THE MICROBIOLOGY OF THE SULFUR-DOMINATED GLACIAL

ECOSYSTEM AT BORUP FIORD PASS, AND HIGH ARCTIC

LOW-TEMPERATURE SPRING ASSOCIATED

MICROBIAL MATS

by

Christopher Bhartendu Trivedi

A thesis submitted to the Faculty and Board of Trustees of the Colorado School of Mines in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Civil and

Environmental Engineering).

Golden, Colorado

Date ______

Signed: ______

Christopher Bhartendu Trivedi

Signed: ______

Dr. John R. Spear Thesis Advisor

Golden, Colorado

Date ______

Signed: ______

Dr. Terri S. Hogue Professor and Department Head Department of Civil and Environmental Engineering

ii

ABSTRACT

Borup Fiord Pass (BFP) located on Ellesmere Island in the Canadian High Arctic is a unique low-temperature environment. BFP is a sulfur-dominated ecosystem that occurs within the Krieger mountains. Here, a subglacial spring emergences from the toe of a coalescence glacier. The presence of high concentrations of sulfur and low temperatures make BFP an excellent terrestrial analog for the study of potential astrobiological targets. High concentrations of sulfur also make BFP an ideal location for studying how microbial organisms can utilize reduced and oxidized forms of sulfur to thrive in cold environments. Yet, due to human influence

Arctic research sites may not remain cold perpetually also making the location ideal to study climate change. Over field campaigns in 2014, 2016, and 2017, numerous samples were collected and analyzed for microbial community composition (16S rRNA gene sequencing), function (metagenomic sequencing), and aqueous geochemistry. 16S rRNA gene sequencing data revealed Flavobacterium was common to almost all samples. Sulfur-oxidizing microorganisms (Sulfurimonas, Sulfurovum, and Sulfuricurvum), and an organism that specializes in the disproportionation of inorganic sulfur compounds (Desulfocapsa) were also abundant in samples with high concentrations of sulfur compounds. To link the identified microbial taxa to function, assembly and binning of metagenomic sequence data was carried out and produced 31 reportable metagenome assembled genomes (MAGs), which revealed that sulfur oxidation may be a cosmopolitan process at BFP. Overall, results suggest that while sulfur cycling organisms dominate during acute events, a basal community structure appears to dominate over time and site type and that functional redundancy may be a key mechanism utilized by microorganisms for energy generation in this low-temperature environment. Finally, to address the changing Arctic climate the community composition and function of microbial

iii mats found across Ellesmere Island was characterized. To my knowledge these are the most northern microbial mats ever identified, potentially acting as a barometer to study climate change and its effect on polar environments. Microbial community composition varied significantly between sample sites. Metagenomes were sequenced from one site, Ice River, revealing the potential for sulfur oxidation, nitrate/nitrite reduction, and carbon fixation via the reductive acetyl-CoA and reverse TCA cycles. Also found were genes that allow microorganisms to adapt to adverse conditions in extreme environments. High Arctic microbial communities help to guide our search for potential life on extraterrestrial worlds, and further enhance our knowledge of low-temperature ecosystems on Earth as our climate continues to change.

