The impact of an active soil microbial community on greenhouse gas emissions in Arctic cryosols.

By

Ianina Altshuler Department of Natural Resource Sciences Faculty of Agriculture and Environmental Sciences McGill University April, 2019

A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Doctor of Philosophy

© Ianina Altshuler 2019

Petit à petit, l'oiseau fait son nid…

- Old French Proverb

ii

Table of Contents

Table of Contents ...... iii

List of Tables ...... vii

List of Figures ...... viii

Acknowledgments...... x

Abstract ...... xi

Résumé ...... xiii

Contributions to knowledge ...... xvi

List of Abbreviations ...... xvii

Introduction and Literature Review: Microbial Life in Permafrost Soils ...... 1

Introduction ...... 1

Permafrost environments ...... 2

Microbial diversity and abundance in permafrost ...... 6

Are they livin’ or just chillin’?...... 11

1.4.1 Soil respiration and laboratory incorporation studies ...... 12

1.4.2 Stable isotope probing in permafrost soils ...... 13

1.4.3 RNA/DNA ratios ...... 14

1.4.4 Subzero growth of permafrost isolates ...... 15

1.4.5 DNA repair...... 17

Live microbes or ancient DNA ...... 19

1.5.1 Strategies for differentiating between old biomarkers and an active microbial community ...... 21

Warming climate and permafrost ...... 22

1.6.1 Methane dynamics in permafrost affected cryosols...... 26

iii

Objectives of this thesis ...... 29

Connecting Text: ...... 34

Species interactions and distinct microbial communities in high Arctic permafrost affected cryosols are associated with the CH4 and CO2 gas fluxes...... 35

Abstract ...... 35

Introduction ...... 37

Materials and Methods ...... 41

2.3.1 Site and sample collection ...... 41

2.3.2 Gas flux measurements ...... 41

2.3.3 DNA extraction and sequencing ...... 42

2.3.4 Isolation of IWP bacteria ...... 43

2.3.5 Bioinformatics...... 43

2.3.6 Hybrid network analysis ...... 45

Results ...... 47

2.4.1 Gas flux ...... 47

2.4.2 IWP terrain microbial community diversity ...... 48

2.4.3 Hybrid network analysis of the microbial community ...... 48

2.4.4 PmoA diversity at the IWP terrain ...... 49

2.4.5 Bacterial isolates from the IWP terrain ...... 50

Discussion ...... 51

Conclusions ...... 58

Supplementary Materials ...... 70

Connecting Text: ...... 77

Denitrifiers, Nitrogen-fixing bacteria, and N2O ...... 78 soil gas flux in high Arctic ice-wedge polygon cryosols...... 78

Abstract ...... 79

iv

Introduction ...... 80

Material and Methods ...... 83

3.3.1 Gas flux measurements, flux calculation, and statistics ...... 83

3.3.2 Site and sample collection ...... 84

- - + 3.3.3 NO3 & NO2 , NH4 , and soil moisture measurements ...... 85

3.3.4 DNA and RNA isolation and sequencing ...... 85

3.3.5 qPCR analysis and statistics...... 86

3.3.6 Bioinformatics and statistical analysis of sequencing data ...... 87

3.3.7 Phylogenetic analysis ...... 88

Results and Discussion ...... 90

- - + 3.4.1 N2O Gas Flux, Soil NO3 & NO2 , NH4 , and Moisture Content ...... 90

3.4.2 Quantitative PCR ...... 91

3.4.3 Community structure of potential denitrifiers and nitrogen fixing bacteria ...... 92

3.4.4 Denitrifying community...... 92

3.4.5 Nitrogen-fixing community ...... 94

3.4.6 Metatranscriptomics ...... 96

Conclusions ...... 98

Supplementary Materials: ...... 109

Connecting Text: ...... 114

In situ DNA-SIP enrichment in concert with genome binning of a high Arctic methanotrophic and methanotroph-associated community...... 115

Abstract ...... 116

Introduction ...... 117

Methods...... 120

Results and Discussion ...... 123

Conclusion ...... 128

v

Discussion and Conclusions ...... 137

Gas flux and microbial community at the IWP terrain ...... 137

Microbial interactions at the IWP terrain ...... 139

Potential, alternative, non-biological methane uptake processes...... 142

Future directions and experimental improvements ...... 145

References ...... 148

vi

List of Tables

Table 1.1. Table outlining the microbial members of permafrost community present, active, and cultivable in Arctic, Antarctic and in high altitude permafrost...... 32

Table 2.1. Microbial community members with the highest number of positive and negative interactions...... 64

Table 2.2. pmoA OTUs and their closest cultured and environmental BLAST matches...... 67

Table 2.3. Table of isolates from the ice-wedge polygon terrain...... 68

Table 3.1. The content of NO3 & NO2 and NH4 as well as soil moisture of the polygon interior and trough soils at 5 cm and 25 cm...... 106

Table 3.2. OTUs recovered in the amplicon sequencing of the nirS gene in the trough and polygon interior soils...... 107

Table 4.1. Select high quality to intermediate MAGs from the 13C 100 ppm CH4-SIP enrichment...... 135

vii

List of Figures

Figure 1.1 Permafrost schematic...... 31

Figure 2.1. CO2 and CH4 flux at the IWP terrain...... 60

Figure 2.2. 16S community profiling of soils at the high Arctic IWP site...... 61

Figure 2.3. Hybrid network analysis of microorganism at the IWP site...... 62

Figure 2.4. pmoA community structure at the IWP terrain...... 63

Figure 2.5. pmoA phylogeny at the IWP terrain...... 64

Figure 3.1. In situ gas concentration measurements...... 100

Figure 3.2. Quantification of nirS in IWP soils...... 101

Figure 3.3. Microbial composition of nitrogen fixing nirS gene at the Order level in the trough soils and polygon interior soils at 5 cm and 25 cm...... 102

Figure 3.4. Phylogenetic reconstruction of the nirS gene...... 103

Figure 3.5. Microbial composition of nitrogen fixing nifH gene at the Order level in the trough soils and polygon interior soils at 5 cm and 25 cm...... 104

Figure 3.6. Relative abundance (%) of transcripts involved in the nitrogen cycle in trough and polygon interior top 5 cm soil metatranscriptomes...... 105

Figure 4.1. Methane gas flux and qPCR at the IWP terrain. (A) Average CH4 gas measurements in the static chambers across the IWP terrain. (B) Quantitative PCR of the pmoA particulate methane monooxygenase gene in the trough (Tr) and polygon interior (PI) soils at 5 cm and 25 cm at the IWP terrain...... 130

Figure 4.2. Visualization of the heavy and light bands. The heavy band is the 13C labelled DNA...... 131

viii

Figure 4.3. The microbial community composition. The community composition is based on

16S profiling in the heavy and light bands from the SIP CH4 enrichment at 100 ppm and 1000 ppm, as well as the composition of the control soils that were not enriched in CH4...... 132

Figure 4.4.Beta-diversity measure of the microbial communities. PCoA Bray-Curtis analysis of the microbial community composition in the heavy and light bands from the SIP CH4 enrichment at 100 ppm and 1000 ppm, as well as the composition of the control soils that were not enriched in CH4...... 133

Figure 4.5. The pmoA containing microbial community profiling in the heavy 13C labeled band from the SIP CH4 enrichment at 100 ppm and 1000 ppm...... 134

Figure 4.6 Schematic of theoretical abiotic CH4 consumption ...... 135

Figure 5.1 Schematic of theoretical abiotic CH4 consumption ...... 147

ix

Acknowledgments

Foremost, I would like to express my gratitude to Dr. Lyle Whyte for his continued support and inspiration, for keeping me focused, and supporting my research directions. Additionally, my co- supervisor Dr. Charles Greer, has been invaluable in the completion of this thesis.

A huge thanks to my family, Mommy, and friends for their love, support, and understanding. Anne

M. Mcleod and Katie Millette for letting me bounce ideas off them. Estefania Bernardo for always being there.

Shaun Turney, Jennifer Ronholm, and especially Isabelle Raymond-Bouchard for their collaborations and indispensable help.

Lastly, I would like to acknowledge my amazing lab mates, that have made this thesis a joy, David

Touchette, Catherine Maggiori, Esteban Gongora, Jacqueline Goordial, Raven Comery, Jesse

Colangelo-Lillis, Brady O’Connor, Evan Marcolefas, Olivia Blenner-Hassett, and Elisse

Magnuson.

x

Abstract

Anthropogenic climate change is thought to have a disproportionately larger impact on polar regions, resulting in permafrost thaw and microorganism mediated greenhouse gas (GHG) emission. Permafrost soils contain between 25-50 % of the total soil organic carbon pool and as permafrost thaws, this carbon will become accessible to microbial degradation. Carbon dioxide

(CO2), methane (CH4), and nitrous oxide (N2O) are the most important GHGs and their flux from permafrost affected soils contributes to a positive feedback loop of climatic warming. However, our understanding of how microorganisms contribute to the biogeochemical cycling and flux of these gases in Arctic soils remains limited.

Topography of the Arctic landscape has a significant impact on GHG emissions as evidenced by the flux at the ice-wedge polygon (IWP) terrain. The wetter tough soils exhibited higher emissions of CO2 and N2O, but lower uptake of CH4, compared to the drier polygon interior soils. The elevated CO2 and N2O fluxes, and the lower CH4 uptake from troughs is concerning from a climate warming perspective since parts of the Arctic are predicted to become warmer and wetter.

Topography also affected the composition of the overall microbial community, with the trough soils having a higher proportion of Betaproteobacteria, Deltaproteobacteria, and Bacterioidetes but a lower proportion of Acidobacteria compared to polygon interior soils. The community of nitrogen fixers, methanotrophs, and denitrifiers was also affected by the topography with all three groups showing unique structures. Overall, members of the nitrogen-fixing and denitrifying bacteria included Rhizobiaceae, Nostocaceae, Cyanothecaceae, Rhodobacteraceae,

Burkholderiaceae, Chloroflexaceae, Azotobacteraceae, and Ectothiorhodospiraceae. Moreover, these organisms appear to be active in the soils, as metatranscriptomic RNA analysis was also able to detect these microbial clades. The active methanotrophs in these soils are likely part of the USCα

xi cluster of currently uncultured high-affinity methanotrophs, as evidenced by stable isotope probing

(SIP) of soils exhibiting CH4 uptake. SIP analysis coupled with metagenome binning lead to the identification of several intermediate-high quality MAGs (metagenome assemble genomes). One

Alphaproteobacterial MAG was identified that contained many of the methane cycling genes including a soluble methane monooxygenase (mmoX) and genes involved in the serine cycle for assimilating formaldehyde characteristic of type II methanotrophs. This MAG also contained genes for ammonia assimilation, biopolymer production, and mercury detoxification. In addition to identifying non-culturable members of the community through metagenome binning, sequencing of culturable isolates reveal presence of carbon cycling genes involved in fermentation,

CO2 fixation, denitrification, polysaccharide and aromatic compound metabolism. Suggesting that the microbial community at the IWP terrain is poised to degrade the thawing carbon stores in permafrost.

In addition to topography affecting the microbial community structure, key microbial members across the IWP terrain also appear to have positive and negative impacts on other microbial species. This was determined by developing a novel hybrid network analysis to determine species interactions within of the microbial community. Overall, members of Proteobacteria, Candidatus

Rokubacteria, and Actinobacteria phyla tended to have a positive impact, while members of

Verrucomicrobia and Acidobacteria had a negative impact on other microbials members. These results indicate that both environmental abiotic parameters and biotic interactions impact the microbial community structure and possibly GHG fluxes from soils.

xii

Résumé

Les changements climatiques d’origine anthropogénique ont un impact plus important dans les régions polaires, provoquant la fonte du pergélisol et l’émission de gaz à effet de serre (GES) par les microorganismes. Le pergélisol contient entre 25 et 50 pourcents du réservoir total de carbone organique et, avec la fonte du pergélisol, ce carbone devient accessible à la dégradation microbienne. Le dioxyde de carbone (CO2), le méthane (CH4) et le protoxyde d’azote (N2O) sont les GES les plus importants et leurs flux provenant de la fonte du pergélisol contribuent à une boucle de rétroaction positive sur le réchauffement du climat. Toutefois, notre compréhension du rôle des microorganismes sur les cycles biogéochimiques et sur les flux de ces gaz dans les sols arctiques est limitée.

La topographie des paysages arctiques a elle aussi un impact important sur l’émission de GES, comme le démontrent les flux de gaz en terrain « ice-wedge polygon » (IWP), topographie retrouvée dans l’Arctique. Les sols des arêtes des polygones sont plus creux et plus humides. Ils présentent un plus haut taux d’émission de CO2 et de N2O, mais une plus faible absorption de CH4 que les sols plus secs de l’intérieur des polygones. Les flux de CO2 et de N2O élevés ainsi que la faible absorption de CH4 provenant des arêtes des polygones sont particulièrement inquiétants dans une perspective de changements climatiques puisque plusieurs régions arctiques devraient devenir plus chaudes et humides. La topographie affecte aussi la composition des communautés microbiennes. Les sols des arêtes des polygones ont une plus grande proportion de

Betaproteobacteria, Deltaproteobacteria, et de Bacterioidetes, mais une plus faible proportion d’Acidobacteria comparé aux sols de l’intérieur des polygones. Les communautés de fixateurs d’azote, de méthanotrophes et de microorganismes dénitrifiants sont aussi affectées par la topographie, puisque chacun de ces groupes présente une structure unique. Les principales

xiii bactéries fixatrices d’azote et bactéries dénitrifiantes sont Rhizobiaceae, Nostocaceae,

Cyanothecaceae, Rhodobacteraceae, Burkholderiaceae, Chloroflexaceae, Azotobacteraceae et

Ectothiorhodospiraceae. De plus, ces organismes semblent être actifs dans les sols, puisque ces clades ont pu être détectés grâce à une analyse de métatranscriptomique. Les méthanotrophes actifs dans ces sols font partie du groupe USCα de microorganismes incultivés, soupçonné d’être des méthanotrophes ayant une haute affinité pour le méthane, démontré par la méthode de sonde à isotope stable (SIP), utilisant des sols démontrant une absorption de CH4. L’analyse SIP couplée au binning de métagénome a conduit à l’identification de plusieurs MAG (metagenome assembled genomes) de qualité moyenne à supérieure. Nous avons identifié un MAG Alphaproteobacterial qui contient plusieurs gènes associés au cycle du méthane, incluant une monooxygénase soluble de méthane (mmoX) et des gènes impliqués dans le cycle de la serine, pour assimiler le formaldéhyde, caractéristique des méthanotrophes de Type II. Ce MAG contient aussi des gènes reliés à l’assimilation de l’ammoniac, la production de polymère et la détoxification du mercure.

En plus d’identifier des membres non cultivables de la communauté microbienne par binning de métagénome, le séquençage des membres cultivables a révélé la présence de gènes du cycle du carbone impliqués dans la fermentation, la fixation du CO2, la dénitrification, ainsi que dans le métabolisme des polysaccharides et le métabolisme des composés aromatiques. Ceci suggère que la communauté microbienne des terrains IWP est prête à dégrader les réserves de carbone disponible suite à la fonte du pergélisol.

En plus de la topographie qui affecte la structure des communautés microbiennes, certains microorganismes clés du terrain IWP semblent aussi avoir un impact, positif ou négatif, sur les autres espèces microbiennes de la communauté. Ces relations ont été déterminé grâce au développement d’un nouveau modèle hydride d’analyse de réseau de la communauté microbienne.

xiv

Dans l’ensemble, les membres des phyla Proteobacteria, Candidatus Rokubacteria, et

Actinobacteria ont tendance à avoir un impact positif, alors que des membres Verrucomicrobia et des Acidobacteria des ont un impact négatif sur les autres microorganismes. Ces résultats indiquent que les paramètres environnementaux abiotiques et les interactions biotiques modifient la structure de la communauté microbienne et ont possiblement un impact sur les flux de GES provenant des sols.

xv

Contributions to knowledge

1. This thesis includes the first comprehensive survey of nitrogen fixation and denitrification genes, N2O flux data, and metatranscriptome analysis of Arctic cryosols at an ice-wedge polygon terrain (IWP). This work is important for understanding N cycling in permafrost affected soils.

2. I determined that the topography of the IWP terrain affects the flux of N2O, CH4, and CO2 gasses and the associated microbial community found in these cryosols. This information is important for modeling future GHG fluxes and climatic change in the Arctic.

3. It contains the first in situ CH4 SIP DNA labeling results from Arctic cryosols that exhibit atmospheric methane uptake. Metagenomic sequencing of the labeled DNA lead to binning of potential active organisms involved with the atmospheric methane sink in Arctic cryosols. The results of this work contribute to our knowledge of the microorganisms involved in the CH4 biogeochemical cycle.

4. I developed a novel hybrid network analysis method to understand how microbial organisms interact within a community. This helped identify potential keystone microbial members and microbial species-species interactions at the IWP site. Most network analysis methods rely on cooccurrence patters of species and are not able to discriminate between true microbe interactions and shared niche spaces. The alternative method presented here aimed to identify microbial interactions independent of the abiotic factors.

5. This thesis also included whole genome sequencing results of 18 culturable isolates from the

IWP terrain to understand the potential types of C and N cycling metabolic pathways present at the ice-wedge polygon terrain.

xvi

List of Abbreviations

BLAST- Basic local alignment search tool IWP- Ice-wedge polygon PCR- Polymerase Chain Reaction qPCR- Quantitative PCR OTU- Operational taxonomic unit SIP- Stable Isotope Probing MAG- Metagenome Assembled Genome

xvii 1

Introduction and Literature Review: Microbial Life in Permafrost Soils

Portions of this literature review appear in:

Altshuler I., Goordial, J., G. and L. Whyte (2017). Microbial Life in Permafrost. Psychrophiles: From Biodiversity to Biotechnology, Springer: 153-179.

Contributions of authors: I.A. developed and wrote this review. J.G. contributed to writing the Antarctica section. L.G.W provided editing support.

Introduction

Permafrost represents a large and extremely challenging environment for microbial life, covering

27% of the terrestrial surface on Earth, in which microorganism must cope with multiple environmental stressors (Goordial et al. 2013). Freezing temperatures, low kinetic energy, and low water and carbon availability limit microbial growth (Nikrad et al. 2016). Despite this, permafrost harbours a diverse microbial community that is viable and metabolically active. Furthermore, there has been a surge of interest into the diversity and activity of permafrost microorganisms as the permafrost thaws due to anthropogenic climate change.

Permafrost microorganisms are generally characterized as cryophiles, organisms that can sustain growth and reproduction at low temperatures ranging from -17 °C to +10 °C (D'amico et al. 2006;

2014). In addition to permanently subfreezing temperatures, any microbial life within permafrost must be able to survive the often oligotrophic conditions, background radiation on geological timescales, and limited liquid water activity; any liquid water present in permafrost is thought to exist in saline brine veins or in special saline niche environments such as cryopegs (Gilichinsky et

2 al. 2003). As such, any organisms which are able to survive these conditions are often polyextremophilic. While many microorganisms are either non-viable or dormant under these harsh conditions, there is clear evidence that globally, diverse and abundant microbial communities within permafrost can be viable and active in situ. In addition to in situ measurements indicative of microbial growth, psychrophilic organisms have been successfully isolated from permafrost environments, with sub-zero growth observed in the laboratory. This Chapter discusses the diversity of microorganisms found in permafrost globally, as well as evidence for viability and activity within the permafrost environment. Finally, I discuss the issue of climate change and how microbial communities in permafrost are expected to respond, with an emphasis on methane dynamics based on research to date. This final section set up my thesis which aims to understand the current active microbial community in Arctic cryosols and their association with the greenhouse gas emissions in these soils.

Permafrost environments

Permafrost, defined as ground material that is at ≤0 °C for two or more consecutive years (van

Everdingen 2005), is found primarily in the Arctic, sub-Arctic, and Antarctic regions, as well as in alpine regions (for example in the Qinghai-Tibet Plateau, South America, and Sweden)

(Bockheim and Munroe 2014). Perennially frozen permafrost is most often overlaid by a seasonally thawed active layer, the depth of which is dependent on air temperatures, moisture content, vegetation, and snow cover (Tarnocai 1980). Permafrost can be hundreds of meters thick

(for example over 500 m in Siberia), while the active layers can range between a few centimeters to several meters in depth (Tarnocai 1980). A transition zone exists between permafrost and the active layer, which acts as a temperature buffer and fluctuates between being seasonally frozen

3 and perennially frozen over decadal time scales (Shur et al. 2005). Permafrost is a heterogeneous environment due to frost heave, front sorting, and cryoturbation, processes which disrupt and mix different soil horizons, sometimes transferring organic carbon from surface to deeper layers and often creating patterned ground (van Everdingen 2005; Hugelius and Kuhry 2009).

Permafrost soils can also contain different geomorphological features such as ice-wedges, taliks, cryopegs, massive ground ice, frost boils, thermokarst lakes, organic accumulations, and broken soil horizons, each providing a unique habitat for microbial growth (Figure 1.1). Ice-wedge polygons are formed when the frozen ground contracts during the cold winter months and cracks the surface, dividing it into polygonal blocks (Kerfoot 1972). During the spring, these cracks are filled with snow melt water, which freezes and forms ice-wedges. As the temperature rises, the ground between the cracks expands and elevates, forming the polygon center (Kerfoot 1972). The ice-wedges are overlaid with active layer soil, creating the polygon trough (Shur et al. 2005). Over subsequent years of freeze and thaw cycles, the ice-wedges grow, resulting in high centered polygon surrounded by lower troughs. Taliks are unfrozen masses of ground soil found within the permafrost; often they are located under lakes due to the water’s ability to transfer heat (Shur and

Jorgenson 2007). Cryopegs are supercooled groundwater brine lenses that remain liquid at below

0 °C due to their high salt content (Gilichinsky et al. 2003). Along with a very steep temperature gradient, soluble nutrients, such as nitrogen, phosphorus, calcium, magnesium, and potassium, can also potentially form a gradient in the soil. As the nutrients are solubilized in meltwater, they move along the thermal gradient, enriching the top frozen permafrost layer (Kokelj and Burn 2005;

Tarnocai 2009). This results in the upper layer of permafrost becoming a sink of soluble materials

4 and nutrients during permafrost formation, and can then conversely become a source of these nutrients during permafrost degradation (Kokelj and Burn 2005).

Arctic permafrost overall is a heterogeneous environment with both mineral and organic soils. The polar deserts of High Arctic permafrost tend to harbour mineral permafrost, without an organic layer over the mineral horizons (Steven et al. 2006; Shur and Jorgenson 2007; Tarnocai 2009).

However, the Arctic does contain large stores of organic carbon frozen in permafrost. Pleistocene

Yedoma permafrost deposits in the North East Siberia are rich in organic carbon and were formed by windblown dust and sediment deposits that were subsequently frozen during the glacial age

(Zimov et al. 2006; Tarnocai 2009; Vonk et al. 2013). These deposits contain large ice-wedges and well preserved organic material, which constitutes roughly one third of organic carbon stored in permafrost globally (Vonk et al. 2013). Due to the cold climate during the formation of the deposits, the organic carbon is well preserved and readily available for biological degradation

(Schuur et al. 2008; Vonk et al. 2013). A prevalent feature of the Yedoma permafrost deposits is the high water content present in the form of ice veins and large ice-wedges that account for ~50

% of the soil content (Vonk et al. 2013). Due to the high ice content, the Yedoma permafrost deposits are highly susceptible to permafrost degradation due to climate change (Vonk et al. 2013).

Permafrost peatlands, primarily in the southern Arctic and sub-Arctic regions also hold large reservoirs of soil organic carbon. The permafrost deposits in these areas initially developed as fens and bogs that contained woody plants, mosses and sages (Routh et al. 2014). These perennially frozen peatlands are vulnerable to climate warming in the Arctic and sub-Arctic regions, the melting of ice in the permafrost will lead to degradation of the peatlands and biodegradation of the currently sequestered carbon (Tarnocai 2006).

5

Antarctic permafrost is less studied that its Arctic counterpart; however, Antarctic permafrost represents a much more extreme environment in terms of the combined freezing temperatures, aridity, and oligotrophy. The Antarctic contains 37 % of the world’s permafrost (Bockheim and

Hall 2002), though compared to Arctic and alpine permafrost, relatively little is known about

Antarctic permafrost communities (Goordial and Whyte 2014). The majority of the Antarctic continent is snow and ice covered, with only 0.35 % of exposed ground on which permafrost occurs

(Campbell and Claridge 2009). The McMurdo Dry Valleys’ polar desert represents the largest ice- free area in the Antarctic, and has been the focus of most permafrost studies to date (Gilichinsky et al. 2007; Goordial et al. 2016). Trough-like depressions may be underlain by sand wedges instead of ice, though ice veins and ice lenses may also be found within sand wedge structures

(Bockheim et al. 2009). The McMurdo Dry Valleys receive very low annual precipitation and is the only known place on Earth where dry permafrost is found (defined as permafrost which contains less than <3 % water by mass) which forms from sublimation of ice-cemented permafrost over time (Bockheim et al. 2007). Ice-cemented permafrost is found primarily on the coastal areas and younger surfaces where the geography facilitates drainage and therefore results in wetter soils.

Dry permafrost is found overlaying ice-cemented permafrost and is found at higher elevations and in older inland arid areas (Campbell and Claridge 2006). In some higher elevation valleys, there is an absence of any active layer that rises above 0 °C seasonally (Marinova et al. 2013). The majority of the water in the permafrost is frozen and the possibility for brine veins or thin films of water depends on temperature, solute concentration, and distance from the coast. Coastal and lower elevation valleys receive higher salt influx and have higher soil chlorine concentrations, as well as have a larger number of thaw days throughout the year (Goordial et al. 2016). As a result, higher

6 elevation valleys, such as University Valley, do not have sufficient solute concentrations or temperatures to form thin films of water for more than a few hours a year (Goordial et al. 2016).

In addition to the poles, permafrost is also present at high altitudes. The largest amount of high altitude (alpine) permafrost is located in China (Ran et al. 2012) with the Tibetan plateau being the largest alpine permafrost region (Chen et al. 2016). Temperate mountain permafrost soils have well-drained coarse sediments, steep slopes that have higher spatial and geothermal variability, more variable snow distribution, lower influence of vegetation, and warmer mean annual temperatures compared to Arctic and Antarctic permafrost (Haeberli and Gruber 2009; Frey et al.

2016). Furthermore, incoming solar radiation is moderated by the slope and shading (Etzelmüller

2013). In more coastal areas, alpine permafrost tends to be located above the tree line; however in continental areas, forests may promote permafrost development (Etzelmüller 2013). Recently, high altitude permafrost has garnered more attention since, in the last decade, monitoring of alpine permafrost has shown warming across the globe, but particularly in colder regions, with unknown consequences on geotechnical stability (Haeberli and Gruber 2009; Etzelmüller 2013).

Microbial diversity and abundance in permafrost

Despite being a hostile environment, permafrost does harbour a microbial community with 105 to

109 cells g-1 in Arctic permafrost, 1010 to 108 in alpine permafrost and a lower amount of 103 to

106 cells g-1 in Antarctic permafrost (Vishnivetskaya et al. 2006; Gilichinsky et al. 2007; Hansen et al. 2007; Steven et al. 2007; Blanco et al. 2012; Hu et al. 2015; Goordial et al. 2016). Across studies in alpine and polar regions, the dominant groups present in permafrost soils tend to be

Actinobacteria, Bacteroidetes, Proteobacteria (Alpha- and Beta-primarily but also Delta- and

7

Gamma-Proteobacteria), Firmicutes, Chloroflexi and Acidobacteria (Steven et al. 2008; Yergeau et al. 2010; Wilhelm et al. 2011; Deng et al. 2015; Stackhouse et al. 2015) (Table 1.1). The fermentative members of Chloroflexi and Bacterioidetes increase with soil depth indicating anaerobic carbon degradation in permafrost (Deng et al. 2015). Archaea and fungi are present, though 200-1000 times lower in abundance (Yergeau et al. 2010; Stackhouse et al. 2015; Frey et al. 2016). Abundant archaeal groups in permafrost are often related to halophilic archaea, part of

Euryarchaeota; likely due to the brine veins thought to host active microbial life within permafrost soils (Steven et al. 2007). However, Crenarchaeota appear to dominate acidic wetland permafrost

(Wilhelm et al. 2011).

Under frozen permafrost conditions, based on RNA data, the most transcriptionally active organisms appear to be the same as the dominant phyla: Proteobacteria, Firmicutes,

Acidobacteria, and Actinobacteria, as well as Euryarchaeota and ascomycetous fungi (Coolen and

Orsi 2015) (Table 1.1). Furthermore, permafrost contains active sulfate and Iron(III) reducers, nitrifying bacteria, methanotrophs, and methanogens (Yergeau et al. 2010; Stackhouse et al. 2015).

This is reflected in the presence of genes, transcripts, and proteins involved in sulphate reduction,

Iron(III) reduction, nitrogen cycle, methanogenesis, and methane oxidation found in the permafrost (Yergeau et al. 2010; Mackelprang et al. 2011; Hultman et al. 2015b). Functional groups of bacteria in high altitude permafrost also include ammonia-oxidizing bacteria and archaea, methane-oxidizing bacteria, nitrifying bacteria, nitrogen-fixing rhizobial symbiont bacteria, sulfur- and sulfate-reducing bacteria and thiosulfate-oxidizing bacteria (Zhang et al.

2007; Zhang et al. 2009; Wu et al. 2012; Yun et al. 2014; Hu et al. 2016). A unique study of the northern slope of Mount Everest showed an interesting comparison of ammonia oxidizers above and below 5800 m a.s.l.: the soil above this threshold is permafrost and is characterized by

8 increasingly colder temperatures, stronger radiation, lower oxygen concentration, and lower nutrients with increase in altitude (Zhang et al. 2009). The increase in altitude also decreased the abundance of both archaeal and bacterial ammonia oxidizers; however, at lower elevation, the soils were dominated by archaeal ammonia oxidizers, but were replaced by bacterial ammonia oxidisers in the higher altitude permafrost soils (Zhang et al. 2009). Overall, caution should be used when taking the results of higher altitude permafrost microbial abundance and function studies at face value, since authors in these papers often confusingly refer to active layer soils as permafrost.

Overall, permafrost contains genes involved in response to stressors, pathogenicity, toxicity, degradation of carbon compounds, methanogenesis, methane oxidation, denitrification, nitrogen fixation, ammonia assimilation, and sulphate reduction (Yergeau et al. 2010; Hultman et al.

2015b), and have shown presence of proteins involved in chemotaxis and motility (Hultman et al.

2015b). This result is also supported by the isolation of motile permafrost microbes from brine veins containing flagella (Shcherbakova et al. 2005). Compared to the overlaying active layer, the permafrost contains lower abundance of transporter proteins and transcripts, but higher presence of cold-shock proteins and other stress response genes (Yergeau et al. 2010; Mackelprang et al.

2011; Hultman et al. 2015b). Microbiota in mineral permafrost horizons have lower carbon degradation capacity compared to organisms from organic active layers; this can be due to lower taxonomic diversity in oligotrophic conditions found in the permafrost (Ernakovich and

Wallenstein 2015; Hultman et al. 2015b). However, the permafrost environment does contain genes for degradation of sugar alcohols, aminosugars, mono-, di- and oligosaccharides, starch, lignocellulose, chitin, trehalose, and cellulose (Yergeau et al. 2010; Hultman et al. 2015b).

Furthermore, Actinobacteria is often a dominant bacterial phyla present in permafrost soils (Table

1.1). This phyla is known to contain members that are facultative anaerobic degraders of complex

9 soil organic matter and are adapted to low carbon availability. In addition, permafrost soils have demonstrated potential enzymatic activities of hydrolytic (cellobiohydrolase, endochitinase, N- acetylglucosaminidase, and leucine aminopeptidase) and oxidative (phenoloxidase and peroxidase) enzymes (Gittel et al. 2014a; Gittel et al. 2014b).

Compared to permafrost soils, ice-wedges and ground ice are relatively low in diversity and abundance of microorganisms (Steven et al. 2008), they reflect the community of the surrounding permafrost soils and include members of Proteobacteria, Actinobacteria, Acidobacteria,

Bacterioidetes, and Firmicutes (Steven et al. 2008; Wilhelm et al. 2012). These are habitable cryoenvironments which show evidence of in situ heterotrophic activity based on occluded gas measurements (Katayama et al. 2007; Lacelle et al. 2011; Wilhelm et al. 2012) and contain cultivable microbiota, despite low viable cell counts (Gilichinsky et al. 1995; Katayama et al. 2007;

Lacelle et al. 2011). Soil particles suspended in the ice-wedges and brine veins are thought to serve as refugia for microorganisms and protect the cells from ice crystals.

Specifically in the Antarctic, the microbial community consists of Proteobacteria

(Gammaproteobacteria), as well as members of Nitrospina, green non-sulfur bacteria and relatives, Fibrobacter, Acidobacterium, and the Flexibacter-Cytophaga- Bacteroides group and low levels of anaerobes including denitrifying bacteria, methanogens, and sulfate reducers

(Gilichinsky et al. 2007). Culturable isolates belonged to Alpha-, Beta- and Gamma-

Proteobacteria, Actinobacteria, Firmicutes, and Methylobacterium, as well as mycelial fungi and yeast groups (Gilichinsky et al. 2007). Interestingly, unicellular green algae including

Mychonastes sp. (Chlorellaceae), Chlorococcum sp. (Chlorococcaceae), and Chlorella sp.

