 mg/L for Fall 2017 and

Summer 2015 sampling windows, respectively. Although oxygen data was not being collected in the Fall 2014 seasonal sampling event, the water level in Fall 2017 was ~1.8 meters higher than the same month three years prior, due to the cumulative effects of increased precipitation over a normal year and the closure of the mouth throughout the entire summer and fall seasons, into the winter. When the mouth finally did open in late

Fall 2017, the water level at the measurement point dropped by ~1.2 meters in only 19 hours, highlighting how rapidly conditions can vary in this wetland.

121

Old Woman Creek Lower Estuary Gauge Station 20 2.5 ) L ) / m

g 15 (

m s ( r

e n t e e g m

y 1.25

10 x n i O

h d t e p v e l o 5 D s s i D

0 0 2014 2015 2016 2017 2018 Dissolved Oxygen Depth Figure 25 Water depth and dissolved oxygen concentrations at the lower estuary monitoring station, Old Woman Creek, 2014 – 2017 Water depth as measured at the Old Woman Creek lower estuary gauge station (blue line) has been much higher in recent years, due to the estuary mouth to Lake Erie being closed for a larger portion of the year. This results in heightened water depths throughout the fall season, leading to a dramatic decrease in dissolved oxygen concentrations (red line) in the water column. When the “mouth” of the wetland becomes opened to Lake Erie, it results in a rapid draining of the wetland, observable by a steep decline in water depth and the dissolved oxygen level does return to “normal” levels (5-15 mg/L).

122

Figure 26 – A comparison of water level and dissolved oxygen concentrations in “open water” soils – Summer 2015 and Fall 2017 data A. A graphical representation of the changing water level at Old Woman Creek. The mean water level during the 6-hour window of the Summer 2015 sampling event (grey line) was ~0.8 meters at the monitoring station. The mean water level during the Fall 2017 sampling event (dashed black line) was ~1.9m, more than a meter greater than the previously highest level at the time of a sampling event. This resulted in all open water, mud, and plant transects being flooded with at least 1 meter of standing water. B. An examination of the dissolved oxygen concentration in the water ecosite demonstrates a marked difference. The previously reported soil dissolved oxygen concentration curve observed in Summer 2015 sampling (white line) was replaced in Fall 2017 by an oxygen gradient that was determined to be anoxic by the water/soil interface, and remained anoxic throughout the measurable soil depth (red line).

123

These effects in the water column can have direct effects upon the availability of dissolved oxygen and substrates impacting microbial communities in wetland soils. To first investigate how the soil dissolved oxygen concentrations were impacted by these hydrological changes, dissolved oxygen (DO) measurements were conducted via the same methodology as previously descried (Chapter 2). For the most direct comparison to previous data, coordinates for a site which had previously been an open water transect were used. This was because effectively all of the site was now an open water environment, with some of the taller plants emerging from the water in spots. This change in water level between the 2 sampling points is depicted in Figure 26A. Because of logistical constraints,

DO sampling was limited to the water column and the first 20cm of the soil. The difference in DO concentrations between the same ecosite under the varied hydrology is dramatic, shown in Figure 26B by the white line (soil DO in Summer 2015) and the red line (water column and soil DO in Fall 2017). While the water column retains oxic conditions for at least the first 10 cm for Fall 2015, this dissolved oxygen was not detectable (~5 uM) at the water/soil interface in Fall 2017. For comparison at this same depth, two years earlier the

DO concentration of soil porewater was ~170 uM in Summer 2015, when the standing water over the site was roughly 1 meter less. This demonstrates just how impactful the hydrologic forcing can be on the potential for microbial metabolisms in the wetland soils.

124

4.2 Preliminary data on the impact of decreased soil dissolved oxygen on in situ methane concentration and methane production potentials

This drastic difference in dissolved oxygen concentration is likely to have significant impact on the methanogenic populations in Old Woman Creek soils. While we know that oxic surface soils had high potential for methane production, dominated by a single acetoclastic methanogen, that paradigm could change significantly given a completely new set of hydrological and anoxic conditions. This has already been shown in paddy field soils, where a shift in the taxonomic estimation of dominant mcrA transcripts shifted between Orders under transition from drained to flooded conditions269. There is also evidence from boreal fens which demonstrate more diverse activity within the community during a high water table (presumably good conditions for methanogens), whereas a relative few are detecting actively transcribing under dryer conditions270. While no data yet exists as the microbial community and specific activities in OWC soils, there is preliminary in situ methane concentration and methane production potential data which provides some insight into how these altered conditions have modified the ability of the methanogenic communities as a whole to produce methane in surface and deep soils (Figures 27 and 28).

An examination of the in situ methane concentrations across the three seasons demonstrates some interesting trends. Firstly, the concentration of methane in surface soils were generally comparable to surface soil concentrations from previous sampling events.

While it would have been tempting to speculate on why this concentration is not higher than previous seasons when oxygen penetration into the soil was much greater, we do not currently have methanogenic substrate data to provide context, nor do we know the impact 125 these changes had the makeup and activity of the methanogenic communities. It is also possible that the in situ methane concentration in surface soils can only reach such a saturation level before it is dissolving upwards into the water column or air (depending on water level), which could possibly limit the total in situ methane concentrations we can observe in these soils. Also of interest, deep plant and mud soils from Fall 2017 contained significantly greater in situ methane concentrations than deep samples from Summer 2015.

While the reason behind this increased methanogenesis activity in deeps soils in this year compared to previous seasons is unknown, this is a particularly interesting finding as it infers that the enzymatic latch hypothesis271 may not be the only factor controlling carbon degradation and subsequent methanogenic substrate availability in this wetland. This hypothesis, first described in peatland soils, states that oxygen limitations on a single phenol oxidase enzyme is capable of limiting release of carbon into the atmosphere. A landmark study combining laboratory mesocosms along with in vitro experimentation and field observations supported this hypothesis, showing drought conditions causing increased phenol oxidase activity and decomposition of organic matter in the soil272, which was exacerbated by rewetting events following drought periods which accelerated carbon loss from these peatland soils. Alternatively, our preliminary data from this chapter show similar patterns of in situ methane concentrations and increased methane production potential in deep soils in relation to previous seasons with greater oxygen penetration. This could indicate that, in these continuously flooded conditions resulting in totally anoxic conditions throughout the soils, anaerobic respiration or fermentation may instead be yielding methanogenic substrate concentrations capable of supporting methanogenesis,

126 consistent with recent findings in permafrost and peatland soils273,274. This represents an additional mechanism, different from those likely occurring under the oxygenic conditions we reported previously and independent of the “enzymatic latch”. This is an interesting finding which could be an area of focus for future works built around the rapid transition between inundated and exposed soils from OWC wetland soils.

Fall 2014 Summer 2015 Fall 2017

10 (µg) 4

CH 5 in situ

0 Figure 27 – in situ methane concentrations following increased water levels and decreased dissolved oxygen concentrations An examination of in situ methane concentrations across the sampled seasons demonstrated that, while similar to previously sampled seasons, Fall 2017 deep soils were significantly greater than deep soils from other seasons. Plant, mud, and open water transects are denoted by green, orange, and blue respectively, with the lighter and darker shades of the colors representing surface and deep soils, respectively.

127

Fall 2014 Summer 2015 Fall 2017 )

-1 80

60

40

20 production rate (µg day 4 CH 0

Figure 28 – 10-day methane production rates following increased water levels and decreased dissolved oxygen concentrations Methane production potential (MPP) experimentation demonstrated that Fall 2017 surface soils generated significantly more methane than either of the previously sampled seasons. Plant, mud, and open water transects are once again denotes by green, orange, and blue respectively, with the lighter and darker shades of the colors representing surface and deep soils, respectively.

128

Like previous seasons, methane production potential experiments were conducted on uninoculated soils from all three ecosites. In general, the trends of surface soils having greater methane production rates over this 10-day period remains consistent (Figure 28).

The production rates of the Fall 2017 surface soils are greater than the previous 2 sampled seasons, however due to the small sample size (n=3), significant differences could not be determined. Of note, one open water vial (OS3) produced methane at ~3x higher a rate than ever previously observed across all seasons. This sample vial was earmarked for future community analyses and will be the basis of enrichment attempts when MPP experimentation has concluded. At a bulk level (n=9 per season), Fall 2017 surface production rates were significantly greater than either of the previously sampled seasons

(ANOVA, p <0.05). All three deep soils from Fall 2014 showed significantly greater methane production rates compared to Fall 2017. Forty-day methane production experiments were also conducted (data not shown), revealing similar trends as the 10-day production rates. As is typical, the 40-day rates are less than that of the 10-day rates as substrate limitation likely begins to curtail the rate of methane production in these vials.

Cumulatively, these new dissolved oxygen, in situ methane, and methane production potential data do not answer the question of how changing water levels affect specific methanogenic community members, but they do provide some context for further exploration. This represents a critical gap in the literature, as very few studies have examined the effect of changing water levels and DO on microbial ecology, especially methanogens. Since it has been demonstrated that both N-cycling and sulfate-reducing

129 communities can show oxygenated-zone dependent shifts in community structure, it is almost a certainty that methanogenic communities could have similar patterns.

As with many of the projects, preliminary geochemical or bulk-activity measurements can act as a springboard to explore interesting findings, and should continue to be incorporated into future work. Collecting and measuring these processes are relatively quick, cheap, and great places for undergraduates or new trainees on the project to begin.