iv

TABLE OF CONTENTS

ABSTRACT …...... iii

LIST OF FIGURES ...... ix

LIST OF TABLES ...... xi

LIST OF ABBREVIATIONS ...... xii

ACKNOWLEDGEMENTS ...... xiii

CHAPTER 1 INTRODUCTION ...... 1

1.1 Sulfur Metabolism in the High Arctic ...... 1

1.2 Permafrost hydrology and transient springs ...... 6

1.3 Challenges of studying microbiology at 81° N 81° W ...... 8

1.4 Evolution of DNA Sequencing Technologies in Cryomicrobiology ...... 9

1.5 Research objectives ...... 11

1.6 Conclusions ...... 15

1.7 Publications associated with CSM graduate degree ...... 15

1.8 References ...... 16

CHAPTER 2 LOW-TEMPERATURE SULFIDIC-ICE MICROBIAL COMMUNITIES, BORUP FIORD PASS, CANADIAN HIGH ARCTIC ...... 22

2.1. Introduction ...... 23

2.1.1 BFP Site Background ...... 25

2.2 Materials and Methods ...... 26

2.2.1 Field Work and Sample Collection ...... 26

2.2.2 DNA Extraction and 16S rRNA Gene Library Sequencing ...... 30

2.2.3 Aqueous Geochemistry ...... 32

v 2.3 Results ...... 34

2.3.1 General Field Observations ...... 34

2.3.2 Estimated Biomass and Sample QC...... 35

2.3.3 Melt Pools and 2016 Spring Fluid Samples ...... 35

2.3.4 Aufeis Samples ...... 39

2.3.5 Mineral Precipitate Samples ...... 41

2.3.6 Glacier Samples ...... 44

2.3.7 Additional Aqueous Geochemistry ...... 46

2.4 Discussion ...... 47

2.5 Conclusion ...... 56

2.6 Data Availability ...... 57

2.7 Acknowledgements ...... 57

2.8 References ...... 58

CHAPTER 3 METAGENOMIC INSIGHTS INTO MICROBIAL METABOLISMS OF A SULFUR-DOMINATED GLACIAL ECOSYSTEM ...... 64

3.1 Introduction ...... 65

3.2 Results ...... 70

3.2.1 Whole shotgun sequencing and assembly ...... 70

3.2.2 Binning, MAG determination, and putative taxonomic identification ...... 70

3.2.3 MAG recruitment by sample...... 72

3.2.4 Sulfur cycling genes ...... 73

3.2.5 Carbon metabolism genes ...... 77

3.2.6 Nitrogen cycling genes...... 79

3.2.7 Hydrogen, Oxygen, and Hopanoids ...... 79

vi 3.3 Discussion ...... 80

3.4 Conclusion ...... 87

3.5 Methods ...... 88

3.5.1 Site Description and Sample Collection ...... 88

3.5.2 DNA extraction and metagenomic library preparation ...... 89

3.5.3 Metagenomic Sequencing Assembly and Analysis ...... 90

3.6 Data availability ...... 93

3.7 References ...... 93

CHAPTER 4 CHARACTERIZATION OF FOUR MICROBIAL MATS ABOVE 80° N IN THE CANADIAN HIGH ARCTIC...... 100

4.1 Introduction ...... 101

4.2 Materials and Methods ...... 103

4.2.1 Site Description and Sample Collection ...... 103

4.2.2 DNA extraction and 16S rRNA gene library preparation ...... 104

4.2.3 16S rRNA gene sequencing analysis ...... 107

4.2.4 Metagenomic sequencing and analysis ...... 108

4.3 Results ...... 109

4.3.1 Mat microbial communities contain distinct assemblages ...... 109

4.3.2 Cape Evans ...... 110

4.3.3 Derberville ...... 114

4.3.4 Ice River ...... 114

4.3.5 Moss Valley ...... 116

4.3.6 Ice River metagenomes ...... 116

4.4 Discussion ...... 119

vii 4.5 Data availability ...... 126

4.6 References ...... 126

CHAPTER 5 CONCLUDING REMARKS ...... 133

5.1 Conclusions ...... 133

5.2 Future Directions ...... 137

5.3 References ...... 139

APPENDIX A SUPPORTING INFORMATION FOR CHAPTER 2 ...... 140

APPENDIX B SUPPORTING INFORMATION FOR CHAPTER 3 ...... 147

APPENDIX C SUPPORTING INFORMATION FOR CHAPTER 4 ...... 151

APPENDIX D CARBONATE-RICH DENDROLITIC CONES: INSIGHTS INTO A MODERN ANALOGUE FOR INCIPIENT MICROBIALITE FORMATION, LITTLE HOT CREEK, LONG VALLEY CALDERA, CALIFORNIA ...... 155

APPENDIX E SUPPLEMENTAL ELECTRONIC FILES ...... 190

APPENDIX F COPYRIGHT FOR PUBLISHED WORKS ...... 192

viii

LIST OF FIGURES

Figure 1.1. Borup Fiord Pass site location...... 4

Figure 1.2. Graphical representation of two-step PCR...... 10

Figure 2.1. Borup Fiord Pass: Sample locations and types...... 27

Figure 2.2. Log-log plot of sulfate vs chloride aqueous geochemistry data...... 33