(Chlorellaceae) have also been isolated from Antarctic permafrost (Gilichinsky et al. 2007). The

McMurdo Dry Valleys’ permafrost ranges from cold and dry in the coastal Taylor Valley (TV;

10 subxerous; 1–75mm precipitation; summer temperatures sometimes above 0 °C in the Valley) to extremely cold, extremely dry, and highly oligotrophic in high inland University Valley (UV; ultraxerous; average summer temperature -5 °C in the Valley; 0.01−0.05% total carbon, undetectable to 0.09% total nitrogen) (Tamppari et al. 2012; Goordial et al. 2016). The diversity of the microbiota is also reflected between the two valleys. The wetter TV permafrost contains a higher abundance of microbes compared to the drier and colder UV (Tamppari et al. 2012). For example, TV permafrost showed presence of both anaerobic and aerobic microbial groups, whereas in the UV permafrost the anaerobic microbial groups were undetected based on PLFA data, however, later sequencing data did show presence of anaerobes in UV permafrost as well; this is likely due to TV permafrost being ‘wetter’ and therefore potentially having anoxic conditions near the permafrost table (Tamppari et al. 2012; Goordial et al. 2016). Furthermore, the dominant microbial groups in TV are Proteobacteria, sulfate-reducing bacteria, anaerobic metal reducers, Acidobacteria, Gemmatimonadetes, and Firmicutes, whereas in UV the dominant groups are Gamma- and Beta-Proteobacteria, though Firmicutes, Actinobacteria (including

Actinomycetes), and Bacteroidetes were also present (Tamppari et al. 2012; Bakermans et al. 2014;

Goordial et al. 2016).

The permafrost microbial community in high elevation permafrost appears to be dominated by bacterial phyla of Proteobacteria and Actinobacteria, and in one case Patescibacteria superphyla

(Frey et al. 2016; Hu et al. 2016). Although, Actinobacteria, Proteobacteria, and Chloroflexi dominate high altitude wetland permafrost (Yun et al. 2014). Arachaeal diversity is dominated by

Thaumarchaeota (Hu et al. 2016). Culturable isolates in high altitude permafrost belonging to

Arthrobacter, Pseudomonas, Alpha-, Beta- and Gamma-Proteobacteria, Firmicutes, and CFB (Bai et al. 2006; Zhang et al. 2007; Hu et al. 2015). Interestingly, the majority of the phylotypes of

11 bacterial and archaeal origin were less than 97 % similar to previously isolated strains (Hu et al.

2016). Frey et al. (2016) conducted an elegant study on alpine permafrost and found that the permafrost compared to the nearby non-permafrost soils was highly enriched in uncultured bacteria. The uncultured members were part of candidate phyla OD1 (proposed Parcubacteria),

TM7 (Saccharibacteria), GN02 (Gracilibacteria), OP11 (Microgenomates), OP5 (Caldiserica),

SR1, MVP-21, WS5, and Kazan-3B-28. The candidate phyla OD1, TM7, GN02, and OP11 are part of a proposed superphylum Patescibacteria (Frey et al. 2016). Members of Patescibacteria have very limited biosynthesis abilities of key macronutrients and are hypothesized to lead a semi parasitic or symbiotic lifestyle by attaching to the surface of other cells. These bacteria can be characterized by their low C-G content, small genome size, and are found in anoxic environments

(Frey et al. 2016). Such features may embed a selective advantage to this superphylum in the harsh, oxygen-limited permafrost environment.

Are they livin’ or just chillin’?

It may be intuitive to imagine that the microbes in the harsh frozen permafrost environments are in a state of dormancy and stasis. However, microbes in permafrost are biologically active. The active layer of the permafrost does boast higher microbial activity and microbial diversity; however, despite the harsh permafrost conditions, active microbes are still found within the permafrost (Yergeau et al. 2010; Coolen and Orsi 2015; Mackelprang et al. 2016). There are several complimentary lines of evidence for this: sub-zero respiration under aerobic and anaerobic conditions measured, isotope incorporation and mineralization under frozen conditions in

12 permafrost soils, isolation of permafrost microbes capable of sub-zero growth, and evidence from

RNA based studies.

1.4.1 Soil respiration and laboratory incorporation studies

Respiration and subsequent release of gases has been measured in frozen soils in situ and in laboratory studies. Some of the early evidence that microbes are living and are active in permafrost soils came from measurements of CO2 and CH4 release to the atmosphere from wintertime frozen tundra and peat bog soils (Fahnestock et al. 1999; Panikov and Dedysh 2000; Elberling and Brandt

2003). However, these in situ soil respiration studies should be interpreted with caution since the gas emissions could be due to releases of trapped gas. Laboratory experiments using permafrost soils have also demonstrated respiration at freezing temperatures as low as -2 °C (Michaelson and

Ping 2003), -4 °C (Larsen et al. 2002), -16 °C (Panikov and Dedysh 2000), -18 °C (Elberling and

Brandt 2003), -39 °C (controversial) (Panikov et al. 2006), and in Antarctic soils down to -5 °C

(Bakermans et al. 2014). Overall, respiration in frozen soils is dependent on water availability, temperature, and carbon content of the soils (Panikov and Dedysh 2000; Michaelson and Ping

2003; Öquist et al. 2009).

Respiration and heterotrophic activity under sub-zero conditions on environmental permafrost samples has also been shown in studies using highly sensitive radiolabelling microbial activity assays. For example, respiration in permafrost soils, active layer soil, and ice-wedges has been demonstrated using C14 labeled glucose and acetate as substrates (Rivkina et al. 2000; Steven et al. 2008; Wilhelm et al. 2012). The briny habitat existing within cryopegs is also thought to host active microorganisms as activity down to -15 °C was demonstrated via uptake of C14 labeled

13 glucose (Gilichinsky et al. 2003). Massive ground ice on the other hand, was not found to host an actively mineralizing community (Steven et al. 2008). Similarly, the highly unique permafrost in the high elevation Dry Valleys do not appear to respire under frozen conditions in laboratory or in situ, as demonstrated in University Valley soils (Goordial et al. 2016). In this instance, the combination of cold temperature (mean annual temp −23 °C with no days above freezing), aridity

(<10 mm snow melt equivalent per year), and low salinity to facilitate brine veins within the permafrost (Goordial et al. 2016) all contribute to an inactive community. In contrast, the permafrost in the more coastal, and therefore wetter and warmer, Taylor Valley of McMurdo Dry

Valleys does show active respiration at sub-zero conditions (Bakermans et al. 2014). These incorporation studies highlight environmental samples and conditions which favour active microbial life – namely the presence of liquid water, facilitated through the presence of solutes or salt. While these studies show microbial activity under ambient permafrost conditions, they do not show active DNA replication and growth, and do not reveal which microbial community members are active in the sub-zero permafrost environment.

1.4.2 Stable isotope probing in permafrost soils

Stable isotope probing (SIP) is used to label the DNA of actively dividing microbes. In an elegantly designed study, Tuorto et al (2014) demonstrated microbial genome replication in permafrost soils in temperatures ranging from 0 °C to -20 °C using 13C acetate. The active community members that were able to perform DNA replication in frozen permafrost conditions were part of

Acidobacteria, Actinobacteria, Chloroflexi, Gemmatimonadetes and Proteobacteria phyla and were distantly subzero related to isolated psychrophilic strains that are able to grow at subzero temperatures (Tuorto et al. 2014). Firmicutes were not detected by SIP in this study, suggesting

14 that spore-forming members of this phyla are dormant and non-metabolically active in an ambient permafrost environment, and their near ubiquitous detection in molecular studies may be due to the increased longevity of spores. Some members of the permafrost community were actively growing across all temperatures, whereas some were limited to specific niche temperature ranges; for example, uncultured members of Actinobacteria and Proteobacteria were only able to synthesize DNA between -9 °C and -20 °C. This could be due to changes in solute concentrations and available water, permitting only the growth of microorganisms that are adapted to very specific niche conditions (Tuorto et al. 2014).

1.4.3 RNA/DNA ratios

Ratios of RNA/DNA can be used to infer metabolic activity of the community or a particular group of microbes. Microbial community members with higher RNA/DNA ratios are thought to be more metabolically active in the environment, as the active microbes would be synthesizing higher amounts of RNA per cell compared to inactive ones, while the DNA content would remain static regardless of activity (Eriksson et al. 2001; Blazewicz et al. 2013). Hultman et al. (2015) looked at the ratios of RNA transcripts in metatranscriptomes (MT) to the DNA in metagenomes (MG) of permafrost soils to determine that the most active groups of microbes in permafrost were

Proteobacteria, Acidobacteria, and Firmicutes, suggesting that these members of the community were more acclimated to life at subzero temperatures (Hultman et al. 2015b). Furthermore, the study showed that permafrost had a higher ration of methanogenesis and methane oxidation transcripts (RNA) to genes (DNA) than the active layer, suggesting that the permafrost had relatively higher ratio of active methanogens and methanotrophs compared to the active layer.

15

However, the active layer appeared to have a higher RNA/DNA ratio of Nif genes, suggesting that it harbours a more active community of nitrogen fixers (Hultman et al. 2015b). Schostag et al

(2015) also performed a DNA and RNA based analysis though only on the active layer of permafrost; their study spanned the winter and the summer season allowing the comparison of the same soils in frozen and thawed conditions. The copy number of rRNA genes and transcripts did not fluctuate between the two seasons, suggesting that a similar relative abundance of microbes continue to be active during the winter with the soil temperatures below -10 °C (Schostag et al.

2015).

1.4.4 Subzero growth of permafrost isolates

Further evidence that microbes in permafrost are active is the ability of isolated permafrost microorganisms to grow at below freezing temperatures. There have been many permafrost isolates, mainly belonging to the Firmicutes, Actinobacteria, Proteobacteria and Bacterioidetes phyla (Goordial et al. 2013; Jansson and Taş 2014). However, few are able to sustain subzero growth, and those that do have evolved key adaptive strategies. For example, Psychrobacter cryopegella is able to grow down to -10 °C and up to 28 °C, with maximum growth at 22 °C. It was isolated from saline cryopegs buried within 40,000 years old Siberian permafrost (Bakermans et al. 2003). The authors demonstrated that growth yield of the isolate peaked at 4 °C; at this temperature, the microbial cells needed the least amount of RNA and proteins to divide. At freezing temperatures, the isolate likely needed to produce more cold acclimation and cold shock proteins as well as initiate other cold adaptive changes; above this temperature, the isolate needed to produce more RNA and proteins due to higher turnover and degradation rate caused by higher temperatures (Bakermans and Nealson 2004). Other Siberian permafrost isolates include a Gram-

16 positive Exiguobacterium sibiricum and a Gram-negative Psychrobacter sp. 273-4. These isolates were able to grow at -2.5 °C, and are thought to cope with the subzero temperatures by lowering fatty acid saturation and chain length in their membranes, changing the composition of exopolysaccharides, and increasing their ice-nucleation activity which is thought to reduce damaging intracellular ice accumulation (Ponder et al. 2005). An obligate anaerobic, spore forming, bacterial isolate that was capable of subzero growth was isolated from a water brine from the Kolyma Lowland region; Clostridium algoriphilum grows down to -5 °C, with optimal growth at +5 °C (Shcherbakova et al. 2005). From Alaskan permafrost, Panikov and Sizova (2007) isolated bacterial and fungal members capable of subzero growth. The bacterial isolates Pseudomonas sp.

3-2005 and Arthrobacter sp. 9-2 grew at temperatures down to -17 °C; Polaromonas sp. strain hydrogenovorans grew at -1 °C. The fungal members Leucosporidium sp. MS-1, -2 and Geomyces sp. FMCC-1, -2, -3, -4 were able to grow down to -35 °C; Mrakia sp. MS-2 grew down to -12 °C.

However, in these isolates, growth between -18 °C and -35 °C was transient and ceased after three weeks, though normal growth dynamics were sustained in subzero growth above -18 °C (Panikov and Sizova 2007). An Antarctic Psychrobacter sp. PAMC 21119 isolate capable of subzero growth was isolated from permafrost soil on Barton Peninsula (Kim et al. 2012). The microbe is able to grow down to at least -5 °C, partially due to an increase in production of proteins involved in metabolite transport, proper protein folding, and membrane fluidity (Koh et al. 2016). Two other

Antarctic isolates (Rhodotorula and Rhodococcus spp.) are also able to grow at −10 °C and −5 °C, respectively, and were isolated from extremely dry University Valley Antarctic permafrost

(Goordial et al. 2016).

Another bacterial permafrost isolate capable of subzero growth of note is Planococcus halocryophilus Or1; it was isolated from high Arctic permafrost and is able to grow down to

17

-15 °C, and sustain low level of metabolic activity down to -25 °C (Mykytczuk et al. 2013). Cells grown under the colder temperatures counter-intuitively contained higher levels of saturated fatty acids over branched ones and developed a crust of dense nodular material (Mykytczuk et al. 2013).

The genome of the isolate showed adaptations to cold and osmotic stress including cold shock proteins, chaperones, genes involved in regulation, repair mechanisms, osmolyte uptake, and membrane alteration. Genome redundancy in P. halocryophilus would also suggest the presence of isozymes that may be adapted to specific temperatures (Mykytczuk et al. 2013). However, despite our ability to culture subzero permafrost isolates, it appears that the uncultivable native consortia of permafrost microorganisms are more adapted to subzero temperatures compared to individual permafrost members currently isolated, based on comparisons of growth yield and utilization of C14 labeled ethanol between permafrost isolates and permafrost soils (Panikov and

Sizova 2007).

1.4.5 DNA repair

Another piece of evidence that microorganisms are likely alive, active, and growing in the permafrost environment is their need and ability to repair DNA damage from background ionizing radiation. Natural background radiation can damage a cell’s DNA over long geological timescales, such as those encountered in permafrost (Johnson et al. 2007). Completely dormant and inactive bacteria frozen in permafrost would continue to accumulate DNA damage and eventually lose viability; thus any viable microbes recovered from an ancient permafrost environment would need to have been at least minimally metabolically active in order to repair DNA damage during their tenure in the subzero permafrost (Price and Sowers 2004). Unchecked DNA damage will cross

18 link the DNA and/or reduce it into 100 bp fragments within 100,000 to 1 million years in frozen conditions (Poinar et al. 1996; Hansen et al. 2006). To date, there have been several permafrost bacteria isolated from ancient permafrost; for example, Exiguobacterium sibiricum was isolated and cultured at subzero temperatures from a depth of 43.6 m from a 2-3 million years old Siberian permafrost (Ponder et al. 2005). Furthermore, experiments on Psychrobacter cryohalolentis K5 and P. arcticus 273-4 showed that the microbes are able to grow under -15 °C conditions while simultaneously exposed to ionizing radiation (Amato et al. 2010). These microbes were metabolically active, as demonstrated by [3H] thymidine incorporation, and showed that microbes were able to sustain enough metabolic activity to repair DNA damage in the permafrost environment (Amato et al. 2010). Indeed, RNA sequencing of permafrost under frozen and thawed conditions has also demonstrated a relative overexpression of genes involved in DNA repair mechanisms in frozen permafrost soils (Coolen and Orsi 2015). In addition, Johnson et al. (2007) demonstrated bacterial survival for at least half a million years in sealed permafrost. The authors amplified a long 4 Kbp DNA fragment from ancient permafrost samples (400,000-500,000 years).

The fragment was 20-fold longer than ancient DNA from dead plant/animal samples (max amplicon lengths of 100-500 bp) of a similar age, suggesting that an active DNA repair mechanism must have been present to yield such a large DNA fragment from the ancient permafrost (Johnson et al. 2007). Sequence diversity greatly decreased with permafrost age, suggesting select few microorganisms are able to sustain long term survival in permafrost (Johnson et al. 2007). In younger permafrost (5,000-30,000 years), endospore forming bacterial members were shown to accumulate DNA damage; in older permafrost samples (400,000-600,000 years), there was no presence of bacteria with capacity for dormancy; instead, members related to non-spore forming

Arthrobacter (Actinobacteria) were dominant. In addition, authors were able to show active

19 respiration in these older permafrost samples at ambient permafrost temperatures. Together, these results suggest that microbes in ancient permafrost can sustain viability by maintaining low levels of metabolic activity and DNA repair; bacterial members that are capable of this strategy may outperform bacterial members capable of dormancy in very old permafrost environments (Johnson et al. 2007).

Live microbes or ancient DNA

The advent of next generation sequencing technologies has greatly propelled the study of microbial diversity, giving us a wealth of information and greater insight into the microbial processes that take place in permafrost environments. However, one uncertainty of studying microbial life and ecology of permafrost through novel molecular means is whether the nucleic acids that are isolated from the permafrost represent the current active microbial community in the ground, DNA within cryopreserved cells, spores, or just extracellular DNA and RNA molecules (i.e. eDNA) adsorbed to soil particles (Willerslev et al. 2004a; Willerslev et al. 2004b; Pietramellara et al. 2009). Do nucleic acids extracted from permafrost represent a frozen snapshot of past life or does it represent current life adapted to a harsh environment? Dry conditions, low temperatures, and salinity promote the persistence of nucleic acids in the environment, with the rate of DNA degradation decreasing by a level of magnitude for every 10 C drop in temperature (Smith et al. 2001;

Willerslev et al. 2004a; Hebsgaard and Willerslev 2009). However, rates of nucleic acid degradation under different environmental factors are not well understood (Hebsgaard and

Willerslev 2009), and permafrost conditions are thought to be favourable for long term microbial and nucleic acid persistence (Johnson et al. 2007). DNA in the environment, resting cells, and

20 endospores degrade over time because of chemical hydrolysis and oxidation, eventually becoming non-viable and non-amplifiable due to lack of active DNA repair. However, even minimally metabolically active cells may retain a functioning DNA repair mechanism and persist in the permafrost environment over longer time periods (Price and Sowers 2004; Johnson et al. 2007).

DNA fragments (100-500 bp) may not persist in colder environments for more than 105 years

(Lindahl 1993; Briggs and Summons 2014).

Finally, contamination can never be ruled out, when working with such low quantities and quality of nucleic acids, as in the case of permafrost, the sensitivity of PCR to contaminants becomes problematic. In addition, when we are studying the current microbial community in the permafrost soils, extracellular eDNA from dead cells can obscure the microbial diversity recovered (Carini et al. 2016). On the other hand, when trying to use ancient nucleic acids as fossil molecules for paleodiversity studies and to reconstruct past ecosystems, there is risk of currently active microbes interfering with the results (Bellemain et al. 2013; Briggs and Summons 2014). Bellemain et al.

(2013) run into this problem in their fungal paleodiversity study of two Siberian Pleistocene aged permafrost samples. The authors used permafrost samples to infer past ecology and environment of the area and through metabarcoding detected presence of plant-associated fungal taxa and fungal insect pathogens; based on this, they concluded that these fungi were an active component of the

Pleistocene environments. However, psychrophilic and psychrotolerant fungal taxa were also detected in their molecular analysis, suggesting presence of metabolically active taxa in the permafrost that were potentially interfering with an accurate paleo-reconstruction (Bellemain et al.

2013). So how is it possible to differentiate between old nucleic acids preserved in the environment and a potentially metabolically active microbial community?

21

1.5.1 Strategies for differentiating between old biomarkers and an active microbial

community

As DNA degrades over time, it is increasingly hard to amplify long stretches of preserved DNA.

Designing primers that flank large portions of 16S or other genes of interest is one way to ensure that only recent DNA is amplified (Johnson et al. 2007). RNA molecules on the other hand are more susceptible to degradations and do not persist in the environment as free molecules like DNA, due to their single stranded nature allowing direct cleavage of the phosphodiester bonds. Because of this, it is thought that molecular reconstruction of current microbial communities through RNA extractions and sequencing is more reliable. However, RNA extractions are notoriously difficult in permafrost soils; this is either due to low biomass in the samples or lack of actual RNA molecules altogether. The latter would imply lack of active cells. It is prudent to note that some studies have been able to extract or show evidence of long-term RNA preservation in permafrost, ice, snow, and other environments aged 50-140,000 years; for now, these studies are limited to viral and plant RNA, though this does not exclude the possibility of microbial and fungal RNA preservation (Guy 2014). Contamination may be an even bigger hurdle in permafrost RNA studies compared to DNA ones, due to ubiquitous prevalence of RNases. Another option is the propidium monoazide (PMA) treatment, which is able to differentiate between live cells’ DNA and eDNA/DNA within dead cells. PMA binds to DNA and inhibits the PCR reaction; however, it is not able to penetrate intact membranes of live cells (Bae and Wuertz 2009). Therefore, permafrost studies that aim to describe the current viable microbial community can use PMA treatment on their samples (Yergeau et al. 2010). However, these studies are limited to amplicon sequencing or any design that requires a PCR amplification step in the protocol. The advantage of this is that differences between the 'live' and total DNA can be used to elucidate the portion of DNA that is

22 environmental and preserved in the permafrost. Though ancient DNA would provide us with the most comprehensive phylogenetic information of past microbial communities, other biomolecules can be more persistent in the environment and may serve better for paleodiversity studies (Briggs and Summons 2014).

Warming climate and permafrost

Permafrost contains large amounts of frozen ancient carbon stores, in the range of 25-50 % of the total soil organic carbon (Tarnocai 2009). These pools are currently mostly inaccessible to microbial metabolism (Mackelprang et al. 2016). However, as the climate warms due to anthropogenic climate change, these pools of frozen carbon thawing are becoming available for heterotrophic microbial decomposition (Schuur et al. 2015). As the carbon pools are degraded, the microbes release greenhouse gases (GHG; CO2, CH4, and N2O) into the atmosphere which can potentially increase the rate of climate change via a positive GHG feedback loop (Marushchak et al. 2011; Graham et al. 2012). Auxiliary effects of climate change such as wildfires further increase permafrost degradation and active layer deepening (Taş et al. 2014) (Figure 1.1). Active layer detachment and permafrost collapse due to thawing further expose formerly buried permafrost and also increase microbial activity and degradation of previously unavailable soil organic matter

(Pautler et al. 2010).

In laboratory warming experiments, permafrost thaw induced an increased CO2 production in both

Arctic and high altitude permafrost soils (Stackhouse et al. 2015; Mu et al. 2016). While carbon emissions from permafrost at subzero temperatures are present, rates of CH4 and CO2 emissions generally significantly increase with permafrost thaw under aerobic and anaerobic conditions

23

(Song et al. 2014). Furthermore, N2O emissions have been reported in laboratory permafrost core melting experiments (Elberling et al. 2010). Interestingly, the initial permafrost thaw released minimal N2O; however, a cycle of drying and rewetting with the meltwater induced high rates of

N2O production in wetland permafrost (Elberling et al. 2010). Understanding the extent of ancient carbon degradation and resulting GHG emissions from permafrost is challenging due to different factors that affect the thaw of permafrost and the heterogeneity of permafrost environments, such as vegetation cover, latitude, soil composition, hydrology and geology of the area, precipitation, types of organic carbon stored, C:N ratios, and permafrost depth (Chen et al. 2016). Indeed, deeper permafrost soils upon thaw may release less CO2 due to these older soils containing higher proportions of recalcitrant carbon and lower enzyme (amylase and cellulose) concentrations and activity (Song et al. 2014). Thus, characterizing the current microbial community and activity in permafrost and understanding how that will shift with permafrost thaw and deepening of the active layer is one of the key current questions of environmental microbiology and the study of permafrost life. Indeed, studies that not only consider the current microbial biodiversity of permafrost, but that directly explore the changes associated with permafrost thaw by looking at environments that are currently undergoing this transition or by directly thawing permafrost samples in the laboratory, are starting to provide us with real insight into what the microbial community, its metabolic potential, and its effects on further GHG production may be, once permafrost thaws.

Coolen et al. (2015) looked at the transcriptional response of a microbial community to permafrost thaw in a moist acidic Arctic tundra location. Thawing permafrost conditions stimulated the growth of Firmicutes, Bacteroidetes, Euryarchaeota, Chloroflexi, Crenarchaeota, and ascomycetous fungi (Coolen and Orsi 2015). However, other studies reported that permafrost melt

24 increased abundance of Actinobacteria upon thawing short term (Mackelprang et al. 2011) and long term (Deng et al. 2015). In mineral permafrost horizons, thaw increased the abundance of

Actinomycetales that degrade complex recalcitrant carbon sources, as well as Chitinophaga (a chitinolytic genus) and Sphingomonadales (degraders of aromatics compounds) (Deng et al. 2015).

Gene expression upon thaw tended towards overexpression of genes involved in amino acid transport and metabolism, energy production, and DNA repair, replication and recombination

(Coolen and Orsi 2015). However, genes encoding for biofilm formation, virulence, and horizontal gene transfer were higher expressed under frozen conditions compared to thawed soil (Coolen and

Orsi 2015). Biofilm formation in the frozen soils could be restricted to the liquid brine vein microhabitat that surrounds the frozen permafrost soil particles (Gilichinsky et al. 2003; Coolen and Orsi 2015). As the permafrost thaws, genes that code for translation, ribosomal structure, and biogenesis are up-regulated, as well as genes involved in extracellular protein degradation, anaerobic metabolism, and the uptake, transport, and degradation of carbohydrates and are thus, likely contributing to permafrost soil organic carbon degradation. However, genes coding for hydrolases responsible for the cleavage of complex carbon polymers into C1 and C2 substrates were expressed in both frozen and thawed permafrost (Coolen and Orsi 2015).

Wildfires in upland Boreal Alaska stimulate near surface permafrost thaw; this thaw can shift the microbial community in the soils (Taş et al. 2014). Tas et al. (2014) conducted a novel study looking at the effects of fire and subsequent permafrost thaw on the microbial community and its metabolic potential for further GHG fluxes. The fire burned the majority of the top organic layer, thawed the permafrost to at least 1m in depth and lowered the C, N, DOC, and moisture content of the soils, but increased the pH and percentage of aromaticity. While fire and the subsequent thawing permafrost had a negative impact on the abundance of Verrumicrobia and Chloroflexi, it

25 had a positive impact on the abundance of candidate phyla AD5, which was one of the most abundance phyla in the deeper thawed and frozen permafrost soils. This is likely due to lowered

C, N, and moisture content and a higher pH, as the abundance of AD5 was also correlated to these soil parameters (Taş et al. 2014). It is likely that this candidate phyla, with no known culturable isolates, thrives in nutrient and moisture poor environments, as it was shown to increase in abundance with both depth and fire disturbance. Furthermore, the permafrost thawed soils had a different functional potential compared to intact permafrost. The thawed permafrost contained more genes for the hydrogenotrophic methanogenesis compared to acetoclassic methanogenesis in intact permafrost (Taş et al. 2014). However, regardless of the metabolic pathways, anaerobic incubations of permafrost soils demonstrated an overall reduction in CH4 production in burned soils; this is potentially due to a reduced moisture content and thus reduced microbial activity and anaerobic niches (Taş et al. 2014). Nitrogen cycling genes were overall more abundant in the thawed permafrost compared to intact permafrost, including genes for nitrate assimilation and denitrification. However, genes coding for nitrite and nitrous oxide reductases were lower in the thawed permafrost soils; this suggests a potential for incomplete denitrification and release of N2O upon thaw (Elberling et al. 2010; Taş et al. 2014).

Effects of Antarctic permafrost thaw on the microbial community and GHG emissions has not been studied. However, Buelow et al. (2016) have looked at the potential effects of permafrost thaw on McMurdo Dry Valley soils by simulating addition of water and organic matter to the arid

Antarctic top soils. Soils were dominated by Actinobacteria, Firmicutes, and Proteobacteria phyla

(Buelow et al. 2016). Water and organic matter supplements increased bacterial abundance but decreased diversity suggesting that the region may experience loss of endemic dry-adapted

26 oligotrophic taxa and replacement by generalist taxa with increasing temperatures due to climate warming (Buelow et al. 2016).

1.6.1 Methane dynamics in permafrost affected cryosols

Methane is ~20 times more potent that CO2 as a GHG. Currently, it is uncertain the portion of stored organic carbon that will be released as CH4 versus CO2 (Mackelprang et al. 2016). The flux of methane from the soils is governed by the equilibrium of methanotrophs and methanogens.

Methanogens are anaerobic archaea (part of Euryarchaeota) that are responsible for the biogenic production of CH4. Methanotrophs are characterized by their ability to oxidize CH4 and assimilate it as organic carbon (Hanson and Hanson 1996), belonging to the phyla Verrucomicrobia and

Proteobacteria and are further classified as either Type I methanotrophs belonging to the

Gammaproteobacteria or Type II methanotrophs belonging to the Alphaproteobacteria (Conrad

2007). Furthermore, anaerobic oxidation of CH4 is also possible via reverse methanogenesis by a group of ANME archaea related to methanogens (Knittel and Boetius 2009). Currently, wetland areas of the Arctic that are acting as methane sources and upland polar desert soils are acting as methane sinks (Christiansen et al. 2015; Lau et al. 2015). It is thought the emissions of CH4 in peat and wetlands are currently offset by CH4 uptake in upland dryer soils (Emmerton et al. 2014;

Christiansen et al. 2015). However, this may not hold up as the climate warms and permafrost degrades. As the permafrost thaws, it not only introduces organic carbon into the deepening active layer, but also causes land surface collapse, changes in soil hydrology, and the formation of thermokarst bogs and wetlands with anoxic conditions (Johansson et al. 2006; Graham et al. 2012).

These anoxic conditions promote the growth of methanogens and are favourable conditions for

27

CH4 production. So the question arises: what is the potential of permafrost to act as a CH4 sink or source once it thaws?

Permafrost does contain an active community of methanotrophs and methanogens (Yergeau et al.

2010; Mackelprang et al. 2011; Allan et al. 2014; Deng et al. 2015), though there are conflicting results in terms of diversity of these organisms and no consensus as to how their abundance will be affected with permafrost thaw. Overall, the permafrost methanogen community is dominated by Methanococcales and Methanomicrobiales in mineral permafrost and Methanosarcina,

Methanoregula, and Methanobacterium in methane containing permafrost (Allan et al. 2014;

Shcherbakova et al. 2016). In contrast, active layer soil microbial community is dominated by

Methanobacteriales, Methanosarcinales, Methanocellaceae, and Methanomicrobiaceae (Ganzert et al. 2007; Barbier et al. 2012). Methanogens’ diversity and abundance increases with soil depth in permafrost, likely due to the anaerobic conditions in the subsurface (Stackhouse et al. 2015;

Shcherbakova et al. 2016). Furthermore, methanogens increase in abundance and diversity with permafrost thaw and creation of wetlands; this may unfortunately increase the ratio of CH4 to CO2 flux from microbial decomposition of stored organic matter (Allan et al. 2014; McCalley et al.

2014). By looking at habitats at different stages of permafrost thaw, Hultman et al. (2015) observed higher abundance of methanogens in the thawed sites, possibly suggesting that even if the current permafrost soil do not harbour a large number of methanogens, they can be colonized by these organisms as the permafrost thaws. In addition to abundance, the thaw also stimulates methanogen activity, likely due to increase in temperature (Allan et al. 2014); this is reflected in a higher abundance of mcrA gene transcripts, a gene that catalyzes the final methanogenesis step (Coolen and Orsi 2015). Furthermore, upon thaw there is an increase in transcripts of the fhs gene which is part of the acetogenic fermentation pathway; this corresponded with an increase in the genes

28 involved in the acetoclassic methanogenesis by Methanosarcina barkeri (able to utilize C1 and C2 compounds including acetate) (Coolen and Orsi 2015). This supported a previous finding by

McCalley et al. (2014), who demonstrated that increased CH4 emissions due to peatland permafrost thaw are associated with a switch from a hydrogenotrophic to acetoclastic methanogenesis. This was also reflected in the shift of dominant novel methanogen taxa from Methanoflorens stordalenmirensis in partially thawed sites to members of Methanosaeta genus in fully thawed sites (McCalley et al. 2014; Mondav et al. 2014). However, this is in contrast to two studies that found higher abundance of hydrogenotrophic Methanobacteria members in the upper permafrost horizons compared to acetoclastic Methanomicrobia members in the lower permafrost horizons

(Barbier et al. 2012; Deng et al. 2015). This could have been also due to a shift from organic more aerobic upper soils to more anoxic mineral lower permafrost soils (Mondav et al. 2014; Deng et al. 2015). In addition to thawing permafrost, there is evidence that methanogenesis is able to occur in frozen permafrost as well (Rivkina et al. 2007; Yergeau et al. 2010). When methane is not able to diffuse through the frozen soil, permafrost ends up accumulating methane that may be released rapidly into the atmosphere as the permafrost thaws (Rivkina et al. 2007) (Mackelprang et al.

2011).

Whether permafrost thaw will have a significant impact on release of CH4 into the atmosphere is also dependent on the response of methanotrophs. Permafrost and active layer soils harbour Type

I, Type II, and Verrucomicrobia methanotrophs; however, these organisms are not ubiquitous across all permafrost soils (Yergeau et al. 2010; Christiansen et al. 2015; Stackhouse et al. 2015).

The relative abundance of methanotrophs decreases with depth, possibly due to an increasingly anoxic environment (Stackhouse et al. 2015). The dominant methanotroph organisms in permafrost soils are Methylococcaceae, Methylocystaceae, Methylocapsa, Methylocella, and

29

Methylacidiphilum (Yergeau et al. 2010; Deng et al. 2015; Stackhouse et al. 2015). Currently, organic rich permafrost and permafrost affected soils act as a methane source and mineral permafrost soils act as a methane sink, possibly due to the activity of high affinity methanotrophs

(Christiansen et al. 2015; Lau et al. 2015). Though many studies predict increases in CH4 releases with permafrost thaw (Mackelprang et al. 2016), Emmerton et al. (2014) predicts that the future changes in temperature and soil moisture content may increase CH4 consumption in mineral well- drained permafrost affected soils, but at the same time will have minimal effects on increasing

CH4 production (Emmerton et al. 2014).

Objectives of this thesis

The impact of permafrost thaw and subsequent emissions of GHGs from permafrost affected soils are poised to have a positive feedback loop of climatic change on permafrost thaw. Understanding the current active microbial community and the GHG flux of these permafrost soils is important in understanding future warming potential of the Arctic. The overarching goal of my thesis is to pinpoint the active microbial community members that contribute to GHG emissions at an ice- wedge polygon Arctic cryosol site.