It should also not be overlooking that it was an initial Day 0 methane reading for a MPP experiment which revealed more methane in surface soils and thus gave us our first idea there may be unique occurring in surface soils. In summation, the examination of this natural, rapid change in water level and dissolved oxygen concentrations in a tight temporal scale would almost certainly lead to novel findings on the response of the methanogenic community as, to the best of my knowledge, this has not been documented previously.

4.3 Towards the future: Isolating and fully characterizing Candidatus Methanothrix paradoxum

Given the stability and dominance of Candidatus Methanothrix paradoxum, ongoing research efforts should be dedicated to cultivating this organism. Based on prior enrichment attempts, it is known that acetate concentrations in enrichment cultures need to be kept low to increase the chances of successful enrichment. If acetate exceeds 1 mM through amendment or homoacetogenesis, outgrowth of Methanosarcina, which remains inactive and at very low relative abundance in wetland soils, occurs in the lab reactors. The current approach used in the laboratory is to perform 16S rRNA gene quantification from unamended MPP experiment soils and monitor these for Candidatus Methanothrix 130 paradoxum and acetate concentrations in real-time. Methods are also underway to develop a specific FISH probe for quantifying Candidatus Methanothrix paradoxum in cultures.

The long-term goal of the work should be to cultivate this dominant wetland methanogen so that thorough biochemical and physiological experimentation can illuminate the potential of this organism for methane production under varying oxygen levels and environmental conditions. These data would be critical in further informing accurate methane emissions modeling efforts at Old Woman Creek.

4.4 Towards the future: Unraveling the effects of dissolved organic carbon upon wetland methanogenesis

A robust approach to understanding the interactions between soil community members and natural dissolved organic carbon involving examination of the DOC through ultra high-resolution mass spectrometry and controlled chamber reactors is being actively developed, with implementation later this year (Figure 29). In collaboration with members of EMSL and PNNL, an informatics pipeline is under construction to integrate organic carbon compound data yielded from FTICR-MS analyses with the presence and activity of necessary degradation pathways detected in wetland soils via metagenomic and metatranscriptomics. This would represent the bridging of two scientific realms which have historically struggled to be connectable. Combined with smaller methanogenic substrates detectable via NMR, this could paint an incredibly detailed picture of the ability of wetland organisms to utilize in situ organic carbon pools, and may help elucidate how these degradation pathways ultimately impact methanogenesis.

131

Additionally, as part of a 2017 Early Career Award to Dr. Wrighton, the laboratory is currently developing water saturated soil reactors, which offer the ability to study these in situ communities in an unprecedentedly natural manner by keeping the soil structure intact while providing natural organic carbon inputs as substrate. Furthermore, these reactors are equipped with microelectrodes and optodes enabling detailed quantification of oxygen gradients in soils. Additionally, these reactors will allow for the manipulation of conditions to examine how variability in temperature, moisture, and carbon input exposure impact methane production and anoxic microsite formation, and allow for non-destructive and consistent sampling without return to the field site each time. This ambitious combination of data on the responses of undisturbed methanogenic communities to in situ dissolved organic carbon inputs in such a controlled manner represent a tantalizing and extremely unique area of exploration for this and no doubt other soils of interest around the world.

CH4 DOC

Oxygen

Porosity

Water

Figure 29 – Sample schematic of proposed bioreactor design This is the simplified schematic of the bioreactor setup proposed for study of dissolved organic carbon utilization by wetland soil communities, as proposed in the awarded United Stated Department of Energy grant (DE-FOA-0001625). 132

4.5 Summary and concluding comments

Globally, there is an ever-growing list of methane paradox sites. Although numerous mechanisms can drive methane production in oxygenated zones, my data suggests that the majority of these are likely due to traditional archaeal methanogens finding a way to live and produce methane in these environments, which was presented for the first time here. When the work detailed here on Old Woman Creek began, we were unaware that it was also a methane paradox environment. Data from this dissertation has not only characterized it as such, but expanded our knowledge of the identity, novelty, and activity of the methanogenic community. It is now known that a single acetoclastic methanogen - Candidatus Methanothrix paradoxum – dominates the methanogenic activity in the oxic surface soils of Old Woman Creek, being ubiquitous across the wetland and consistently active across spatial and temporal scales examined. Expanding to a larger scale, this organism is globally distributed and detected in a number of other methane paradox sites. It is also now known that the this detected biological activity at OWC creates an observable increase in the methane concentrations in oxic surface soils, and that surface soils account for up to 80% of the wetland methane budget. Further analyses of the metagenomic and metatranscriptomic data from wetland soils revealed a diverse and differentiated active methanogenic community between surface and deep habitats. The reconstruction of numerous other methanogen genomes from both surface and deep not only verify the genera we expected to see based of mcrA sequence recovery, but has allowed us to assess the novelty of the organisms that inhabit this wetland, and reinforce the value of these metagenomic techniques being single marker genes analyses. Methane 133 production potential experimentation allowed for another metric of evaluating the potent methane producing power of OWC soils, and also allowed us to distinguish substrate limitations in deep soils, a principle which set the stage for downstream organic carbon characterization and utilization which is currently underway. Lastly, the recent increased water levels at OWC and subsequent anoxic conditions which ensued undoubtedly alter the methanogenic community activity in the soil. Just how these stresses manifest and if these alter greenhouse gas emissions remains a critically important area of research moving forward, especially as variable precipitation and temperature increases are predicted for the

Great Lakes area in the future. The combination of ongoing intensive field campaigns with highly temporally and spatially resolved sampling campaigns, combined with laboratory soil reactors, will surely offer new mechanistic understanding on the factors and organisms sustaining the methane paradox in this wetland, findings that can be critically important to many other wetlands worldwide.

134

References

1. Climate Change 2013 - The Physical Science Basis. (Cambridge University Press, 2014). doi:10.1017/CBO9781107415324 2. Etiope, G. & Sherwood Lollar, B. Abiotic Methane on Earth. Rev. Geophys. 51, 276–299 (2013). 3. Riley, W. J. et al. Barriers to predicting changes in global terrestrial methane fluxes: analyses using CLM4Me, a methane biogeochemistry model integrated in CESM. Biogeosciences 8, (1925). 4. Grossart, H.-P., Frindte, K., Dziallas, C., Eckert, W. & Tang, K. W. Microbial methane production in oxygenated water column of an oligotrophic lake. Proc. Natl. Acad. Sci. U. S. A. 108, 19657–61 (2011). 5. Karl, D. M. et al. Aerobic production of methane in the sea. Nat. Geosci. 1, 473– 478 (2008). 6. Angel, R., Matthies, D. & Conrad, R. Activation of Methanogenesis in Arid Biological Soil Crusts Despite the Presence of Oxygen. PLoS One 6, e20453 (2011). 7. Bogard, M. J. et al. Oxic water column methanogenesis as a major component of aquatic CH4 fluxes. Nat. Commun. 5, 5350 (2014). 8. Angle, J. C. et al. Methanogenesis in oxygenated soils is a substantial fraction of wetland methane emissions. Nat. Commun. 8, (2017). 9. Boone, D. R., Whitman, W. B. & Rouvière, P. Diversity and Taxonomy of Methanogens. in Methanogenesis 35–80 (Springer US, 1993). doi:10.1007/978-1- 4615-2391-8_2 10. Thauer, R. K. Biochemistry of methanogenesis : a tribute to Marjory Stephenson. Microbiology 1928, 27–2377 (1998). 11. Barnes, R. O. & Goldberg, E. D. Methane production and consumption in anoxic marine sediments. Geology 4, 297 (1976). 12. King, G. M. Utilization of hydrogen, acetate, and ‘noncompetitive’; substrates by methanogenic bacteria in marine sediments. Geomicrobiol. J. 3, 275–306 (1984). 13. Martens, C. S. & Goldhaber, M. B. Early diagenesis in transitional sedimentary environments of the White Oak River Estuary, North Carolina. Limnol. Oceanogr. 23, 428–441 (1978). 14. Oremland, R. S. & Polcin, S. Methanogenesis and sulfate reduction: competitive and noncompetitive substrates in estuarine sediments. Appl. Environ. Microbiol. 44, 1270–6 (1982). 15. Lovley, D. R. & Klug, M. J. Methanogenesis from methanol and methylamines and acetogenesis from hydrogen and carbon dioxide in the sediments of a eutrophic lake. Appl. Environ. Microbiol. 45, 1310–5 (1983). 16. Winfrey, M. R., Nelson, D. R., Klevickis, S. C. & Zeikus, J. G. Association of hydrogen metabolism with methanogenesis in Lake Mendota sediments. Appl. Environ. Microbiol. 33, 312–8 (1977). 135