Figure 2.3. BFP melt pool bar charts...... 38

Figure 2.4. Aufeis bar charts ...... 41

Figure 2.5. Mineral precipitate bar charts ...... 43

Figure 2.6. Glacier bar charts ...... 45

Figure 3.1. Borup Fiord Pass satellite view...... 66

Figure 3.2. BFP Sampling locations...... 68

Figure 3.3. Anvi’o diagram and gene presence/absence table...... 74

Figure 3.4. Observed gene sequences within queried MAG contigs...... 76

Figure 3.5. Cladogram of MAGs...... 78

Figure 4.2. NMDS of Arctic mat sites...... 111

Figure 4.3. SSU rRNA Gene Bar Chart Organized by Location...... 112

Figure 4.4. Top 15 OTUs by Site broken down by Phylum, Class, and Genus...... 113

Figure 4.5. Summary of Ice River metagenomes...... 118

Figure A.1. Melt Pool site locations...... 141

Figure A.2. Aufeis site locations...... 142

Figure A.3. BFP17 Site location...... 143

Figure A.4. Principal Component Analysis (PCoA) of BFP aqueous geochemistry...... 144

ix Figure A.5. Alpha rarefaction plot of BFP site samples...... 145

Figure A.6. Rank abundance plot of BFP site types...... 146

Figure B.1. BFP16 Spring location...... 149

Figure B.2. Examples of sample types used for metagenomes...... 150

Figure C.1. Alpha diversity of Arctic mat sites...... 153

Figure C.2. Heatmap of top 12 phyla by site...... 154

Figure D.1. Cone Pool at Little Hot Creek...... 158

Figure D.2. Microscopic characterization of dendrolitic cones...... 161

Figure D.3. Analysis of structure of dendrolitic cone filament...... 163

Figure D.4. Bacterial community composition from Cone Pool East...... 164

Figure D.5. Bacterial community composition of mats from Little Hot Creek...... 166

Figure D.6. Calcium carbonate saturation in water of Cone Pool...... 171

Figure D.7. Phylogenetic tree of the top 25 OTUs represented across LHC sample types. 174

x

LIST OF TABLES

Table 2.1. Aqueous geochemistry for selected BFP sites...... 37

Table 3.1. Metadata for all reported MAGs...... 71

Table 4.1. Location data of microbial mat sites...... 104

Table 4.2. SSU rRNA Gene Library Summary...... 106

Table A.1. Sample information table including sequencing counts, charge balance error, and calculated biomass...... 140

Table B.1. Co-assembly metadata...... 147

Table B.2. JSpeciesWS results...... 148

Table C.1. Mat sample PCR amplification parameters...... 151

Table C.2. ADONIS Results between mat sites...... 152

Table C.3. Ice River MAG putative taxonomy...... 152

Table D.1. Physical and chemical properties of stream water at Cone Pool and Outflow Pool...... 159

Table E.1. Table of electronic files associated with Chapters and Appendices...... 190

xi

LIST OF ABBREVIATIONS

BFP Borup Fiord Pass

OTU Operational Taxonomic Unit

SOM Sulfur-oxidizing microorganism

SRM Sulfate-reducing microorganism

PCSP Polar Continental Shelf Program

BSR Biological Sulfate Reduction

EPS Extra Polymeric Substances

DOC Dissolved Organic Carbon

xii

ACKNOWLEDGEMENTS

I’ve found the graduate research process to be fun and enlightening, but also strenuous and extremely challenging at times. This has been an adventure for me, and I couldn’t have done it alone. This work was made possible through funding from the NASA Exobiology and

Evolutionary Biology Program (NNX13AJ32G). Additional support was provided by the NASA

Astrobiology Institute Rock Powered Life grant (Cooperative Agreement NNA15BB02A). None of my field work would have been possible without the awesome support of Natural Resources

Canada’s (NRC) Polar Continental Shelf Program (PCSP) - thanks for getting me to Borup!

Further thanks to Dr. Karsten Piepjohn and the Federal Institute for Geosciences and Natural

Resources at BGR, Hanover, Germany for all logistics for the 2017 field campaign of the

Circum-Arctic Structural Events-19 (CASE-19) expedition.