Specifically, in Chapter 2, I aim to understand the CH4 and CO2 emission dynamics at the IWP terrain with focus on effects of topography and association with the microbial community of the soils. This is achieved through measuring the CH4 and CO2 gas flux at the IWP terrain over two consecutive years, metagenomic sequencing, and whole genome sequencing of bacterial species isolated from the terrain. In addition, to looking at the effects of topography and its associated abiotic variables on microbial species abundances, I try to identify microbe-microbe interactions

30 to pinpoint microbial keystone species and their effects on abundance of other species at the IWP site. For this the approach I developed a hybrid method that incorporated classical co-occurrence correlation-based network analysis and a general linear model (GLM) approach to determine which species positively or negatively affected the abundance of other species at the site. In the third chapter I explore microbial contribution to nitrogen cycling at the IWP terrain and the subsequent nitrous oxide (N2O) flux at the site. Nitrous oxide is the third most important GHG, with the warming potential of 300X of CO2. I achieved this through a combination of in situ N2O gas flux measurements, targeted nitrogen fixation and denitrification gene sequencing, and

Quantitative PCR. Furthermore, I use metatranscriptomics to determine the active members of the microbial community involved in the soil nitrogen biogeochemical cycle. Finally, in my fourth

Chapter, I aim to identify microbial members involved in methane metabolism and pinpoint active organism involved in the methane sink observed at the IWP site. My approach was to use in situ

13 Stable Isotope Probing (SIP) with labeled CH4 at the IWP site, in combination with methane monooxygenase (pmoA) amplicon sequencing, metagenomic sequencing, and genome binning.

This would allow labeling of the DNA of the active organisms involved in methane metabolism at the site. In Chapter 5, I aim to synthesize the results of my three data chapters in the context of global GHG biogeochemical cycling and future implication on the warming Arctic and permafrost thaw.

31

Figure 1.1 Permafrost schematic Schematic diagram of permafrost, including microbial and abiotic processes involved in permafrost thaw and GHG emissions.

32

Table 1.1. Table outlining the microbial members of permafrost community present, active, and cultivable in Arctic, Antarctic and in high altitude permafrost.

(sub)Arctic Antarctic High Altitude Permafrost microbial Presenta In situ Cultivable References Presenta In situ Cultivable References Presenta In situ Cultivable References member Activeb isolates activeb Isolates activeb isolates Proteobacteria + + + 1-15 + + 16-19 + + 20-24 Alpha + + + 1, 4, 9, 13, 14, 15 + + 16, 17 + + 20-23 Beta + + + 1, 4, 9, 12, 14, 15 + + 16, 17 + + 20-23 Gamma + + 3, 4, 9-15 + + 16, 17, 25 + + 20-23 Delta + 3,4,6,9,13-15 + 20, 21

Firmicutes + + + 1,3,5-9, + + 16-18 + 20-23 11,13,14,26,27 Acidobacteria + + 1, 2, 5, 7, 8, 14, + 19 + 21, 24 15 Actinobacteria + + + 1-3, 5, 7, 8, 9, + + 16, 17, 19 + 20-24 12-15, 28 Chloroflexi + + 6, 7, 13-15 + 19 + 21, 24 Gemmatimonadetes + 2, 4, 7, 14, 15 + 19 + 24

Verrucomicrobia + + 2, 7, 14, 15 + 19 + 21, 24

Bacteroidetes + + 1, 3-7, 9, 13-15 + 16, 17, 19 + + 22-24, 29 Planctomycetes + + 4, 7 + 19 + 21, 24 OP11 + 7 + 24 (Microgenomates) OP5 (Caldiserica) + 24 OD1 (Parcubacteria) + 24 TM7 + 9 + 24 (Saccharibacteria) GN02 + 24 (Gracilibacteria), AD3 candidate + 15 division

33

Spirochaetes + + 7 Chlorobi + + 7 Cyanobacteria + 21 Nitrospirae + 21, 24 Euryarchaeota + + + 1, 3, 4, 7, 8, 30- + 17 + 20, 33 32, 34 Crenarchaeota + 2, 4 + 33 Thaumarchaeota + 34 + 20 Bathyarchaeota + 34 Woesearchaeota + 34

aDNA based evidence; bRNA based evidence 1, (Yergeau et al. 2010); 2, (Wilhelm et al. 2011); 3, (Steven et al. 2008); 4, (Steven et al. 2007); 5, (Stackhouse et al. 2015); 6, (Deng et al. 2015); 7, (Hultman et al. 2015a); 8, (Coolen and Orsi 2015); 9, (Hansen et al. 2007); 10, (Bakermans et al. 2003); 11, (Ponder et al. 2005); 12, (Panikov and Sizova 2007); 13, (Gittel et al. 2014a); 14, (Gittel et al. 2014b); 15, (Taş et al. 2014); 16, (Gilichinsky et al. 2007); 17, (Goordial et al. 2016); 18, (Tamppari et al. 2012); 19, (Bakermans et al. 2014); 20, (Hu et al. 2016); 21, (Yun et al. 2014); 22, (Bai et al. 2006); 23, (Zhang et al. 2007); 24, (Frey et al. 2016); 25, (Kim et al. 2012); 26, (Shcherbakova et al. 2005); 27, (Mykytczuk et al. 2013); 28, (Finster et al. 2009); 29, (Zhao et al. 2011); 30, (Rivkina et al. 2007); 31, (Shcherbakova et al. 2011); 32, (Mackelprang et al. 2011); 33, (Wei et al. 2014); 34, (Shcherbakova et al. 2016)

34

Connecting Text:

Understanding the flux of CO2 and CH4 from permafrost affected soils and its relationship to the microbial community present in Arctic soils is important for future climate models. However, most of studies either focus on carbon flux from soils or the microbial community and very few have reported on mineral cryosols. Here, I aim to understand how the flux of the two most important GHG is related to the microbial community of the soils and the topography of the IWP terrain. Furthermore, I use a hybrid network analysis approach to understand how microbial members interact with each other independently of abiotic factors.

Portions of this chapter appear in:

Ianina Altshuler, Jeremie Hamel, Shaun Turney, Elisse Magnuson, Roger Levesque, Charles Greer, Lyle G. Whyte (2018). Species interactions and distinct microbial communities in high Arctic permafrost affected cryosols are associated with the CH4 and

CO2 gas fluxes. Environmental Microbiology (Accepted Pending Revisions).

Contributions of authors: I.A. wrote the manuscript, performed gas flux analysis, DNA extractions, phylogenetic analysis, and isolation of IWP culturable organisms. I.A., S.T., E.M. conceptualized and performed the network analysis. J.H and R.L. performed DNA extractions and whole genome sequencing of the 18 IWP isolates. All authors contributed to editing of the manuscript.

34

35

Species interactions and distinct microbial communities in high Arctic permafrost affected cryosols are associated with the CH4 and CO2 gas fluxes. Authors: Ianina Altshuler1, Jeremie Hamel2, Shaun Turney1, Elisse Magnuson1, Roger Levesque2, Charles Greer1,3, Lyle G. Whyte1

1 Department of Natural Resource Sciences, Faculty of Agricultural and Environmental Sciences, Macdonald Campus, McGill University, 21111 Lakeshore Rd, Ste Anne-de-Bellevue, QC, H9X 3V9, CANADA 2 Institut de Biologie Intégrative et des Systèmes (IBIS), Université Laval, Québec, Québec, CANADA 3 National Research Council of Canada, 6100 Royalmount Avenue, Montreal, QC, H4P 2R2 CANADA

Abstract

Microbial metabolism of the thawing organic carbon stores in permafrost, results in a positive feedback loop of greenhouse gas emissions. CO2 and CH4 fluxes and the associated microbial communities in Arctic cryosols are important in predicting future warming potential of the Arctic.

We demonstrate that topography had an impact on CH4 and CO2 flux at a high Arctic ice-wedge polygon terrain site, with higher CO2 emissions and lower CH4 uptake at troughs compared to polygon interior soils. The pmoA sequencing suggested that the USCα cluster of uncultured methanotrophs is likely responsible for the observed methane sink. Community profiling revealed distinct assemblages across the terrain at different depths. Deeper soils contained higher abundances of Verrucomicrobia and Gemmatimonadetes, while the polygon interior had higher

Acidobacteria and lower Betaproteobacteria and Deltaproteobacteria abundances. Genome sequencing of isolates from the terrain revealed the presence of carbon cycling genes including ones involved in serine and ribulose monophosphate pathways. A novel hybrid network analysis identified key members that had positive and negative impacts on other species. OTUs with

35

36 numerous positive interactions corresponded to Proteobacteria, Candidatus Rokubacteria, and

Actinobacteria phyla, while Verrucomicrobia and Acidobacteria members had negative impacts on other species. Results indicated that topography and microbial interactions impact community composition.

36

37

Introduction

Anthropogenic climate change is having a disproportionately higher impact on Arctic habitats resulting in an increased rate of permafrost thaw and deepening of the active layer (Shur and

Jorgenson 2007). This leads to releases of greenhouse gases (GHG) to the atmosphere due to heterotrophic microbial decomposition (Schuur et al., 2015) of the ancient carbon stored in permafrost (Tarnocai et al., 2009). These stores have accumulated over 1,000 Pg of carbon (C) in the northern circumpolar permafrost region during the last glacial cycle due to low temperatures and oxygen conditions (Hugelius et al., 2014). Releases of carbon dioxide (CO2) and methane

(CH4) gases from thawing permafrost are poised to have a large impact on further climate warming and subsequent permafrost thaw via a positive GHG emissions feedback loop (Marushchak et al.,

2011; Graham et al., 2012). Permafrost carbon represents roughly half of global belowground C stocks and may add up to 50 ppm CO2 into the atmosphere by 2100 (Hugelius et al., 2014). Indeed, laboratory and field warming experiments predict net positive C losses from Arctic permafrost affected soils (Natali et al., 2015, Stackhouse et al., 2015, Mauritz et al., 2017 and Raz‐Yaseef et al., 2017). These results mirror large scale studies demonstrating that permafrost affected soils are current net sources of C, despite increased atmospheric C uptake due to a longer growing season

(Belshe et al., 2013, Lara et al., 2015, Natali et al., 2015 and Kittler et al., 2017).

Majority of the stored carbon in permafrost soils across the Arctic is released into the atmosphere as CO2 (Drake et al., 2015 and Stackhouse et al., 2015). However, CH4 is the second most important GHG, having ~25 times the capacity of CO2 in trapping heat. Soils play an influential role in the dynamic of atmospheric CH4 and act as its sources or sinks of the gas. Wetland and peat bog soils provide an anaerobic habitat that can favor the growth of methanogens alongside aerobic methane oxidizers (methanotrophs) (Yavitt, et al., 1995). In these habitats, the flux of methane is

37

38 dependent on whether the methanogens or the methanotrophs are more active (Yavitt, et al., 1995 and Conrad 2007), with the methanotrophs potentially able to oxidize >90% of the methane produced (Wagner and Liebner 2009). Alternatively, well-drained, aerated upland soils can act a methane sinks. These soils are thought to harbour a phylogenetically distinct consortia of high- affinity methanotrophs capable of atmospheric methane oxidation (~1.8 ppm CH4) (Lau, et al.,

2015)(Kolb et al. 2005). These organisms are only phylogenetically identified by their pmoA methane oxidation gene and include both α and γ-Proteobacteria belonging to clades of USCα and

USCγ (Kolb et al. 2005; Martineau et al. 2014).

Arctic warming and wetting/drying will influence whether these soils act as CH4 sinks or sources

(Lara et al., 2015 and Natali et al., 2015). Warming will likely increase the amount of C lost to the atmosphere from permafrost soils; however, the ratio of CH4:CO2 released is currently uncertain and would likely be determined by the geomorphology, hydrology, vegetation, and microbiology of these soils (Lara et al., 2015, Natali et al., 2015 and Mackelprang et al., 2016). Permafrost thaw would affect soil topography and increase moisture content in areas with high ice content in the permafrost but thaw can also induce drying via drainage and changes in hydrology in parts of the

Arctic (Natali et al., 2015 and Kittler et al., 2017). Drainage disturbance and drying in Arctic tundra habitats increase CO2 emissions, but decrease CH4 emissions (Kittler et al., 2017), possibly due to anaerobic habitat loss that favors methanogesis. Therefore, a warmer climate with drier soils may decrease CH4 emissions and increase CH4 uptake in parts of the Arctic, likely due to aerated soils promoting methanotrophy (Emmerton et al., 2014 and Lara et al., 2015). However, this could also result in areas of saturated ground that could promote methanogenesis and act as CH4 sources

(Kittler et al., 2017).

38

39

In addition to these topological changes across Arctic soils that can impact future microbial metabolic processes, microbe abundances are also governed by microbial species interactions. Co- occurrence networks have often been used in microbial ecology to understand interactions of species interactions, especially since most interactions cannot be empirically observed. The rationale underlying microbial network analyses is that if two species co-occur in space and time together more often that by random chance, then they likely interact. Hence, instances of significant species co-occurrence have been used as evidence for positive interactions and instances of significant co-exclusion as evidence for negative interaction between species (Faust and Raes 2012). In microbial systems positive interactions can include cooperation (such as biofilm production), mutualism (such as syntrophy or cross-feeding) where microorganisms exchange metabolic products to the benefit of both organisms, and commensalism where organisms use by-products of other microbes (Faust and Raes 2012 and Seth and Taga 2014). On the other hand, negative interactions include predation (in cases of bacterial predators of other microorganisms such as Bdellovibrio and Lysobacter), production of metabolic waste products that negatively alter the environment or inhibit growth of other organisms, production of antibiotics that specifically or generally target other microbes, and competition where microorganisms competing for the same niche exclude each other (Faust and Raes 2012 and Pérez et al., 2016). These network analyses go beyond simple alpha and beta diversity indexes that are often associated with microbial community analysis and can help illuminate direct and indirect interactions between microbial taxa (Barberán et al., 2012) in human microbiomes (Friedman and

Alm 2012), oceans (Steele et al., 2011 and Lima-Mendez et al., 2015). This type of analysis has been used in assessing microbial interactions in soils (Barberán et al., 2012). However, caution

39

40 should be taken when interpreting correlation co-occurrence matrixes, as simple correlation between species can be due to niche overlap (Freilich et al., 2018).

Understanding current in situ microbial community composition and how it might impact the flux of CO2 and CH4 is important for modeling future warming potential of the Arctic (Graham et al.,

2012). The IWP terrain is comprised of mineral cryosol; this type of soil is common in the high

Arctic, comprising ~26% of ground cover and totaling 1,358,000 km2 of polar deserts (Walker et al., 2002). The effects of topography on the microbial community and on the GHG flux is not well understood. The topography of the IWP terrain allows us to study GHG emissions and the associated microbial community at differential soils moisture regimes, since the depressed trough soils are wetter than the well-drained polygon interior soil (Allan et al., 2014 and McCann et al.,

2016). As microorganisms are not isolated but exist in complex ecological interaction webs (Faust and Raes 2012), interactions of species also play an important role in species distribution and microbial community structure in addition to topography. However, teasing out these abiotic and biological effects on species diversity and distribution is challenging.

Here we measured CH4 and CO2 gas flux across an IWP mineral cryosol site in the Canadian high

Arctic during the 2015 and 2016 summers and 2016 winter to, for the first time, determine the effects of soil topography on the GHG flux at the site. Furthermore, we analyzed the diversity of the microbial community found in these soils via 16S rRNA gene sequencing to try to understand the effect of this landscape topography on the microbial community structure. We developed a novel hybrid network analysis approach that involves combining co-occurrence analyses with a generalized linear model (GLM) analysis to identify species interactions independent from topological factors that are driving species distribution. In addition, we performed targeted sequencing of a gene involved in methane oxidation, the pmoA, to help understand methane

40

41 cycling at the IWP terrain. Lastly, to gauge functional potential the warming Arctic soils and the diversity of the carbon cycling genes present in the environment we isolated and sequenced 18 bacterial genomes of viable bacterial strains from the IWP terrain. These genomes were also screened for genes associated with methane and methanol oxidation and pathways responsible for the resulting formaldehyde assimilation, namely the serine cycle and the ribulose monophosphate pathway (RuMP) (Hanson and Hanson 1996).

Materials and Methods

2.3.1 Site and sample collection

The study site is located near the McGill Arctic Research Station (MARS), at Expedition Fjord,

Nunavut on Axel Heiberg Island in the Canadian high Arctic (coordinates- 79º26'N, 90º46'W).

The active soil layer at the high centered ice-wedge polygon terrain site ranges from 60-73 cm in depth during the summer (Lau et al., 2015). Soil sampling for 16S and pmoA amplicon sequencing was done in the summer of 2016 from the active layer soil at the top 5 cm and bottom 25-30 cm belowground depths. The soil was collected in sterile 50 ml Falcon tubes and immediately frozen at -20⁰C.

2.3.2 Gas flux measurements

The in situ soil CH4 and CO2 flux was measured using the static chamber system (Collier et al.,

2014), at the trough and polygon interior soils of the IWP terrain. Measurements were taken over two days in summer (mid-July) of 2016, one day in summer of 2017, and two days in winter (early-

May) of 2016. Three replicates were performed for each soil type/day for up to 24 h. The gas samples were collected in 20 ml evacuated glass vials and brought back to the laboratory for 41

42 analysis on the greenhouse gas analyzer GC 450 (Bruker, Germany). Each sample was simultaneously injected onto a thermal conductivity detector (TCD) and flame ionization detector

(FID). The concentration of CH4 and CO2 vs. time was separately plotted and a linear regression of concentrations was used to calculate the slope (Collier et al., 2014). The slope of the regression was then used to calculate the CH2 and CO2 flux (Collier et al., 2014). A t-test was used to check significance between the mean fluxes of the soils.

2.3.3 DNA extraction and sequencing

DNA extractions were performed with Power Soil DNA extraction kit from Qiagen using 0.2 g of soil. Amplicon sequencing of the 16S rRNA and pmoA genes was done for biological triplicates for each soil type using Illumina Nextera V2 chemistry to generate libraries and pair-end sequenced on an Illumina MiSeq sequencer with 500 cycles. Whole genome sequencing of bacterial isolates from the IWP terrain was performed by Laval University. To obtain whole genome sequencing data, the bacterial strains were inoculated from single isolated colonies into

R2A broth (Teknova, Hollister, CA, USA) and grown at room temperature for periods ranging from 12 to 72 hours. The genomic DNA was extracted using the E-Z 96 Tissue DNA Kit (Omega

Bio-tek, Norcross GA, USA). Between 250 and 700 ng of DNA were fragmented via sonication using a Covaris M220 (Covaris, Woburn MA, USA) for 40 seconds at 18-22°C. The peak power was set at 50.0, duty factor at 20.0 and the number of burst at 200. The librairies were prepared using the NEBNext Ultra II DNA library prep kit for Illumina (New England Biolabs, Ipswich

MA, USA) following the manufacturer’s instructions. The libraries were barcoded with TruSeq

HT adapters (Illumina, SanDiego CA, USA) and sequenced via an Illumina MiSeq 300 bp paired- end run at the Plateforme d’Analyses Génomiques at the Institut de Biologie Intégrative et des

42

43

Systèmes (Laval University, Quebec, Canada). Sequencing data have been submitted to the

GenBank database under bioproject number BioProject PRJNA496474.

2.3.4 Isolation of IWP bacteria

We used a combination of in situ iChips, ex situ iChips (Goordial et al., 2017), and laboratory soil microcosm incubations along with standard isolation techniques to cultivate 18 microbial isolates from the IWP terrain. Method for how each isolate was cultivated is summarized in Table S1. The iChip isolation methodology (construction, dilutions, etc.) we followed Goordial et al 2017. In situ iChips were incubated in July 2016 for 2 weeks in field. Ex situ iChips were incubated for 6 months in the laboratory using IWP soils. Laboratory microcosm incubations were performed in the lab in

100ml stoppered vials with 1000 CH4 methane to encourage the growth of organisms involved in the methane biogeochemical cycle.

2.3.5 Bioinformatics

The raw reads from the 16S and pmoA targeted-amplicon sequencing were analyzed using the CLC

Genomics Workbench software with the Microbial Genomics Module plugin (Qiagen, Denmark).

Quality and chimera filtering was done using default settings, OTUs clustering and taxonomic assignment were done using Greengenes database for the 16S rRNA gene using 97% identity, the assignments were then corroborated using the SILVA database. For pmoA, OTU clustering at 90% identity was performed without taxonomic assignments. The closest uncultured and cultured relatives of pmoA OTU representatives were then determined using BLAST (Altschul et al., 1990 and Wheeler et al., 2007). PERMANOVA analysis was performed using R version 3 to assess whether soil depth (top 5 cm vs. bottom 25 cm) or soil type (polygon interior vs. trough soils) had a significant effect on microbial community composition (Dixon 2003). The Weighted UniFrac

43

44 model was used to preform principal component analysis (PCoA) of the functional pmoA gene and the 16S rRNA gene, distance matrix, and PCoA plots were generated using the Microbial

Genomics Module of the CLC Genomics Workbench version 11.

For phylogenetic reconstruction of the pmoA gene, alignment of the amplicon pmoA sequences and NCBI available pmoA sequence was performed using the ClustalW algorithm in

MEGA software version 7.0 (Kumar et al., 2016) with manual refinement of the alignments. The unrooted phylogenetic tree was constructed using the Maximum-likelihood statistical method based on the Tamura-Nei mode with 1000 Bootstrap replicates. The phylogenetic tree was then visualized using FigTree 1.4.3 (Rambaut 2016).

The whole genome sequencing data of the bacterial isolates was assembled and scaffolded using the A5 assembly pipeline (Tritt et al., 2012). To search for genes involved in methane metabolism, a custom database was built by fetching the amino acid sequences of all genes involved in methane metabolism according to the Gene Ontology Consortium from Uniprot

(Ashburner et al., 2000; Apweiler et al., 2004). These sequences were then submitted to a BLAST query against all proteins deposited in Genbank to expand the database (Altschul et al., 1990 and

Benson et al., 2017). All proteins with at least 40% identify with the query and an E-value lower than 1E-30 were added to the database. Two pathways involved in formaldehyde detoxification in methylotrophs were also added to the analysis, the serine cycle and the pentose monophosphate pathway (Kato et al., 2006 and Šmejkalová et al., 2010). For these databases, amino acid sequences of the involved enzymes from Methylobacterium exortens (serine cycle) or Methylomonas aminofaciens (RuMP pathway) along with those of proteins showing 50% identity were fetched from Uniprot. The genomes were annotated using PROKKA (Seemann 2014). Each gene was first submitted to a BLAST query against our three custom databases. The proteins producing 44

45 significant hits were again submitted to a BLAST query against all proteins deposited in NCBI to remove database bias. The results of the two queries were manually curated to check for matching annotations. These verification precautions were performed to account for the evolutionary pressure that permafrost affected cryosols can have on microorganisms (Feller 2007; Margesin and Miteva 2011). In addition, genomes were annotated using RAST (http://rast.nmpdr.org/) (Aziz et al., 2008).

2.3.6 Hybrid network analysis

The microbial network analysis was performed on the processed 16S amplicon data (see above) in

R version 3 (Team 2013) (Supplementary Information). Relative abundance of each OTU were used to correct for sequencing depth between samples. The OTUs that had less than 100 combined reads across all the samples and replicates were filtered out, leaving 506 OTUs for community network analysis. This filtering step was performed to remove low abundance OU

TUs and reduce complexity of the network to allow for understanding of the core microbial community at the IWP terrain (Barberán et al., 2012 and Weiss et al., 2016). We first calculated

Spearman's rank correlation coefficients between each of the OTU pairs. Significant co-occurrence and co-exclusion (positive and negative correlation, respectively) between OTUs were determined if correlation coefficient was ≥ 0.7 or ≤-0.7 with a p value of ≤0.01 (Barberán et al., 2012).

Alongside the Spearman correlations, we performed two general linear models (GLM)

(McCullagh and Nelder, 1989), with Poisson distributions on the 16S abundance data for each

OTU combination. Total abundances were corrected to account for sequencing depth between samples by including an offset of the log abundance. The first model predicted the abundance of

45

46 each OTUi based on the topography (polygon interior vs. trough), the soil depth (top 5 cm vs. bottom 25 cm), and the interaction between these two environmental niche variables.

GLM Model 1:

퐀퐛퐮퐧퐝퐚퐧퐜퐞 퐨퐟 퐎퐓퐔풊 ~ (퐓퐨퐩퐨퐥퐨퐠퐲) ∗ (퐃퐞퐩퐭퐡) + 퐨퐟퐟퐬퐞퐭 (퐥퐨퐠ퟏퟎ(퐒퐚퐦퐩퐥퐞 퐀퐛퐮퐧퐝퐚퐧퐜퐞))

The second model predicted the abundance of each individual OTUi based on these two environmental niche valuables and their interaction, same as the first model, but with the addition of the abundance of a second OTUj. The second OTUj abundance has an additive contribution to the variation in the abundance of OTUi in GLM Model 2 and does not consider the potential interaction between OTUj and the other variables on the abundance of OTUi, though this may be considered for future studies.

GLM Model 2:

퐀퐛퐮퐧퐝퐚퐧퐜퐞 퐨퐟 퐎퐓퐔풊 ~ (퐓퐨퐩퐨퐥퐨퐠퐲) ∗ (퐃퐞퐩퐭퐡) + 퐨퐟퐟퐬퐞퐭 (퐥퐨퐠ퟏퟎ(퐒퐚퐦퐩퐥퐞 퐀퐛퐮퐧퐝퐚퐧퐜퐞)) + 퐀퐛퐮퐧퐝퐚퐧퐜퐞 퐨퐟 퐎퐓퐔풋

This was repeated for all OTU combinations. For each OTU combination, we then used an

ANOVA model to perform a Chi-square test on the two models to determine if the addition of the second OTUj to the model significantly (p ≤ 0.001) reduced the residual sum of squares, meaning that we tested whether the model that included the second OTUj was significantly better at explaining the abundance of the first OTUi compared to the environmental variables only model.

We used a lower p-value compared to previous published network analysis studies, to be conservative in identifying microbial interactions (Barberán et al., 2012). The β (standardized co- efficient) was calculated to determine the strength of the relationship and if it was positive or negative. If the addition of the second OTUj was statistically significant, this implied that the

46

47 second OTUj influenced the first OTUi’s abundance independently from environmental variables that are associated with topography and soil depth. The advantage of this method over a simple correlation co-occurrence analysis is that we can determine if species have significant interactions with other species while controlling for environmental niche variables, allowing us to potentially detect true species interactions. Finally, to be conservative in our interpretation and limit false positives we filtered for those OTUi - OTUj interactions that were significant and had the same type of effect (positive vs. negative) in both the classical co-occurrence correlation analysis and in the GLM analysis (Figure S1). Another advantage of this hybrid network analysis over a correlation only approach is that the correlation co-occurrence analysis cannot distinguish between a OTUj - OTUi interaction versus a OTUi - OTUj interaction. The hybrid approach on the other hand, models a directional relationship, where OTUj can have a positive/negative impact on OTUj without the inverse being true.

Results

2.4.1 Gas flux

-2 -1 The summer (July) CO2 flux at trough soils (7.382 (±1.050) g CO2●m ●day ) was significantly

-2 -1 higher than at the polygon interior soils (2.517 (±0.671) g CO2●m ●day ; p=0.00037) (Figure

-2 -1 2.1). The flux of CH4 at the trough soils (-1.07 (± 0.41) mg CH4 ●m ●day ) was significantly

-2 -1 lower than in the polygon interior soils (-5.01 (± 1.10) mg CH4●m ●day ); p= 0.0047) (Figure

2.1). Trough and polygon interior soils did not exhibit a flux of CO2 or CH4 in the late winter (May) of 2016 when the soil was frozen (Figure 2.1).

47

48

2.4.2 IWP terrain microbial community diversity

High throughput 16S rRNA gene sequencing of the 5 cm and 25 cm polygon interior and trough soils revealed distinct microbial communities at the two different depths and between trough and polygon interior soils at the IWP terrain (Figure 2.3). PERMANOVA analysis demonstrated that depth (5 cm and 25 cm), topography (trough and polygon interior), and the interaction between the two variables had a significant effect on the microbial community composition (Table S2). The deeper soils (25 cm) of the polygon interior and the through had higher relative abundance of

Verrumicrobia and Gemmatimonadetes compared to their top soil (5 cm) counterparts (Figure 2.2).

Polygon interior soils at both depths had higher relative abundance of Chloracidobacteria

(Acidobacteria) compared to the trough soils. The top trough soil had a higher portion of

Betaproteobacteria and Deltaproteobactria compared to all the other soil types. Within these classes, the top trough soils also had a higher proportion of Desulforomonanadales,

Desulfobulbaceae (both Gammaproteobacteria), and Gallionellales (Betaproteobacteria) compared to all the other soil types. These bacterial taxa were mostly absent in both top and bottom polygon interior soils.

2.4.3 Hybrid network analysis of the microbial community

Out of 127,765 possible species interactions ((5062-506)/2, because self-interactions and causal direction are not considered), the correlation analysis identified 24,428 positive interactions and

21,520 negative interactions where the correlation coefficient was strong (≥ 0.7 or ≤-0.7) and significant (≤0.01) (Barberán et al. 2012). Out of 255,530 possible species interactions (5062 –

506, because causal direction is considered, and self-interactions are not), the GLM approach

48

49 identified 423 positive and 407 negative interactions, where an OTU had a significant effect on another OTU once environmental niche variables (topography and depth) were controlled for.

Once we filtered for interactions that were significant in both approaches, we were left with 296 positive and 92 negative interactions, which involved a total of 228 OTUs. Figure 3 depicts the network of negative and positive interactions separately; the full network containing both positive and negative interactions together is depicted in Figure S2.2. Several OTU nodes (Table 2.1) had numerous interactions, primarily OTUs corresponding to Proteobacteria (Burkholderiales), the

Candidate Phylum Rokubacteria (Uncultured Phylum, previously known as SPAM), and

Actinobacteria (Family- Actinomycetales) had positive impact on the abundance of multiple other species (Table 2.1). Several other OTU nodes, identified as Acidobacteria (Classes-

Chloracidobacteria, Solibacteres, and Acidobacteriia), and Verrucomicrobia (Order-

Chthoniobacteraceae) members, had numerous negative impacts on the abundance of other species (Table 2.1). We detected only 8 interactions, involving 12 species, in which the two species had mutually positive effects on each other; otherwise most species had a positive or a negative effect on the abundance of others without the reciprocal being true. These mutually positive interactions are summarized in Table S2.3. Overall, members of Burkholderiales were involved in five of the 8 mutually positive relationships.

2.4.4 PmoA diversity at the IWP terrain

Amplicon sequencing of the methane monooxygenase gene pmoA, recovered five OTUs, with

OTU1 dominating the community of potential methanotrophs (Figure 2.4) and was most closely related to uncultured methanotrophs from the receding glacier fore-field in Southeast Greenland

49

50

(Table 2; Barcena et al., 2011). Based on the phylogenetic reconstruction of the pmoA genes at the IWP terrain, all the OTUs grouped within the USCα cluster of suspected high-affinity methanotrophs (Figure 2.5). PCo analysis of the pmoA OTUs demonstrated that the three components accounted for 97% of the difference between the samples (Figure 2.4). The trough 5 cm samples grouped together, as did the polygon interior top 5 cm samples, while the deeper 25 cm soil samples grouped together irrespective of the location on the IWP terrain.

2.4.5 Bacterial isolates from the IWP terrain

The cultures isolates spanned 3 phyla (Actinobacteria, Proteobacteria, and Bacteroidetes) and eleven families based on their 16S sequences (Table 2.3). The GC content of the isolates ranged from 33.6-71.9% and the genome sizes ranged between 3,089,412 – 7,300,892 bp. Many of the culturable isolates contained genes for degradation of polysaccharides and aromatic compounds

(Table 2.3) as well as genes for photosynthesis, CO2 fixation, denitrification, ammonification and ammonia assimilation. Isolates also contained genes for sporulation and dormancy.

Several sequences isolated belonged to genera that were also present in our network analysis. Four sequenced isolates (E6.1, S11.1Y, S5.1, and S12) belonged to the genius Sphingomonas. We were also able to identify three OTUs in our network analysis that belonged to this genus. These OTUs had a positive impact on other microbial species, and one of the OTUs (#234) appeared to have a mutualistic relationship with a member of the Burkholderiales order (OTU #380) (Figure 2.3).

One sequenced isolate belonged to the Rhodanobacter genus, we were also able to identify one

OTU (#221) in our network analysis belonging to this genus. This OTU tended to have a positive

50

51 effect on other Proteobacteria species and appeared to also have a mutualistic relationship with a member of the Burkholderiales order (OTU #380) (Figure 2.3).

In combination, the culturable isolates contained many of the genes involved in the methane metabolism (Table S2.4). However, the genes coding for a methane monooxygenase, were absent in the cultured isolates. Of the 18 strains isolated, four possessed homologues of pyrroloquinoline quinone-dependent alcohol dehydrogenases (exaF) and 2 had a formaldehyde activating enzyme

(Fae). Additionally, 17 isolates had genes putatively encoding for enzymes involved in the serine cycle, with some genes present in multiple copies (Table S4). None of the strains contained the complete serine cycle. The GlyA (serine hydroxymethyl transferase), Hpr (hydroxypyruvate reductase) and Ppc (phosphoenolpyruvate carboxylase) genes were present in most of the strains, whereas none of the strains appeared to have the Eno (enolase) and Mtk (malata-CoA dehydrogenase) genes (Table S4). Two isolates (S9.2P and S5.26) both had two genes (rmpA and rmpB) indicative and essential to the RuMP pathway of assimilating formaldehyde on two consecutive ORFs, suggesting this pathway is present in these organisms.