17. Ward, T. E. & Frea, J. I. Sediment distribution of methanogenic bacteria in lake erie and cleveland harbor. Appl. Environ. Microbiol. 39, 597–603 (1980). 18. Holzapfel-Pschorn, A., Conrad, R. & Seiler, W. Production, oxidation and emission of methane in rice paddies. FEMS Microbiol. Lett. 31, 343–351 (1985). 19. Svensson, B. H. & Rosswall, T. In situ Methane Production from Acid Peat in Plant Communities with Different Moisture Regimes in a Subarctic Mire. Oikos 43, 341 (1984). 20. DeLaune, R. D., Smith, C. J. & Patrick, W. H. Methane release from Gulf coast wetlands. Tellus B 35B, 8–15 (1983). 21. Jørgensen, B. B. Bacteria and Marine Biogeochemistry. in Marine Geochemistry 169–206 (Springer-Verlag, 2006). doi:10.1007/3-540-32144-6_5 22. Lovley, D. R., Dwyer, D. F. & Klug, M. J. Kinetic analysis of competition between sulfate reducers and methanogens for hydrogen in sediments. Appl. Environ. Microbiol. 43, 1373–9 (1982). 23. Mountfort, D. O., Asher, R. A., Mays, E. L. & Tiedje, J. M. Carbon and electron flow in mud and sandflat intertidal sediments at delaware inlet, nelson, new zealand. Appl. Environ. Microbiol. 39, 686–94 (1980). 24. Sørensen, J. & Jeørgensen, B. B. Early diagenesis in sediments from Danish coastal waters: Microbial activity and Mn-Fe-S geochemistry. Geochim. Cosmochim. Acta 51, 1583–1590 (1987). 25. Alewell, C., Paul, S., Lischeid, G. & Storck, F. R. Co-regulation of redox processes in freshwater wetlands as a function of organic matter availability? Sci. Total Environ. 404, 335–342 (2008). 26. Storz, G. & Imlayt, J. A. Oxidative stress. Curr. Opin. Microbiol. 2, 188–194 (1999). 27. Fridovich, I. Superoxide radical and superoxide dismutates. Annu. Rev. Biochem 64, 97–112 (1995). 28. Imlay, J. A. Iron-sulphur clusters and the problem with oxygen. Mol. Microbiol. 59, 1073–1082 (2006). 29. Walsh, C. T. & Orme-Johnson, W. H. Perspectives in Biochemistry Nickel Enzymes. (1987). 30. Jasso-Chávez, R. et al. Air-Adapted Methanosarcina acetivorans Shows High Methane Production and Develops Resistance against Oxygen Stress. PLoS One 10, e0117331 (2015). 31. Repeta, D. J. et al. Marine methane paradox explained by bacterial degradation of dissolved organic matter. Nat. Geosci. 9, 884–887 (2016). 32. Yao, M., Henny, C. & Maresca, J. A. Freshwater bacteria release methane as a byproduct of phosphorus acquisition. Appl. Environ. Microbiol. 82, 6994–7003 (2016). 33. Lamontagne, R. A., Swinnerton, J. W. & Linnenbom, V. J. Nonequilibrium of carbon monoxide and methane at the air-sea interface. J. Geophys. Res. 76, 5117– 5121 (1971). 34. Lamontagne, R. A., Swinnerton, J. W., Linnenbom, V. J. & Smith, W. D. Methane concentrations in various marine environments. J. Geophys. Res. 78, 5317–5324 136

(1973). 35. Scranton, M. I. & Brewer, P. G. Occurrence of methane in the near-surface waters of the western subtropical North-Atlantic. Deep Sea Res. 24, 127–138 (1977). 36. Teh, Y. A., Silver, W. L. & Conrad, M. E. Oxygen effects on methane production and oxidation in humid tropical forest soils. Glob. Chang. Biol. 11, 1283–1297 (2005). 37. Tang, K. W., McGinnis, D. F., Ionescu, D. & Grossart, H.-P. Methane Production in Oxic Lake Waters Potentially Increases Aquatic Methane Flux to Air. Environ. Sci. Technol. Lett. 3, 227–233 (2016). 38. Luton, P. E., Wayne, J. M., Sharp, R. J. & Riley, P. W. The mcrA gene as an alternative to 16S rRNA in the phylogenetic analysis of methanogen populations in landfill b. Microbiology 148, 3521–3530 (2002). 39. Steinberg, L. M. & Regan, J. M. mcrA-targeted real-time quantitative PCR method to examine methanogen communities. Appl. Environ. Microbiol. 75, 4435–42 (2009). 40. Kallistova, A. Y., Merkel, A. Y., Tarnovetskii, I. Y. & Pimenov, N. V. Methane formation and oxidation by prokaryotes. Microbiology 86, 671–691 (2017). 41. McKay, L. J., Hatzenpichler, R., Inskeep, W. P. & Fields, M. W. Occurrence and expression of novel methyl-coenzyme M reductase gene (mcrA) variants in hot spring sediments. Sci. Rep. 7, 7252 (2017). 42. Vanwonterghem, I. et al. Methylotrophic methanogenesis discovered in the archaeal phylum Verstraetearchaeota. Nat. Microbiol. 1, 16170 (2016). 43. Narrowe, A. B. et al. High-resolution sequencing reveals unexplored archaeal diversity in freshwater wetland soils. Environ. Microbiol. 19, (2017). 44. Harms, U. & Thauer, R. K. Methylcobalamin:Coenzyme M Methyltransferase Isoenzymes MtaA and MtbA from Methanosarcina barkeri. Cloning, Sequencing and Differential Transcription of the Encoding Genes, and Functional Overexpression of the mtaA Gene in Escherichia coli. Eur. J. Biochem. 235, 653– 659 (1996). 45. Krätzer, C., Carini, P., Hovey, R. & Deppenmeier, U. Transcriptional profiling of methyltransferase genes during growth of Methanosarcina mazei on trimethylamine. J. Bacteriol. 191, 5108–15 (2009). 46. Kaster, A.-K. et al. More Than 200 Genes Required for Methane Formation from H2O and CO2 and Energy Conservation Are Present in Methanothermobacter marburgensis and Methanothermobacter thermautotrophicus. Archaea 2011, 1–23 (2011). 47. Welte, C. & Deppenmeier, U. Membrane-bound electron transport in Methanosaeta thermophila. J. Bacteriol. 193, 2868–70 (2011). 48. Ferry, J. G. & Lessner, D. J. Methanogenesis in Marine Sediments. Ann. N. Y. Acad. Sci. 1125, 147–157 (2008). 49. Zhu, J. et al. The Genome Characteristics and Predicted Function of Methyl-Group Oxidation Pathway in the Obligate Aceticlastic Methanogens, Methanosaeta spp. PLoS One 7, e36756 (2012). 50. Angel, R., Claus, P. & Conrad, R. Methanogenic archaea are globally ubiquitous 137

in aerated soils and become active under wet anoxic conditions. ISME J. 6, 847– 862 (2012). 51. Ma, K., Conrad, R. & Lu, Y. Responses of methanogen mcrA genes and their transcripts to an alternate dry/wet cycle of paddy field soil. Appl. Environ. Microbiol. 78, 445–54 (2012). 52. Fredette, V., Planté, C. & Roy, A. Numerical data concerning the sensitivity of anaerobic bacteria to oxygen. J. Bacteriol. 94, 2012–7 (1967). 53. Loesche, W. J. Oxygen sensitivity of various anaerobic bacteria. Appl. Microbiol. 18, 723–7 (1969). 54. Morris, J. G. The Physiology of Obligate Anaerobiosis. Adv. Microb. Physiol. 12, 169–246 (1975). 55. McCord, J. M., Keele, B. B. & Fridovich, I. An Enzyme-Based Theory of Obligate Anaerobiosis: The Physiological Function of Superoxide Dismutase. Proc. Natl. Acad. Sci. 68, (1971). 56. Kirby, T. W., Lancaster, J. R. & Fridovich, I. Isolation and characterization of the iron-containing superoxide dismutase of Methanobacterium bryantii. Arch. Biochem. Biophys. 210, 140–148 (1981). 57. Takao, M., Oikawa, A. & Yasui, A. Characterization of a superoxide dismutase gene from the archaebacterium Methanobacterium thermoautotrophicum. Arch. Biochem. Biophys. 283, 210–216 (1990). 58. Bult, C. J. et al. Linked references are available on JSTOR for this article : Complete Genome Sequence of the Methanogenic Archaeon , Methanococcus jannaschii. Science (80-. ). 273, 1058–1073 (2016). 59. Smith, D. R. et al. Complete genome sequence of Methanobacterium thermoautotrophicum deltaH: functional analysis and comparative genomics. J. Bacteriol. 179, 7135–55 (1997). 60. Shima, S. et al. Purification, characterization, and primary structure of a monofunctional catalase from Methanosarcina barkeri. Arch. Microbiol. 171, 317– 323 (1999). 61. Brioukhanov, A., Netrusov, A., Sordel, M., Thauer, R. K. & Shima, S. Protection of Methanosarcina barkeri against oxidative stress: identification and characterization of an iron superoxide dismutase. Arch. Microbiol. 174, 213–216 (2000). 62. Lyu, Z. & Lu, Y. Metabolic shift at the class level sheds light on adaptation of methanogens to oxidative environments. ISME J. 12, 411–423 (2018). 63. Seedorf, H., Dreisbach, A., Hedderich, R., Shima, S. & Thauer, R. F420H2 oxidase (FprA) from Methanobrevibacter arboriphilus, a coenzyme F420- dependent enzyme involved in O2 detoxification. Arch. Microbiol. 182, 126–137 (2004). 64. Tholen, A., Pester, M. & Brune, A. Simultaneous methanogenesis and oxygen reduction by Methanobrevibacter cuticularis at low oxygen fluxes. FEMS Microbiol. Ecol. 62, 303–312 (2007). 65. Liu, C.-T., Miyaki, T., Aono, T. & Oyaizu, H. Evaluation of Methanogenic Strains and Their Ability to Endure Aeration and Water Stress. Curr. Microbiol. 56, 214– 138