My sincerest gratitude to all of the wonderful people in the GEM lab and beyond who helped to make me the scientist I am today. Thank you Dr. Stephanie Carr, Dr. Tess Weathers,

Dr. Zackary Jones, Dr. Dina Drennan, Dr. Robert Robert Almstrand, Dr. Dong Li, Dr. Kristin

Mikkelson, and Dr. Lisa Gallagher for welcoming me into the lab and teaching me the ropes.

Thank you Emily Kraus, Kalen Rasmussen, Kevin Chan, Laura Leonard, Michael Vega, Alex

Honeyman, Dr. Andrew Pfluger, Chelsea Bokman, Dr. Brent Brouillard and the rest of the GEM lab elite for making lab enjoyable and fun. A huge thanks to Dr. Blake Stamps and Dr. Gary

Vanzin - without your individual guidance I never would have made it this far.

Thanks to the CEE department staff for their continued support. Thank you Tim

VanHaverbeke for your guidance back in 2012 when I was trying to figure out a new path in life, and thank you Angela Knighton for your ever present smile and unending patience.

I also want to thank the wonderful people in the Templeton lab who have supported me

xiii over the years. Thank you to Dr. Graham Lau who is my Artic partner in crime, and thank you to

Katie Rempfert and Eric Ellison whom I consider close friends. I also want to thank all of the amazing people I befriended at the 2015 International Geobiology Course. There are too many to name here, but I count them among some of my closest friends and colleagues.

Wonderful Talla “Tee” Talyai, the light of my life, I can’t express in words what your support has meant to me during this process. This degree is as much yours as it is mine. I love you and I could never have accomplished any of this without you (and Nola).

To my thesis committee, thank you for your guidance and always steering me back on track when I got lost. Thank you Dr. Josh Sharp and Dr. Junko Munakata Marr for your advice over the years. Thank you Dr. Alexis Navarre-Sitchler for challenging me to think differently.

Thank you Dr. Stephen Grasby for reminding me to push myself (Be the Steve!), and thank you

Dr. Alexis Templeton for your never-ending support and guidance. You were my co-advisor without the fancy title. Lastly, thank you to my advisor Dr. John Spear who taught me how to view the world from 30,000 feet before diving in. Thank you for never hand-holding me and teaching me to grow as a confident, independent scientist. For that I am eternally grateful.

I think most graduate students would agree that obtaining a graduate degree tests the human mind and spirit on almost every level. I do know that I am a much different person now than I was almost six years ago when I started this journey. I can happily say that I’ve grown as a person and a scientist and am better off because of it. I want to give one last thank you to everyone, friends and family alike, who have helped me along this journey. I couldn’t have done it without you.

xiv

DEDICATION

To all my parents, thank you for always believing in me and allowing me to find my own path, even when I may have been on the wrong one. Without your support and love I would not be the

person or scientist that I am today. I love you.

xv

CHAPTER 1

INTRODUCTION

1.1 Sulfur Metabolism in the High Arctic

Sulfur is the sixth most abundant element on Earth and plays a major role in many abiotic and biotic chemical pathways. It is an essential component for all extant life and is a basic component of cysteine and methionine, which are necessary for protein synthesis (Mandeville,

2010). Sulfur geochemistry can be used to understand the early evolution of Earth’s atmosphere and hydrosphere, and recent advances have allowed scientists to begin using sulfur stable isotopes as a tracer to better understand how sulfur plays a role in inorganic and biogenic processes. Because sulfur is ubiquitous across the planet it plays a role in a number of different biogeochemical pathways. Sulfur can occur as sulfide minerals in the Earth’s mantle, for example oceanic basalts and diamond mineral inclusions (Kiseeva et al., 2017), and as sulfide or sulfate minerals in crustal rocks (e.g. FeS; Wedepohl, 1984). Sulfur can also be found as elemental sulfur near volcanoes, hydrothermal vents and some high latitude, Arctic, sulfide-rich springs (Grasby et al., 2003), which even includes some rare forms not predicted to persist at low temperatures (Lau et al., 2017; see Appendix E). Furthermore, sulfur is also found as dissolved sulfate or dimethylsulfide in ocean waters, and as a trace gas in the atmosphere (Mandeville,