Discussion

Historically, soils across the Arctic have acted as sinks of C, this is evidence by the presence of sequestered carbon stocks (Shur and Jorgenson 2007 and Belshe et al., 2013). Currently, Arctic soils vary in their contributions as sources or sinks of C. This is governed by many factors, including soils organic content, moisture, temperature, and vegetation cover (Christensen et al.,

2003 and Olivas et al., 2011). Despite increases in photosynthesis and primary productivity rates in the Arctic with recent increases in temperature, the increase in CO2 respiration rates have resulted in CO2 emissions exceeding CO2 uptake (Olivas et al., 2011 and Belshe et al., 2013).

51

52

Furthermore, the flux often varies across fine-scale microtopography of the Arctic (Olivas et al.,

2011). Soil CH4 flux also varies across the Arctic, ranging from wet, low centered polygonal

Siberian tundra that act as a source of CH4 over growing season (Sachs et al., 2010), to upland polar desert soils that act as CH4 sinks (Brummell et al., 2014 and Emmerton et al., 2014). Overall, due to extensive permafrost coverage, Arctic soils are modeled to become future CO2 and CH4 sources (Ringeval et al., 2010).

Based on this study, the soils at the IWP terrain are currently acting as sources of CO2 and sinks of CH4 during the summer when the active layer is thawed. The topography of the IWP terrain seems to impact the flux of these gases, with the wetter trough soils having higher CO2 emissions, but reduced CH4 uptake. This may be due to the aerated polygon interior environment being more conducive to methanotrophy, an aerobic process. Methanotrophy rates peak at ~20% soil moisture content, with higher water content likely creating more anaerobic zones and lower moisture content inhibiting microbial activity (Whalen and Reeburgh 1996). Our results are in accordance with a study by Lara et al (2015) that measured CH4 flux across the Alaska Barrow Peninsula landscape, the authors found positive CH4 fluxes tended to be highest in wetter anaerobic

-2 -1 environments (ranging 114.5-43.2 mgC-CH4●m ●day ) compared to drier environments (< 16.0

-2 -1 mgC-CH4●m ●day ) with the positive flux decreasing with increasing elevation (Lara et al.,

-2 -1 2015). Contrarily, the CO2 fluxes all ranged between 1.0-3.4 g C-CO2●m ●day ), with no obvious effect of the different geomorphic soil types on the flux, suggesting other abiotic drivers responsible for the CO2 flux pattern (Lara et al., 2015). Furthermore, CH4 uptake (-0.15 to -0.23

-2 mgC-CH4●m ●day) has been previously observed in ice-wedge polygonal terrain, though the effect of terrain topography was not studied (Allan et al., 2014).

52

53

Warming is thought to increase the volume of thawed and saturated soils and thus likely to support the methanogenic community and subsequent CH4 production (Natali et al., 2015), and would also likely suppress methanotrophic activity in soils that are current CH4 sinks. On the other hand, as the active layer deepens, surface waters may drain into subsurface soil layers overall reducing wetland areas (Avis et al., 2011). Therefore, in upland soils, permafrost thaw may increase drying in some areas but increase wetting in others due to ground collapse producing localized saturated soil areas interspersed with dryer areas (Jorgenson et al., 2001). This leads to increased heterogeneity of the Arctic landscape, resulting in potentially unpredicted GHG emissions scenarios (Jorgenson et al., 2001). Thus, dry soils that currently act as CH4 sinks may be reduced in their methane oxidation potential and switch to being sources of methane.

Differing GHG fluxes between the trough and the polygon interior soils in this study corresponded to differences in the microbial community found in these soils (Figure 2.2). Overall, dominant taxa in our results reflect those from other Arctic studies, which often tend to be Actinobacteria,

Bacteroidetes, Proteobacteria (Alpha- and Beta- primarily, as well as also Delta- and Gamma-

Proteobacteria), Firmicutes, Chloroflexi and Acidobacteria (Steven et al., 2008, Yergeau et al.,

2010, Wilhelm et al., 2011 and Stackhouse et al., 2015). We also detected sequences belonging to the candidate phylum AD3, with a higher proportion of these sequences in the deeper layers of the trough and polygon interior soils. The candidate phylum AD3 was also detected in active layer and permafrost soils of an upland Alaskan boreal forest, similar to our study the prevalence of this phylum increased with depth (Taş et al., 2014). Little is known about this uncultured candidate phylum, other than that it is often present in the active layers of permafrost affected Arctic, alpine, and Antarctic desert soils (Taş et al., 2014, Frey et al., 2016 and Ji et al., 2017). Verrumicrobia and Gemmatimonadetes were at a higher relative abundance at the IWP site than in other

53

54 permafrost affected soils (Gittel et al. 2014a; Taş et al. 2014), these phyla were also at a relatively higher abundance in the deeper soils. This is partially in contrast to the Gittel et al (2014) study that found Verrumicrobia sequences more prevalent in surface soils compared to deeper soils although Gemmatimonadetes did have the same pattern in Siberian tundra as we observed here, with higher prevalence of this clade in deeper soil compared to surface soils (Gittel et al., 2014a).

Interpreting network analysis results based on correlation co-occurrence matrixes is challenging, since correlation between species can be due to niche overlap or actual interaction between the species. A recent study by Freilich et al. (2018), assessed the efficacy of co-occurrence networks in identifying species relationships by comparing the inferred correlation network interactions to empirically based macroorganisms species interactions. Co-occurrence based network analyses were determined to be inadequate at detecting true species interactions, instead they were more likely to detect species with shared niche spaces and similar responses to environmental variables

(Freilich et al., 2018), thus detecting many false positive species-species interactions. Co- occurrence based network analyses of microbial communities alone are therefore also unlikely to separate true microbe-microbe interactions from external environmental variables. To resolve this weakness, our modified hybrid network analysis approach, which controlled for environmental variables, aimed to filter out majority of the interactions initially detected based only on correlation. This was achieved by combing the correlation analysis with a GLM method to control for topography and soil depth. Ideally, our hybrid network analysis revealed true interactions between microbes, separate from effects of shared niche space, although we have yet to empirically show this. Overall, our results suggest that, in addition to the heterogeneity of the IWP terrain, microbe-microbe interactions also affect species distribution and abundance. This suggests that microbe-microbe interactions should also be considered when trying to understand potential future

54

55

GHG emissions. Majority of the microbe-microbe interactions appear to be positive, meaning the presence of some species had positive effects on other species, though negative interactions were also detected (Figure 2.3; Table 2.1). Members of Burkholderiales, Actinomycetales and Candidate

Phylum Rokubacteria tended to have numerous positive effects on the abundance of other species, while Acidobacteria, and Verrucomicrobia had negative impact on other species. Little is known of the currently uncultured Candidate Phylum Rokubacteria, so it is interesting to see its high centrality in our hybrid network analysis. This candidate phylum is globally present in terrestrial environments, thought at low abundances (Figuerola et al., 2015). Recent single cell sequencing revealed that this phylum has high genomic heterogeneity among individual and a mixotrophic metabolism (Becraft et al., 2017). This candidate phylum also appears to have a large genome size in combination with small cell size, implying extensive DNA packaging or dormancy, which would be useful in permafrost affected soils such as a the IWP terrain (Becraft et al., 2017).

The network analysis detected that relatively few species seemed to be engaged in mutually positive relationships (ie. mutualism), with Burkholderiales members engaged in five of eight of these mutually positive interactions. Interestingly, we were not able to detect any microbial members that simultaneously had positive effects on some species, but negative effects on others.

This suggests that the microbes behave (e.g., consuming nutrients, generating by-products) in ways that are either widely beneficial or harmful to other species, rather than having specific effects on different species. Members of Firmicutes, while present in our samples and in other permafrost affected soils, did not overall seem to have an impact (positive or negative) on the abundance of other species, this may be due to members of this phyla being spore-forming and potentially dormant in these soils and therefore potentially not impacting the abundance of other species. A previous stable isotope probing based study that looked at active genome replication in Arctic soils

55

56 also suggested that spore forming members of Firmicutes are present, but not active in Arctic soils

(Tuorto et al., 2014). Overall, we believe this is an interesting approach that could be applied on a larger meta-scale analysis to help reveal key in situ microbe-microbe interactions and provide targets for potential microbial keystone species. A meta-analysis with enough samples using our hybrid approach could potentially be used to also reveal three-way interactions by adjusting the

GLM to include abundances of multiple species. Our study aimed to develop this approach on a smaller scale using sequencing data from the IWP terrain to help understand the microbial community, the variables that control the composition of this community, and how it is related to the variation in GHG emissions we observe at the IWP terrain.

Arctic soils, including mineral cryosols have previously shown atmospheric methane uptake. This has been attributed to an unculturable group of high affinity methanotrophs that can oxidize methane to methanol at atmospheric levels (~1.8 ppm CH4) (Martineau et al., 2014, Christiansen et al., 2015 and Lau et al., 2015). However, some studies suggest that conventional, low-affinity methanotrophs are able to perform atmospheric methane oxidation as well (Cai et al., 2016).

However, the only methanotrophs at the IWP cryosol cite that we could detect were those closely related, based on their pmoA sequences, to uncultured methanotrophs belonging to the USCα cluster of suspected high-affinity methanotrophs (Pratscher et al., 2018). This is similar to our previous studies that have shown these proposed high-affinity methanotrophs in Arctic soils

(Martineau et al., 2014; Lau et al., 2015). In the current study, we also demonstrate that despite the methanotrophic community across the IWP terrain being dominated by one OTU, there does appear to be some differences in the community composition based on topography and soil depth

(Figure 2.4). The drier polygon interior soils with higher methane uptake had relatively higher number of OTUs related to upland forest soils compared to wetter troughs that had lower rates of

56

57 methane uptake. Methanotrophs were not detected in our 16S sequencing results, this is likely due to the only methanotrophs at the site being related to uncultured high-affinity methanotrophs, with unknown 16S phylogeny.

Genomic analyses of the 18 IWP isolated strains provided insights into their identity and functional capacity. The IWP isolates contained genes for degradation of complex carbohydrates and aromatic compounds. This corresponds to studies that have found extracellular enzyme activities in mineral cryosols, including cellobiohydrolase and phenol oxidases (Gittel et al., 2014b and

Schnecker et al., 2014). In addition, Arctic mineral soil horizons contain relatively high levels polysaccharides and phenolic compounds compared to other organic carbon sources (Xu et al.,

2009). As the decomposition of soil organic material depends partially on microbial extracellular carbon acquiring enzymes (Schnecker et al., 2014), the microbial community at the IWP terrain in poised to take advantage of the thawing carbon stored in the permafrost. Not surprisingly these isolates also contained genes for sporulation and dormancy, suggesting that even if they are currently dormant in soils, they are likely to become active with warming of the soil and permafrost thaw (Wagner 2008).

In combination, the culturable isolates contained most of the genes involved in methanotrophy and formaldehyde assimilation, though none of the isolates contained complete known methane oxidation pathways (Table S2.4); methane monooxygenase genes (mmoX, pmoA), encoding for the initial step in the metabolism, were not detected in any of the cultured isolate genomes indicating that none of these isolates are capable of methanotrophic activity. Based on the complimentary analysis of pmoA amplicon sequencing and previous studies (Martineau et al.,

2014 and Lau et al., 2015), it is likely that microorganisms responsible for the first step of methane metabolism and for the methane sink observed in this IWP site are part of the USCα clade of high- 57

58 affinity uncultivable methanotrophs. Serine cycle and RuMP pathway are two ways that carbon gets assimilated into biomass following methane and methanol oxidation. The presence of multiple

GlyA gene copies indicates that these isolates are well adapted to one-carbon chemistry, as this enzyme is accepts variety substrates involved in multiple one-carbon chemistry pathways including the serine pathway of assimilating formaldehyde (Schirch and Szebenyi 2005 and Florio et al., 2011). The presence of two genes essential to the RuMP pathway on two consecutive ORFs is a good indication that S9.2P and S5.26 have the capacity to assimilate formaldehyde this way

(Sakai et al., 1999), whereas the rest of the isolates appear to potentially assimilate formaldehyde using the serine pathway. Thus, these isolates are likely involved in the overall sequestration of

CH4 from the atmosphere into soils.

Conclusions

Increased warming in the Arctic due to global warming will affect GHG fluxes from these cryosol environments. Understanding how the topography of the ecosystem effects both the microbial community and GHG flux is likely important for future climate models. Wetter trough soils demonstrated higher emissions of CO2 and lower uptake of CH4 when compared to the drier polygon interior soils. This difference in GHG flux also corresponded to differences in the microbial composition in the soils, suggesting that soil topography effects both GHG flux and the microbial community. Genomic analysis of the 18 IWP terrain isolates also showed that most bacteria from these cryosols possess genes enabling them to assimilate methane oxidation products and degrade thawing permafrost carbon stores. Furthermore, soil microbes within the active layer appear to be affected by both the abiotic factors in the environment, such as topography, and by biotic interactions with other microbes. Therefore, future studies that are focused on teasing out

58

59 which species influence the abundance of others, may be important in understanding how microbial communities will shift with the warming of the Arctic.

59

60

Figure 2.1. CO2 and CH4 flux at the IWP terrain.

CO2 and CH4 flux at the ice-wedge polygon site, during the 2016 and 2017 seasons. Blue markers denote flux at the trough soils. Orange markers denote the flux at polygon interior soils. Grey and black markers denote the flux at the trough and polygon interior soils in the Spring of 2016 when the ground was frozen. Error bars are based on SEM.

60

61

Figure 2.2. 16S community profiling of soils at the high Arctic IWP site.

(A) Microbial diversity and community composition by phylum based on 16S rRNA amplicon sequencing in the trough soils and polygon interior soils at 5 cm and 25 cm. (B) PCo analysis of the microbial community, orange circles denote polygon soils at 5 cm; yellow circles denote trough 5 cm soils; blue circles denote polygon soils at 25cm; green circles denote trough soils at 25 cm.

61

62

Figure 2.3. Hybrid network analysis of microorganism at the IWP site. Network analysis of the microbial community at the IWP terrain based on 16S sequencing. The networks are based on a combination of classical correlation-based network analysis and a GLM approach that tests if the interaction between species is significant once environmental variables are controlled for. Arrows depict the direction of the relationship. Green connections (A) depict positive relationships, red connections (B) depict negative relationships. The saturation of the connecting arrow corresponds to the relative strength of the relationship and is based on the β calculation. The combined full network is depicted in Figure S2.2.

62

63

Figure 2.4. pmoA community structure at the IWP terrain. Functional gene (pmoA) sequencing of soils at the ice wedge polygon site. (A) Diversity of the pmoA OTUs in the trough soils and polygon interior soils at 5 cm and 25 cm. (B) PCoA based on the pmoA gene of the microbial community, orange circles denote polygon soils at 5 cm; yellow circles denote trough 5 cm soils; blue circles denote polygon soils at 25cm; green circles denote trough soils at 25 cm.

63

64

Figure 2.5. pmoA phylogeny at the IWP terrain. Phylogenetic reconstruction of pmoA sequences at the ice-wedge polygon site. OTUs recovered in the analysis are denoted by the stars.

Table 2.1. Microbial community members with the highest number of positive and negative interactions.

Organism Type of Interaction Effect Affected Organism Candidate Phylum Rokubacteria Positive Chloroflexi (11) OTU 462 Acidobacteria (8) (46) Candidate Phylum AD3 (5) Proteobacteris (5) Gemmatimonadetes (5) Bacterioidetes (3) Planctomycetes (3) Verrucomicrobia (2) Candidate Phylum WPS-2 (1) Cyanobacteria (1) Crenarchaeota (1) Armatimonadetes (1)

64

65

Proteobacteria Positive Proteobacteria (14) OUT 62 Acidobacteria (7) (Order- Comamonadaceae) Planctomycetes (4) (36) Actinobacteria (3) Armatimonadetes (3) Bacterioidetes (3) Elusimicrobia (1) Verrucomicrobia (1) Acidobacteria Negative Proteobacteris (13) OUT 155 Chloroflexi (3) (Class- ChlorAcidobacteria) Actinobacteria (3) (24) Acidobacteria (2) Gemmatimonadetes (1) Elusimicrobia (1) Planctomycetes (1) Actinobacteria Positive Chloroflexi (4) OUT 34 Actinobacteria (4) (Order- Intrasporangiaceae) Candidate Phylum AD3 (3) (20) Gemmatimonadetes (2) Proteobacteria (2) Armatimonadetes (1)

Bacterioidetes (1)

Acidobacteria (1)

Candidate Phylum WPS-2 (1)

Candidate Phylum (1)

Actinobacteria Positive Proteobacteria (7) OUT 422 Acidobacteria (4) (Family- Actinomycetales) Chloroflexi (3) (20) Actinobacteria (2) Gemmatimonadetes (1) Planctomycetes (1) Verrucomicrobia (1) Elusimicrobia (1) Acidobacteria Negative Candidate Phylum AD3 (4) OUT 203 Actinobacteria (2) (Family- Solibacterales) Chloroflexi (2) (13) Bacterioidetes (1) Gemmatimonadetes (1) Planctomycetes (1) Proteobacteria (1) Candidate Phylum WPS-2 (1) Acidobacteria Negative Chloroflexi (3) OUT 231 Candidate Phylum AD3 (2)

65

66

(Order- Acidobacteriaceae) Actinobacteria (2) (11) Bacterioidetes (1) Proteobacteria (1)

Planctomycetes (1)

Candidate Phylum WPS-2 (1)

Verrucomicrobia Negative Proteobacteria (5)

OTU 278 Acidobacteria (2)

(Order- Chthoniobacteraceae) Verrucomicrobia (1)

(11) Chloroflexi (1)

Firmicutes(1)

Actinobacteria (1)

aOnly organisms with more that 20 positive connections or more that 10 negative connections are displayed. Numbers in brackets beside the “Organism” show the number of other species that organism is positively or negatively impacting. Numbers in brackets beside “Affected Organisms” display the number of organisms in that group that are being affected.

66

67

Table 2.2. pmoA OTUs and their closest cultured and environmental BLAST matches.

OTU Closest NCBI Accession % Location/ Reference Closest NCBI BLAST Accession % numbe BLAST match number match description cultured match number match r 1 Uncultured FN651815 99 Receding Glacier (Barcena et Methylocapsa FN433470 79 methanotrophic Forefield, al. 2011) aurea type strain bacterium clone Southeast KYGT D12-240_p5 Greenland 2 Uncultured KX534036 95 lichen-dominated (Belova et Methylocapsa KP715290 79 bacterium clone forested tundra, al. 2016) palsarum strain LFT89 near Nadym, Unpublishe NE2 Russia d 3 Uncultured FN651657 98 European beech (Degelman Methylocystis sp. EU275143 75 methanotrophic and Norway n et al. 0510-P-6 bacterium clone spruce soils 2010) G7-215_p1 4 Uncultured FN564624 99 European beech (Degelman Methanocapsa AJ278727 76 bacterium clone and Norway n et al. acidophila B2 of SW-F-31 spruce soils 2010) 5 Uncultured KF757100 93 Sanjiang Plain (Yun and Methylocapsa KP715290 73 methanotrophic wetland, Zhang palsarum strain bacterium clone northeastern 2013) NE2 SJ-mb661-CL7 China Unpublishe d

67

68

Table 2.3. Table of isolates from the ice-wedge polygon terrain.

N

SporilationDormancy &

3

Serine

MotylityChomotaxis &

O & O

Ammonia assimilation

Metabol Arom. Comp.

Closest 16S Match (%)

Formald.Assim:

Polygo

Mono

N

-

2

Glyoxylatecycl

Photo

O

Di

Dinitrification

NCBI/RAST ID

Fermentation

Ammonificat.

Closest Strain

Sulfur Assim.

- -

CO

-

saccharides saccharides

GC ContentGC

oligosacch.

-

16S genes

synthesis

2

Size (bp)

Isolate fixation

Family

RuMP

e

7,300 Pseudomonas 1 - - - + +(4) +(11) +(2) +(2) +(14) - + + in/ - + OWC3 58.5 318524 ,892 99.3 caspiana Pseudomonadaceae org 5,440 Hymenobacter 2 - + + + +(4) +(7) +(2) +(6) +(8) + + + org + + S9.2P 64.2 318525 ,343 98.2 nivis Hymenobacteraceae 6,938 Pseudomonas 1 - - - + +(5) +(11) +(2) +(2) +(16) + + + in/ + + OWC5 58.7 318526 ,776 99.5 mandelii Pseudomonadaceae org 4,284 Rhodanobacter 1 - - - + +(5) +(3) - +(1) +(7) + - + in + - S10 64.4 318531 ,014 98.7 glycinis Xanthomonadaceae 5,126 Sphingomonas 1 + + - + +(4) +(6) +(1) +(2) +(9) - + + org + + E6.1 66 318533 ,166 98.4 glacialis Sphingomonadaceae 5,351 Ewingella 1 - - - + +(6) +(16) +(2) +(5) +(6) + + + in/ + + E4 51.7 318534 ,618 99.2 americana Yersiniaceae org 6,717 Pseudomonas 1 - - - + +(5) +(10) +(1) +(2) +(12) - + + in/ + + E3 58.8 318539 ,681 99.7 arsenicoxydans Pseudomonadaceae org S11.1 3,595 Sphingomonas 1 + + - - +(4) +(6) - +(2) +(6) - - + in + - Y 65.3 318543 ,775 97.2 koreensis Sphingomonadaceae 6,384 Mycolicibacteriu 2 - + - + +(5) +(6) +(2) +(5) +(9) - - + in/ + + S5.20 65.9 318544 ,412 98.9 m hodleri Mycobacteriaceae org 6,463 Flavobacterium 1 - - - + +(4) +(7) +(2) +(4) +(7) - + + in - + 42 33.6 318546 ,631 99.1 pectinovorum Flavobacteriaceae 4,556 Variovorax 1 - - - + +(3) +(8) +(2) +(1) +(15) - + + In/ + + S06.C 65.5 318548 ,448 98.4 guangxiensis Comamonadaceae org 4,554 Variovorax 1 - - - + +(3) +(8) +(2) +(1) +(15) - + + in/ + + S09.D 66.5 318552 ,645 99.5 ginsengisoli Comamonadaceae org Sphingomonas 1 + - - + +(3) +(2) - - +(7) - - + org + - 65.8 319773 3,089 S5.1 ,412 97.3 oligophenolica Sphingomonadaceae 3,812 Curtobacterium 1 - - + - +(3) +(8) +(2) +(7) +(4) - - + org + + S5.26 70.7 319774 ,089 100 flaccumfaciens Microbacteriaceae 4,223 Phycicoccus 1 + - - - +(4) +(6) +(2) +(4) +(8) - + + org + + S9.3A 70 319768 ,623 99.3 bigeumensis Intrasporangiaceae 3,715 Sphingomonas 1 + + - - +(3) +(6) - +(2) +(8) - - + in/ + - S12 65.4 319776 ,333 97.3 koreensis Sphingomonadaceae org

68

69

N

SporilationDormancy &

3

Serine

Motylity

O & O

Ammonia assimilation

Metabol Arom. Comp.

Closest 16S Match (%)

Formald.Assim:

Polygo

Mono

N

-

2

Glyoxylatecycle

Photo

O

& Chomotaxis

Di

Dinitrification

NCBI/RAST ID

Fermentation

Ammonificat.

Closest Strain

Sulfur Assim.

- -

CO

-

saccharides saccharides

GC ContentGC

oligosacch.

-

16S genes

synthesis

2

Size (bp)

Isolate fixation

Fa

RuMP

mily

4,256 Rhodanobacter 2 - - - + +(5) +(3) - (+)1 +(7) + - + in + - S13Y 64.2 319778 ,908 98.7 glycinis Xanthomonadaceae 6,535 Roseomonas 2 - + - + +(4) +(5) +(2) +(3) +(13) - + + org + - S9.3B 71.9 319770 ,210 98.6 nepalensis Acetobacteraceae

This data was generated based on RAST annotation. The table summarizes the isolates sequenced in the study. The + marks indicate presence of genes involved in the indicated metabolic process, with brackets representing the number of sub-functions present within each proses. The - marks indicate absence of genes involved in the indicated metabolic possess. Under the “Sulfur Assim.” column the shorthand “in” refers to inorganic sulfur assimilation and “org” refers to organic sulfur assimilation.

69

70

Supplementary Materials

Supplementary Data Tables 1-3 Supplementary Figures 1-3

Table S2.1. Techniques used to isolate the 18 sequences bacterial isolates from the IWP terrain.

Isolate Isolation method Notes

OWC3 Microcosm isolate 10-5 dilution before plating soil S9.2P Microcosm isolate 10-4 dilution before plating soil OWC5 Microcosm isolate 10-5 dilution before plating soil S10 Microcosm isolate 10-4 dilution before plating soil E6.1 Ex situ iChip 318533 E4 Ex situ iChip 318534 E3 Ex situ iChip 318539 S11.1Y Microcosm isolate 10-4 dilution before plating soil S5.20 Microcosm isolate 10-4 dilution before plating soil 42 In situ iChip 318546 S06.C Ex situ iChip 318548 S09.D Microcosm isolate 10-4 dilution before plating soil S5.1 Microcosm isolate 10-4 dilution before plating soil S5.26 Microcosm isolate 10-4 dilution before plating soil S9.3A Microcosm isolate 10-4 dilution before plating soil S12 Microcosm isolate 10-4 dilution before plating soil S13Y Microcosm isolate 10-4 dilution before plating soil S9.3B Microcosm isolate 10-4 dilution before plating soil

70

71

Table S2. 2. PERMANOVA analysis of microbial community based on the relative abundance of the species.

Df Sum of Sqs Mean Sqs F. Model R2 P Topography (Polygon Interior and 1 5.6118 0.17952 0.006 Trough) 0.25284 0.25284 Depth (5 cm and 25 cm) 1 0.57489 0.57489 12.7599 0.40819 0.001 Topography : Depth 1 0.22022 0.22022 4.8879 0.15636 0.006 Residuals 8 0.36044 0.04505 0.25592 Total 11 1.00000

71

72

Table S2. 3. Table summarizing the mutually positive interactions between the microbial members.

OTU pair Mutually positive taxon interaction (Family) 380 6 Burkholderiales & Saprospirales 380 221 Burkholderiales & Xanthomonadales 380 234 Burkholderiales & Sphingomonadales 422 62 Actinomycetales & Burkholderiales 80 355 Chthoniobacterales & Burkholderiales 471 141 Rhodospirillales & Uncultured Acidobacteria, iii1-8 471 253 Rhodospirillales & Uncultured Betaproteobacteria, A21b 34 253 Actinomycetales & Uncultured Betaproteobacteria, A21b

72

73

Figure S2.1. Three network analysis plots. Three plots that aimed to identify significant interactions between species at the IWP terrain. A total of 506 OTUs were used for the analysis. Each OTU was checked for significant interaction against every other OTU. Plot A depicts the correlation co-occurrence only approach, plot B depicts GLM only approach, plot C combines both approaches and only depicts those interactions that were significant in both the correlation and the GLM approaches

73

74

Figure S2.2. Full hybrid network analysis. Full network analysis of the microbial community at the IWP terrain based on 16S metagenomic sequencing. The network is based on a combination of classical correlation-based network analysis and a general linear model approach that tests if the interaction between species is significant once niche overlap is controlled for. Arrows depict the direction of the relationship. Green connections depict positive relationships, red connections depict negative relationships. The saturation of the connecting arrow corresponds to the relative strength of the relationship and is based on the β calculation.

74

75

Table S2.4. Homology-inferred annotation of genes involved in the serine cycle and ribulose monophosphate pathways

Isolate exaF Fae GlyA Sga Hpr Gck Eno Ppc Mdh Mtk Mcl1 rmpA rmpB

99.0% OWC5 99.5%, 99.0%, 99.8% 97.5% 98.6% 99.7%

S10 96.2% 93.0% 89.7% 88.5%

99.0% 100% OWC3 99.0%, 99.5%, 99.8% 98.1%, 97.5% 99.1% 99.7%

E6.1 95.4% 84.9% 98.5% 94.8% 97.8% 93.9%, 81.6%

E4 97.4% 94.8% 95.2% 93.6%

99.7% 100% E3 99.8%, 99.5%, 98.8% 95.7%, 99.1% 98.1% 99.4%

S9.2P 97.2% 92.6% 98.7% 95.7% 80.2%

S5.20 92.9% 92.3% 90.3%

42 98.8% 96.8% 100%

83.6%, 80.6%, S06.C 97.6% 91.5% 86.7% 76.2%

83.6%, 80.6%, S09.D 97.6% 91.5% 86.7% 79.2%

75

76

Isolate exaF Fae GlyA Sga Hpr Gck Eno Ppc Mdh Mtk Mcl1 rmpA rmpB

S9.3A 85.2%, 94.6% 93.1%

76.7%, 76.5%, S9.3B 94.7% 88.8%, 90.1% 88.6% 97.8% 86.5%

S5.1 93.8% 92.0% 83.7% 82.3% 72.7%, 59.6%

S5.26 99.8% 99.8% 100% 100%

S12 90.3% 78.8% 66.8% 77.3% 92.2% 59.6%, 83.6%

S13Y 96.2% 93.0% 89.7% 88.5%

S11.1Y

76

77

Connecting Text:

The study in Chapter 2 explored the CH4 and CO2 dynamics at the IWP terrain site. While these two gases are poised to contribute the most to future Arctic warming potential, N2O is considered the third more important GHG, with the warming potential of ~300X compared to CO2. To better understand the full picture of GHG emissions in Arctic cryosols, in Chapter 3 I explore the N2O flux at the IWP site and identify the active microbial community present in the soils that is likely responsible for the observed flux of this gas.

Portions of this chapter appear in:

Ianina Altshuler, Jennifer Ronholm, Alice Layton, Tullis C Onstott, Charles W. Greer, Lyle

G. Whyte (2018). Denitrifiers, Nitrogen-fixing bacteria, and N2O soil gas flux in high Arctic ice-wedge polygon cryosols. FEMS Microbiology Ecology. (Accepted, in press)

Contributions of authors: IA wrote the manuscript and performed gas flux collection, phylogenetic and bioinformatic analysis, and the soil moisture and content analysis. I.A. and J.R. performed the samples collection and DNA/RNA extractions. IA and AL performed the DNA library preparation and sequencing. I.A., L.G.W., C.G. and T.C.O contributed to conception and editing of the manuscript.

77

78

Denitrifiers, Nitrogen-fixing bacteria, and N2O soil gas flux in high Arctic ice-wedge polygon cryosols. Authors: Ianina Altshuler1, Jennifer Ronholm2, 3, Alice Layton4, Tullis C Onstott5, Charles W. Greer1,6 , Lyle G. Whyte1*

Affiliations: 1 Department of Natural Resource Sciences, Faculty of Agricultural and Environmental Sciences, Macdonald Campus, McGill University, 21111 Lakeshore Rd, Ste Anne-de-Bellevue, QC, H9X 2 Department of Food Science and Agricultural Chemistry, Faculty of Agricultural and Environmental Sciences, Macdonald Campus, McGill University, 21,111 Lakeshore Ste Anne- de-Bellevue, QC, H9X 3V9, Canada 4 Center for Environmental Biotechnology, University of Tennessee – Knoxville, 676 Dabney- Buehler Hall, 1416 Circle Drive. Tennessee, United States of America 3 Department of Animal Science, Faculty of Agricultural and Environmental Sciences, Macdonald Campus, McGill University, 21,111 Lakeshore Ste Anne-de-Bellevue, QC, H9X 3V9, Canada 5 Geomicrobiology, Geosciences, Princeton University, Princeton, NJ 08544, United States of America 6 National Research Council Canada, Montreal, QC, Canada

78

79

Abstract

Climate warming and subsequent permafrost thaw may result in frozen organic carbon and nutrient stores being metabolized by microbial communities and resulting in a positive feedback loop of greenhouse gas (GHG) soil emissions. As the third most important GHG, understanding nitrous oxide (N2O) flux in Arctic mineral ice-wedge polygon cryosols and its relationship to the active microbial community is potentially a key parameter to understanding future GHG emissions and climatic warming potential. In the present study, metatranscriptomic analyses of active layer

Arctic cryosols, at a representative ice-wedge polygon site, identified active nitrogen-fixing and denitrifying bacteria that included members of Rhizobiaceae, Nostocaceae, Cyanothecaceae,

Rhodobacteraceae, Burkholderiaceae, Chloroflexaceae, Azotobacteraceae, and

Ectothiorhodospiraceae. Unique microbial assemblages with a higher proportion of

Rhodobacteriales and Rhocyclales were identified by targeted functional gene sequencing at locations with higher N2O emissions in the wetter trough soils compared to the dryer polygon interior soils. This coincided with a higher relative abundance of the denitrification nirS gene and higher nitrate/nitrite concentrations in trough soils. The elevated N2O flux observed from wetter trough soils compared to drier polygon interior soils is concerning from a climate warming perspective, since the Arctic is predicted to become warmer and wetter.

79

80

Introduction

The disproportional rapid warming of the Arctic compared to the rest of the globe due to climatic change is resulting in permafrost thaw and subsequent decomposition of soil organic matter

(SOM)(Graversen et al. 2008; Gittel et al. 2014a). Arctic soil harbors ~50% of the belowground global organic carbon pool, 88% of which occurs in the permafrost (Tarnocai et al. 2009).

Microbial community composition and availability of decomposers is thought to partially control the decomposition of the SOM (Gittel et al. 2014a). Currently, frozen carbon stores are inaccessible to decomposition or are being metabolized at extremely low rates. However, as the

Arctic warms and precipitation increases the active soil layer will deepen and these stores may become biologically available to microbial metabolism, resulting in further GHG emissions (CO2,

CH4, and N2O) via a positive feedback loop (Lawrence et al. 2008; Tarnocai et al. 2009;

Stackhouse et al. 2015).

Nitrous oxide is the third most important greenhouse gas, after methane (CH4) and carbon dioxide

(CO2). It is an ozone-depleting gas with a 100-year global warming potential of 298 times higher than CO2 (Davidson and Kanter 2014). Studies on natural and anthropogenic emissions of N2O from soils have mainly concentrated on agricultural and tropical soils, as these tend to be well- known sources of N2O to the atmosphere (Butterbach-Bahl et al. 2013). However, models based only on agricultural and tropical soils as sources of N2O may miss important sources of N2O arising from Artic environments which are disproportionately affected by climate change (Davidson and

Kanter 2014).