218 (2008). 66. McCarver, A. C. & Lessner, D. J. Molecular characterization of the thioredoxin system from Methanosarcina acetivorans. FEBS J. 281, 4598–4611 (2014). 67. Zhang, W., Culley, D. E., Nie, L. & Brockman, F. J. DNA microarray analysis of anaerobic Methanosarcina barkeri reveals responses to heat shock and air exposure. J. Ind. Microbiol. Biotechnol. 33, 784–790 (2006). 68. Brioukhanov, A. L., Netrusov, A. I. & Eggen, R. I. L. The catalase and superoxide dismutase genes are transcriptionally up-regulated upon oxidative stress in the strictly anaerobic archaeon Methanosarcina barkeri. Microbiology 152, 1671–1677 (2006). 69. Kiener, A. & Leisinger, T. Oxygen Sensitivity of Methanogenic Bacteria. Syst. Appl. Microbiol. 4, 305–312 (1983). 70. Kato, M. T., Field, J. A. & Lettinga, G. Methanogenesis in granular sludge exposed to oxygen. FEMS Microbiol. Lett. 114, 317–323 (1993). 71. Arah, J. R. M. & Vinten, A. J. A. Simplified models of anoxia and denitrification in aggregated and simple-structured soils. Eur. J. Soil Sci. 46, 507–517 (1995). 72. Sobolev, D. & Roden, E. E. Evidence for rapid microscale bacterial redox cycling of iron in circumneutral environments. Antonie Van Leeuwenhoek 81, 587–597 (2002). 73. Sexstone, A. J., Revsbech, N. P., Parkin, T. B. & Tiedje, J. M. Direct Measurement of Oxygen Profiles and Denitrification Rates in Soil Aggregates. Soil Sci. Soc. Am. J. 49, 645 (1985). 74. Zausig, J., Stepniewski, W. & Horn, R. Oxygen Concentration and Redox Potential Gradients in Unsaturated Model Soil Aggregates. Soil Sci. Soc. Am. J. 57, 908 (1993). 75. Verchot, L. V., Davidson, E. A., Cattânio, J. H. & Ackerman, I. L. Land-Use Change and Biogeochemical Controls of Methane Fluxes in Soils of Eastern Amazonia. Ecosystems 3, 41–56 (2000). 76. Schropp, S. J., Schwarz, J. R. & LaRock, P. A. Hydrogen production potential of fermentative microorganisms from the Sargasso Sea. Geomicrobiol. J. 5, 149–158 (1987). 77. Bianchi, M., Marty, D., Teyssié, J.-L. & Fowler, S. W. Strictly aerobic and anaerobic bacteria associated with sinking particulate matter and zooplankton fecal pellets. Marine Ecology Progress Series 88, 55–60 (1992). 78. de Angelis, M. A. & Lee, C. Methane production during zooplankton grazing on marine phytoplankton. Limnol. Oceanogr. 39, 1298–1308 (1994). 79. Karl, D. M. & Tilbrook, B. D. Production and transport of methane in oceanic particulate organic matter. Nature 368, 732–734 (1994). 80. Marty, D., Nival, P. & Yoon, W. Oceanologica acta. Oceanologica Acta 20, (Éditions scientifiques et médicales Elsevier SAS, 1997). 81. Sasakawa, M. et al. Carbon isotopic characterization for the origin of excess methane in subsurface seawater. J. Geophys. Res. 113, C03012 (2008). 82. Ditchfield, A. et al. Identification of putative methylotrophic and hydrogenotrophic methanogens within sedimenting material and copepod faecal 139

pellets. Aquat. Microb. Ecol. 67, 151–160 (2012). 83. Oremland, R. S. Methanogenic activity in plankton samples and fish intestines A mechanism for in situ methanogenesis in oceanic surface waters. Limnol. Oceanogr. 24, 1136–1141 (1979). 84. Maarel, M. J. E. C., Sprenger, W., Haanstra, R. & Forney, L. J. Detection of methanogenic archaea in seawater particles and the digestive tract of a marine fish species. FEMS Microbiol. Lett. 173, 189–194 (1999). 85. Finlay, B. J. et al. Some rumen ciliates have endosymbiotic methanogens. FEMS Microbiol. Lett. 117, 157–161 (1994). 86. Hook S, Wright A, M. B. Methanogens: Methane Producers of the Rumen and Mitigation Strategies - ProQuest. Methanogens Methane Prod. Rumen Mitig. Strateg. 945785 (2010). 87. Fenchel, T. & Finlay, B. J. Free-Living Protozoa with Endosymbiotic Methanogens. in 1–11 (Springer, Berlin, Heidelberg, 2010). doi:10.1007/978-3- 642-13615-3_1 88. Narayanan, N., Krishnakumar, B. & Manilal, V. B. Oxygen tolerance and occurrence of superoxide dismutase as an antioxidant enzyme in Metopus es. Res. Microbiol. 161, 227–233 (2010). 89. Schulz, K. et al. Methanogenic food web in the gut contents of methane-emitting earthworm Eudrilus eugeniae from Brazil. ISME J. 9, 1778–1792 (2015). 90. Depkat-Jakob, P. S. et al. Emission of methane by Eudrilus eugeniae and other earthworms from Brazil. Appl. Environ. Microbiol. 78, 3014–9 (2012). 91. Kernecker, M., Whalen, J. K. & Bradley, R. L. Endogeic earthworms lower net methane production in saturated riparian soils. Biol. Fertil. Soils 51, 271–275 (2015). 92. WEST, W. E., COLOSO, J. J. & JONES, S. E. Effects of algal and terrestrial carbon on methane production rates and methanogen community structure in a temperate lake sediment. Freshw. Biol. 57, 949–955 (2012). 93. Schwarz, J. I. K., Eckert, W. & Conrad, R. Response of the methanogenic microbial community of a profundal lake sediment (Lake Kinneret, Israel) to algal deposition. Limnol. Oceanogr. 53, 113–121 (2008). 94. Lenhart, K. et al. Evidence for methane production by the marine algae <i>Emiliania huxleyi</i> Biogeosciences 13, 3163–3174 (2016). 95. Damm, E., Thoms, S., Beszczynska-Möller, A., Nöthig, E. M. & Kattner, G. Methane excess production in oxygen-rich polar water and a model of cellular conditions for this paradox. Polar Sci. 9, 327–334 (2015). 96. Kamat, S. S. & Raushel, F. M. The enzymatic conversion of phosphonates to phosphate by bacteria. Curr. Opin. Chem. Biol. 17, 589–596 (2013). 97. Villarreal-Chiu, J. F., Quinn, J. P. & McGrath, J. W. The genes and enzymes of phosphonate metabolism by bacteria, and their distribution in the marine environment. Front. Microbiol. 3, 19 (2012). 98. Carini, P., White, A. E., Campbell, E. O. & Giovannoni, S. J. Methane production by phosphate-starved SAR11 chemoheterotrophic marine bacteria. Nat. Commun. 5, 4346 (2014). 140

99. Fox, A., Kwapinski, W., Griffiths, B. S. & Schmalenberger, A. The role of sulfur- and phosphorus-mobilizing bacteria in biochar-induced growth promotion of Lolium perenne. FEMS Microbiol. Ecol. 90, 78–91 (2014). 100. Wang, Q., Dore, J. E. & McDermott, T. R. Methylphosphonate metabolism by Pseudomonas sp. populations contributes to the methane oversaturation paradox in an oxic freshwater lake. Environ. Microbiol. 19, 2366–2378 (2017). 101. Wolfe, R. S. Microbial Formation of Methane. Adv. Microb. Physiol. 6, 107–146 (1971). 102. Tilbrook, B. D. & Karl, D. M. Methane sources, distributions and sinks from California coastal waters to the oligotrophic North Pacific gyre. Mar. Chem. 49, 51–64 (1995). 103. Martínez, A., Ventouras, L.-A., Wilson, S. T., Karl, D. M. & DeLong, E. F. Metatranscriptomic and functional metagenomic analysis of methylphosphonate utilization by marine bacteria. Front. Microbiol. 4, 340 (2013). 104. Murase, J., Sakai, Y., Kametani, A. & Sugimoto, A. Dynamics of methane in mesotrophic Lake Biwa, Japan. Ecol. Res. 20, 377–385 (2005). 105. Hofmann, H., Federwisch, L. & Peeters, F. Wave-induced release of methane: Littoral zones as source of methane in lakes. Limnol. Oceanogr. 55, 1990–2000 (2010). 106. Schmidt, U. & Conrad, R. Hydrogen, carbon monoxide, and methane dynamics in Lake Constance. Limnol. Oceanogr. 38, 1214–1226 (1993). 107. Schulz, M., Faber, E., Hollerbach, A., Schröder, H. G. & Güde, H. The methane cycle in the epilimnion of Lake Constance. Fundam. Appl. Limnol. 151, 157–176 (2001). 108. Owens, N. J. P., Law, C. S., Mantoura, R. F. C., Burkill, P. H. & Llewellyn, C. A. Methane flux to the atmosphere from the Arabian Sea. Nature 354, 293–296 (1991). 109. Damm, E. et al. Methane production in aerobic oligotrophic surface water in the central Arctic Ocean. Biogeosciences 7, 1099–1108 (2010). 110. Tang, K. W., McGinnis, D. F., Frindte, K., Brüchert, V. & Grossart, H.-P. Paradox reconsidered: Methane oversaturation in well-oxygenated lake waters. Limnol. Oceanogr. 59, 275–284 (2014). 111. Michmerhuizen, C. M., Striegl, R. G. & McDonald, M. E. Potential methane emission from north-temperate lakes following ice melt. Limnol. Oceanogr. 41, 985–991 (1996). 112. Bastviken, D., Cole, J., Pace, M. & Tranvik, L. Methane emissions from lakes: Dependence of lake characteristics, two regional assessments, and a global estimate. Global Biogeochem. Cycles 18, n/a-n/a (2004). 113. Prairie, Y. T. & Del Giorgio, P. A. A new pathway of freshwater methane emissions and the putative importance of microbubbles. Inl. Waters 3, 311–320 (2013). 114. Donis, D. et al. Full-scale evaluation of methane production under oxic conditions in a mesotrophic lake. Nat. Commun. 8, 1661 (2017). 115. Fetzer, S. & Conrad, R. Effect of redox potential on methanogenesis by 141