2010). Many organisms are capable of using sulfur compounds for energy, including plants, phytoplankton, and a number of different microorganisms. Sulfur has at least eight redox states,

2- from the most reduced, sulfide, H2S, to the fully oxidized sulfate, SO4 , and microorganisms can take advantage of electron transfer across the entire redox series to metabolize sulfur, using it

1 2- either as an electron donor (H2S) or as an electron acceptor (SO4 ). Microfossil evidence suggests that the earliest microbial metabolisms were sulfur-based (Wacey et al., 2011), and reduced species such as hydrogen bisulfide (HS-) would have to have been oxidized prior to oxygenation of the world’s oceans (Hanson et al., 2013).

Microorganisms that use sulfur as a source of energy are abundant and found in many diverse, extreme environments (Dahl and Friedrich, 2008; Klotz et al., 2011). Sulfur is key to microbial metabolisms, such as sulfur oxidation and sulfate reduction, across extreme environments including hydrothermal vents (Dahl and Friedrich, 2008; Gilhooly et al., 2014), marine sediments (Wasmund et al., 2017), cave ecosystems (Jones et al., 2016; Macalady et al.,

2008), sea ice (Boetius et al., 2015), Arctic saline springs (Niederberger et al., 2009), and subglacial ecosystems (Mikucki and Priscu, 2007; Purcell et al., 2014). Two predominant forms of sulfur utilizing microorganisms exist: sulfur-oxidizing microorganisms (SOMs), and sulfate- reducing microorganisms (SRMs). Microbially-mediated sulfur oxidation occurs in both oxic and anoxic environments, and is highly dependent on what electron acceptors are present. In oxic environments, oxygen is the preferred electron acceptor, whereas within anaerobic environments, by definition devoid of oxygen, alternative electron acceptors must be utilized

(e.g. nitrate or sulfate). Sulfate reducing microorganisms help to facilitate the opposite set of

2- reactions where highly oxidized sulfur (e.g. sulfate or sulfite (SO3 )) is reduced towards

0 elemental sulfur (S ) or even hydrogen sulfide (H2S). SRMs can utilize organic carbon (OC) or

2- inorganic compounds (such as iron) as an electron donor during the reduction of sulfate (SO4 ) as their terminal electron acceptor (Barton and Fauque, 2009) under anoxic (i.e. anaerobic) conditions. Sulfate reduction by SRMs is critical within marine environments where they contribute to over 50% of total marine OC oxidation globally (Bowles et al., 2014; Thullner et

2 al., 2009). Beyond simple sulfate reduction or sulfur oxidation, some microorganisms within the class Deltaproteobacteria are capable of sulfur disproportionation, where compounds such as sulfur, thiosulfate, and sulfite can serve as both electron donor and acceptor (Finster, 2008), generating both hydrogen sulfide and sulfate (Barker Jøgensen, 1990). Due to its ability to serve as both an electron donor and/or electron acceptor across an eight electron transfer redox state (-

2 to +6), sulfur is an important ingredient for life on Earth, and in particular to microorganisms in extreme environments.

Some research on sulfur cycling in low-temperature environments has been conducted, primarily in polar ecosystems such as Antarctica’s Blood Falls (Mikucki and Priscu, 2007) and

Lake Whillans (Purcell et al., 2014), and in Arctic locations such as Gypsum Hill and Coulour

Peak (Niederberger et al., 2009; Perreault et al., 2008); however, there exists a dearth of knowledge towards sulfur cycling in low-temperature environments, primarily in terms of microbial contributions to these processes. An good example is the low-temperature, sulfur- dominated supraglacial spring system found at Borup Fiord Pass (BFP; Figure 1.1), Ellesmere

Island, Nunavut, Canada (Grasby et al., 2003). BFP is one of the few known low-temperature springs in the Canadian Arctic, along with Gypsum Hill, Colour Peak on Axel Heiberg Island

(Niederberger et al., 2009; Perreault et al., 2007, 2008; Pollard et al., 1999), and Ice River on