Nitrous oxide is a by-product of the nitrogen-cycle, specifically denitrification in soils, sediments, and bodies of water, with soils accounting for ~70% of N2O emissions (Butterbach-Bahl et al.

80

81

- - 2013). During denitrification, nitrate (NO3 ) is stepwise reduced to nitrite (NO2 ), nitric oxide

(NO), nitrous oxide (N2O) and finally to dinitrogen (N2) through a series of reactions catalyzed by oxidoreductases. These are encoded by narGHI and napAB (nitrate reductases), nirS and nirK

(nitrite reductases), norB and norVW (nitric oxide reductases), and nosZ (nitrous oxide reductase).

However, depending on abiotic factors such as temperature, pH, water and oxygen availability,

- - carbon to nitrogen ratio, and NO3 /NO2 availability, the denitrification process may be incomplete and result in significant N2O production in addition to N2 (ŠImek and Cooper 2002; Dalal and

Allen 2008). The ratio of N2O to N2 tends to increase in environments with lower pH and low O2 availability (ŠImek and Cooper 2002; Dalal and Allen 2008). Because of this, topography plays an important factor in driving nitrogen soil-atmosphere exchanges in Arctic terrestrial ecosystems as topography is tightly linked to soil moisture patterns, SOM quality/quantity, soil nutrient content and chemistry, and vegetation (Stewart et al. 2014).

As the Arctic soils warm and microbial activity increases, so do emissions of CH4 and CO2

(Graham et al. 2012). However, the low levels of nitrogen in these soils may limit microbial metabolism and the magnitude of carbon emissions. In Arctic permafrost affected soils, nitrogen may enter the system through a combination of permafrost thaw leading to depolymerisation of organic nitrogen containing compounds and ammonia mineralization, atmospheric nitrogen deposition, and nitrogen fixation (Finger et al. 2016; Voigt et al. 2017). Nitrogen fixation is thought to be a primary source of nitrogen into terrestrial Arctic ecosystems (Deslippe et al. 2005), and plays an important role in N2O emissions, with higher N-fixation rates increasing the rate of N cycling in Arctic soils (Stewart et al. 2014). The rates of nitrogen fixation will also likely increase as Arctic soils warm and dampen as a result of climatic change. This is due to higher rates of N- fixation in wetter and warmer soils, with the estimated optimal temperature for N-fixation in Arctic

81

82 ecosystems being between 15-30oC (Hobara et al. 2006; Stewart et al. 2014). This may potentially diminish the nitrogen limitation barrier to microbial growth (Chapin et al. 1991; Stewart et al.

2011a; Stewart et al. 2014). Indeed, Hultman et al. (2015) gauged the activity of the microbial community in active layer soils and the underlying permafrost by looking at the ratios of

RNA:DNA. They reported that the active layer appeared to have a higher RNA/DNA ratio of nif genes, suggesting that it harbours a more active community of nitrogen fixers (Hultman et al.

2015). With the deepening of the active layer, these nitrogen-fixing communities may become significant sources of nitrogen influx to permafrost affected soils and further stimulate microbial metabolism and degradation of frozen carbon stores.

The denitrifying and nitrogen fixing communities in Arctic polar regions are not well characterized, nor is their effect on GHG emissions, including N2O, well understood. Production and emission of N2O have been reported in laboratory permafrost core melting experiments

(Elberling et al. 2010) and cryoturbation (a process resulting in the mixing of soil layers due to repeated freeze-thaw cycles) increases N2O emissions (Palmer et al. 2012). Permafrost soils already harbour organisms and genes involved in denitrification, nitrogen fixation, and ammonia assimilation (Yergeau et al. 2010; Hultman et al. 2015). Furthermore, these nitrogen-cycling genes are overall more abundant in the thawed permafrost compared to intact permafrost, including genes for nitrate assimilation and denitrification. However, genes coding for nitrite and nitrous oxide reductases are relatively lower in abundance in the thawed permafrost soils, suggesting a potential for incomplete denitrification and release of N2O upon thaw (Elberling et al. 2010; Taş et al. 2014).

In this study, we investigated the active microbial community which may be responsible for a potential positive N2O gas flux at a mineral (carbon poor) ice-wedge polygon cryosol site in two soil locations in an ice-wedge polygon (IWP) terrain; the drier soils of the polygon interior and the 82

83 wetter trough soils that overlay the ice-wedges. These mineral soils represent the majority of Arctic permafrost cryosols (Hugelius et al. 2014) and approximately 26% of land not covered by ice is composed of polar deserts (Walker et al. 2002). However, majority of studies to date on GHG flux from Arctic soils, and, specifically N2O flux, have focused on carbon rich organic peat soils

(Marushchak et al. 2011; Palmer et al. 2012; Li et al. 2016; Voigt et al. 2016) and these studies do not link N2O flux to the microbial community present in the soils. Microbial composition and abundance is one of the main drivers that govern GHG fluxes from soils (Graham et al. 2012); thus, understanding how the underlying microbial community in Arctic soils is related to N2O flux and how that community shifts as conditions become warmer and wetter in the Arctic will help in predicting future GHG emissions from Arctic soils. Our approach to link the observed N2O flux with the active microbial community included metatranscriptome sequencing, quantitative PCR of denitrification gene (nirS), functional gene (nirS, nifH) targeted-amplicon sequencing, soil moisture and nitrogen soil content measurements, and in situ soil N2O gas flux measurements.

This study is also the first comprehensive sequencing survey targeting denitrifying and nitrogen- fixing bacteria in mineral cryosols of an Arctic polar desert.

Material and Methods

3.3.1 Gas flux measurements, flux calculation, and statistics

The study site is located near the McGill Arctic Research Station (MARS), at Expedition Fjord,

Nunavut on Axel Heiberg Island in the Canadian high Arctic (coordinates- 79º26'N, 90º46'W).

The active soil layer ranges from 60-73 cm in depth during the summer with the top 5 cm at

9⁰C±0.8⁰C (Lau et al. 2015). The site is characterized by a high centered ice-wedge polygon

83

84 terrain. Soil sampling for metatranscriptome sequencing was done in the summer of 2013, the top

5 cm of the active layer surface soils from the trough and polygon interior were collected and immediately placed in RNA-Lifeguard Soil Preservation Solution (MoBio) and frozen at -20⁰C to maintain the integrity of RNA in the samples. Soils for functional gene amplicon sequencing (nirS, nifH), soil nitrogen/ammonia and moisture content, and qPCR analysis were collected in the summer of 2015 from the active layer trough and polygon interior soils at the top 5 cm and bottom

25-30 cm belowground depths. The soil for functional gene amplicon sequencing, and qPCR analysis was collected in sterile 50 ml Falcon tubes and immediately frozen at -20⁰C. Soil for nitrogen/ammonia and moisture content was collected in 500 ml Whirl-Pak bags and stored at -

20⁰C.

3.3.2 Site and sample collection

The in situ N2O soil gas flux was measured using a static chamber system (Collier et al. 2015) using inverted buckets outfitted with rubber stoppers for gas sampling, at both the trough and polygon interior soils. Gas flux measurements were performed in the summer of 2015. The buckets were placed over bare, non-vegetated soils. Four replicates were collected for each soil type (at two separate polygons/troughs and over two consecutive days) over an eight hour period.

The gas samples were collected in 20 ml evacuated glass vials and brought back to the laboratory to be analyzed by greenhouse gas analyzer GC 450 (Bruker, Germany). Each sample was simultaneously injected onto an electron capture detector (ECD), thermal conductivity detector

(TCD) and flame ionization detector (FID). The N2O was analyzed on the ECD with Argon as the carrier gas. The detection limit for on the GC is 0.1 N2O ppm. The concentration of N2O vs. time was plotted and a linear regression with four time points in the time series of concentrations 84

85 was used to calculate the slope (Collier et al. 2014). The slope of the regression was then used to calculate the N2O flux for each replicate (Collier et al. 2014). A t-test was used to check for significance between the trough and polygon interior soil fluxes.

- - + 3.3.3 NO3 & NO2 , NH4 , and soil moisture measurements

- - + Soils collected for soil moisture, NO3 + NO2 , and NH4 measurements were placed in Whirlpack bags and kept frozen at -20°C during transportation. Soil moisture was measured in triplicate using

- - + ~5 g of soil dried overnight in an oven at 105°C. The NO3 + NO2 and NH4 measurements were performed in triplicate. First, 2 M KCl was added to the soils and shaken for 30 min, supernatant

- - + - - was then filtered and measured for extractible nitrogen (NO3 + NO2 , and NH4 ). The NO3 + NO2 content was measured by colorimetry on a Flow Injection analyzer at 520 nm using a copperized

+ cadmium column. The NH4 content was measured by colorimetry on a Flow Injection analyzer using the sodium salycilate method at 660nm.

3.3.4 DNA and RNA isolation and sequencing

The RNA for metatranscriptome sequencing was isolated using the MoBio RNA Power Soil Total

RNA Isolation Kit with 15 g of soil. The total RNA was used to generate Illumina TruSeq libraries and metatranscriptome sequencing was performed on an Illumina MiSeq sequencer on a V3 reagent cartridge with 150 cycles, generating single 150 bp reads. The DNA for quantitative PCR, and 16S rRNA and functional gene sequencing was isolated using the MoBio Power Soil DNA

Isolation kit with 2 g of soil. Paired end sequencing of the nirS and nifH genes in triplicate biological replicates for each soil type was performed by generating Illumina Nextera libraries

85

86 using the V2 chemistry and sequencing on an Illumina MiSeq sequencer with 500 cycles. Though both nirS and nirK are often used to assess the denitrifying microbial community, we focused on nirS, since nirK was shown to be of a much lower (~1000x) abundance in Arctic soils compared to nirS (Palmer et al. 2012). Primers and annealing temperature used for amplicon preparation are summarized in Table S1. All sequencing data was deposited to GenBank under BioProject

PRJNA421174.

3.3.5 qPCR analysis and statistics

Quantitative PCR of the nirS and 16S rRNA genes was performed using iQ SYBR Green Supermix from BioRad using the manufacturer’s specifications. The qPCR was performed with DNA extracted from the top 5 cm of the soil (top) and soil at 25-30 cm (bottom) of the trough and polygon interior soils. This resulted in a total of four soil types; trough top 5 cm; polygon interior top 5 cm; trough bottom 25-30 cm; polygon interior bottom 25-30 cm. Quantitative PCR was performed on three biological replicates per soil type with three total technical replicates per biological replicate. The qPCR reactions were performed on the BioRad iQ5 Multicolour qPCR

Detection System and the CT values were acquired. The cycler program used was 3 min at 95.0

C, followed by 40X of 10 seconds at 95.0 C; 45 seconds at 54.0 sec; 45 seconds at 72.0 C. The melt curve protocol included increasing the temperature by 0.5 C every 30 sec from 55.0 C-95.0

C and was ran following the amplification protocol to check for fidelity of the primers and ensure that only one product was produced during the amplification. Analysis of the qPCR data included first normalizing the target gene nirS to the 16S rRNA gene in each sample to control for DNA quality and quantity, thus producing the ∆CT (delta cycle threshold) values (Livak and Schmittgen

86

87

2001; Spanier et al. 2010). The CT used was an average of the three technical replicates for each sample. Secondly, the trough soil at 5 cm was used to calibrate and calculate the relative changes in gene abundance between all the soil types via the standard 2−∆∆CT method (Livak and

Schmittgen 2001). This approach does not provide the total gene copy numbers of the target genes in the soil, rather it is a method to more accurately identify any relative differences in the amount of each target gene between the soil samples. Two-way ANOVA was used to check if depth of soil sampling (5 cm vs. 25-30 cm) or the location of the soil (trough vs. polygon interior) had a statistically significant effect on nirS gene abundance in the soil, following this, t-tests were performed between each soil sample type to test which soil location was statistically different in nirS abundance.

3.3.6 Bioinformatics and statistical analysis of sequencing data

For functional (nirS, nifH) targeted-amplicon sequencing the raw reads were analyzed using the

CLC Genomics Workbench software with the Microbial Genomics Module plugin (Qiagen,

Denmark). Quality and chimera filtering was done using default settings, OTUs clustering and taxonomic assignment were done using custom databases constructed from the FunGene functional gene database repository (http://fungene.cme.msu.edu/) (Fish et al. 2013) for the nifH and nirS genes (clustered at 90% see (Palmer et al. 2012), on fixed length trimmed sequences (390 bp for nirS, 390 bp for nifH). The closest uncultured and cultured relatives of OTU representatives were determined using BLAST for those OTU that were not able to cluster with cultured organisms from the databases (Altschul et al. 1990; Wheeler et al. 2007). PERMANOVA analysis was performed using CLC Genomics Workbench to assess whether soil depth (top 5 cm vs. bottom 25

87

88 cm) or soil type (polygon interior vs. trough soils) had a significant effect on microbial community composition. Alpha and Beta diversity was measured using Shannon and Bray-Curtis models respectfully and the Bray-Curtis transformed data was used to preform principal component analysis of the functional nifH and nirS genes, distance matrix, and PCoA plots were then generated using the Microbial Genomics Module of the CLC Genomics Workbench version

10.1.1.

Metatranscriptomic analysis of the raw reads was performed using the Meta Genome Rapid

Annotation (MG-RAST) pipeline. The raw reads were quality controlled, filtered for artificial duplicates, and annotated using the MG-RAST (Meyer et al. 2008). For the MG-RAST pipeline the raw sequences were uploaded, the pipeline options were left as default, low quality bases were trimmed, artificial replicate sequences removed, sequences with greater than 5 ambiguous bases were filtered. SEED subsystems were used to produce the hierarchical functional annotations

(Overbeek et al. 2014). Organismal annotations were made using the MG-RAST and the SSU database, the taxonomy is assigned based on ribosome-encoding genes from the SILVA database

(Wilke et al. 2016) (Wilke et al. 2012). All metatranscriptome sequence data have been deposited to GenBank under Bioproject PRJNA421174.

3.3.7 Phylogenetic analysis

Alignment of the amplicon nirS sequences was performed using the ClustalW algorithm in MEGA software version 7.0 (Kumar et al. 2016) with manual refinement of the alignments. The unrooted phylogenetic tree was constructed using the Maximum-likelihood statistical method based on the

88

89

Tamura-Nei mode with 1000 Bootstrap replicates. The phylogenetic tree was then visualized using

FigTree 1.4.3 (Rambaut 2012).

89

90

Results and Discussion

- - + 3.4.1 N2O Gas Flux, Soil NO3 & NO2 , NH4 , and Moisture Content

At the ice-wedge polygon site on Axel Heiberg Island (AHI), the N2O gas flux at the trough soils overlaying ice-wedges was higher than in the interior of the raised polygon soils (p=0.053); the

-2 -1 flux was measured to be 0.291 (±0.086) mg N2O●m ●day at the troughs and 0.121 (±0.16) mg

-2 -1 - - + N2O●m ●day in the polygon interiors (Figure 3.1). The content of NO3 & NO2 and NH4 as well as soil moisture of the polygon interior and trough soils is summarized in Table 1. The in situ measurements of N2O gas flux in other parts of the Arctic and in laboratory experiments suggest that denitrification and release of N2O are occurring in Arctic permafrost affected soils and such flux has the potential to increase with permafrost thaw (Rodionow et al. 2006; Elberling et al.

2010; Marushchak et al. 2011; Brummell et al. 2014). The majority of Arctic N2O soil flux studies have concentrated on organic and peat soils; however, the flux at our mineral AHI ice-wedge polygon site are comparable (Marushchak et al. 2011; Palmer et al. 2012; Li et al. 2016; Voigt et al. 2016). Together, laboratory and field studies demonstrate that several factors such as vegetation, topography, sunlight, temperature, electron donor availability, carbon content, water content, freeze thaw cycles, and cryoturbation govern N2O emissions from these Arctic soils (Repo et al. 2009; Elberling et al. 2010; Marushchak et al. 2011; Palmer et al. 2012; Stewart et al. 2014;

Li et al. 2016; Voigt et al. 2016). Marushchak et al. (2011) demonstrated that N2O emissions are occurring from Arctic peat permafrost soils in Russia and Finland, with the non-vegetated peat

-2 -1 soils having much higher N2O emissions (10.3 – 1.08 mg N2O●m ●day ) compared to the vegetated soils that had negligible N2O releases: the authors, however, did not detect N2O flux from non-vegetated mineral soils. This is in contract to our results, as we were able to detect N2O

90

91

fluxes in bare, non-vegetated mineral soils. Our results are also comparable to N2O flux at the

-2 -1 Spitsbergen, Svalbard site in the high Arctic where 0.17- 0.37 mg N2O●m ●day flux was detected from soils in the presence of light, but a much lower flux and in some cases even uptake of N2O in the absence of sunlight (Li et al. 2016). Voigt et al (2016) demonstrated that experimental warming increased N2O emissions from vegetated and non-vegetated peat plateaus in the Finnish Lapland but had negligible effect on N2O flux from upland tundra in a discontinuous permafrost zone. In contrast to our acidic mineral cryosols, N2O fluxes in three polar deserts with alkaline soils on Ellesmere Island, Canada were more varied and shifted between production and

-2 -1 consumption depending on specific site and ranged between 0.35 to -1.02 mg N2O●m ●day , with production of N2O occurring not only at the surface of the soils, but throughout the soil profile down to the permafrost table (Brummell et al. 2014). Although N2O flux from agricultural and tropical soils has been the focus of most studies, understanding the drivers that govern the flux of

N2O in polar Arctic soils, both organic peat and mineral, is important for predicting future climate models due to the large size (1,358,000 km2) of Arctic polar deserts (McCann et al. 2016).

3.4.2 Quantitative PCR

Quantitative PCR was used to measure the relative differences in the denitrification gene abundance between the two soils types, the polygon interior active layer soil and the trough active layer soils at depths of 5 cm and 25-30 cm. Quantitative PCR of the nirS genes showed that the trough soils overall had a higher relative abundance of the gene compared to the polygon interior soils, though significance was only detected in the top 5 cm soil samples (Figure 3.2). ANOVA analysis determined that the soil location (p= 0.005; trough vs. polygon interior) had a significant effect on relative nirS abundance, but not soil depth (p=0.107). These results coincide with the 91

92

higher flux of N2O that we observed from the trough soils compared to the polygon interior soils.

- - Furthermore, the trough soils appear to contain higher amounts of nitrogen as NO3 + NO2 , though only in the deeper (25-30 cm) soils (Table 3.1). This is consistent with current knowledge that higher levels of nitrate are known to stimulate denitrification and subsequent N2O releases (Yu et al. 2000; Kemp and Dodds 2002) and that higher abundance of denitrification genes correlates with higher rates of denitrification (Dong et al. 2009). Together with the gas flux data, these results suggest that N2O generation is occurring primarily in the top 5 cm of the trough soils at the IWP terrain.

3.4.3 Community structure of potential denitrifiers and nitrogen fixing bacteria

We analyzed the community composition of potential denitrifiers and nitrogen fixers in the microbial community of the polygon interior and trough soils at the IWP site via amplicon sequencing of the nirS and the nifH genes. In addition, we looked at the functionally active microbial community by analyzing the nitrogen cycling gene transcripts and 16S rRNA transcript sequences in the polygon interior and trough soil metatranscriptomes.

3.4.4 Denitrifying community

Based on targeted gene sequencing of the nirS denitrification gene marker, the microbial community was distinct at the different depths and between soil types (Figure 3 and 4). A total of only 10 nirS OTUs were recovered between all the samples. Unique OTUs were recovered from the deeper soils compared to their top soil counterparts (Figure 3.3). The majority of the recovered

92

93

OTUs did not cluster with nirS genes from known cultured organisms (Table 3.2). PERMANONA analysis of the OTU abundance tables determined that both depth (p=0.010) and soil location

(p=0.002) had a significant impact on the diversity of the nirS containing denitrifying microbial community. Principal component analysis (PCA) was performed on the nirS OTU abundance table, with the first three principal components accounting for 89% of the variation between the samples and replicates. The PCA grouped the polygon interior nirS containing denitrifying community together (Figure S2), with the interior polygon soils and replicates clustering together by depth. The top trough soil replicates also clustered together and separately from the polygon interior soils. However, the deeper trough soils did not appear to cluster together (Figure S2). Both the 25-30 cm and the top 5 cm soils of the trough had high abundances of OTU 6, OTU 3, and

OTU 2 which were closely related to uncultured denitrifiers detected in the Tibetan Plateau (Xie et al. 2014), Finish fen soil (Palmer and Horn 2015), and rice patty fields in Tokyo, Japan (Yoshida et al. 2009) respectively (Table 3.2). Furthermore, many of the denitrifying microbial members were also recovered in both trough and polygon interior soil metatranscriptomes (Table S2). The protein coding and 16S rRNA transcript sequences in the metatranscriptomes were related to known denitrifying bacteria, including members of the genera Thiobacillus, Denitrovibrio,

Pseudomonas, Azospirillum and Azorhizobium, were detected in both trough and polygon interior soils. Contrasting denitrifier communities have previously been linked to differential N2O emissions from soils in discontinuous permafrost Arctic peat soils of Russia (Palmer et al. 2012).

Palmer et al (2012) is one of the few other surveys of denitrifiers in Arctic soils, the study looked at the phylogenetic diversity of the nirS gene in Arctic peat soils that were affected and not affected by cryoturbation. The authors found that the cryoturbated soils emitting N2O in laboratory microcosms had distinct communities of denitrifiers compared to the unturbated soils that were

93

94

not producing N2O (Palmer et al. 2012). Many of the OTUs recovered from the soils were closely related to uncultured bacteria (Palmer et al. 2012). Our findings suggest that this is a similar case in situ at the ice-wedge polygon site: the community of denitrifying bacteria were distinct between the trough soil that appears to have a higher N2O flux in situ and the drier polygon interior soils which had lower N2O flux. Furthermore, phylogenetic analysis of nirS OTUs demonstrates that much of the denitrifying community at the ice-wedge polygon site are currently uncultured bacteria that are phylogenetically distinct from cultured denitrifiers (Figure 3.4). In contrast to our study, a long-term fertilization, warming, and water addition experiment in Alexandra Fiord, Nunavut, did not show increases in N2O production in wetter plots, nor did the authors observer diversity of nosZ (another functional marker of denitrification) between plots (Lamb et al. 2011).

3.4.5 Nitrogen-fixing community

High throughput amplicon sequencing of the nitrogenase (nifH) genes in the top 5 cm and bottom

25-30 cm of the polygon interior and trough soils revealed distinct nifH containing microbial communities between the soil types (Figure 3.5). A total of 127 distinct OTUs were recovered between all the samples. PERMANOVA analysis demonstrated that soil location significantly influenced the taxonomic distribution of the potential nitrogen fixing microbial community

(p=0.002), but soil depth did not (p=0.183). The PCA analysis was performed on the nifH OTU abundance table, with the first three principal components accounting for 93% of the variation between the samples and replicates. The PCA grouped the polygon interior soil nitrogen fixing community together regardless of depth (Figure S3.2), with all the interior polygon soils and replicates clustering together. However, the trough soil replicates clustered together based on depth

94

95 and were separate from the interior soils (Figure S3.2). Polygon interior soils at both depths were dominated by Rhizobiales, though there was also a presence of Chromatiales (Figure 3.5).

Rhizobiales sequences were also present at both soil depths in troughs, but were relatively less abundant overall (Figure 3.5). The trough soils at both depths also contained nifH sequences belonging to Gallionelales, Desulfuromondales and Desulfobacteriales classes. Furthermore, majority of the microbial orders recovered from the targeted amplicon functional gene sequencing of nifH were also present in both soil metatatranscriptomes (Table S3.2). Both protein coding and

16S rRNA transcripts related to known free living nitrogen fixing genera such as Azotobacter,

Beijerinckia, and Nostoc were also detected in both trough and polygon interior soil metatranscriptomes, as well as the symbiotic nitrogen fixing genera Frankia, Azorhizobium,

Bradyrhizobium and Rhizobium. This supports the notion that High Arctic polar desert cryosols may support an active community of nitrogen fixing bacteria. There have been few surveys of the nifH containing microbial communities in the Arctic overall and fewer still in Arctic soils.

Cyanobacteria, such as Nostoc commune have previously been found to be the dominant nitrogen fixers on Devon Island (high Arctic), although this work was done in salt marches, meadows, and hummocks (Chapin et al. 1991). The authors found that soil moisture and temperature were the prominent controlling factors for nitrogen fixation (estimated from acetylene reduction rates), more so than nitrogen and phosphorus content (Chapin et al. 1991). A limited survey of nifH genes in Arctic soil of Cambridge Bay, Canada, revealed nifH genes related to Alphaproteobacteria,

Gammaproteobacteria, Deltaproteobacteria/Spirochaetae groups, and OTUs belonging to an uncultured clone cluster (Izquierdo and Nüsslein 2006). The authors also surveyed tropical and temperate soils, but only found nifH genes belonging to few microbial clades, suggesting that the clone sequencing likely did not capture the full diversity of the nitrogen fixers (Izquierdo and

95

96

Nüsslein 2006). Next generation sequencing data from our study suggests a much greater diversity of nifH containing microbial community in Arctic soils compared to previous surveys. A survey in the lowlands near Alexandra Fiord, Ellesmere Island using terminal restriction fragment length polymorphism (T-RFLP) analysis, identified Alphaproteobacteria as the most common microbial members containing the nifH gene (Deslippe et al. 2005). In addition, nifH containing members of the Betaproteobacteria and Cyanobacteria (Nostocales), along with several unidentified taxa were also present in the soils (Deslippe et al. 2005). Though our study demonstrates unique nifH containing microbial communities between the different soil location of the IWP site, this may not have any effect on nitrogenase activity and nitrogen fixing potential of the soil, as previous studies in Arctic soils demonstrated a lack of correlation between changes in the taxonomic composition of the nitrogen fixing community and nitrogenase activity in the soils (Deslippe et al. 2005).

Nitrogenase activity assays would be the next logical step in understanding the effects of the microbial community and abiotic factors such as temperature and moisture content on the nitrogen fixing potential of the soils at the IWP terrain. This was previously demonstrated by Stewart et al.

(2011) in a low-arctic ecosystem which reported higher N-fixation rates in biological soil crusts that had higher moisture content (Stewart et al. 2011b).

3.4.6 Metatranscriptomics

Metatranscriptome sequencing and analysis was performed on the top 5 cm of the polygon interior and trough soils. Thought transcriptional gene abundances do not always correlate to protein abundances and real activity, both soils showed a presence of gene transcripts involved in the nitrogen cycle including genes for denitrification, nitrogen fixation, and ammonia utilization and

96

97 transport (Figure 3.6). This indicates that the microbes in these soils are actively producing transcripts of genes that are part of the denitrification and nitrogen fixing pathways. Specific microbial members that produced RNA transcripts of genes involved in denitrification included

Paracoccus denitrificans, Pseudomonas aeruginosa, Rhodopseudomonas palustris, Bacillus cereus, Chloroflexus sp., and Sinorhizobium medicae. Since single stranded RNA molecules are more susceptible to degradation, higher turnover rate, and do not persist in the environment as free molecules like DNA, these results suggest that the IWP soils support an active community of these denitrifying bacteria. This is consistent with previous RNA-based sequencing studies that demonstrated organisms capable of denitrification and nitrogen fixation are active members of the microbial community in other permafrost affected soils (Tveit et al. 2014; Hultman et al. 2015b).

Transcripts of genes involved in nitrogen fixation included members of Nostoc punctiforme,

Cyanothece sp., Rhodobacter sphaeroides, Burkholderia xenovorans, Azotobacter vinelandii,

Chlorobium luteolum, Anabaena variabilis, and Halorhodospira halophila, whereas gene transcripts involved in ammonia utilization and transport belonged to members of Pedobacter heparinus, Candidatus Koribacter versatilis, Saccharopolyspora erythraea, Candidatus Solibacter usitatus, Acidithiobacillus ferrooxidans, Chlorobium tepidum, and Desulfovibrio vulgaris.

Together our results suggest that the input of nitrogen into the ecosystem at the site is potentially from a combination of permafrost thaw, which may supply the mineral ammonia and be assimilated by ammonia oxidizers, and through atmospheric nitrogen fixation by diazotrophs. As temperatures warm in the Arctic due to climatic change, both sources of nitrogen would potentially increase. With increasing temperatures, the rate of permafrost thaw would increase, potentially further adding sources of nitrogen through decomposition of organic matter (mineralization).

Meltwater from permafrost soils contains higher levels of ammonium which may be utilized by

97

98

ammonium oxidising bacteria and further contribute to N2O via nitrifier denitrification (Elberling et al. 2010). Indeed, ammonium levels and fertilization with ammonium have shown positive

- - correlations to N2O releases while levels of NO3 and augmentations with NO3 have not in wetter and lowland Arctic permafrost soils (Ma et al. 2007; Elberling et al. 2010), though this is not always the case (Lamb et al. 2011). Previous studies have demonstrated active expression of hydrolases involved in the initial breakdown of complex soil biopolymers in moist Alaskan acidic tundra permafrost (Coolen and Orsi 2015) as well as the presence of active gene transcripts involved in nitrate reduction, denitrification, ammonia uptake and nitrogen fixation in active layer and permafrost soils of Alaskan bog systems (Hultman et al. 2015b), all suggesting that the microbial community in thawing permafrost is primed to utilize the carbon and nitrogen sources that become available with permafrost thaw. Furthermore, with the warming temperatures, the activity of nitrogenase would also be increased (Liengen and Olsen 1997; Stewart et al. 2014) providing an additional input of nitrogen into this ecosystem.

Conclusions

The effects of warming temperatures on Arctic soils and nitrogen availability on SOM decomposition rates are complex and involve the interplay responses and feedback loops of microbial populations, plants, microbial grazers, as well as the inherent magnitude of SOM stocks and their composition. These results demonstrated that mineral cryosols of Arctic polar deserts contain an active community potentially capable of denitrification and nitrogen fixation.

Furthermore, we demonstrated that topography of the land has a significant effect on the composition of the microbial community and N2O flux, with the wetter trough soils having a more prominent flux of N2O and a distinct community of denitrifiers and potential nitrogen fixing

98

99 bacteria. With the warming temperatures and increased moisture content, nitrogen availability in

Arctic soils may potentially increase through increased rates of nitrogenase activity and permafrost thaw. Our results suggested that the active microbial community in these soils is poised to take advantage of these nitrogen inputs into the ecosystem. This may in turn reduce the nitrogen limitation of the entire soil microbial community resulting in increased microbial growth, further exasperating the positive feedback loop of GHG emissions and climatic warming.

99

100

Figure 3.1. In situ gas concentration measurements.

The measurements were made using static chambers at the trough (A) and polygon interior (B) soils. Four replicates were done for each soils type (at two sites and on two consecutive days).

Gas flux was then determined to be 0.291 (±0.086) mg N2O●m-2●day-1 in wetter trough soils and 0.121 (±0.16) mg N2O●m-2●day-1 in the polygon interior soils. Error bars represent standard error of the mean (SEM).

100

101

Figure 3.2. Quantification of nirS in IWP soils. qPRC of denitrification genes nirS in trough soils (Tr) and Polygon interior soils (P Int), in the top 5cm and bottom 25cm. Expression is represented as fold change relative to trough soil at 5cm. All gene expression is relative to 16S rRNA genes to control for quality and quantity of DNA extractions. Error bars represent SEM. Letters represent statistically different gene abundances between the soils

101

102

Figure 3.3. Microbial composition of nitrogen fixing nirS gene at the Order level in the trough soils and polygon interior soils at 5 cm and 25 cm.

102

103

Figure 3.4. Phylogenetic reconstruction of the nirS gene. Phylogenetic reconstruction of the nirS gene including the OTUs recovered from the study and their closest cultured and uncultured NCBI matches. The shapes beside the OTU names indicate OTUs that were present in at least two of the three replicates from each soil type. Light shapes denote that the OTU was present in the top 5 cm of the soils, while dark shapes denote deeper soils. Circles denote OTUs that were present in trough soils, while squares denote polygon interior soils. The numbers at branch nodes denote portion of replicate trees in which the taxa clustered together in the bootstrap test (1,000 replicates), only bootstrap values over 0.5 are shown. The tree is unrooted.

103

104

Figure 3.5. Microbial composition of nitrogen fixing nifH gene at the Order level in the trough soils and polygon interior soils at 5 cm and 25 cm.

104

105

Figure 3.6. Relative abundance (%) of transcripts involved in the nitrogen cycle in trough and polygon interior top 5 cm soil metatranscriptomes.

105

106

Table 3.1. The content of NO3 & NO2 and NH4 as well as soil moisture of the polygon interior and trough soils at 5 cm and 25 cm.

Soil type N mg/kg dry soil as N mg/kg dry soil as NH4 % soil moisture content NO3+NO2

Top 5 cm trough soil aBDL 2.572 (0.294) 35.5 (6.6)

Top 5 cm polygon aBDL 2.173 (0.301) 13.0 (1.6) interior soil

Bottom 25 cm trough 3.440 (0.133) 1.182 (0.143) 18.8 (0.29) soil

Bottom 25 cm polygon b0.791 1.435 (0.052) 16.2 (3.5) interior soil aBDL denotes measurements below the detection limit of >0.7mg/kg. Standard deviation (SD) is given in brackets. bSD could not be calculated, as only one replicate was above the detection limit of 0.7mg/kg NO3+NO2.