Methanosarcina barkeri. Arch. Microbiol. 160, 108–113 (1993). 116. Fetzer, S., Bak, F. & Conrad, R. Sensitivity of methanogenic bacteria from paddy soil to oxygen and desiccation. FEMS Microbiol. Ecol. 12, 107–115 (1993). 117. Peter Mayer, H. & Conrad, R. Factors influencing the population of methanogenic bacteria and the initiation of methane production upon flooding of paddy soil. FEMS Microbiol. Lett. 73, 103–111 (1990). 118. Peters, V. & Conrad, R. Methanogenic and other strictly anaerobic bacteria in desert soil and other oxic soils. Appl. Environ. Microbiol. 61, 1673–6 (1995). 119. Nicol, G. W., Glover, L. A. & Prosser, J. I. Molecular analysis of methanogenic archaeal communities in managed and natural upland pasture soils. Glob. Chang. Biol. 9, 1451–1457 (2003). 120. Lee, C. G., Watanabe, T., Murase, J., Asakawa, S. & Kimura, M. Growth of methanogens in an oxic soil microcosm: Elucidation by a DNA-SIP experiment using 13C-labeled dried rice callus. Appl. Soil Ecol. 58, 37–44 (2012). 121. Bridgham, S. D., Cadillo-Quiroz, H., Keller, J. K. & Zhuang, Q. Methane emissions from wetlands: biogeochemical, microbial, and modeling perspectives from local to global scales. Glob. Chang. Biol. 19, 1325–1346 (2013). 122. Melton, J. R. et al. Present state of global wetland extent and wetland methane modelling: conclusions from a model inter-comparison project (WETCHIMP). Biogeosciences 10, 753–788 (2013). 123. Melton, J. & Bohn, T. WETCHIMP-WSL: Intercomparison of wetland methane emissions models over West Siberia. EGU Gen. Assem. 2015, held 12-17 April. 2015 Vienna, Austria. id.4064 17, (2015). 124. Ito, A. & Inatomi, M. Use of a process-based model for assessing the methane budgets of global terrestrial ecosystems and evaluation of uncertainty. Biogeosciences 9, 759–773 (2012). 125. Kaiser, S. Process-based modelling of the methane balance in periglacial landscapes. Jena (2018). 126. Wania, R., Ross, I. & Prentice, I. C. Geoscientific Model Development Implementation and evaluation of a new methane model within a dynamic global vegetation model: LPJ-WHyMe v1.3.1. Geosci. Model Dev 3, 565–584 (2010). 127. Tang, J., Zhuang, Q., Shannon, R. D. & White, J. R. Quantifying wetland methane emissions with process-based models of different complexities. Biogeosciences 7, 3817–3837 (2010). 128. Grant, R. F. & Roulet, N. T. Methane efflux from boreal wetlands: Theory and testing of the ecosystem model Ecosys with chamber and tower flux measurements. Global Biogeochem. Cycles 16, 2-1-2–16 (2002). 129. Uchida, Y., Clough, T. J., Kelliher, F. M. & Sherlock, R. R. Effects of aggregate size, soil compaction, and bovine urine on N2O emissions from a pasture soil. Soil Biol. Biochem. 40, 924–931 (2008). 130. Khalil, K., Mary, B. & Renault, P. Nitrous oxide production by nitrification and denitrification in soil aggregates as affected by O2 concentration. Soil Biol. Biochem. 36, 687–699 (2004). 131. Butterbach-Bahl, K., Baggs, E. M., Dannenmann, M., Kiese, R. & Zechmeister- 142

Boltenstern, S. Nitrous oxide emissions from soils: how well do we understand the processes and their controls? Philos. Trans. R. Soc. Lond. B. Biol. Sci. 368, 20130122 (2013). 132. Bodelier, P. L. E. & Dedysh, S. N. Microbiology of wetlands. Front. Microbiol. 4, 79 (2013). 133. Cedervall, P. E., Dey, M., Pearson, A. R., Ragsdale, S. W. & Wilmot, C. M. Structural Insight into Methyl-Coenzyme M Reductase Chemistry Using Coenzyme B Analogues. Biochemistry 49, 7683–7693 (2010). 134. Bethke, C. M., Sanford, R. A., Kirk, M. F., Jin, Q. & Flynn, T. M. The thermodynamic ladder in geomicrobiology. Am. J. Sci. 311, 183–210 (2011). 135. Yuan, Y., Conrad, R. & Lu, Y. Transcriptional response of methanogen mcrA genes to oxygen exposure of rice field soil. Environ. Microbiol. Rep. 3, 320–328 (2011). 136. Rey-Sanchez, A. C., Morin, T. H., Stefanik, K. C., Wrighton, K. & Bohrer, G. Determining total emissions and environmental drivers of methane flux in a Lake Erie estuarine marsh. Ecol. Eng. (2017). doi:10.1016/J.ECOLENG.2017.06.042 137. Grosskopf, R., Janssen, P. H. & Liesack, W. Diversity and structure of the methanogenic community in anoxic rice paddy soil microcosms as examined by cultivation and direct 16S rRNA gene sequence retrieval. Appl. Environ. Microbiol. 64, 960–9 (1998). 138. He, S. et al. Patterns in wetland microbial community composition and functional gene repertoire associated with methane emissions. MBio 6, e00066-15 (2015). 139. Kelley, L. A., Mezulis, S., Yates, C. M., Wass, M. N. & Sternberg, M. J. E. The Phyre2 web portal for protein modeling, prediction and analysis. Nat. Protoc. 10, 845–858 (2015). 140. Morris, J. J., Lenski, R. E. & Zinser, E. R. The Black Queen Hypothesis: evolution of dependencies through adaptive gene loss. MBio 3, e00036-12 (2012). 141. Fenchel, T. & Finlay, B. J. Production of methane and hydrogen by anaerobic ciliates containing symbiotic methanogens. Arch. Microbiol. 157, 475–480 142. Jetten, M. S. M., Stams, A. J. M. & Zehnder, A. J. B. Acetate threshold values and acetate activating enzymes in methanogenic bacteria. FEMS Microbiol. Lett. 73, 339–344 (1990). 143. Kemnitz, D., Chin, K.-J., Bodelier, P. & Conrad, R. Community analysis of methanogenic archaea within a riparian flooding gradient. Environ. Microbiol. 6, 449–461 (2004). 144. Yavitt, J. B., Yashiro, E., Cadillo-Quiroz, H. & Zinder, S. H. Methanogen diversity and community composition in peatlands of the central to northern Appalachian Mountain region, North America. Biogeochemistry 109, 117–131 (2012). 145. Cadillo-Quiroz, H., Yashiro, E., Yavitt, J. B. & Zinder, S. H. Characterization of the archaeal community in a minerotrophic fen and terminal restriction fragment length polymorphism-directed isolation of a novel hydrogenotrophic methanogen. Appl. Environ. Microbiol. 74, 2059–68 (2008). 146. Itoh, H. et al. Seasonal Transition of Active Bacterial and Archaeal Communities in Relation to Water Management in Paddy Soils. Microbes Environ. 28, 370–380 143