Ellesmere Island (Grasby et al., 2014). The spring at BFP is unique as it discharges from a glacier and has very high concentrations of dissolved hydrogen sulfide (H2S). Biotic and/or abiotic oxidation of H2S to elemental sulfur creates visually striking yellow glacial ice, as well as yellow colored aufeis (whereby aufeis is ice formed by freezing of spring discharge into sub-zero air temperatures). This sulfur covered ice extends for tens of square meters to a few square kilometers down valley (Gleeson et al., 2012; Grasby et al., 2003; Lau et al., 2017). Spring water

3

Figure 1.1. Borup Fiord Pass site location. The top panel is a Google satellite image showing the location (denoted by a red dot) of Borup Fiord Pass on Ellesmere Island in the Canadian High Arctic. The bottom panel is a picture of the site as seen from above on 19 June, 2014.

4 discharging from the glacier is briny, sulfide-rich, and hosts a diverse microbial community that may make use of sulfur metabolisms for growth (Gleeson et al., 2011; Grasby et al., 2003;

Wright et al., 2013). When discharging over cold winter months the spring water freezes, forming sulfidic aufeis as well as associated cryogenic minerals (Grasby, 2003; Grasby et al.,

2003; Lau et al., 2017). During the subsequent summer months surface melt pools form, releasing trapped hydrogen sulfide gas that is then oxidized to elemental sulfur that accumulates as sulfur-rich cryogenic materials (Lau et al., 2017). It remains unclear to what degree, if at all,

H2S oxidation is driven by microbial interaction (Gleeson et al., 2011; Grasby et al., 2003;Wright et al., 2013), or what the overarching microbial metabolic functionality is in these complex sulfidic-ice environments are.

In addition to glacial and spring-derived fluids within low-temperature environments, microbial mats found at high latitudes are poorly understood and are worth investigating due to their complexity and diversity, but also due to their ability to grow and thrive in inhospitable habitats as well as to be the ‘first responders’ to climate change. Microbial mats are found globally and are often associated with extreme environments like geothermal hot springs such as those found in Yellowstone (Pepe-Ranney et al., 2012; Reyes et al., 2013), the Long Valley

Caldera (Bradley et al., 2017), Iceland (Aguilera et al., 2010), and Mexico (Prieto-Barajas et al.,

2018). They have also been observed in other extreme environments like the hypersaline

Guerrero Negro mats in Mexico (Feazel et al., 2008; Harris et al., 2013; Kunin et al., 2008; Ley et al., 2006; Robertson et al., 2009; Spear et al., 2003) and even in low-temperature polar regions like Antarctica (Andriuzzi et al., 2018; Jungblut et al., 2016; Varin et al., 2010, 2012), and Axel

Heiberg Island in the Canadian Arctic (Sapers et al., 2017). While information concerning their formation, biogeography, and microbiological structure is abundant, literature detailing their

5 growth in polar locales in terrestrial locations is sparse. Understanding microbial mats at these latitudes could help scientists to understand how these layered assemblages are able to cope (or not) with the absence of sunlight (in winter months), as well as potentially be used to indicate changes in the global climate cycle.

1.2 Permafrost hydrology and transient springs

While the focus of this dissertation is not hydrology, the fact that it plays such an important role in the BFP system deserves some thought. The spring system at BFP is believed to be transient in nature due to its presence and absence, randomly, from year to year. The most current conceptual models of these transient springs utilize glacial retreat and a shifting permafrost table as possible explanations for development (Grasby et al., 2012; Scheidegger et al., 2012). Glacial hydrology is complicated by an increased number of variables when compared to traditional terrestrial hydrology. One of these variables is the intrusion of permafrost to the hydraulics of a system, which creates an even greater challenge, especially when dealing with thermal and polythermal glacial systems (Mcintosh et al., 2012; Scheidegger et al., 2012).