106

107

Table 3.2. OTUs recovered in the amplicon sequencing of the nirS gene in the trough and polygon interior soils. OTU Closest NCBI Closest %match Location/description Reference of Closest %match Name number BLAST match NCBI the match cultured match NCBI match 1 Uncultured KC468737 90 Qinghai-Tibet (Xie et al. AB542309 80 Herbaspirillum bacterium clone F- Plateau 2014) sp. TSA20w 5-1-33 2 Uncultured HE995567 83 Fen denitrifier (Palmer and AB545713 83 Herbaspirillum bacterium community in Horn 2015) sp. TSO46-2 Finnish Lapland

3 Uncultured AB377855 93 Rice paddy field soil, (Yoshida et AM230903 74 Paracoccus sp. bacterium 518S64 Tokyo al. 2009) R-24665 4 Uncultured AB377951 92 Rice paddy field soil, (Yoshida et AJ626840 77 Pseudomonas bacterium clone Tokyo al. 2009) sp. R125 727S039 5 Uncultured KC468727 98 Qinghai-Tibet (Xie et al. AB480490 88 Rhodanobacter bacterium clone F- Plateau 2014) sp. D206a 5-1-23 6 Uncultured KC469026 95 Qinghai-Tibet (Xie et al. AB696890 78 Rhodocyclales bacterium clone F- Plateau 2014) bacterium 7-3-177 UNPF16 7 Uncultured KC469053 92 Qinghai-Tibet (Xie et al. AP014879 77 Sulfuricaulis bacterium clone F- Plateau 2014) limicola 7-3-217 8 Pseudomonas sp. CP006852 100 Isolated from (Ohtsubo et CP006852 100 Pseudomonas TKP sediments at al. 2014) sp. TKP Kyushu, Japan

107

108

9 Uncultured KX378403 88 N/A Unpublished- AB937702 88 Cupriavidus sp. prokaryote clone Direct NCBI NC3H-95a nirS1-13 submission 10 Uncultured KJ498556 99 Wastewater Unpublished- AP012547 78 Sulfuritalea denitrifying treatment systems Direct NCBI hydrogenivorans bacterium clone submission sk43H J30927555_4_5_T7 .

108

109

Supplementary Materials:

Figure S 3.1. The ice-wedge polygon terrain from the ground and birds eye view.

109

110

Figure S 3.2. Principal Component Analysis of the nirS genes present in trough and polygon interior soils at top 5 cm and bottom 25 cm. Trough soil replicate samples at 5 cm are in blue; trough soil replicate samples at 25 cm are in green; polygon interior soil replicates at 5 cm are in orange; polygon interior soil replicates at 25 cm are in red.

110

111

Figure S 3.3. Principal Component Analysis of the nifH genes present in trough and polygon interior soils at top 5 cm and bottom 25 cm. Trough soil replicate samples at 5cm are in blue; trough soil replicate samples at 25 cm are in green; polygon interior soil replicates at 5cm are in orange; polygon interior soil replicates at 25 cm are in red.

Table S 3.1. Primers used the study. Name Primer sequence 5’-3’ Gene Tm used Application Reference EubR GGACTACCAGGGTATCTAATCCTGTT 16S 54 qPCR (Nadkarni et al. 2002) EubF TCCTACGGGAGGCAGCAGT 16S 54 qPCR (Nadkarni et al. 2002) nirSR3cd (mod) GASTTCGGRTGSGTCTTSAYGAA nirS 54 qPCR (Kandeler et al. 2006) nirSCd3aF (mod) AACGYSAAGGARACSGG nirS 54 qPCR (Kandeler et al. 2006) nirSR3cd GASTTCGGRTGSGTCTTGA nirS 53 Sequencing (Throbäck et al. 2004) nirScd3aF GTSAACGTSAAGGARACSGG nirS 53 Sequencing (Throbäck et al. 2004) DVV ATIGCRAAICCICCRCAIACIACRTC nifH 56 Sequencing (Gaby and Buckley 2012) IGK3 GCIWTHTAYGGIAARGGIGGIATHGGIAA nifH 56 Sequencing (Gaby and Buckley 2012)

111

112

Table S 3.2. The percent of 16S rRNA transcripts belonging to each bacterial phylum/class from the trough and polygon interior top 5 cm soil metatranscriptomes.

Polygon interior soil Trough soil Phylum (and -class) (% bacterial SSU reads) (% bacterial SSU reads) Bacteroidetes 28.35 4.45 -Sphingobacteria 23.49 3.66 -Cytophagia 2.40 0.40 -Bacteroidia 1.10 0.16 -Flavobacteria 1.07 0.12 -unclassified 0.28 0.10 Proteobacteria 14.23 14.40 -Alphaproteobacteria 3.54 5.18 -Betaproteobacteria 2.12 1.37 -Gammaproteobacteria 5.85 2.33 -DeltaProteobacteria 1.77 3.54 -unclassified 0.54 1.36 -EpsilonProteobacteria 0.41 0.61 -ZetaProteobacteria 0.01 0.02 Acidobacteria 10.84 17.21 Firmicutes 6.74 6.18 Actinobacteria 4.03 15.75 Verrucomicrobia 3.49 4.73 Cyanobacteria 3.30 0.46 Planctomycetes 0.59 2.74 Chlorobi 0.50 0.02 Synergistetes 0.48 0.39 Chloroflexi 0.32 2.31 Spirochaetes 0.23 0.31 Thermotogae 0.18 0.21 Gemmatimonadetes 0.13 0.48 Chlamydiae 0.09 0.36 Other 26.50 29.99

112

113

Table S 3.3. The percent of 16S rRNA transcripts from the trough and polygon interior top 5 cm soil metatranscriptomes belonging to bacterial orders that were also dominant in the functional gene sequencing of nirS and nifH genes.

Order Polygon Interior Trough Xanthomonadales 1.340649794 0.362393251 Opitutales 0.818003235 0.360915101 Burkholderiales 0.798203895 0.436591931 Rhizobiales 0.634792258 1.095354374 Rhodospirillales 0.436262169 0.734976783 Pseudomonadales 0.100980131 0.08539679 Desulfovibrionales 0.084330952 0.344946595 Desulfuromonadales 0.061463123 0.101589258 Rhodobacterales 0.054252757 0.036416255 Desulfobacterales 0.052817684 0.152137529 Rhodocyclales 0.024781259 0.07099602 Gallionellales 0.006031973 0.008824111 Nitrosomonadales 0.002881813 0.047524781 Acidiferrobacterales 0 0

114

Connecting Text:

Chapter 3 aimed to identify active microbial organisms that are potentially responsible for the observed N2O emissions at the IWP site. Unlike N2O, CH4 at this site is being taken up by the soils, as demonstrated in Chapter 2 and in previous studies. Here in Chapter 4, I aim to determine the in situ active microbial community that is involved in the CH4 metabolism at the IWP terrain 13 through direct stable isotope labeling of the soils with CH4.

Contributions of authors: The overall conseptualization and writing of this manuscript was performed by I.A., C.G., and L.G.W. In situ enrichments were performed by I.A. and I.R-B. Bioinformatic analysis was done by I.A., J.T., C.G. and E.M. Editing of the manuscript was performed by all the authors.

115

In situ DNA-SIP enrichment in concert with genome binning of a high Arctic methanotrophic and methanotroph- associated community.

Ianina Altshuler1, Julien Tremblay2, Elisse Magnusson1, Isabelle Raymond-Bouchard1, C. W. Greer1,2, L.G. Whyte1

1McGill University, 21,111 Lakeshore Rd., Ste. Anne de Bellevue, QC, Canada, H9X 3V9

2 National Research Council Canada

116

Abstract

Greenhous gas (GHG) emissions from Arctic permafrost soils are poised to create a positive feedback loop of climatic warming and further GHG emissions. Active methane uptake in soils can reduce the impact of GHG emissions on future Arctic warming potential. Aerobic methane oxidizers are thought to be responsible for this apparent methane sink, though these organisms have resisted all culturing efforts. In this study, we labelled organisms in situ using

13 Stable Isotope Probing (SIP) with CH4 (at 100 ppm and 1000 ppm) to identify organisms involved with CH4 metabolism at an ice-wedge polygon (IWP), Arctic cryosol site. The labeled microbial community was enriched for Proteobacteria and Verrucomicrobia phyla. Sequencing of 13C-labelled pmoA genes demonstrated that Type II methanotroph are the dominant active methanotrophs at this terrain, with Type I methanotrophs being only labeled in the 100 ppm SIP treatment. Genome binning of the labeled metagenomic DNA resulted in 28 total bins. From these we identified nine high to intermediate quality metagenome assembled genomes (MAGs), belonging to α and β-Proteobacteria and Gemmatimonadetes, with three of these MAGs containing genes associated with methanotrophy based on hidden markov model scans (HMMR).

Furthermore, we identified an Alphaproteobacterial MAG, that contained both an mmoX methanotrophy gene and serine cycle genes associated with Type II methanotrophs.

117

Introduction

The bulk of microbial species are resistant to isolation and laboratory cultivation (Ferrari, et al. 2008; Nichols, et al. 2010). Nevertheless, this “microbial dark matter” likely has a significant contribution to biogeochemical cycling in many ecosystems (Torsvik et al. 2002; Gies et al. 2014).

This remains one of the biggest hurdles in linking function and biological processes to specific microorganisms and furthering the field of microbial ecology. Genome binning has been used to reconstruct genomes from metagenomic data of organisms that have resisted laboratory culturing.

This allows researchers to characterize and study the identity, physiology, and metabolism of these uncultivated microbial species. Unfortunately, as high community diversity is not conducive to robust genome binning, the majority of successful genome binning efforts have come from low complexity microbial communities or studies that have in other ways reduced the diversity of the sample, (Albertsen et al. 2013) (Becraft et al. 2015; Brown et al. 2015). One way in to reduce sample complexity in highly diverse samples, is to target organisms using Stable Isotope Probing

(SIP). SIP relies on the incorporation of stable isotope species, for example 13C, into biological molecules of microorganisms, labeling them in the process (Manefield, et al. 2002). This way, organisms that can grow and utilize a target substrate or are involved in the downstream metabolism of the substrate can be identified and analyzed (Dumont and Murrell 2005). SIP labeling of nucleic acids (RNA and DNA-SIP) is useful in microbial ecology as it allows for phylogenetic identification of active organisms that can utilize a target substrate. The labeled and unlabeled nucleic acids are separated using cesium chloride (CsCl) density gradient centrifugation

(Dumont and Murrell 2005). The organisms which incorporated the heavy isotope into their nucleic acids are thus the ones using the substrate of interest or its downstream by-products. SIP

118 ex situ labeling has been previously used to identify methanotrophic communities in High Arctic wetlands (Graef, et al. 2011), Arctic lake sediments (He, et al. 2012), and Arctic soil (Martineau,

13 et al. 2010), though to date in situ SIP labeling has not been reported. Using CH4-SIP coupled to genome binning allows us to first reduce the diversity of the DNA sample by targeting organisms involved in the CH4 metabolism and to then acquire potential genomes of these organism through genome binning, which works best with low diversity samples.

Methanotrophic organisms play a key role in regulating the overall methane (CH4) emissions to the atmosphere and reducing the CH4 atmospheric load. Methanotrophs are characterized by their ability to oxidize CH4 and assimilate it as organic carbon (Hanson and Hanson 1996). These organisms are phylogenetically diverse, belonging to the phyla Verrucomicrobia and

Proteobacteria and are further classified as either Type I methanotrophs within the

Gammaproteobacteria or Type II methanotrophs within the Alphaproteobacteria (Conrad 2007).

All methanotrophs utilize a methane monooxygenase (MMO) enzyme to oxidize methane into methanol. The two forms of the enzyme are a soluble cytoplasmic form (sMMO) coded by the mmoX gene, and a particulate membrane bound form (pMMO) (Ricke, et al. 2005). The pMMO is encoded by three consecutive conserved open reading frames: pmoC, pmoA, and pmoB, with pmoA coding for the active site. While the soluble methane monooxygenase is found in several methanotrophs, the pMMO is ubiquitous to all methanotrophs (apart from Methylocella and

Methyloferula species) (Knief, et al. 2003; Martineau, et al. 2014).

Arctic soils, including the Axel Heiberg ice-wedge polygon site (IWP), act as methane sinks

(Emmerton, et al. 2014; Martineau, et al. 2014; Jørgensen, et al. 2015; Lau, et al. 2015) (Martineau et al. 2010). This has largely been attributed to a group of putative methanotrophs with divergent pmoA genes part of the USCα (Alphaproteobacteria) and USCϒ (Gammaproteobacteria) clusters

119 of high affinity methanotrophs (Holmes, et al. 1999; Kolb, et al. 2005) (Pratscher et al. 2018).

Studies in rice paddy soils have shown that culturable low-affinity methanotrophs could also be responsible for atmospheric methane oxidation (Cai et al. 2016). Arctic soils range in their negative methane flux from -3.1±1.4 to -0.02±0.01mg CH4 m-2d-1 (Emmerton, et al. 2014; Jørgensen, et al. 2015; Lau, et al. 2015). The soils are projected to potentially increase in their methane consumption because of projected increases in temperatures of Arctic soils coupled to increased methanotrophy rates (Curry 2009; Jørgensen, et al. 2015). The pmoA genes detected in high Arctic mineral cryosols acting as methane sinks are phylogenetically related to other high affinity methane oxidisers (USCα and USCϒ) from upland forest soils (Martineau, et al. 2014; Lau, et al.

2015). However, to date, these and all other atmospheric methane oxidizers have been resistant to culturing and isolation, thus challenging our understanding their physiology and other metabolic potential.

Here we focused on identifying active in situ methanotrophs in soils acting as methane sinks at the

IWP terrain. We used both qPCR and gas flux measurements to first identify candidate soils with

13 high methane oxidation rates, the identified soils were then used for in situ CH4- SIP labeling with to identify methanotrophic organisms that are responsible for the negative methane flux at in

Arctic cryosols. We performed pmoA and 16S rRNA targeted functional gene amplicon sequencing of the 13C-labelled extracted DNA from the cryosols to identify organisms involved in the methane metabolism at the site. In addition, .etagenome sequencing of the 13C-labelled DNA in concert with genome binning was used to yield several high to intermediate quality MAGs

(Metagenome Assembled Genomes). These MAGs were used to identify the metabolic potential of non-culturable organisms involved in the methane cycle at the IWP site.

120

Methods

Selecting target soils via gas flux measurements and quantitative PCR

The in situ CH4 soil gas flux measurements were performed using a static chamber system

(Collier et al. 2015) by means of inverted buckets outfitted with rubber stoppers for gas sampling.

Gas flux measurements were performed at both the trough and polygon interior soils to identify which soils were hot spots of methane oxidation. Four replicates were collected per each soil type, at two polygons/troughs and over two separate days, the samples were collected over an eight- hour period into 20 ml evacuated glass vials and brought back to the laboratory. The samples were then analyzed Gas chromatograph analyzer 450-GC (Bruker, Germany). Each sample was simultaneously injected onto an electron capture detector (ECD), thermal conductivity detector

(TCD), and flame ionization detector (FID). The CH4 and CO2 was analyzed on the FID with

Helium as the carrier gas.

Quantitative PCR of the pmoA and 16S rRNA genes was performed using iQ SYBR Green

Supermix from BioRad using the manufacturer’s specifications. The qPCR was performed with

DNA extracted from the top 5 cm of the soil (top) and soil at 25 cm (bottom) using the trough and polygon interior soils. This resulted in a total of four soil types; trough top 5 cm; polygon interior top 5 cm; trough bottom 25 cm; polygon interior bottom 25 cm. Quantitative PCR was performed on three biological replicates per soil type with three total technical replicates per biological replicate. The qPCR reactions were performed on the BioRad iQ5 Multicolour qPCR Detection

System and the CT values were acquired. The cycler program used was 3 min at 95.0 C, followed by 40X of 10 seconds at 95.0 C; 45 seconds at 54.0 sec; 45 seconds at 72.0 C. The melt curve protocol included increasing the temperature by 0.5 C every 30 sec from 55.0 C-95.0 C and was

121 ran following the amplification protocol to check for fidelity of the primers and ensure that only one product was produced during the amplification. Analysis of the qPCR data included first normalizing the target gene pmoA to the 16S rRNA gene in each sample to control for DNA quality and quantity, thus producing the ∆CT (delta cycle threshold) values (Livak and Schmittgen 2001;

Spanier et al. 2010). The CT used was an average of the three technical replicates for each sample.

Secondly, the trough soil at 5 cm was used to calibrate and calculate the relative changes in gene expression between all the soil types via the standard 2−∆∆CT method (Livak and Schmittgen 2001).

This approach does not provide the total gene copy numbers of the target genes in the soil, rather it is a method to more accurately identify any relative differences in the amount of each target gene between the soil samples.

13 In situ CH4-SIP enrichments and soil collection

In situ enrichments of the trough soils were performed by placing a closed chamber over the trough soils and injecting CH4 into the headspace. The chambers were constructed by outfitting inverted plastic flour pots with rubber stoppers for injecting the gas. The headspace of the soils was injected with ≥99.0% CH4 gas (Sigma-Aldrich) to a final concentration of 100 ppm and 1000

13 12 ppm of CH4 gas in triplicate for each treatment, with CH4 and no CH4 augmentation as controls.

The gas was flushed and refiled every other day for a total incubation time of 12 days in situ. The soils from the enrichments and controls were collected in 50 ml Falcon tubes and immediately frozen at -20⁰C for transport to the laboratory.

DNA extraction, separation on a CsCl gradient, and sequencing

The DNA was extracted from 10 gr of soil per replicate using the DNeasy PowerMax Soil

Kit (MoBio) following the manufacturer’s instructions. Separation of the CsCl gradient followed

122 the method from Martineau et al. (2010). The heavy and light DNA bands were visualized using the Safe Imager blue light transilluminator and extracted with an 18-gauge needle. The DNA samples were further purified to get rid of residual salt using the QIAEX II Gel Extraction Kit

(Qiagen). Illumina libraries for pmoA and 16S targeted gene sequencing were prepared using the

V2 chemistry with the 500-cycle kit. Sequencing was performed on an Illumina MiSeq, generating

2X250 paired end reads. Nextera XT DNA Library Preparation Kit was used to prepare three

13 13 metagenome libraries using the heavy C-DNA from the 100ppm CH4-SIP enrichment with the

600-cycle kit. Sequencing was performed on an Illumina MiSeq, generating 2X300 paired end reads.

Bioinformatics

Metagenomic reads were trimmed with Trimmomatic (v0.36, with settings LEADING:3,

TRAILING:3, SLIDINGWINDOW:4:15, MINLEN:36) (Bolger et al. 2014). Trimmed reads from the triplicate samples were merged and assembled with Megahit (v1.1.3, default settings) (Li et al.

2016). Assembled contigs were binned with MetaBAT2 (v2.12.1, default settings) (Kang et al.

2015) and MaxBin (v2.2.4, default settings) (Wu et al. 2016). Bin completeness, contamination, and taxonomy were estimated with AII similarity (Rodriguez-R and Konstantinidis 2014).

Taxonomy was also checked by with BLASTn alignment of binned contigs against the Greengenes database (v13_5) and NCBI nt database. Gene prediction in the Megahit assembly was carried out with MetaGeneMark (v3.25) (Zhu et al. 2010). HMM models of pmoA, pmoB, and pmoC genes were created with HMMER (v3.2.1) (hmmer.org). These hmm models were added to the existing

Pfam database (v31.0) to create a custom database. Predicted genes identified by MetaGeneMark were annotated with this database using the HMMER hmmscan function. Separately, an HMM model of the mmoX gene was created and aligned to the predicted genes. The results were merged.

123

Results and Discussion

Identifying methane oxidation hotspot for the 2015 season

The flux of CH4 at the IWP site was negative, with CH4 concentrations in some of the replicates dipping below the detection limit of the GC at 0.1 ppm. The CH4 flux from the soils at

2 the IWP site was on average -6.23 (±1.39) mg CH4-m -day across the terrain (Figure 4.1). The

2 negative flux was more pronounced in trough soils at -8.47 mg CH4-m -day, compared to polygon

2 interior soils at -3.99 mg CH4-m -day. This deviated from previous studies in Arctic soils and other years at the IWP terrain (Chapter 2). In the 2015 year the negative flux was higher at the IWP terrain overall and more pronounced at the trough soils compared to the polygon interior soils.

This is opposite of what we observed in the 2016 and 2017 seasons. The qPCR analysis revealed that trough soils had relatively higher levels of pmoA genes compared to polygon interior soils

(Figure 4.1) and that the top 5 cm of both soils had relatively higher levels of the pmoA gene as well (Figure 4.1). Therefore, the qPCR results in combination with gas flux results point to the trough top 5 cm soils as being hotspots for atmospheric methane oxidation at the IWP terrain during the 2015 summer season. Therefore, we used these soils for in situ SIP analyses with 13C- labelled CH4.

The 16S and pmoA analysis of the 13C labeled DNA

The in situ 100ppm and 1000ppm 13CH4-SIP enrichment resulted in a clear separation of the heavily labeled DNA from the light DNA band via the CsCl gradient (Figure 4.2). The

13 12 microbial community in the C heavy and the C light bands of the SIP CH4 enrichments were different from each other and from the control, non-enriched samples (Figure 4.3, Figure 4.4)

124 indicating that we successfully separated the 13C enriched DNA from the 12C-DNA during the extraction. Despite the atmospheric methane oxidation (~ 1.8 ppm CH4) occurring in these soils, we performed the in situ labeling experiments at 100 ppm and 1000 ppm of CH4, to ensure sufficient labeling of any active methanotrophs in the soils. Beta diversity analysis separated the communities based on two principal components that explained 85.89% of the variation between the samples (Figure 4.4). The heavy bands appeared to be enriched for Proteobacteria and

Verrucomicrobia (Figure 4.3) likely due to only members of these phyla are known to oxidize methane. Similar results were previously reported in other studies targeting methanotrophs, though

13 with laboratory CH4-SIP (Qiu et al. 2008; Esson et al. 2016).

The pmoA gene was only amplifiable in the heavy 13C-DNA and was not detected in the light non- labled DNA. The majority of the pmoA sequences in the 100 ppm and 1000 ppm SIP CH4 enrichments matched to Alphaproteobacteria (type II methanotrophs) (Figure 4.5). However, the

100 ppm enrichment also contained Gammaproteobacteria (type I methanotrophs) (Figure 4.5).

The most dominant Alphaproteobacteria type II pmoA sequences most closely matched the

Methylocapsa genus from cultured methanotrophs. This genus is phylogenetically closest to the uncultured USCα cluster, members of which are hypothesized to be high affinity methanotrophs capable of atmospheric oxidation (Knief et al. 2003). The most dominant Gammaproteobacteria

Type I pmoA sequences most closely matched the genus of Methylomarinovum (Figure 4.5). This apparent low diversity of methanotrophs, based on pmoA sequences, has been also demonstrated in Arctic wetlands, compared to their more temperate counterparts (Graef et al. 2011). A previous laboratory incubation ex situ SIP study also showed low pmoA diversity, with higher proportion of Type II methanotrophs than Type I, though it was done with pmoA clone libraries instead of amplicon sequencing (Martineau et al. 2014).

125

Metagenomics and Genome Binning from SIP labelled DNA

To further characterize the functional potential in general and methane metabolism

13 specifically, we performed metagenomic sequencing of the 100ppm, CH4 labeled DNA. Since

13 we were able to label the DNA in situ with CH4 concentrations of 100 ppm we decided to focus on these samples because they had a higher diversity of methanotrophs based on their pmoA sequences and they were closer to atmospheric CH4 concentrations compared to the 1000 ppm treatment. The SIP metagenomes yielded a total of ~8 Gbp of sequence data. The metagenomes contained genes involved in nitrogen fixation, denitrification, ammonia assimilation, inorganic sulfur assimilation, degradation of aromatic compounds, and fermentation processes. Genes for formaldehyde assimilation via both the serine and the ribulose monophosphate pathways were also

13 13 detected. Genome binning with MetaBAT from the C-DNA band of the 100 ppm CH4 SIP labeled metagenome resulted in 28 bins, with eight bins of intermediate to very high quality. These

MAGs (Metagenome Assembled Genomes) ranged in completeness between 88.3-30.6%, with contamination ranging between 0.9-4.5% based on copy number of essential genes (Rodriguez-R et al. 2018). (Table 4.1). The taxonomic assignments of the MAGs included members of

Nitrosomonadaceae, Burkholderiales, Alphaproteobacteria, Thiobacillus, Gemmatimonadales, and Ramlibacter. These assignments were based on average amino acid identity (AAI), which has been recommended to phylogenetically assign distantly related genomes, as the 16S genes are not usually detected in binned genomes and likely constituted of non-cultured microbial clades. (Table

4.1).

The HMMer algorithm uses probabilistic models to detect new homologous sequences and novel proteins with similar function to previously identified proteins. It is therefore able to identify divergent homologs that may not be detected with BLAST (Doxey et al. 2018). The sequence

126 similarity significance is assessed using E-values with low values (E < 0.001) used to infer homology (Pearson 2013; Doxey et al. 2018). The SIP CH4 labelled metagenomes contained both pmoA and mmoX methane monooxygenase genes based on HMMR scans with the pfam database.

A total of three of the high – intermediate quality MAGs (# 8, 16, and 21) contained methane oxidation (mmoX, pmoB) genes were identified with the HMMR algorithm, but were not detected by annotating the MAGs with RAST. We did not identify any high-quality MAGs with all three pmoABC genes, though MAG # 21 was identified to contain an mmoX gene (Tabel 4.1).

Mag # 21 contained an mmoX gene was classified as belonging to Proteobacteria phylum (p-value:

0.011) and potentially the Alphaproteobacteria class (p-value: 0.28), with a GC content of 63.4%.

The closest relative was determined to be Candidatus Hodgkinia cicadicola with 40.67% AAI; this organism is an intracellular insect symbiote with a drastically reduced genome (McCutcheon and

Moran 2012). It is possible that this organism may not be capable of active methanotrophy and instead acquired the heavy 13C carbon through syntrophy (cross-feeding) with other active methanotrophs in the soil (Morris et al. 2013). Through a combination of RAST and HMMer analysis we determined that this MAG also contains other genes involved in the methane metabolism and 1-C carbon assimilation, including genes (Eno, Mdh, Hpr, Ppc, Mcl1, and sgaA) for the serine pathway of assimilating formaldehyde (Figure 4.6). This pathway is typically utilized by other Type II Alphaproteobacteria methanotrophs, which is consistent with the phylogenetic classification of the organism. Nitrogen acquisition in this MAG appears to be through ammonia uptake and assimilation. Genes for uptake and utilization of amino peptides and amino acids were also present, indicating that if this organism does utilize CH4, it likely also relies on other carbon sources as well, indicating a mixotrophic metabolism strategy. This MAG also contained genes coding for export of cadmium and arsenic and detoxification of mercury via a mercuric reductase

127

(coded by merA). Mercury(II) can interfere with methanotrophy rates (Contin et al. 2012). It irreversibly inhibits the sMMO enzyme (coded by mmoX) (Boden and Murrell 2011), therefore detoxification of Mercury (II) via merA is advantageous for the mmoX containing methanotrophs.

Conversely, genes coding for a copper importing p-type ATPase were present. sMMO expression is tightly linked to the availability of copper, with some mutants unable to have an active sMMO in the absence of copper (Yan et al. 2016). This MAG also contained genes for synthesis of polyhydroxyalkanoate biopolymers such as poly-3-hydroxybutyrate (PHB) (Strong et al. 2016).

This pathway is present and expressed in other methanotrophic organisms under nutrient limited conditions, where instead of entering the TCA cycle, serine is funneled into the PHB cycle to generate intracellular storage granules that serve as a C source. (Strong et al. 2016). Specifically,

N-limiting and P-limiting conditions increase the PHB production as a way to store carbon for future growth once N and P limitation is removed. The stored PHBs can serve as a source of carbon for synthesis or they can facilitate further methane consumption by providing a source of reducing power (Strong et al. 2016).

Two recent previous metagenome binning attempts to identify a high affinity methanotroph genomes were performed on Antarctic mineral soils (Taylor Dry Valley) and forest soil (Marburg,

Germany), though neither study used SIP labeling to target methanotrophs (Edwards et al. 2017;

Pratscher et al. 2018). Antarctic mineral soil contained a MAG likely belonging to

Gammaproteobacteria (Type I methanotroph), while the forest soil contained a MAG likely belonging to Alphaproteobacteria (Edwards et al. 2017; Pratscher et al. 2018). In contrast to our

IWP MAG #21 which contained an mmoX gene, the forest and the Antarctic soil MAGs contained a pmoABC genes. Methanotroph SIP labeling previously identified methanotrophs in High Arctic wetlands (Graef et al. 2011), Arctic lake sediments (He et al. 2012), soil (Martineau et al. 2010),

128 and peatlands (Gupta et al. 2012); however, these studies all involved taking the soil out of the natural environment and preforming laboratory incubations in sealed glass microcosms.

Nevertheless, these studies are valuable in helping us understand the biological component of the biogeochemical methane cycle and the active diversity of methanotrophs found across different ecological niches. Both genome binning and pmoA sequencing suggest that Type II methanotrophs are the dominant active methanotrophs in Arctic cryosols. Type II methanotrophs utilize the serine pathway of formaldehyde assimilation versus the RuMP pathway that Type I methanotrophs utilize

(Conrad 2007). Genes indicative of the serine pathway were more abundant in the SIP metagenomes and in the binned genomes. This is consistent with previous SIP studies that showed

Type II methanotrophs as the dominate methane oxidizers in acidic peat soils (Qiu et al. 2008;

Gupta et al. 2012; Esson et al. 2016). Though type I methanotrophs have also been shown to contribute to methane consumption, albeit in smaller proportion (Esson et al. 2016). However, in

Arctic wetlands Gammaproteobacteria (type I) Methylocacter methanotrophs appear to be dominant active methane oxidizers (Graef et al. 2011). Type I methenotrophs were also more abundant in flooded rice field paddy soils (Qiu et al. 2008).

Conclusion

Metagenomic binning and stable isotope probing have both been used to target the uncultured microbial dark matter, while SIP is useful in linking phylogeny to physiological function. Here we used both approaches in concert to first label the microbial community involved in the methane oxidation cycle with SIP, thus reducing the DNA diversity in the sample. This allowed a more robust and targeted approach for metagenome binning. Several high to

129 intermediate quality MAGs were recovered with this tactic, a few of which appeared to contain methane oxidation genes. The majority of the pmoA sequences based on amplicon sequencing in the SIP labeled DNA were related to the USCα Type II methanotrophs, though Type I methanotrophs were also detected. Majority of the MAGs contained genes for the serine cycle of assimilating formaldehyde, which is also indicative of Type II methanotrophs. An

Alphaproteobacterial MAG that contained many of the methane cycling genes including mmoX, and genes for serine cycle of assimilating formaldehyde, was also determined to contain genes for biopolymer production and mercury detoxification.

130

Figure 4.1. Methane gas flux and qPCR at the IWP terrain.

(A) Average CH4 gas measurements in the static chambers across the IWP terrain. (B) Quantitative PCR of the pmoA particulate methane monooxygenase gene in the trough (Tr) and polygon interior (PI) soils at 5 cm and 25 cm at the IWP terrain.

131

Figure 4.2. Visualization of the heavy and light bands. The heavy band is the 13C labelled DNA.

132

Figure 4.3. The microbial community composition. The community composition is based on 16S profiling in the heavy and light bands from the SIP CH4 enrichment at 100 ppm and 1000 ppm, as well as the composition of the control soils that were not enriched in CH4.

133

Figure 4.4.Beta-diversity measure of the microbial communities. PCoA Bray-Curtis analysis of the microbial community composition in the heavy and light bands from the SIP CH4 enrichment at 100 ppm and 1000 ppm, as well as the composition of the control soils that were not enriched in CH4.

134

Figure 4.5. The pmoA containing microbial community profiling in the heavy 13C labeled band from the SIP CH4 enrichment at 100 ppm and 1000 ppm.

135

Figure 4.6 Schematic of theoretical abiotic CH4 consumption

Table 4.1. Selection of high quality to intermediate MAGs from the 13C 100 ppm CH4-SIP enrichment. This MAG contained genes for the methane oxidation cycle, including a potential mmoX and genes for serine cycle of formaldehyde assimilation.

136

MetaBAT Completeness Contamination Lowest resolved classification based on Quality Bin AAI

6 79.3 0.9 Burkholderiales (p-value 0.332) High

8 80.2 3.6 Nitrosomonadaceae (p-value 0.365) High

15 88.3 0.9 Ramlibacter tataouinensis (p-value 0.463) Very High

16 53.2 2.7 Thiobacillus (p-value 0.293) Intermediate

20 30.6 0.9 Gemmatimonadales (p-value 0.449) Intermediate

21 62.2 4.5 Alphaproteobacteria (p-value 0.283) Intermediate

22 33.3 1.8 Betaproteobacteria (p-value 0.231) Intermediate

24 30.6 0.9 Burkholderiales (p-value 0.176) Intermediate

27 20.7 0 Alphaproteobacteria (p-value 0.259) Intermediate

137

Discussion and Conclusions

Permafrost affected soils account for ~27% of terrestrial environments and will be some of the most affected habitats due to anthropogenic climate change, with the thawing permafrost contributing to a positive feedback loop of GHG emissions. This thesis aimed to help understand the flux of the three major gases, CO2, CH4, and N2O, and to determine the microbial community responsible for the biogeochemical cycling of these gases in Arctic mineral cryosols.

Gas flux and microbial community at the IWP terrain

Trying to understand the future fate of Arctic soils and whether they will be sources or sinks of the three major greenhouse gases (GHG) is difficult and requires knowledge across a multitude of fields including, geology, biogeochemistry, soil science, microbiology, atmospheric science, hydrology, plant science, etc. The microbial contribution to the hypothesized positive feedback loop of GHG emissions is thought to play an important role in predicting future warming potential of the Arctic (Schuur et al. 2015). Mineral polar deserts are often overlooked when studding GHG emissions, even though they cover ~26% of ice-free land in the Arctic (Walker et al., 2002). These mineral cryosols soils appear to harbor an active microbial community of methanotrophs, nitrogen-fixing bacteria, denitrifiers, and organisms capable degrading a variety of C compounds stored in the soils (Chapters 2, 3, and 4). This has also been previously demonstrated in organic Arctic peat soils (Tveit et al. 2013; Coolen and Orsi 2015; Hultman et al.