(2013). 147. Chan, O. C. et al. Vertical distribution of structure and function of the methanogenic archaeal community in Lake Dagow sediment. Environ. Microbiol. 7, 1139–1149 (2005). 148. Reis, M. P., Barbosa, F. A. R., Chartone-Souza, E. & Nascimento, A. M. A. The prokaryotic community of a historically mining-impacted tropical stream sediment is as diverse as that from a pristine stream sediment. Extremophiles 17, 301–309 (2013). 149. Stoeva, M. K. et al. Microbial Community Structure in Lake and Wetland Sediments from a High Arctic Polar Desert Revealed by Targeted Transcriptomics. PLoS One 9, e89531 (2014). 150. Morin, T. H. et al. Combining eddy-covariance and chamber measurements to determine the methane budget from a small, heterogeneous urban floodplain wetland park. Agric. For. Meteorol. 237–238, 160–170 (2017). 151. Patel, G. B. Characterization and nutritional properties of Methanothrix concilii sp. nov., a mesophilic, aceticlastic methanogen. Can. J. Microbiol. 30, 1383–1396 (1984). 152. Ma, K., Liu, X. & Dong, X. Methanosaeta harundinacea sp. nov., a novel acetate- scavenging methanogen isolated from a UASB reactor. Int. J. Syst. Evol. Microbiol. 56, 127–131 (2006). 153. Gomez-Alvarez, V., Revetta, R. P. & Domingo, J. W. Metagenome analyses of corroded concrete wastewater pipe biofilms reveal a complex microbial system. BMC Microbiol. 12, 122 (2012). 154. Fernández, N., Díaz, E. E., Amils, R. & Sanz, J. L. Analysis of Microbial Community during Biofilm Development in an Anaerobic Wastewater Treatment Reactor. Microb. Ecol. 56, 121–132 (2008). 155. McKeown, R. M. et al. Psychrophilic methanogenic community development during long-term cultivation of anaerobic granular biofilms. ISME J. 3, 1231–1242 (2009). 156. Brazelton, W. J., Mehta, M. P., Kelley, D. S. & Baross, J. A. Physiological differentiation within a single-species biofilm fueled by serpentinization. MBio 2, e00127-11 (2011). 157. Zhou, L. et al. Transcriptomic and physiological insights into the robustness of long filamentous cells of Methanosaeta harundinacea, prevalent in upflow anaerobic sludge blanket granules. Appl. Environ. Microbiol. 81, 831–9 (2015). 158. Mandal, A. K., Cheung, W. D. & Argüello, J. M. Characterization of a thermophilic P-type Ag+/Cu+-ATPase from the extremophile Archaeoglobus fulgidus. J. Biol. Chem. 277, 7201–8 (2002). 159. Zhang, G. et al. Acyl homoserine lactone-based quorum sensing in a methanogenic archaeon. ISME J. 6, 1336–1344 (2012). 160. Flemming, H.-C. & Wingender, J. The biofilm matrix. Nat. Rev. Microbiol. 8, 623–633 (2010). 161. Ma, L. et al. Assembly and Development of the Pseudomonas aeruginosa Biofilm Matrix. PLoS Pathog. 5, e1000354 (2009). 144

162. Watnick, P. I. & Kolter, R. Steps in the development of a Vibrio cholerae El Tor biofilm. Mol. Microbiol. 34, 586–595 (1999). 163. Franchini, A. G., Henneberger, R., Aeppli, M. & Zeyer, J. Methane dynamics in an alpine fen: a field-based study on methanogenic and methanotrophic microbial communities. FEMS Microbiol. Ecol. 91, (2015). 164. Liu, D. et al. Effect of Paddy-Upland Rotation on Methanogenic Archaeal Community Structure in Paddy Field Soil. Microb. Ecol. 69, 160–168 (2015). 165. Duenas, C., Fernandez, M. C., Carretero, J., Perez, M. & Liger, E. Consumption of methane by soils. Environ. Monit. Assess. 31–31, 125–130 (1994). 166. Rudd, J. W. M. & Hamilton, R. D. Methane cycling in a cutrophic shield lake and its effects on whole lake metabolism. Limnol. Oceanogr. 23, 337–348 (1978). 167. Vigano, I., Weelden, H. Van, Holzinger, R., Keppler, F. & Röckmann, T. Effect of UV radiation and temperature on the emission of methane from plant biomass and structural components. Biogeosciences Discuss. 5, 243–270 (2008). 168. Yarlett, N. & Hackstein, J. Hydrogenosomes: One Organelle, Multiple Origins. Bioscience 55, 657–668 (2005). 169. Scott, R. I., Norman Williams, T., Whitmore, T. N. & Lloyd, D. Direct measurement of methanogenesis in anaerobic digestors by membrane inlet mass spectrometry. Eur. J. Appl. Microbiol. Biotechnol. 18, 236–241 (1983). 170. Chae, K.-J. et al. Methanogenesis control by employing various environmental stress conditions in two-chambered microbial fuel cells. Bioresour. Technol. 101, 5350–5357 (2010). 171. Keiluweit, M., Nico, P. S., Kleber, M. & Fendorf, S. Are oxygen limitations under recognized regulators of organic carbon turnover in upland soils? Biogeochemistry 127, 157–171 (2016). 172. Chin, Y.-P., Traina, S. J., Swank, C. R. & Backhus, D. Abundance and properties of dissolved organic matter in pore waters of a freshwater wetland. Limnol. Oceanogr. 43, 1287–1296 (1998). 173. Bodegom, P. M. va. & Stams, A. J. M. Effects of alternative electron acceptors and temperature on methanogenesis in rice paddy soils. Chemosphere 39, 167–182 (1999). 174. Miller, K. E., Lai, C.-T., Friedman, E. S., Angenent, L. T. & Lipson, D. A. Methane suppression by iron and humic acids in soils of the Arctic Coastal Plain. Soil Biol. Biochem. 83, 176–183 (2015). 175. Pramanik, P. & Kim, P. J. Effect of limited nickel availability on methane emission from EDTA treated soils: Coenzyme M an alternative biomarker for methanogens. Chemosphere 90, 873–876 (2013). 176. Roden, E. E. & Wetzel, R. G. Competition between Fe(III)-Reducing and Methanogenic Bacteria for Acetate in Iron-Rich Freshwater Sediments. Microb. Ecol. 45, 252–258 (2003). 177. Kotsyurbenko, O. R., Glagolev, M. V., Nozhevnikova, A. N. & Conrad, R. Competition between homoacetogenic bacteria and methanogenic archaea for hydrogen at low temperature. FEMS Microbiol. Ecol. 38, 153–159 (2001). 178. Seth, E. C. & Taga, M. E. Nutrient cross-feeding in the microbial world. Front. 145

Microbiol. 5, 350 (2014). 179. Lai, M. C. & Chen, S. C. Methanofollis aquaemaris sp. nov., a methanogen isolated from an aquaculture fish pond. Int. J. Syst. Evol. Microbiol. 51, 1873–80 (2001). 180. Grissa, I., Vergnaud, G. & Pourcel, C. The CRISPRdb database and tools to display CRISPRs and to generate dictionaries of spacers and repeats. BMC Bioinformatics 8, 172 (2007). 181. Stefanik, K. C. & Mitsch, W. J. Metabolism and methane flux of dominant macrophyte communities in created riverine wetlands using open system flow through chambers. Ecol. Eng. 72, 67–73 (2014). 182. Chan, A. S. K. & Parkin, T. B. Methane Oxidation and Production Activity in Soils from Natural and Agricultural Ecosystems. J. Environ. Qual. 30, 1896 (2001). 183. Daly, R. A. et al. Microbial metabolisms in a 2.5-km-deep ecosystem created by hydraulic fracturing in shales. Nat. Microbiol. 1, 16146 (2016). 184. MacDonald, L. H., Paull, J. S. & Jaffé, P. R. Enhanced semipermanent dialysis samplers for long-term environmental monitoring in saturated sediments. Environ. Monit. Assess. 185, 3613–3624 (2013). 185. Reid, M. C., Tripathee, R., Schäfer, K. V. R. & Jaffé, P. R. Tidal marsh methane dynamics: Difference in seasonal lags in emissions driven by storage in vegetated versus unvegetated sediments. J. Geophys. Res. Biogeosciences 118, 1802–1813 (2013). 186. Pal, D. S., Reid, M. C. & Jaffé, P. R. Impact of Hurricane Sandy on CH4 Released from Vegetated and Unvegetated Wetland Microsites. Environ. Sci. Technol. Lett. 1, 372–375 (2014). 187. Wright, J. J., Konwar, K. M. & Hallam, S. J. Microbial ecology of expanding oxygen minimum zones. Nat. Rev. Microbiol. 10, 381–394 (2012). 188. Haario, H., Saksman, E. & Tamminen, J. Adaptive proposal distribution for random walk Metropolis algorithm. Comput. Stat. 14, 375 (1999). 189. Haario, H., Laine, M., Mira, A. & Saksman, E. DRAM: Efficient adaptive MCMC. Stat. Comput. 16, 339–354 (2006). 190. Morin, T. H. et al. Environmental drivers of methane fluxes from an urban temperate wetland park. J. Geophys. Res. Biogeosciences 119, 2188–2208 (2014). 191. Morin, T. H. et al. The seasonal and diurnal dynamics of methane flux at a created urban wetland. Ecol. Eng. 72, 74–83 (2014). 192. Lee, X., Finnigan, J. & Paw U, K. T. Coordinate Systems and Flux Bias Error. in Handbook of Micrometeorology 33–66 (Kluwer Academic Publishers, 2004). doi:10.1007/1-4020-2265-4_3 193. Detto, M., Verfaillie, J., Anderson, F., Xu, L. & Baldocchi, D. Comparing laser- based open- and closed-path gas analyzers to measure methane fluxes using the eddy covariance method. Agric. For. Meteorol. 151, 1312–1324 (2011). 194. Webb, E. K., Pearman, G. I. & Leuning, R. Correction of flux measurements for density effects due to heat and water vapour transfer. Q. J. R. Meteorol. Soc. 106, 85–100 (1980). 146