Hinzman et al. (2006) define permafrost as earth material that has had a temperature at or below

0°C for at least two consecutive summers. This creates a dual system where the permafrost is below while the active layer sits on top and freezes and thaws each year. Flow through permafrost has the potential to be very transient, due to the presence or absence of subsurface voids and taliks (an area of unfrozen ground surrounded by permafrost) within permeable sediments (Moorman, 2003). Flow must be constant (especially for groundwater) through a talik to prevent freeze-back through heat loss to the surrounding frozen ground. This energy balance and requirement for heat can be expressed by:

6

∗ 푄 = 푄푆 + 푄퐿 − 푄퐺

* where Q is the net heat flux, QS is the sensible heat with the unfrozen water in the talik, QL is the latent heat of fusion of the unfrozen water within the talik, and QG is the heat being conducted away from the talik through the permafrost (Moorman, 2003). QS is a function of temperature and the flow rate of the water, while QL is a function of the porosity of the unfrozen sediment, and lastly QG is a function of the surrounding permafrost and its thermal conductivity.

When Q* is negative the talik will begin to shrink in size by freeze-back.

Talik formation and flow has been found in a number of areas including Axel Heiberg

Island in the Canadian High Arctic. Heldmann et al. (2005) and other groups have previously reported on the hydrology of a perennial saline spring that is capable of discharging through thick continuous permafrost in the absence of volcanic heat sources (Heldmann et al., 2005;

Pollard, 2005; Woo et al., 2008). Another example of this are the springs of Gypsum Hill studied by Perreault et al. (2008) while looking at microbial populations, including sulfur-oxidizing

Proteobacteria, that live in these unique systems. Furthermore, this can also been seen on

Ellesmere Island at Borup Fiord Pass (BFP) and is the currently supported hypothesis for spring formation at this site.

Subglacial recharge, especially in zones of thick permafrost (Arctic deserts) is dominated by glacial meltwater or meltwater fed by surface water bodies. This is due to very low amounts of precipitation being unlikely able to recharge spring flows (Scheidegger et al., 2012; Wainstein et al., 2014). Furthermore, Woo (2012) explains that permafrost regions may also allow for groundwater recharge through fissures in the bedrock, although these diminish with depth. It is

7 also of note that localized recharge of intrapermafrost or subpermafrost groundwater may be facilitated by deep fractures, as well as passages through soluble rocks such as gypsum and carbonates (Woo, 2012). This is especially applicable to the BFP site as Grasby et al. (2003) mention that the bedrock of the system is “dominated by Carboniferous to Triassic marine carbonates, evaporates, and clastics of the Sverdrup Basin”. This supports the theory that the

BFP hydrologic system may indeed be recharged primarily via glacial meltwater moving down through the bedrock of the site where it then interacts with sulfur-rich geological evaporite horizons, and then emerges onto the glacier in the form of springs and seeps. While the current hydrological models developed for BFP include a lot of speculation, it would be highly beneficial to be able to constrain fluid flow under the glacier to gain a better understanding of subsurface geologic interactions. This would then strengthen our information about potential carbon sources for microbial metabolisms among other things.

1.3 Challenges of studying microbiology at 81° N 81° W

It should be noted that studying extreme and hard to access field sites can be difficult and presents a number of challenges. At Borup Fiord Pass, it is unknown whether flowing springs are an event that only occurs during the summer melt season, or are perennial as has been documented in other Arctic spring systems (Grasby et al., 2014; Niederberger et al., 2009;

Perreault et al., 2007, 2008; Pollard et al., 1999). To answer this can be logistically difficult due to the transient nature of the system and limited site access in the winter. Being able to constrain whether or not the site is perennial in nature would add valuable knowledge to the underlying hydrology of the system, as well as further our understanding of potential changes in microbial community structure in a more temporally complete fashion. To that end, the possibility for very

8 low biomass in collected samples in glacial environments (Boetius et al., 2015; Stibal et al.,

2015) was another challenge we faced. Due to low microbial biomass in a number of samples the extraction of genomic DNA (gDNA) to be used in downstream sequencing (16S rRNA gene & whole genome) was difficult. Many samples were below detection when looking at extracted gDNA concentrations.