2015b).

The topography of the Arctic cryosols appears to affect both GHG flux and the microbial community in the soils. Across the ice-wedge polygon (IWP) terrain, soils that exhibited

138

differential flux of CH4, CO2, and N2O also contain unique microbial assemblages of the overall microbial community (Chapter 2) and of specific microbial clades, such as nitrogen fixing bacteria, denitrifiers, and methanotrophs (Chapters 2 and 3). Overall the wetter soils in the troughs at the

IWP terrain appear to be greater sources of CO2 and NO2 gases, but a weaker sink of CH4 gas compared to drier polygon interior soils (Chapter 2 and 3). The overall microbial community in the troughs contained a higher proportion of Proteobacteria and Bacterioidetes compared to the polygon interior soils (Chapters 2). The denitrifying members community contained higher proportion of Rhodobacterales and Rhodocyclales, while the nitrogen-fixing members contained higher proportion of Desulfovibrionales, Gallionellales, and uncultured Desulfobacterales in the trough soils compared to polygon interior soils (Chapter 3). However, the unique nifH containing microbial communities between the different soils may not affect the nitrogen-fixing potential of the soil since there is lack of correlation between changes in the taxonomic composition of the nitrogen fixing community and nitrogenase activity in Arctic soils (Deslippe et al. 2005). This suggests that there is likely some level of functional redundancy in the nitrogen-fixing communities of the soils. However, this contrasts with another study showing links between unique denitrifier communities and differential N2O emissions from Arctic peat soils in Russia (Palmer et al. 2012).

Methanotrophic microorganisms in the soils were related to the USCα clade of potential high affinity methanotrophs (Chapter 2); this was confirmed with stable isotope probing (SIP) which

13 also identified pmoA genes of this clade through amplicon sequencing of the heavy CH4 labeled

DNA (Chapter 3) and has previously been reported in these soils (Lau et al. 2015). Unfortunately, metagenome binning from the heavily labeled DNA did not result in a metagenome assembled genome (MAG) containing the same pmoA related to the USCα clade. However, a potential

139 methanotroph MAG containing a mmoX gene coding for a soluble methane monooxygenase

(sMMO) was identified in the binning analysis (Alphaproteobacterial MAG #21). This potential organism also contained other genes in the methane metabolism pathway, including ones for the assimilation of formaldehyde via the serine pathway. This suggested that active methanotrophs at the IWP terrain are likely Type II methanotrophs, which is consistent with another recent genome binning study that identified another Type II methanotroph (Alphaproteobacterial MAG) in upland forest soils (Pratchscher et al. 2018). However, this contrasts with an earlier study that identified a Type I methanotroph (Gammaproteobacterial MAG) in Antarctic mineral soils (Edwards et al.

13 2017). The potential methanotrophic MAG identified via CH4 SIP in this thesis also contained genes for detoxification of mercury. This was consistent with what we know of mercury’s ability to irreversibly inhibit the sMMO enzyme (Contin et al. 2012) and interfere with methanotrophy rates (Boden and Murrell 2011).

Microbial interactions at the IWP terrain

In addition to being affected by the topography of the terrain, the hybrid-network analysis revealed that IWP microbial communities appear to also be self-regulating (i.e. feedback regulation of the community composition) (Chapter 2), with several key members affecting the relative abundance of other species. For example, members of the Proteobacteria, Candidatus

Rokubacteria, and Actinobacteria phyla tended to have a positive effect by increasing the relative abundance of other species, while members of the Verrucomicrobia and Acidobacteria tended to have a negative impact by decreasing the relative abundance of other species (Chapter 2).

140

It was interesting to note that very few positive microbial relationships appeared to be directly mutualistic in nature, where two species were both benefiting from the interaction (i.e. mutualism)

(Chapter 2). This implies that overall the microbial community functions together as a whole by sharing ‘public goods’ (e.g. metabolites, exoenzymes, siderophores) and influencing habitat (e.g. changing the pH of the ecosystem) (Xavier 2011; Harrington and Sanchez 2014; Braga et al. 2016), instead of having distinct and direct mutualistic relationships seen in higher eukaryote-eukaryote interactions and microorganisms-eukaryote interactions (Baumgarten et al. 2015; Zipfel and

Oldroyd 2017; Hembry et al. 2018). Therefore, network analyses of microbial systems can potentially be used to study evolution of cooperative behavior. Cooperative behavior increases the fitness of a (sub)population at a fitness cost to the individual (Hauert et al. 2006). Furthermore, sharing of resources is prone to individual exploitation of the resources (i.e. tragedy of the commons (Hardin 1968)). Due to these reasons, the benefit of engaging in cooperative behavior is often unclear, yet cooperative behavior is widespread in biological systems, including microbial ones (West et al. 2007).

One of the high-quality MAGs that was identified in the SIP enrichment (Chapter 4) belonged to the Ramlibacter genus; this organism was also shown to be part of the hybrid network analysis

(Chapter 2) and appears to be positively affected by a representative of Sideroxydans

(Betaproteobactria). Sideroxydans are iron-oxidizing lithoautotrophs (Beckwith et al. 2015), and also had a positive effect on the abundance on other β-γ-δ-Proteobacteria, Verrucomicrobia,

Bacterioidetes, and Acidobacteria. The lithoautotrophic Sideroxydans would potentially be providing a source of fixed organic carbon to support the heterotrophic growth of other microorganisms. Furthermore, Sideroxydans appear to have a suite of nitrogen fixation genes and

141 are likely contributing to bioavailable nitrogen in this overall nitrogen-poor system (Emerson et al. 2013).

Other nitrogen-fixing bacteria identified in the metatranscriptopme and amplicon sequencing analysis were also detected in the hybrid network analysis (Chapter 2 and 3). For example, members of Gallionellales tended to have a positive impact on each other and other microbial members, but in turn were negatively impacted by members of Verrucomicrobia. Members of

Rhyzobiales overall did not affect the abundance of other microbial members but were themselves negatively impacted by Acidobacterial members and positively affected by α-β-Proteobacterial and Candidatus Phylum Rokubacteria members (Chapter 2 and 3). Overall the hybrid network analysis in this thesis aimed to disentangle abiotic and biotic drivers affecting microbial species composition (Chapter 2). Many of the previous studies have relied on species co-occurrence to gauge the microbial interactions in ecosystems (Barberán et al. 2012; Berry and Widder 2014;

Williams et al. 2014). However, this approach is limited in distinguishing true species interactions from habitat filtering (i.e. niche overlap) (Freilich et al., 2018). Nevertheless, there are some patterns that are common across studies, including this one. For example, the Barberán et al. (2012) study also demonstrated that members of Actinobacteria and Verrucomicrobia did not tend to have an active positive effect on other species. The hybrid analysis in this thesis was then able to demonstrate that these two phyla actually have a negative effect on a multitude of other species

(Chapter 2).

The hybrid network analysis could also potentially identify keystone species that drive/support the overall composition of the microbial community. A keystone species is often defined as having a disproportionately larger effect on the abundance/composition of other species relative to its abundance (Berry and Widder 2014). For example, a microbe that excretes extracellular enzymes

142 that degrade recalcitrant carbon and support the growth of other organisms would likely be a keystone species (Berry and Widder 2014). It is possible to calculate the ‘degree of keystoneness’ of a species using a network of microbial interactions by considering the number of other species that are directly and indirectly influenced by that species (Berry and Widder 2014). For example, based on the hybrid network analysis performed with the organism at the IWP terrain, organisms that had a high degree of “keystoneness” (Berry and Widder 2014) included an uncultured member of Burkholderiaceae, uncultured Knoellia sp. (Actinobacteria/Micrococcales), Candidatus

Udaeobacter sp. (Verrucomicrobia), and uncultured Acidothermus sp.

(Actinobacteria/Frankiales), since they had high centrality in the network and many direct and indirect interactions with other species (Chapter 2).

Potential, alternative, non-biological methane uptake processes

The role of microbial methane oxidation is well established in reducing total methane emissions in wetland soils (Graef et al. 2011; Kip et al. 2011; Chowdhury and Dick 2013) and taking up methane in well drained soils (Kolb et al. 2003; Zhuang et al. 2004; Kolb et al. 2005;

Martineau et al. 2010; Lau et al. 2015). There is a possibility that this methane oxidation is not only biological in nature but is partially also abiotically mediated.

In the mid to upper troposphere, hydroxyl radicals (•OH) reacts with CH4 to generate •CH3, which eventually gets oxidized all the way to CO2. This initial reaction is often spontaneous and constitutes the largest global CH4 sink (~90%) (Kirschke et al. 2013). Hydroxyl radicals are highly reactive and unselectively oxidizing. The •OH radicals are produced in both freshwater and seawater and are generated via absorbance of UV light by dissolved organic carbon (DOM),

143 photolysis of nitrite and nitrate in natural waters, ligand-to-metal charge-transfer reactions and photo-Fenton chemistry in waters with high metal concentration (Mopper and Zhou 1990; Zellner et al. 1990; Vaughan and Blough 1998; Faust 1999). More recently, Arctic soils and surface waters have demonstrated the ability to generate •OH radicals in the dark (Page et al. 2013).

Environmental conditions switch between aerobic and anaerobic in soils experiencing cycles of wetting and drying (as during the freeze/thaw cycle). During the aerobic phase the •OH radicals form during oxidation of reduced dissolved organic matter (DOM) and iron (Rx. 1). Therefore,

•OH formation is coupled to oxidation of Fe(II) and DOMred (Page et al. 2013). Once anoxic conditions are reestablished, the Fe(III) and DOMox can become reduced and be replenished (Page et al. 2013) (Rx. 2). This could be mediated by iron-reducing bacteria, which inhabit Arctic soils, including the IWP terrain (Lipson et al. 2010; Stackhouse et al. 2015).

+2 +3 O2 + Fe /DOMred → •OH + Fe /DOMox (Aerobic phase) (Rx. 1)

+3 +2 Fe /DOMox → Fe /DOMred (Anaerobic phase via Fe(III) reducing bacteria) (Rx. 2)

Since the •OH radicals are extremely reactive, they can then oxidize recalcitrant organic matter and perhaps dissolved CH4 in the soils/pore water. The greatest •OH formation is predicted at oxic–anoxic boundaries in soil and surface waters (Page et al. 2013). It is possible that a similar process is happening at the IWP terrain between the permafrost/ice-wedge and the thawed active layer. Because permafrost reduces drainage of the active layer soils (Lawrence et al. 2015) (Lewis et al. 2012), the soils are relatively dry in the top 5 cm but have much higher moisture content closer to the permafrost table where the liquid water gets trapped and accumulates above the permafrost (Chapter 3). This scenario could also theoretically create conditions suitable for dark formation of •OH radicals. The •OH radical could potentially react with the CH4 dissolved in the soil pore water creating a •CH3 radical (Rx. 3) (Figure 5.1), it can also oxidize DOM and

144 recalcitrant carbon, recalcitrant organic matter (thereby increasing its bioavailability) (Rx. 4), or another hydroxyl radical (Rx. 5) (Rimmer 2006; Page et al. 2013). The •CH3 radical can then participate in a termination reaction with another •OH radical to produce methanol (Rx. 6), possibly react with O2 to produce formaldehyde (Rx. 7), or also react with DOM/recalcitrant carbon (Rx. 4) (Anastasi et al. 1991; Jasper et al. 2007)

•OH + CH4 → CH3• + H2O (Rx. 3)

•OH/•CH3 + DOM/recalcitrant carbon → Eventual low molecular weight acids (Rx. 4)

•OH + •OH → H2O2 (Rx. 5)

•CH3 + •OH → CH3-OH (Rx. 6)

•CH3 + O2 → HCHO + •OH (Rx. 7)

Therefore, it may be theoretically possible that abiotic processes, in addition to microbial metabolism, partially contribute to the oxidation of CH4 at the IWP terrain and in other permafrost affected areas. Through this abiotic process the oxidized CH4 could still end up in microbial biomass from metabolism of downstream products such as methanol (Figure 5.1). Based on the metatranscriptomic (Chapter 3) and SIP metagenomic (Chapter 4) analysis in this thesis, as well as previous studies, Arctic soils do contain genes for methanol and formaldehyde assimilation (He et al. 2012; Tveit et al. 2014; Deng et al. 2015). In addition, this same abiotic process would also contribute to degrading recalcitrant organic matter in the soils, thus allowing for further microbial degradation of soil organic matter by heterotrophic microorganisms (Page et al. 2013).

One way to check if the abiotic consumption of CH4 is occurring at the IWP terrain in tandem with biological oxidation, would be to first establish if •OH radicals are being formed in the soils using a similar methodology as Page et al. (2013). The authors incubated soils and quantified the

145 formation of •OH by measuring its reaction with a probe compound to a stable compound (Page et al. 2013). Since recycling of Fe(III) to Fe(II) is potentially mediated by iron-reducing bacteria and is necessary for •OH production (Figure 5.1), performing sterile microcosms controls to rule out abiotic methane consumption would not work since the Fe(III) reducing bacteria would also be eliminated. One way around this would be to have controls with sterile soil that is inoculated with Fe(III) reducing bacteria or by continuously adding Fe(II) to the soil.

Future directions and experimental improvements

There are three main avenues of future research that stem from this thesis. The first is a more comprehensive survey of GHG fluxes across the different terrains and soil types in the Arctic.

This would be beneficial in helping model future GHG emissions and Arctic warming potential.

The second is an improved SIP enrichment experimental design. One flaw with the SIP approach is that, given enough time, in addition to labeling of the target organism that can use the substrate of interest, the peripheral, non-targeted microbial community also gets labeled through syntropy, cross feeding, and using up metabolic waste products of the targeted organisms. This limitation could be exploited to study the flow of carbon through the microbial system. A SIP experiment designed with a time series sampling of the enrichment incubations would be able to differentiate between the organisms originally using up the targeted substrate and ones that are getting peripherally labeled. The organisms using the substrate of interest will be the first to be sufficiently labeled to allow detection and should show up in earlier time series samples. As the incubation progresses, other microbes would begin to be labeled starting with those in direct syntrophic relationship with the target organism. Given enough time, a wider community would be labeled as

146 the heavy carbon cycles through the system. Lastly, improving and validating the hybrid network analysis method is important for studying the contribution of biotic microbe-microbe interactions on the distribution and abundance of the non-culturable portion of the microbial community.

Models that can differentiate between direct and indirect, as well as three-way microbial interactions, would be the next step in improving the hybrid network analysis model presented in this thesis. Validating the hybrid network analysis is challenging since the majority of the taxa are not cultivable. Validation of the models may have to be done using higher eukaryotic organisms, where direct interactions are either known or could be observed (Freilich et al. 2018). Novel culturing techniques and the ability to form artificial microbial communities would allow for high- throughput co-culturing and analysis of different species combinations, allowing for model validation in microbial systems (Faust and Raes 2012).

147

Figure 5.1 Schematic of theoretical abiotic CH4 consumption

Abiotic methane consumption could theoretically take place at the sub-oxic boundary of the IWP terrain in the soils, or during permafrost thaw and cycles of wetting and drying. When O2 is available (solid arrows) •OH radicals could be produced through a dark reaction coupled to reduction of Fe(II) and DOMox (Page et al., 2013). The •OH radicals are extremely and unbiasedly oxidizing and would be able to oxidize recalcitrant carbon compounds in the soils,

DOM, and perhaps dissolved CH4 in the pore water. Once oxidized, the resulting methyl radical can react with another •OH radical to methanol, DOM, or recalcitrant carbon, and eventually end up in microbial biomass. During the anoxic phase (dashed arrow) the Fe(III) could be replenished via iron reducing bacteria present in Arctic soils (Stackhouse et al. 2015).

148

References

Albertsen M, Hugenholtz P, Skarshewski A, Nielsen KL, Tyson GW, Nielsen PH. 2013. Genome sequences of rare, uncultured bacteria obtained by differential coverage binning of multiple metagenomes. Nature Biotechnology 31: 533.

Allan J, Ronholm J, Mykytczuk N, Greer C, Onstott T, Whyte L. 2014. Methanogen community composition and rates of methane consumption in Canadian High Arctic permafrost soils. Environmental Microbiology Reports 6: 136-144.

Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. 1990. Basic local alignment search tool. Journal of Molecular Biology 215: 403-410.

Amato P, Doyle SM, Battista JR, Christner BC. 2010. Implications of subzero metabolic activity on long-term microbial survival in terrestrial and extraterrestrial permafrost. Astrobiology 10: 789-798.

Anastasi C, Beverton S, Ellermann T, Pagsberg P. 1991. Reaction of CH3 radicals with OH at room temperature and pressure. Journal of the Chemical Society, Faraday Transactions 87: 2325-2329.

Bae S, Wuertz S. 2009. Discrimination of viable and dead fecal Bacteroidales bacteria by quantitative PCR with propidium monoazide. Applied and Environmental Microbiology 75: 2940-2944.

Bai Y, Yang D, Wang J, Xu S, Wang X, An L. 2006. Phylogenetic diversity of culturable bacteria from alpine permafrost in the Tianshan Mountains, northwestern China. Research in Microbiology 157: 741-751.

Bakermans C, Nealson KH. 2004. Relationship of critical temperature to macromolecular synthesis and growth yield in Psychrobacter cryopegella. Journal of Bacteriology 186: 2340-2345.

Bakermans C, Skidmore ML, Douglas S, McKay CP. 2014. Molecular characterization of bacteria from permafrost of the Taylor Valley, Antarctica. FEMS Microbiology Ecology 89: 331-346.

Bakermans C, Tsapin AI, Souza‐Egipsy V, Gilichinsky DA, Nealson KH. 2003. Reproduction and metabolism at− 10 C of bacteria isolated from Siberian permafrost. Environmental Microbiology 5: 321-326.

Barberán A, Bates ST, Casamayor EO, Fierer N. 2012. Using network analysis to explore co- occurrence patterns in soil microbial communities. The ISME Journal 6: 343.

Barbier BA, Dziduch I, Liebner S, Ganzert L, Lantuit H, Pollard W, Wagner D. 2012. Methane- cycling communities in a permafrost-affected soil on Herschel Island, Western Canadian Arctic: active layer profiling of mcrA and pmoA genes. FEMS Microbiology Ecology 82: 287-302.

149

Baumgarten S, Simakov O, Esherick LY, Liew YJ, Lehnert EM, Michell CT, Li Y, Hambleton EA, Guse A, Oates ME. 2015. The genome of Aiptasia, a sea anemone model for coral symbiosis. Proceedings of the National Academy of Sciences 112: 11893-11898.

Beckwith CR, Edwards MJ, Lawes M, Shi L, Butt JN, Richardson DJ, Clarke TA. 2015. Characterization of MtoD from Sideroxydans lithotrophicus: a cytochrome c electron shuttle used in lithoautotrophic growth. Frontiers in Microbiology 6: 332.

Becraft ED, Dodsworth JA, Murugapiran SK, Ohlsson JI, Briggs BR, Kanbar J, De Vlaminck I, Quake SR, Dong H, Hedlund BP. 2015. Single-cell genomics-facilitated read-first binning of candidate phylum EM19 genomes from geothermal spring metagenomes. Applied and Environmental Microbiology 82: 992-1003

Bellemain E, Davey ML, Kauserud H, Epp LS, Boessenkool S, Coissac E, Geml J, Edwards M, Willerslev E, Gussarova G. 2013. Fungal palaeodiversity revealed using high‐throughput metabarcoding of ancient DNA from arctic permafrost. Environmental Microbiology 15: 1176-1189.

Berry D, Widder S. 2014. Deciphering microbial interactions and detecting keystone species with co-occurrence networks. Frontiers in Microbiology 5: 219.

Blanco Y, Prieto-Ballesteros O, Gómez MJ, Moreno-Paz M, García-Villadangos M, Rodríguez- Manfredi JA, Cruz-Gil P, Sánchez-Román M, Rivas LA, Parro V. 2012. Prokaryotic communities and operating metabolisms in the surface and the permafrost of Deception Island (Antarctica). Environmental Microbiology 14: 2495-2510

Blazewicz SJ, Barnard RL, Daly RA, Firestone MK. 2013. Evaluating rRNA as an indicator of microbial activity in environmental communities: limitations and uses. The ISME Journal 7: 2061-2068.

Bockheim JG, Campbell IB, McLeod M. 2007. Permafrost distribution and active‐layer depths in the McMurdo Dry valleys, Antarctica. Permafrost and Periglacial Processes 18: 217- 227.

Bockheim JG, Hall KJ. 2002. Permafrost, active-layer dynamics and periglacial environments of continental Antarctica: Periglacial and Permafrost Research in the Southern Hemisphere. South African Journal of Science 98: 82-90.

Bockheim JG, Munroe JS. 2014. Organic carbon pools and genesis of alpine soils with permafrost: a review. Arctic, Antarctic, and Alpine Research 46: 987-1006.

Boden R, Murrell JC. 2011. Response to mercury (II) ions in Methylococcus capsulatus (Bath). FEMS Microbiology Letters 324: 106-110.

Braga RM, Dourado MN, Araújo WL. 2016. Microbial interactions: ecology in a molecular perspective. Brazilian Journal of Microbiology 47: 86-98.

150

Briggs DE, Summons RE. 2014. Ancient biomolecules: their origins, fossilization, and role in revealing the history of life. BioEssays 36: 482-490.

Brown CT, Hug LA, Thomas BC, Sharon I, Castelle CJ, Singh A, Wilkins MJ, Wrighton KC, Williams KH, Banfield JF. 2015. Unusual biology across a group comprising more than 15% of domain Bacteria. Nature 523: 208.

Brummell ME, Farrell RE, Hardy SP, Siciliano SD. 2014. Greenhouse gas production and consumption in High Arctic deserts. Soil Biology and Biochemistry 68: 158-165.

Buelow HN, Winter AS, Van Horn DJ, Barrett JE, Gooseff MN, Schwartz E, Takacs-Vesbach CD. 2016. Microbial Community Responses to Increased Water and Organic Matter in the Arid Soils of the McMurdo Dry Valleys, Antarctica. Frontiers in Microbiology 7: 1040

Butterbach-Bahl K, Baggs EM, Dannenmann M, Kiese R, Zechmeister-Boltenstern S. 2013. Nitrous oxide emissions from soils: how well do we understand the processes and their controls? Philosophical Transactions of the Royal Society B 368: 20130122.

Cai Y, Zheng Y, Bodelier PL, Conrad R, Jia Z. 2016. Conventional methanotrophs are responsible for atmospheric methane oxidation in paddy soils. Nature Communications 7: 11728.

Campbell IB, Claridge GG. 2006. Permafrost properties, patterns and processes in the Transantarctic Mountains region. Permafrost and Periglacial Processes 17: 215-232.

Campbell IB, Claridge GG. 2009. Antarctic permafrost soils. In Permafrost Soils, (ed. R Margesin), pp. 17-31. Springer.

Carini P, Marsden PJ, Leff JW, Morgan EE, Strickland MS, Fierer N. 2016. Relic DNA is abundant in soil and obscures estimates of soil microbial diversity. bioRxiv: 043372.

Chapin DM, Bliss L, Bledsoe L. 1991. Environmental regulation of nitrogen fixation in a high arctic lowland ecosystem. Canadian Journal of Botany 69: 2744-2755.

Chen L, Liang J, Qin S, Liu L, Fang K, Xu Y, Ding J, Li F, Luo Y, Yang Y. 2016. Determinants of carbon release from the active layer and permafrost deposits on the Tibetan Plateau. Nature Communications 7: 13046.

Chowdhury TR, Dick RP. 2013. Ecology of aerobic methanotrophs in controlling methane fluxes from wetlands. Applied Soil Ecology 65: 8-22.

Christiansen JR, Romero AJB, Jørgensen NO, Glaring MA, Jørgensen CJ, Berg LK, Elberling B. 2015. Methane fluxes and the functional groups of methanotrophs and methanogens in a young Arctic landscape on Disko Island, West Greenland. Biogeochemistry 122: 15-33.

151

Collier SM, Ruark MD, Oates LG, Jokela WE, Dell CJ. 2014. Measurement of greenhouse gas flux from agricultural soils using static chambers. Journal of Visualized Experiments: JoVE 90: e52110.

Conrad R. 2007. Microbial Ecology of Methanogens and Methanotrophs. Advances in Agronomy 96: 1-63.

Contin M, Rizzardini CB, Catalano L, De Nobili M. 2012. Contamination by mercury affects methane oxidation capacity of aerobic arable soils. Geoderma 189: 250-256.

Coolen MJ, Orsi WD. 2015. The transcriptional response of microbial communities in thawing Alaskan permafrost soils. Frontiers in Microbiology 6: 197.

D'amico S, Collins T, Marx JC, Feller G, Gerday C. 2006. Psychrophilic microorganisms: challenges for life. EMBO reports 7: 385-389.

Dalal RC, Allen DE. 2008. Greenhouse gas fluxes from natural ecosystems. Australian Journal of Botany 56: 369-407.

Davidson EA, Kanter D. 2014. Inventories and scenarios of nitrous oxide emissions. Environmental Research Letters 9: 105012.

De Maayer P, Anderson D, Cary C, Cowan DA. 2014. Some like it cold: understanding the survival strategies of psychrophiles. EMBO reports: e201338170.

Deng J, Gu Y, Zhang J, Xue K, Qin Y, Yuan M, Yin H, He Z, Wu L, Schuur EA. 2015. Shifts of tundra bacterial and archaeal communities along a permafrost thaw gradient in Alaska. Molecular Ecology 24: 222-234.

Deslippe JR, Egger KN, Henry GH. 2005. Impacts of warming and fertilization on nitrogen- fixing microbial communities in the Canadian High Arctic. FEMS microbiology ecology 53: 41-50.

Doxey AC, Mansfield MJ, Montecucco C. 2018. Discovery of novel bacterial toxins by genomics and computational biology. Toxicon 147: 2-12.

Edwards CR, Onstott TC, Miller JM, Wiggins JB, Wang W, Lee CK, Cary SC, Pointing SB, Lau MC. 2017. Draft genome sequence of uncultured upland soil cluster Gammaproteobacteria gives molecular insights into high-affinity methanotrophy. Genome Announcements 5: e00047-00017.

Elberling B, Brandt KK. 2003. Uncoupling of microbial CO2 production and release in frozen soil and its implications for field studies of arctic C cycling. Soil Biology and Biochemistry 35: 263-272.

Elberling B, Christiansen HH, Hansen BU. 2010. High nitrous oxide production from thawing permafrost. Nature Geoscience 3: 332-335.

152

Emerson D, Field E, Chertkov O, Davenport K, Goodwin L, Munk C, Nolan M, Woyke T. 2013. Comparative genomics of freshwater Fe-oxidizing bacteria: implications for physiology, ecology, and systematics. Frontiers in Microbiology 4: 254.

Emmerton CA, St Louis V, Lehnherr I, Humphreys ER, Rydz E, Kosolofski HR. 2014. The net exchange of methane with high Arctic landscapes during the summer growing season. Biogeosciences 11: 3095-3106.

Eriksson M, Ka J-O, Mohn WW. 2001. Effects of low temperature and freeze-thaw cycles on hydrocarbon biodegradation in Arctic tundra soil. Applied and Environmental Microbiology 67: 5107-5112.

Ernakovich JG, Wallenstein MD. 2015. Permafrost microbial community traits and functional diversity indicate low activity at in situ thaw temperatures. Soil Biology and Biochemistry 87: 78-89.

Esson KC, Lin X, Kumaresan D, Chanton JP, Murrell JC, Kostka JE. 2016. Alpha-and gammaproteobacterial methanotrophs codominate the active methane-oxidizing communities in an acidic boreal peat bog. Applied and Environmental Microbiology 82: 2363-2371.

Etzelmüller B. 2013. Recent advances in mountain permafrost research. Permafrost and Periglacial Processes 24: 99-107.

Fahnestock JT, Jones MH, Welker JM. 1999. Wintertime CO2 efflux from arctic soils: implications for annual carbon budgets. Global Biogeochemical Cycles 13: 775-779.

Faust B. 1999. Aquatic photochemical reactions in atmospheric, surface, and marine waters: influences on oxidant formation and pollutant degradation. In Environmental Photochemistry (Ed Pierre Boule), pp. 101-122. Springer.

Faust K, Raes J. 2012. Microbial interactions: from networks to models. Nature Reviews Microbiology 10: 538.

Feller G. 2007. Life at low temperatures: is disorder the driving force? Extremophiles 11: 211- 216.

Finger RA, Turetsky MR, Kielland K, Ruess RW, Mack MC, Euskirchen ES. 2016. Effects of permafrost thaw on nitrogen availability and plant–soil interactions in a boreal Alaskan lowland. Journal of Ecology 104: 1542-1554.

Finster KW, Herbert RA, Kjeldsen KU, Schumann P, Lomstein BA. 2009. Demequina lutea sp. nov., isolated from a high Arctic permafrost soil. International Journal of Systematic and Evolutionary Microbiology 59: 649-653.

Fish JA, Chai B, Wang Q, Sun Y, Brown CT, Tiedje JM, Cole JR. 2013. FunGene: the functional gene pipeline and repository. Frontiers in Microbiology 4: 291.

153

Freilich MA, Wieters E, Broitman BR, Marquet PA, Navarrete SA. 2018. Species co‐occurrence networks: Can they reveal trophic and non‐trophic interactions in ecological communities? Ecology 99: 690-699.

Frey B, Rime T, Phillips M, Stierli B, Hajdas I, Widmer F, Hartmann M. 2016. Microbial diversity in European alpine permafrost and active layers. FEMS Microbiology Ecology 92: fiw018.

Gaby JC, Buckley DH. 2012. A comprehensive evaluation of PCR primers to amplify the nifH gene of nitrogenase. PloS one 7: e42149.

Ganzert L, Jurgens G, Münster U, Wagner D. 2007. Methanogenic communities in permafrost- affected soils of the Laptev Sea coast, Siberian Arctic, characterized by 16S rRNA gene fingerprints. FEMS Microbiology Ecology 59: 476-488.

Gies EA, Konwar KM, Beatty JT, Hallam SJ. 2014. Illuminating microbial dark matter in meromictic Sakinaw Lake. Applied and Environmental Microbiology 80: 6807-6818.

Gilichinsky D, Rivkina E, Shcherbakova V, Laurinavichuis K, Tiedje J. 2003. Supercooled water brines within permafrost-an unknown ecological niche for microorganisms: a model for astrobiology. Astrobiology 3: 331-341.

Gilichinsky D, Wagener S, Vishnevetskaya T. 1995. Permafrost microbiology. Permafrost and Periglacial Processes 6: 281-291.

Gilichinsky D, Wilson G, Friedmann E, McKay C, Sletten R, Rivkina E, Vishnivetskaya T, Erokhina L, Ivanushkina N, Kochkina G. 2007. Microbial populations in Antarctic permafrost: biodiversity, state, age, and implication for astrobiology. Astrobiology 7: 275-311.

Gittel A, Bárta J, Kohoutová I, Mikutta R, Owens S, Gilbert J, Schnecker J, Wild B, Hannisdal B, Maerz J. 2014a. Distinct microbial communities associated with buried soils in the Siberian tundra. The ISME Journal 8: 841-853.

Gittel A, Bárta J, Kohoutová I, Schnecker J, Wild B, Čapek P, Kaiser C, Torsvik VL, Richter A, Schleper C. 2014b. Site-and horizon-specific patterns of microbial community structure and enzyme activities in permafrost-affected soils of Greenland. Frontiers in Microbiology 5: 541.

Goordial J, Davila A, Lacelle D, Pollard W, Marinova MM, Greer CW, DiRuggiero J, McKay CP, Whyte LG. 2016. Nearing the cold-arid limits of microbial life in permafrost of an upper dry valley, Antarctica. The ISME Journal 10: 1613-1624.

Goordial J, Lamarche-Gagnon G, Lay C-Y, Whyte L. 2013. Left out in the cold: life in cryoenvironments. In Polyextremophiles, pp. 335-363. Springer.

Goordial J, Whyte L. 2014. Microbial life in Antarctic permafrost environments. In Antarctic Terrestrial Microbiology (Ed. Don A. Cowan), pp. 217-232. Springer.

154

Graef C, Hestnes AG, Svenning MM, Frenzel P. 2011. The active methanotrophic community in a wetland from the High Arctic. Environmental Microbiology Reports 3: 466-472.

Graham DE, Wallenstein MD, Vishnivetskaya TA, Waldrop MP, Phelps TJ, Pfiffner SM, Onstott TC, Whyte LG, Rivkina EM, Gilichinsky DA. 2012. Microbes in thawing permafrost: the unknown variable in the climate change equation. The ISME Journal 6: 709-712.

Graversen RG, Mauritsen T, Tjernström M, Källén E, Svensson G. 2008. Vertical structure of recent Arctic warming. Nature 451: 53-56.

Gupta V, Smemo KA, Yavitt JB, Basiliko N. 2012. Active methanotrophs in two contrasting North American peatland ecosystems revealed using DNA-SIP. Microbial Ecology 63: 438-445.

Guy PL. 2014. Prospects for analyzing ancient RNA in preserved materials. Wiley Interdisciplinary Reviews: RNA 5: 87-94.

Haeberli W, Gruber S. 2009. Global warming and mountain permafrost. In Permafrost Soils, (ed. R Margesin), pp. 205-218. Springer.

Hansen AA, Herbert RA, Mikkelsen K, Jensen LL, Kristoffersen T, Tiedje JM, Lomstein BA, Finster KW. 2007. Viability, diversity and composition of the bacterial community in a high Arctic permafrost soil from Spitsbergen, Northern Norway. Environmental Microbiology 9: 2870-2884.

Hansen AJ, Mitchell DL, Wiuf C, Paniker L, Brand TB, Binladen J, Gilichinsky DA, Rønn R, Willerslev E. 2006. Crosslinks rather than strand breaks determine access to ancient DNA sequences from frozen sediments. Genetics 173: 1175-1179.