195. Massman, W. J. A simple method for estimating frequency response corrections for eddy covariance systems. Agric. For. Meteorol. 104, 185–198 (2000). 196. Reichstein, M. et al. On the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm. Glob. Chang. Biol. 11, 1424–1439 (2005). 197. Moffat, A. M. et al. Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes. Agric. For. Meteorol. 147, 209–232 (2007). 198. Papale, D. & Valentini, R. A new assessment of European forests carbon exchanges by eddy fluxes and artificial neural network spatialization. Glob. Chang. Biol. 9, 525–535 (2003). 199. Detto, M., Montaldo, N., Albertson, J. D., Mancini, M. & Katul, G. Soil moisture and vegetation controls on evapotranspiration in a heterogeneous Mediterranean ecosystem on Sardinia, Italy. Water Resour. Res. 42, (2006). 200. Hsieh, C.-I., Katul, G. & Chi, T. An approximate analytical model for footprint estimation of scalar fluxes in thermally stratified atmospheric flows. Adv. Water Resour. 23, 765–772 (2000). 201. Solden, L. M. et al. New roles in hemicellulosic sugar fermentation for the uncultivated Bacteroidetes family BS11. ISME J. 11, 691–703 (2017). 202. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359 (2012). 203. Brown, C. T. et al. Unusual biology across a group comprising more than 15% of domain Bacteria. Nature 523, 208–211 (2015). 204. Wrighton, K. C. et al. Fermentation, hydrogen, and sulfur metabolism in multiple uncultivated bacterial phyla. Science 337, 1661–5 (2012). 205. Hyatt, D., LoCascio, P. F., Hauser, L. J. & Uberbacher, E. C. Gene and translation initiation site prediction in metagenomic sequences. Bioinformatics 28, 2223–2230 (2012). 206. Edgar, R. C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461 (2010). 207. Quevillon, E. et al. InterProScan: protein domains identifier. Nucleic Acids Res. 33, W116–W120 (2005). 208. Kang, D. D., Froula, J., Egan, R. & Wang, Z. MetaBAT, an efficient tool for accurately reconstructing single genomes from complex microbial communities. PeerJ 3, e1165 (2015). 209. Wu, M. & Scott, A. J. Phylogenomic analysis of bacterial and archaeal sequences with AMPHORA2. Bioinformatics 28, 1033–1034 (2012). 210. Benedict, M. N., Henriksen, J. R., Metcalf, W. W., Whitaker, R. J. & Price, N. D. ITEP: An integrated toolkit for exploration of microbial pan-genomes. BMC Genomics 15, 8 (2014). 211. Trapnell, C. et al. Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat. Biotechnol. 28, 511–515 (2010). 212. Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32, 1792–1797 (2004). 147

213. Sassera, D. et al. Phylogenomic Evidence for the Presence of a Flagellum and cbb3 Oxidase in the Free-Living Mitochondrial Ancestor. Mol. Biol. Evol. 28, 3285–3296 (2011). 214. Darriba, D., Taboada, G. L., Doallo, R. & Posada, D. ProtTest 3: fast selection of best-fit models of protein evolution. Bioinformatics 27, 1164–1165 (2011). 215. Stamatakis, A. RAxML version 8: a tool for phylogenetic analysis and post- analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014). 216. Letunic, I. & Bork, P. Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucleic Acids Res. 44, W242–W245 (2016). 217. Ludwig, W. et al. ARB: a software environment for sequence data. Nucleic Acids Res. 32, 1363–1371 (2004). 218. Pruesse, E., Peplies, J. & Glöckner, F. O. SINA: Accurate high-throughput multiple sequence alignment of ribosomal RNA genes. Bioinformatics 28, 1823– 1829 (2012). 219. Liu, D., Ding, W., Yuan, J., Xiang, J. & Lin, Y. Substrate and/or substrate-driven changes in the abundance of methanogenic archaea cause seasonal variation of methane production potential in species-specific freshwater wetlands. Appl. Microbiol. Biotechnol. 98, 4711–4721 (2014). 220. Yang, S. et al. In-depth analysis of core methanogenic communities from high elevation permafrost-affected wetlands. Soil Biol. Biochem. 111, 66–77 (2017). 221. Zhang, G., Jiang, N., Liu, X. & Dong, X. Methanogenesis from methanol at low temperatures by a novel psychrophilic methanogen, & Methanolobus psychrophilus; sp. nov., prevalent in Zoige wetland of the Tibetan plateau. Appl. Environ. Microbiol. 74, 6114–20 (2008). 222. Evans, P. N. et al. Methane metabolism in the archaeal phylum Bathyarchaeota revealed by genome-centric metagenomics. Science 350, 434–8 (2015). 223. Hultman, J. et al. Multi-omics of permafrost, active layer and thermokarst bog soil microbiomes. Nature 521, 208–212 (2015). 224. Mackelprang, R., Saleska, S. R., Jacobsen, C. S., Jansson, J. K. & Taş, N. Permafrost Meta-Omics and Climate Change. Annu. Rev. Earth Planet. Sci. 44, 439–462 (2016). 225. Taş, N. et al. Landscape topography structures the soil microbiome in arctic polygonal tundra. Nat. Commun. 9, 777 (2018). 226. Scheller, S., Goenrich, M., Boecher, R., Thauer, R. K. & Jaun, B. The key nickel enzyme of methanogenesis catalyses the anaerobic oxidation of methane. Nature 465, 606–608 (2010). 227. Hallam, S. J. et al. Reverse methanogenesis: testing the hypothesis with environmental genomics. Science 305, 1457–62 (2004). 228. Timmers, P. H. A. et al. Reverse Methanogenesis and Respiration in Methanotrophic Archaea. Archaea 2017, 1–22 (2017). 229. Mondav, R. et al. Discovery of a novel methanogen prevalent in thawing permafrost. Nat. Commun. 5, 3212 (2014). 230. Parks, D. H. et al. Recovery of nearly 8,000 metagenome-assembled genomes 148

substantially expands the tree of life. Nat. Microbiol. 2, 1533–1542 (2017). 231. Tveit, A., Schwacke, R., Svenning, M. M. & Urich, T. Organic carbon transformations in high-Arctic peat soils: key functions and microorganisms. ISME J. 7, 299–311 (2013). 232. Tveit, A. T., Urich, T., Frenzel, P. & Svenning, M. M. Metabolic and trophic interactions modulate methane production by Arctic peat microbiota in response to warming. Proc. Natl. Acad. Sci. U. S. A. 112, E2507-16 (2015). 233. Söllinger, A. et al. Phylogenetic and genomic analysis of Methanomassiliicoccales in wetlands and animal intestinal tracts reveals clade-specific habitat preferences. FEMS Microbiol. Ecol. 92, fiv149 (2016). 234. Yuan, J., Ding, W., Liu, D., Xiang, J. & Lin, Y. Methane production potential and methanogenic archaea community dynamics along the Spartina alterniflora invasion chronosequence in a coastal salt marsh. Appl. Microbiol. Biotechnol. 98, 1817–1829 (2014). 235. Lloyd, K. G., Alperin, M. J. & Teske, A. Environmental evidence for net methane production and oxidation in putative ANaerobic MEthanotrophic (ANME) archaea. Environ. Microbiol. 13, 2548–2564 (2011). 236. Raghoebarsing, A. A. et al. A microbial consortium couples anaerobic methane oxidation to denitrification. Nature 440, 918–921 (2006). 237. Haroon, M. F. et al. Anaerobic oxidation of methane coupled to nitrate reduction in a novel archaeal lineage. Nature 500, 567–570 (2013). 238. Arshad, A. et al. A Metagenomics-Based Metabolic Model of Nitrate-Dependent Anaerobic Oxidation of Methane by Methanoperedens-Like Archaea. Front. Microbiol. 6, 1423 (2015). 239. Ettwig, K. F. et al. Archaea catalyze iron-dependent anaerobic oxidation of methane. Proc. Natl. Acad. Sci. U. S. A. 113, 12792–12796 (2016). 240. Ferry, J. G. The chemical biology of methanogenesis. Planet. Space Sci. 58, 1775– 1783 (2010). 241. Jetten, M. S. M., Stams, A. J. M. & Zehnder, A. J. B. Methanogenesis from acetate: a comparison of the acetate metabolism in Methanothrix soehngenii and Methanosarcina spp. FEMS Microbiol. Lett. 88, 181–197 (1992). 242. Tonner, P. D., Pittman, A. M. C., Gulli, J. G., Sharma, K. & Schmid, A. K. A Regulatory Hierarchy Controls the Dynamic Transcriptional Response to Extreme Oxidative Stress in Archaea. PLoS Genet. 11, e1004912 (2015). 243. Li, D.-C., Yang, F., Lu, B., Chen, D.-F. & Yang, W.-J. Thermotolerance and molecular chaperone function of the small heat shock protein HSP20 from hyperthermophilic archaeon, Sulfolobus solfataricus P2. Cell Stress Chaperones 17, 103–108 (2012). 244. Ling, J. & Söll, D. Severe oxidative stress induces protein mistranslation through impairment of an aminoacyl-tRNA synthetase editing site. Proc. Natl. Acad. Sci. U. S. A. 107, 4028–33 (2010). 245. Freestone, P., Nyström, T., Trinei, M. & Norris, V. The universal stress protein, UspA, of Escherichia coli is phosphorylated in response to stasis. J. Mol. Biol. 274, 318–324 (1997). 149

246. Wu, J.-H. et al. Community and proteomic analysis of methanogenic consortia degrading terephthalate. Appl. Environ. Microbiol. 79, 105–12 (2013). 247. Nyström, T. & Neidhardt, F. C. Effects of overproducing the universal stress protein, UspA, in Escherichia coli K-12. J. Bacteriol. 178, 927–30 (1996). 248. Maupin-Furlow, J. A. & Ferry, J. G. A proteasome from the methanogenic archaeon Methanosarcina thermophila. J. Biol. Chem. 270, 28617–22 (1995). 249. Forouzan, D. et al. The archaeal proteasome is regulated by a network of AAA ATPases. J. Biol. Chem. 287, 39254–62 (2012). 250. Chen, Z., Yu, H., Li, L., Hu, S. & Dong, X. The genome and transcriptome of a newly described psychrophilic archaeon, Methanolobus psychrophilus R15, reveal its cold adaptive characteristics. Environ. Microbiol. Rep. 4, n/a-n/a (2012). 251. Reuter, C. J., Kaczowka, S. J. & Maupin-Furlow, J. A. Differential regulation of the PanA and PanB proteasome-activating nucleotidase and 20S proteasomal proteins of the haloarchaeon Haloferax volcanii. J. Bacteriol. 186, 7763–72 (2004). 252. Cerletti, M., Paggi, R. A., Guevara, C. R., Poetsch, A. & De Castro, R. E. Global role of the membrane protease LonB in Archaea: Potential protease targets revealed by quantitative proteome analysis of a lonB mutant in Haloferax volcanii. J. Proteomics 121, 1–14 (2015). 253. Dueholm, M. S. et al. The Tubular Sheaths Encasing Methanosaeta thermophila Filaments Are Functional Amyloids. J. Biol. Chem. 290, 20590–600 (2015). 254. Li, L. et al. Significant performance enhancement of a UASB reactor by using acyl homoserine lactones to facilitate the long filaments of Methanosaeta harundinacea 6Ac. Appl. Microbiol. Biotechnol. 99, 6471–6480 (2015). 255. Imachi, H. et al. Methanolinea tarda gen. nov., sp. nov., a methane-producing archaeon isolated from a methanogenic digester sludge. Int. J. Syst. Evol. Microbiol. 58, 294–301 (2008). 256. Browne, P. et al. Genomic composition and dynamics among Methanomicrobiales predict adaptation to contrasting environments. ISME J. 11, 87–99 (2017). 257. Garcia, J.-L., Ollivier, B. & Whitman, W. B. The Order Methanomicrobiales. The Prokaryotes (2006). 258. Wu, Y.-W. ezTree: an automated pipeline for identifying phylogenetic marker genes and inferring evolutionary relationships among uncultivated prokaryotic draft genomes. BMC Genomics 19, 921 (2018). 259. Haft, D. H., Selengut, J. D. & White, O. The TIGRFAMs database of protein families. Nucleic Acids Res. 31, 371–373 (2003). 260. Bateman, A. et al. UniProt: the universal protein knowledgebase. Nucleic Acids Res. 45, D158–D169 (2017). 261. Kanehisa, M. & Goto, S. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res. 28, 27–30 (2000). 262. Li, D., Liu, C.-M., Luo, R., Sadakane, K. & Lam, T.-W. MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics 31, 1674–1676 (2015). 263. Soding, J. Protein homology detection by HMM-HMM comparison. 150

Bioinformatics 21, 951–960 (2005). 264. Kearse, M. et al. Geneious Basic: An integrated and extendable desktop software platform for the organization and analysis of sequence data. Bioinformatics 28, 1647–1649 (2012). 265. Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P. & Tyson, G. W. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–55 (2015). 266. Wickham, H. ggplot2 – Elegant Graphics for Data Analysis (2nd Edition). J. Stat. Softw. 77, (2017). 267. Tanner, C. C., D’Eugenio, J., McBride, G. B., Sukias, J. P. S. & Thompson, K. Effect of water level fluctuation on nitrogen removal from constructed wetland mesocosms. Ecol. Eng. 12, 67–92 (1999). 268. Sasikala, S., Tanaka, N., Wah Wah, H. S. Y. & Jinadasa, K. B. S. N. Effects of water level fluctuation on radial oxygen loss, root porosity, and nitrogen removal in subsurface vertical flow wetland mesocosms. Ecol. Eng. 35, 410–417 (2009). 269. Watanabe, T., Kimura, M. & Asakawa, S. Distinct members of a stable methanogenic archaeal community transcribe mcrA genes under flooded and drained conditions in Japanese paddy field soil. Soil Biol. Biochem. 41, 276–285 (2009). 270. Kotiaho, M. et al. Methanogen activity in relation to water table level in two boreal fens. Biol. Fertil. Soils 46, 567–575 (2010). 271. Freeman, C., Ostle, N. & Kang, H. An enzymic ‘latch’ on a global carbon store. Nature 409, 149–149 (2001). 272. Fenner, N. & Freeman, C. Drought-induced carbon loss in peatlands. Nat. Geosci. 4, 895–900 (2011). 273. Schmidt, O., Horn, M. A., Kolb, S. & Drake, H. L. Temperature impacts differentially on the methanogenic food web of cellulose-supplemented peatland soil. Environ. Microbiol. 17, 720–734 (2015). 274. Yang, Z., Wullschleger, S. D., Liang, L., Graham, D. E. & Gu, B. Effects of warming on the degradation and production of low-molecular-weight labile organic carbon in an Arctic tundra soil. Soil Biol. Biochem. 95, 202–211 (2016). 275. Jahnke, R. A. A simple, reliable, and inexpensive pore-water sampler. Limnol. Oceanogr. 33, 483–487 (1988).

151

Appendix A: porewater collection and concentration protocol

Porewater collection:

Soil cores retrieved from OWC for the purpose of porewater collections should be clearly marked as such prior to hydraulic extrusion to avoid incident. When ready to proceed with core-squeezing (described in detail previously172,275, follow these steps:

1) Hydraulically extrude the soil directly into Jahnke whole-core squeezer

sleeves (perforated with holes for porewater removal). Temporarily cap the

sleeve ends using ethanol-sterilized black rubber stoppers.

2) Take the core sleeve to the wall-mounted squeezer apparatus.

3) Screw in the 70-um Porex filter cylinders to each of the open holes (inbed the

filter securely in the threaded plastic connector previously).

4) Replace the top and bottom black rubber stoppers with the plungers (note: the

top plunger has a special release valve screw) and mount machine.

5) Attached this baked glass syringes to the back of each plastic filter connector,

trying to keep the glass syringe as anoxic as possible throughout the exchange.

6) Pressurize the bottom piston upwards. Fluid should eventually be forced

through the Porex filter and out into the syringes.

7) When a syringe is full or ready to be removed, quickly unscrew it from the

connector and attach a 0.2 um filter to the syringe. 152

8) Mount a needle onto the 0.2 um filter, and push syringe contents into the

anoxic, butyl-stoppered baked vial which will house the porewaters.

9) Depending on the downstream use of the porewater, you may acidify them

with HCl to pH 2.0 immediately or wait if the porewaters will be used for

inoculation experiments.

10) Immediately move the porewaters to storage at 4C until ready for transport.

11) If oxygen was accidently introduced at any point, unfull vials can be gassed

out using filters for input gas line if needed.

Porewater concentration:

Porewaters returned from field collection can be filtered and concentrated for downstream analyses such as FTICR-MS analysis. Special note: you are only taking a subsample of the whole porewater sample for this – not modifying the main stock of porewater. To keep porewater samples anoxic all the way to their containers to shipment, the work is performed in the anaerobic chamber.

1) Move all required equipment into the anaerobic chamber. This includes porewater

vials to be extracted from, sterile syringes, intermediate baked vials for mixing

extracted porewaters, metal top removal tool, crimper, crimp tops, baked

glassware housing final concentrated solutions, concentrated H3PO4, methanol

container, waste container, tape, sharpies, and Bond Elut 100mg filters for

filtering and concentrating the solutions.

2) To begin, remove the metal tops from each vial currently storing porewater.

153

3) One at a time, remove the rubber butyl cap carefully as to not spill the liquid

inside.

4) Using a fresh baked glass pipette, remove 3 mL of a given sample and dispense it

into the intermediate housing baked glass vial you will be using for acidification

and mixing.

5) Put the blue butyl stopper back in the porewater-containing vial and recrimp it.

6) Repeat for each of the porewater samples you want to concentrate.

7) Activate each cartridge in sequential order with 3 mL of MeOH. This will take a

moment to pass through.

8) While the cartridges are passing the methanol, add 1 to 2 drops of H3PO4 to the 3

mL of each sample to lower the pH. Be extremely careful working with the

concentrated H3PO4.

9) When all methanol has passed through the columns, they are ready to be used to

concentrate samples. Using another fresh baked glass pipette, move each 3 mL of

acidified porewater sample to it’s corresponding column and let it begin to move

through the column.

10) Once the acidified sample has passed through the column, wash the columns with

3 rounds of 10 mM HCl.

11) Carefully parafilm each of the cartrigdes and label them so you know which is

which. Remove the cartridges collectively from the anaerobic chamber and take

them to the gassing station for the N2 drying step.

12) Using the appropriate attachments, dry the cartridges with N2 gas.

154

13) Collectively move the dried cartridges back into the anaerobic chamber, deparfilm

them, and hang them into the vials where the final concentration porewater

samples will be stored.

14) Elute the organic carbon from the cartridge w 3 mL of MeOH (or 1.5 mL if you

want to concentrate the carbon content of the sample).

15) Discard the cartridges and seal the purified organic carbon in methanol samples

tightly. Parafilm around the opening of the lid as tightly as possible. Ensure that

each storage tube is properly labeled.

16) Remove the storage tubes with the concentrated organic carbon in methanol from

the anaerobic chamber and store them immediately at -20°C.

17) Clean all materials out of the anaerobic chamber, making sure to properly dispose

of the acidified methanol waste that will remain in the containers. Also, handle

with extreme care any remaining acid you have left until you can dispose of it

properly.

155