Hanson RS, Hanson TE. 1996. Methanotrophic bacteria. Microbiological Reviews 60: 439-471.

Hardin G. 1968. The tragedy of the commons. Science 162: 1243-1248.

Harrington KI, Sanchez A. 2014. Eco-evolutionary dynamics of complex social strategies in microbial communities. Communicative & Integrative Biology 7: e28230.

Hauert C, Holmes M, Doebeli M. 2006. Evolutionary games and population dynamics: maintenance of cooperation in public goods games. Proceedings of the Royal Society of London B: Biological Sciences 273: 2565-2571.

He R, Wooller MJ, Pohlman JW, Catranis C, Quensen J, Tiedje JM, Leigh MB. 2012. Identification of functionally active aerobic methanotrophs in sediments from an arctic lake using stable isotope probing. Environmental Microbiology 14: 1403-1419.

Hebsgaard MB, Willerslev E. 2009. Very old DNA. In Permafrost Soils, (ed. R Margesin), pp. 47-57. Springer.

155

Hembry DH, Raimundo RL, Newman EA, Atkinson L, Guo C, Guimarães Jr PR, Gillespie RG. 2018. Does biological intimacy shape ecological network structure? A test using a brood pollination mutualism on continental and oceanic islands. Journal of Animal Ecology 87: 1160-1171.

Hobara S, McCalley C, Koba K, Giblin AE, Weiss MS, Gettel GM, Shaver GR. 2006. Nitrogen fixation in surface soils and vegetation in an Arctic tundra watershed: a key source of atmospheric nitrogen. Arctic, Antarctic, and Alpine Research 38: 363-372.

Hu W, Zhang Q, Tian T, Cheng G, An L, Feng H. 2015. The microbial diversity, distribution, and ecology of permafrost in China: a review. Extremophiles 19: 693-705.

Hu W, Zhang Q, Tian T, Li D, Cheng G, Mu J, Wu Q, Niu F, An L, Feng H. 2016. Characterization of the prokaryotic diversity through a stratigraphic permafrost core profile from the Qinghai-Tibet Plateau. Extremophiles 20: 337-349.

Hugelius G, Kuhry P. 2009. Landscape partitioning and environmental gradient analyses of soil organic carbon in a permafrost environment. Global Biogeochemical Cycles 23: DOI: 10.1029/2008GB003419

Hugelius G, Strauss J, Zubrzycki S, Harden JW, Schuur E, Ping C-L, Schirrmeister L, Grosse G, Michaelson GJ, Koven CD. 2014. Improved estimates show large circumpolar stocks of permafrost carbon while quantifying substantial uncertainty ranges and identifying remaining data gaps. Biogeosciences Discussions 11: 4771-4822.

Hultman J, Waldrop MP, Mackelprang R, David MM, McFarland J, Blazewicz SJ, Harden J, Turetsky MR, McGuire AD, Shah MB. 2015a. Multi-omics of permafrost, active layer and thermokarst bog soil microbiomes. Nature 521: 208-212.

Hultman J, Waldrop MP, Mackelprang R, David MM, McFarland J, Blazewicz SJ, Harden J, Turetsky MR, McGuire AD, Shah MB. 2015b. Multi-omics of permafrost, active layer and thermokarst bog soil microbiomes. Nature 521: 208-212.

Izquierdo JA, Nüsslein K. 2006. Distribution of extensive nifH gene diversity across physical soil microenvironments. Microbial Ecology 51: 441-452.

Jansson JK, Taş N. 2014. The microbial ecology of permafrost. Nature Reviews Microbiology 12: 414-425.

Jasper AW, Klippenstein SJ, Harding LB, Ruscic B. 2007. Kinetics of the reaction of methyl radical with hydroxyl radical and methanol decomposition. The Journal of Physical Chemistry A 111: 3932-3950.

Johansson T, Malmer N, Crill PM, Friborg T, Aakerman JH, Mastepanov M, Christensen TR. 2006. Decadal vegetation changes in a northern peatland, greenhouse gas fluxes and net radiative forcing. Global Change Biology 12: 2352-2369.

156

Johnson SS, Hebsgaard MB, Christensen TR, Mastepanov M, Nielsen R, Munch K, Brand T, Gilbert MTP, Zuber MT, Bunce M. 2007. Ancient bacteria show evidence of DNA repair. Proceedings of the National Academy of Sciences 104: 14401-14405.

Kandeler E, Deiglmayr K, Tscherko D, Bru D, Philippot L. 2006. Abundance of narG, nirS, nirK, and nosZ genes of denitrifying bacteria during primary successions of a glacier foreland. Applied and Environmental Microbiology 72: 5957-5962.

Katayama T, Tanaka M, Moriizumi J, Nakamura T, Brouchkov A, Douglas TA, Fukuda M, Tomita F, Asano K. 2007. Phylogenetic analysis of bacteria preserved in a permafrost ice wedge for 25,000 years. Applied and Environmental Microbiology 73: 2360-2363.

Kemp MJ, Dodds WK. 2002. The influence of ammonium, nitrate, and dissolved oxygen concentrations on uptake, nitrification, and denitrification rates associated with prairie stream substrata. Limnology and Oceanography 47: 1380-1393.

Kerfoot DE. 1972. Thermal contraction cracks in an arctic tundra environment. Arctic: 142-150.

Kim SJ, Shin SC, Hong SG, Lee YM, Choi I-G, Park H. 2012. Genome sequence of a novel member of the genus Psychrobacter isolated from Antarctic soil. Journal of Bacteriology 194: 2403-2403.

Kip N, Fritz C, Langelaan E, Pan Y, Bodrossy L, Pancotto V, Jetten M, Smolders A, Camp H. 2011. Methanotrophic activity and diversity in different Sphagnum magellanicum dominated habitats in the southernmost peat bogs of Patagonia. Biogeosciences 9: 47-55.

Kirschke S, Bousquet P, Ciais P, Saunois M, Canadell JG, Dlugokencky EJ, Bergamaschi P, Bergmann D, Blake DR, Bruhwiler L. 2013. Three decades of global methane sources and sinks. Nature Geoscience 6: 813.

Knief C, Lipski A, Dunfield PF. 2003. Diversity and activity of methanotrophic bacteria in different upland soils. Applied and Environmental Microbiology 69: 6703-6714.

Knittel K, Boetius A. 2009. Anaerobic oxidation of methane: progress with an unknown process. Annual Review of Microbiology 63: 311-334.

Koh HY, Park H, Lee JH, Han SJ, Sohn YC, Lee SG. 2016. Proteomic and transcriptomic investigations on cold‐responsive properties of the psychrophilic Antarctic bacterium Psychrobacter sp. PAMC 21119 at subzero temperatures. Environmental microbiology 19: 628-644.

Kokelj S, Burn C. 2005. Geochemistry of the active layer and near-surface permafrost, Mackenzie delta region, Northwest Territories, Canada. Canadian Journal of Earth Sciences 42: 37-48.

Kolb S, Knief C, Dunfield PF, Conrad R. 2005. Abundance and activity of uncultured methanotrophic bacteria involved in the consumption of atmospheric methane in two forest soils. Environmental Microbiology 7: 1150-1161.

157

Kolb S, Knief C, Stubner S, Conrad R. 2003. Quantitative detection of methanotrophs in soil by novel pmoA-targeted real-time PCR assays. Applied and Environmental Microbiology 69: 2423-2429.

Kumar S, Stecher G, Tamura K. 2016. MEGA7: Molecular Evolutionary Genetics Analysis version 7.0 for bigger datasets. Molecular Biology and Evolution 33: 1870-1874.

Lacelle D, Radtke K, Clark ID, Fisher D, Lauriol B, Utting N, Whyte LG. 2011. Geomicrobiology and occluded O 2–CO 2–Ar gas analyses provide evidence of microbial respiration in ancient terrestrial ground ice. Earth and Planetary Science Letters 306: 46-54.

Lamb EG, Han S, Lanoil BD, Henry GH, Brummell ME, Banerjee S, Siciliano SD. 2011. A High Arctic soil ecosystem resists long‐term environmental manipulations. Global Change Biology 17: 3187-3194.

Larsen KS, Jonasson S, Michelsen A. 2002. Repeated freeze–thaw cycles and their effects on biological processes in two arctic ecosystem types. Applied Soil Ecology 21: 187-195.

Lau M, Stackhouse B, Layton A, Chauhan A, Vishnivetskaya T, Chourey K, Ronholm J, Mykytczuk N, Bennett P, Lamarche-Gagnon G. 2015. An active atmospheric methane sink in high Arctic mineral cryosols. The ISME Journal 9: 1880–1891.

Lawrence DM, Koven C, Swenson SC, Riley W, Slater A. 2015. Permafrost thaw and resulting soil moisture changes regulate projected high-latitude CO2 and CH4 emissions. Environmental Research Letters 10: 094011.

Lawrence DM, Slater AG, Romanovsky VE, Nicolsky DJ. 2008. Sensitivity of a model projection of near‐surface permafrost degradation to soil column depth and representation of soil organic matter. Journal of Geophysical Research: Earth Surface (2003–2012) 113: DOI: 10.1029/2007JF000883.

Lewis K, Zyvoloski G, Travis B, Wilson C, Rowland J. 2012. Drainage subsidence associated with Arctic permafrost degradation. Journal of Geophysical Research: Earth Surface 117.

Li F, Zhu R, Bao T, Wang Q, Xu H. 2016. Sunlight stimulates methane uptake and nitrous oxide emission from the High Arctic tundra. Science of The Total Environment 572: 1150-1160.

Liengen T, Olsen RA. 1997. Nitrogen fixation by free-living cyanobacteria from different coastal sites in a high arctic tundra, Spitsbergen. Arctic and Alpine Research 29: 470-477.

Lindahl T. 1993. Instability and decay of the primary structure of DNA. Nature 362: 709-715.

Lipson DA, Jha M, Raab TK, Oechel WC. 2010. Reduction of iron (III) and humic substances plays a major role in anaerobic respiration in an Arctic peat soil. Journal of Geophysical Research: Biogeosciences 115: DOI: 10.1029/2009JG001147

158

Ma WK, Schautz A, Fishback L-AE, Bedard-Haughn A, Farrell RE, Siciliano SD. 2007. Assessing the potential of ammonia oxidizing bacteria to produce nitrous oxide in soils of a high arctic lowland ecosystem on Devon Island, Canada. Soil Biology and Biochemistry 39: 2001-2013.

Mackelprang R, Saleska SR, Jacobsen CS, Jansson JK, Taş N. 2016. Permafrost Meta-Omics and Climate Change. Annual Review of Earth and Planetary Sciences 44: 439-462.

Mackelprang R, Waldrop MP, DeAngelis KM, David MM, Chavarria KL, Blazewicz SJ, Rubin EM, Jansson JK. 2011. Metagenomic analysis of a permafrost microbial community reveals a rapid response to thaw. Nature 480: 368-371.

Margesin R, Miteva V. 2011. Diversity and ecology of psychrophilic microorganisms. Research in Microbiology 162: 346-361.

Marinova MM, Mckay CP, Pollard WH, Heldmann JL, Davila AF, Andersen DT, Jackson WA, Lacelle D, Paulsen G, Zacny K. 2013. Distribution of depth to ice-cemented soils in the high-elevation Quartermain Mountains, McMurdo Dry Valleys, Antarctica. Antarctic Science 25: 575-582.

Martineau C, Pan Y, Bodrossy L, Yergeau E, Whyte LG, Greer CW. 2014. Atmospheric methane oxidizers are present and active in Canadian high Arctic soils. FEMS Microbiology Ecology 89: 257-269.

Martineau C, Whyte LG, Greer CW. 2010. Stable isotope probing analysis of the diversity and activity of methanotrophic bacteria in soils from the Canadian High Arctic. Applied and Environmental Microbiology 76: 5773-5784.

Marushchak M, Pitkämäki A, Koponen H, Biasi C, Seppälä M, Martikainen P. 2011. Hot spots for nitrous oxide emissions found in different types of permafrost peatlands. Global Change Biology 17: 2601-2614.

McCalley CK, Woodcroft BJ, Hodgkins SB, Wehr RA, Kim E-H, Mondav R, Crill PM, Chanton JP, Rich VI, Tyson GW. 2014. Methane dynamics regulated by microbial community response to permafrost thaw. Nature 514: 478-481.

McCann CM, Wade MJ, Gray ND, Roberts JA, Hubert CR, Graham DW. 2016. Microbial communities in a High Arctic polar desert landscape. Frontiers in Microbiology 7: 419

McCutcheon JP, Moran NA. 2012. Extreme genome reduction in symbiotic bacteria. Nature Reviews Microbiology 10: 13.

Meyer F, Paarmann D, D'Souza M, Olson R, Glass EM, Kubal M, Paczian T, Rodriguez A, Stevens R, Wilke A. 2008. The metagenomics RAST server–a public resource for the automatic phylogenetic and functional analysis of metagenomes. BMC Bioinformatics 9: 386.

159

Michaelson G, Ping C. 2003. Soil organic carbon and CO2 respiration at subzero temperature in soils of Arctic Alaska. Journal of Geophysical Research: Atmospheres 108: DOI: 10.1029/2001JD000920.

Mondav R, Woodcroft BJ, Kim E-H, McCalley CK, Hodgkins SB, Crill PM, Chanton J, Hurst GB, VerBerkmoes NC, Saleska SR. 2014. Discovery of a novel methanogen prevalent in thawing permafrost. Nature Communications 5: 3212.

Mopper K, Zhou X. 1990. Hydroxyl radical photoproduction in the sea and its potential impact on marine processes. Science 250: 661-664.

Morris BE, Henneberger R, Huber H, Moissl-Eichinger C. 2013. Microbial syntrophy: interaction for the common good. FEMS Microbiology Reviews 37: 384-406.

Mu C, Zhang T, Zhang X, Cao B, Peng X. 2016. Sensitivity of soil organic matter decomposition to temperature at different depths in permafrost regions on the northern Qinghai‐Tibet Plateau. European Journal of Soil Science 67: DOI: 10.1111/ejss.12386.

Mykytczuk NC, Foote SJ, Omelon CR, Southam G, Greer CW, Whyte LG. 2013. Bacterial growth at− 15 C; molecular insights from the permafrost bacterium Planococcus halocryophilus Or1. The ISME Journal 7: 1211-1226.

Nadkarni MA, Martin FE, Jacques NA, Hunter N. 2002. Determination of bacterial load by real- time PCR using a broad-range (universal) probe and primers set. Microbiology 148: 257- 266.

Nikrad MP, Kerkhof LJ, Häggblom MM. 2016. The subzero microbiome: Microbial activity in frozen and thawing soils. FEMS Microbiology Ecology 92: fiw081.

Ohtsubo Y, Kishida K, Sato T, Tabata M, Kawasumi T, Ogura Y, Hayashi T, Tsuda M, Nagata Y. 2014. Complete genome sequence of Pseudomonas sp. strain TKP, isolated from a γ- hexachlorocyclohexane-degrading mixed culture. Genome Announcements 2: e01241- 01213.

Öquist MG, Sparrman T, Klemedtsson L, Drotz SH, Grip H, Schleucher J, Nilsson M. 2009. Water availability controls microbial temperature responses in frozen soil CO2 production. Global Change Biology 15: 2715-2722.

Page SE, Kling GW, Sander M, Harrold KH, Logan JR, McNeill K, Cory RM. 2013. Dark formation of hydroxyl radical in Arctic soil and surface waters. Environmental Science & Technology 47: 12860-12867.

Palmer K, Biasi C, Horn MA. 2012. Contrasting denitrifier communities relate to contrasting N2O emission patterns from acidic peat soils in arctic tundra. The ISME Journal 6: 1058- 1077.

Palmer K, Horn MA. 2015. Denitrification activity of a remarkably diverse fen denitrifier community in Finnish Lapland is N-oxide limited. PloS one 10: e0123123.

160

Panikov N, Flanagan P, Oechel W, Mastepanov M, Christensen T. 2006. Microbial activity in soils frozen to below− 39 C. Soil Biology and Biochemistry 38: 785-794.

Panikov NS, Dedysh S. 2000. Cold season CH4 and CO2 emission from boreal peat bogs (West Siberia): Winter fluxes and thaw activation dynamics. Global Biogeochemical Cycles 14: 1071-1080.

Panikov NS, Sizova MV. 2007. Growth kinetics of microorganisms isolated from Alaskan soil and permafrost in solid media frozen down to− 35 C. FEMS Microbiology Ecology 59: 500-512.

Pautler BG, Simpson AJ, Mcnally DJ, Lamoureux SF, Simpson MJ. 2010. Arctic permafrost active layer detachments stimulate microbial activity and degradation of soil organic matter. Environmental Science & Technology 44: 4076-4082.

Pearson WR. 2013. An introduction to sequence similarity (“homology”) searching. Current Protocols in Bioinformatics 42: 3.1. 1-3.1. 8.

Pietramellara G, Ascher J, Borgogni F, Ceccherini M, Guerri G, Nannipieri P. 2009. Extracellular DNA in soil and sediment: fate and ecological relevance. Biology and Fertility of Soils 45: 219-235.

Poinar HN, Hoss M, Bada JL, Paabo S. 1996. Amino acid racemization and the preservation of ancient DNA. Science 272: 864.

Ponder MA, Gilmour SJ, Bergholz PW, Mindock CA, Hollingsworth R, Thomashow MF, Tiedje JM. 2005. Characterization of potential stress responses in ancient Siberian permafrost psychroactive bacteria. FEMS Microbiology Ecology 53: 103-115.

Pratscher J, Vollmers J, Wiegand S, Dumont MG, Kaster AK. 2018. Unravelling the Identity, Metabolic Potential and Global Biogeography of the Atmospheric Methane‐Oxidizing Upland Soil Cluster α. Environmental Microbiology 20: 1016-1029.

Price PB, Sowers T. 2004. Temperature dependence of metabolic rates for microbial growth, maintenance, and survival. Proceedings of the National Academy of Sciences of the United States of America 101: 4631-4636.

Qiu Q, Noll M, Abraham W-R, Lu Y, Conrad R. 2008. Applying stable isotope probing of phospholipid fatty acids and rRNA in a Chinese rice field to study activity and composition of the methanotrophic bacterial communities in situ. The ISME Journal 2: 602-614.

Rambaut A. 2012. FigTree v1.4.3. Computer program distributed by the author, website: http://tree.bio.ed.ac.uk/software/figtree

Ran Y, Li X, Cheng G, Zhang T, Wu Q, Jin H, Jin R. 2012. Distribution of permafrost in China: an overview of existing permafrost maps. Permafrost and Periglacial Processes 23: 322- 333.

161

Repo ME, Susiluoto S, Lind SE, Jokinen S, Elsakov V, Biasi C, Virtanen T, Martikainen PJ. 2009. Large N2O emissions from cryoturbated peat soil in tundra. Nature Geoscience 2: 189-192.

Rimmer D. 2006. Free radicals, antioxidants, and soil organic matter recalcitrance. European Journal of Soil Science 57: 91-94.

Rivkina E, Friedmann E, McKay C, Gilichinsky D. 2000. Metabolic activity of permafrost bacteria below the freezing point. Applied and Environmental Microbiology 66: 3230- 3233.

Rivkina E, Shcherbakova V, Laurinavichius K, Petrovskaya L, Krivushin K, Kraev G, Pecheritsina S, Gilichinsky D. 2007. Biogeochemistry of methane and methanogenic archaea in permafrost. FEMS Microbiology Ecology 61: 1-15.

Rodionow A, Flessa H, Kazansky O, Guggenberger G. 2006. Organic matter composition and potential trace gas production of permafrost soils in the forest tundra in northern Siberia. Geoderma 135: 49-62.

Rodriguez-R LM, Gunturu S, Harvey WT, Rosselló-Mora R, Tiedje JM, Cole JR, Konstantinidis KT. 2018. The Microbial Genomes Atlas (MiGA) webserver: taxonomic and gene diversity analysis of Archaea and Bacteria at the whole genome level. Nucleic Acids Research 46: W282-W288.

Rodriguez-R LM, Konstantinidis KT. 2014. Bypassing cultivation to identify bacterial species. Microbe 9: 111-118.

Routh J, Hugelius G, Kuhry P, Filley T, Tillman PK, Becher M, Crill P. 2014. Multi-proxy study of soil organic matter dynamics in permafrost peat deposits reveal vulnerability to climate change in the European Russian Arctic. Chemical Geology 368: 104-117.

Schostag M, Stibal M, Jacobsen CS, Bælum J, Taş N, Elberling B, Jansson JK, Semenchuk P, Priemé A. 2015. Distinct summer and winter bacterial communities in the active layer of Svalbard permafrost revealed by DNA-and RNA-based analyses. Frontiers in Microbiology 6: 399.

Schuur E, McGuire A, Schädel C, Grosse G, Harden J, Hayes D, Hugelius G, Koven C, Kuhry P, Lawrence D. 2015. Climate change and the permafrost carbon feedback. Nature 520: 171-179.

Schuur EA, Bockheim J, Canadell JG, Euskirchen E, Field CB, Goryachkin SV, Hagemann S, Kuhry P, Lafleur PM, Lee H. 2008. Vulnerability of permafrost carbon to climate change: implications for the global carbon cycle. BioScience 58: 701-714.

Shcherbakova V, Chuvilskaya N, Rivkina E, Pecheritsyna S, Laurinavichius K, Suzina N, Osipov G, Lysenko A, Gilichinsky D, Akimenko V. 2005. Novel psychrophilic anaerobic spore-forming bacterium from the overcooled water brine in permafrost: description Clostridium algoriphilum sp. nov. Extremophiles 9: 239-246.

162

Shcherbakova V, Rivkina E, Pecheritsyna S, Laurinavichius K, Suzina N, Gilichinsky D. 2011. Methanobacterium arcticum sp. nov., a methanogenic archaeon from Holocene Arctic permafrost. International Journal of Systematic and Evolutionary Microbiology 61: 144- 147.

Shcherbakova V, Yoshimura Y, Ryzhmanova Y, Taguchi Y, Segawa T, Oshurkova V, Rivkina E. 2016. Archaeal communities of Arctic methane-containing permafrost. FEMS Microbiology Ecology 92: fiw135.

Shur Y, Hinkel KM, Nelson FE. 2005. The transient layer: implications for geocryology and climate‐change science. Permafrost and Periglacial Processes 16: 5-17.

Shur Y, Jorgenson M. 2007. Patterns of permafrost formation and degradation in relation to climate and ecosystems. Permafrost and Periglacial Processes 18: 7-19.

ŠImek M, Cooper J. 2002. The influence of soil pH on denitrification: progress towards the understanding of this interaction over the last 50 years. European Journal of Soil Science 53: 345-354.

Smith CI, Chamberlain AT, Riley MS, Cooper A, Stringer CB, Collins MJ. 2001. Neanderthal DNA: not just old but old and cold? Nature 410: 771-772.

Song C, Wang X, Miao Y, Wang J, Mao R, Song Y. 2014. Effects of permafrost thaw on carbon emissions under aerobic and anaerobic environments in the Great Hing'an Mountains, China. Science of the Total Environment 487: 604-610.

Stackhouse BT, Vishnivetskaya TA, Layton A, Chauhan A, Pfiffner S, Mykytczuk NC, Sanders R, Whyte LG, Hedin L, Saad N. 2015. Effects of simulated spring thaw of permafrost from mineral cryosol on CO2 emissions and atmospheric CH4 uptake. Journal of Geophysical Research: Biogeosciences 120: 1764-1784.

Steven B, Briggs G, McKay CP, Pollard WH, Greer CW, Whyte LG. 2007. Characterization of the microbial diversity in a permafrost sample from the Canadian high Arctic using culture-dependent and culture-independent methods. FEMS microbiology ecology 59: 513-523.

Steven B, Leveille R, Pollard WH, Whyte LG. 2006. Microbial ecology and biodiversity in permafrost. Extremophiles 10: 259-267.

Steven B, Pollard WH, Greer CW, Whyte LG. 2008. Microbial diversity and activity through a permafrost/ground ice core profile from the Canadian high Arctic. Environmental microbiology 10: 3388-3403.

Stewart KJ, Coxson D, Grogan P. 2011a. Nitrogen inputs by associative cyanobacteria across a low arctic tundra landscape. Arctic, Antarctic, and Alpine Research 43: 267-278.

163

Stewart KJ, Coxson D, Siciliano S. 2011b. Small-scale spatial patterns in N2-fixation and nutrient availability in an arctic hummock–hollow ecosystem. Soil Biology and Biochemistry 43: 133-140.

Stewart KJ, Grogan P, Coxson DS, Siciliano SD. 2014. Topography as a key factor driving atmospheric nitrogen exchanges in arctic terrestrial ecosystems. Soil Biology and Biochemistry 70: 96-112.

Strong PJ, Laycock B, Mahamud SNS, Jensen PD, Lant PA, Tyson G, Pratt S. 2016. The opportunity for high-performance biomaterials from methane. Microorganisms 4: 11.

Tamppari L, Anderson R, Archer P, Douglas S, Kounaves S, McKay C, Ming D, Moore Q, Quinn J, Smith P. 2012. Effects of extreme cold and aridity on soils and habitability: McMurdo Dry Valleys as an analogue for the Mars Phoenix landing site. Antarctic Science 24: 211-228.

Tarnocai C. 1980. Summer temperatures of cryosolic soils in the north-central Keewatin, NWT. Canadian Journal of Soil Science 60: 311-327.

Tarnocai C. 2006. The effect of climate change on carbon in Canadian peatlands. Global and Planetary Change 53: 222-232.

Tarnocai C. 2009. Arctic permafrost soils. In Permafrost Soils, (ed. R Margesin), pp. 3-16. Springer.

Tarnocai C, Canadell J, Schuur E, Kuhry P, Mazhitova G, Zimov S. 2009. Soil organic carbon pools in the northern circumpolar permafrost region. Global Biogeochemical Cycles 23: DOI:10.1029/2008GB003327

Taş N, Prestat E, McFarland JW, Wickland KP, Knight R, Berhe AA, Jorgenson T, Waldrop MP, Jansson JK. 2014. Impact of fire on active layer and permafrost microbial communities and metagenomes in an upland Alaskan boreal forest. The ISME Journal 8: 1904-1919.

Throbäck IN, Enwall K, Jarvis Å, Hallin S. 2004. Reassessing PCR primers targeting nirS, nirK and nosZ genes for community surveys of denitrifying bacteria with DGGE. FEMS Microbiology Ecology 49: 401-417.

Torsvik V, Øvreås L, Thingstad TF. 2002. Prokaryotic diversity--magnitude, dynamics, and controlling factors. Science 296: 1064-1066.

Tuorto SJ, Darias P, McGuinness LR, Panikov N, Zhang T, Häggblom MM, Kerkhof LJ. 2014. Bacterial genome replication at subzero temperatures in permafrost. The ISME Journal 8: 139-149.

Tveit A, Schwacke R, Svenning MM, Urich T. 2013. Organic carbon transformations in high- Arctic peat soils: key functions and microorganisms. The ISME Journal 7: 299-311.

164

Tveit A, Urich T, Svenning MM. 2014. Metatranscriptomic analysis of Arctic peat soil microbiota. Applied and Environmental Microbiology 80: 5761–5772. van Everdingen R. 2005. Multi-Language Glossaru of Permafrost and Related Ground-Ice Terms. International Permafrost Association.

Vaughan PP, Blough NV. 1998. Photochemical formation of hydroxyl radical by constituents of natural waters. Environmental Science & Technology 32: 2947-2953.

Vishnivetskaya TA, Petrova MA, Urbance J, Ponder M, Moyer CL, Gilichinsky DA, Tiedje JM. 2006. Bacterial community in ancient Siberian permafrost as characterized by culture and culture-independent methods. Astrobiology 6: 400-414.

Voigt C, Lamprecht RE, Marushchak ME, Lind SE, Novakovskiy A, Aurela M, Martikainen PJ, Biasi C. 2016. Warming of subarctic tundra increases emissions of all three important greenhouse gases–carbon dioxide, methane, and nitrous oxide. Global Change Biology 23: DOI: 10.1111/gcb.13563.

Voigt C, Marushchak ME, Lamprecht RE, Jackowicz-Korczyński M, Lindgren A, Mastepanov M, Granlund L, Christensen TR, Tahvanainen T, Martikainen PJ. 2017. Increased nitrous oxide emissions from Arctic peatlands after permafrost thaw. Proceedings of the National Academy of Sciences 114: 6238-6243.

Vonk J, Mann P, Dowdy K, Davydova A, Davydov S, Zimov N, Spencer R, Bulygina E, Eglinton T, Holmes R. 2013. Dissolved organic carbon loss from Yedoma permafrost amplified by ice wedge thaw. Environmental Research Letters 8: 035023.

Walker D, Gould W, Maier H, Raynolds M. 2002. The Circumpolar Arctic Vegetation Map: AVHRR-derived base maps, environmental controls, and integrated mapping procedures. International Journal of Remote Sensing 23: 4551-4570.

Wei S, Cui H, He H, Hu F, Su X, Zhu Y. 2014. Diversity and Distribution of Archaea Community along a Stratigraphic Permafrost Profile from Qinghai-Tibetan Plateau, China. Archaea 2014: 240817.

West SA, Diggle SP, Buckling A, Gardner A, Griffin AS. 2007. The social lives of microbes. Annu Rev Ecol Evol Syst 38: 53-77.

Wheeler DL, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, Church DM, DiCuccio M, Edgar R, Federhen S. 2007. Database resources of the national center for biotechnology information. Nucleic acids research 36: D13-D21.

Wilhelm RC, Niederberger TD, Greer C, Whyte LG. 2011. Microbial diversity of active layer and permafrost in an acidic wetland from the Canadian High Arctic. Canadian Journal of Microbiology 57: 303-315.

165

Wilhelm RC, Radtke KJ, Mykytczuk NC, Greer CW, Whyte LG. 2012. Life at the wedge: the activity and diversity of Arctic ice wedge microbial communities. Astrobiology 12: 347- 360.

Willerslev E, Hansen AJ, Poinar HN. 2004a. Isolation of nucleic acids and cultures from fossil ice and permafrost. Trends in Ecology & Evolution 19: 141-147.

Willerslev E, Hansen AJ, Rønn R, Brand TB, Barnes I, Wiuf C, Gilichinsky D, Mitchell D, Cooper A. 2004b. Long-term persistence of bacterial DNA. Current Biology 14: R9-R10.

Williams RJ, Howe A, Hofmockel KS. 2014. Demonstrating microbial co-occurrence pattern analyses within and between ecosystems. Frontiers in Microbiology 5: 358.

Wu X, Zhang W, Liu G, Yang X, Hu P, Chen T, Zhang G, Li Z. 2012. Bacterial diversity in the foreland of the Tianshan No. 1 glacier, China. Environmental Research Letters 7: 014038.

Xavier JB. 2011. Social interaction in synthetic and natural microbial communities. Molecular Systems Biology 7: 483.

Xie Z, Le Roux X, Wang C, Gu Z, An M, Nan H, Chen B, Li F, Liu Y, Du G. 2014. Identifying response groups of soil nitrifiers and denitrifiers to grazing and associated soil environmental drivers in Tibetan alpine meadows. Soil Biology and Biochemistry 77: 89- 99.

Yan X, Chu F, Puri AW, Fu Y, Lidstrom ME. 2016. Electroporation-based genetic manipulation in type I methanotrophs. Applied and Environmental Microbiology 82: 2062-2069.

Yergeau E, Hogues H, Whyte LG, Greer CW. 2010. The functional potential of high Arctic permafrost revealed by metagenomic sequencing, qPCR and microarray analyses. The ISME Journal 4: 1206-1214.

Yoshida M, Ishii S, Otsuka S, Senoo K. 2009. Temporal shifts in diversity and quantity of nirS and nirK in a rice paddy field soil. Soil Biology and Biochemistry 41: 2044-2051.

Yu K, Chen G, Struwe S, Kjøller A. 2000. Production and reduction of nitrous oxide in agricultural and forest soils. Ying yong sheng tai xue bao. The Journal of Applied Ecology 11: 385-389.

Yun J, Ju Y, Deng Y, Zhang H. 2014. Bacterial community structure in two permafrost wetlands on the Tibetan Plateau and Sanjiang Plain, China. Microbial Ecology 68: 360-369.

Zellner R, Exner M, Herrmann H. 1990. Absolute OH quantum yields in the laser photolysis of nitrate, nitrite and dissolved H2O2 at 308 and 351 nm in the temperature range 278–353 K. Journal of Atmospheric Chemistry 10: 411-425.

166

Zhang G, Niu F, Ma X, Liu W, Dong M, Feng H, An L, Cheng G. 2007. Phylogenetic diversity of bacteria isolates from the Qinghai-Tibet Plateau permafrost region. Canadian Journal of Microbiology 53: 1000-1010.

Zhang L-M, Wang M, Prosser JI, Zheng Y-M, He J-Z. 2009. Altitude ammonia-oxidizing bacteria and archaea in soils of Mount Everest. FEMS Microbiology Ecology 70: 208- 217.

Zhao Q, Bai Y, Zhang G, Zhu S, Sheng H, Sun Y, An L. 2011. Chryseobacterium xinjiangense sp. nov., isolated from alpine permafrost. International journal of systematic and evolutionary microbiology 61: 1397-1401.

Zhuang Q, Melillo JM, Kicklighter DW, Prinn RG, McGuire AD, Steudler PA, Felzer BS, Hu S. 2004. Methane fluxes between terrestrial ecosystems and the atmosphere at northern high latitudes during the past century: A retrospective analysis with a process‐based biogeochemistry model. Global Biogeochemical Cycles 18: DOI: 10.1029/2004GB002239

Zimov SA, Schuur EA, Chapin III FS. 2006. Permafrost and the global carbon budget. Science 312: 1612-1613.

Zipfel C, Oldroyd GE. 2017. Plant signalling in symbiosis and immunity. Nature 543: 328-336.