1.4 Evolution of DNA Sequencing Technologies in Cryomicrobiology

High throughput DNA sequencing technology is rapidly evolving. Within the timespan of my degree, pyrosequencing was replaced by sequencing by synthesis (SBS). The transition to

SBS also heralded an enormous increase in sequencing depth, or the number of sequencing reads per instrument run (Luo et al., 2012). In response to the increasing depth of sequencing available to us, we emulated a dual-index amplicon sequencing protocol developed by Kozich et al. (2013) that maximized the number of samples sequenced in a single batch. The primer set for this protocol was based on primers utilized in the Earth Microbiome Project, 515F-806R, from

Caparaso et al. (2011). The design of the dual-index primers included (on both forward and reverse primers) an 8-basepair (bp) index sequence, a padding sequence (to help achieve a desired melting temperature in the polymerase chain reaction (PCR)), a 2-bp linker sequence and then the 16S rRNA gene primer. While allowing for a large number of samples to be sequenced simultaneously, primer design of this sort has costly drawbacks: 1) primer costs can be quite high due to their long length; 2) primer-dimers created during PCR amplification are greater than 100- bp which can complicate downstream clean-up processes and sequencing; and 3) the specific amplified gene fragment (or amplicon) of interest (in this case the 16S rRNA gene) cannot be changed. If another amplicon such as the 18S rRNA gene or the internally transcribed spacer

9 (ITS) region that sits between ribosomal RNA genes were desired, an entirely new set of dual- index primers would have to be purchased.

Advances in molecular biology and microbial ecology are rapid – soon after the implementation of the dual-index primer set, a new and potentially better primer set and indexing approach was developed. Instead of including barcodes on either primer, a sequence derived from the M13 phage was included on the forward primer. Using this M13-derived sequence a two-step PCR approach was used, where initial amplification of the gene of interest is carried out, followed by a second limited cycle PCR to add indices to each sample (Figure 1.2).

Figure 1.2. Graphical representation of two-step PCR. Initial amplification highlights the addition of the M13 sequence onto the original amplicon of choice. The second step limited- cycle PCR shows the addition of the barcode to be used in downstream high-throughput sequencing.

10 The benefit of this method is that the gene primer set is separate from 12-bp indices, or barcodes,

(Hamady et al., 2008) which reduces the primer size, and allows for the barcodes to be used with any amplicon primer set. In this way the majority of cost is focused on the barcodes themselves which can be used interchangeably with any gene primer set of interest. Another change to the amplicon protocol used throughout this dissertation is the use of a larger V4-V5 16S rRNA gene fragment, rather than the more commonly used V4 amplicon (Caporaso et al., 2011). This new primer set 515F (5’- GTGYCAGCMGCCGCGGTAA-3’) 926R (5’-

CCGYCAATTYMTTTRAGTTT-3’) was developed after limitations in the more commonly used 515F-806R primer set were discovered that may bias analyses due to an underestimation of certain groups (e.g. SAR11) and overestimate (e.g. Gammaproteobacteria) others (Parada et al.,

2016). The redesigned primer set more accurately estimated the relative abundance of all amplified taxa and also amplifies the Eucarya. Due to the increased benefits of this primer set previous BFP samples that yielded adequate sequencing coverage from the dual-index primer set were re-sequenced using the Parada primer set modified with the M13 universal primer on the

515F primer.

1.5 Research objectives

While changes in technology can drive a project one way or another, the overall goals for the research being conducted at BFP remained true to form. The aim of the research described in this dissertation was three fold: 1) to characterize microbial communities over time paired with aqueous geochemistry; 2) to explore the metabolic and functional potential of microbial communities at BFP; and 3) to characterize microbial mat communities found in the high Arctic.

This was achieved through three separate avenues: 1) characterize BFP communities via targeted

11 16S rRNA gene sequencing paired to aqueous geochemistry; 2) Query the metabolic potential of

BFP via whole genome shotgun metagenomic sequencing; and 3) Characterize a group of microbial mats collected in high latitudes in the Canadian Arctic and compare these to previously characterized microbial mats in other extreme environments across the world.

Chapter 2 covers the first aim of this dissertation in detailing the characterization of microbial communities paired with aqueous geochemistry at BFP. To address this, field campaigns to BFP over multiple years were undertaken to collect samples from multiple different site types: 1) spring discharge - sulfidic saline water flowing at high rate (8 L s-1) from a hole in the glacier surface, 2) aufeis - sulfidic ice in the proglacial area formed from freezing spring discharge, 3) minerals precipitates - small (up to 10 cm high) piles of toothpaste like consistency on the glacier and aufeis surface as well as in surrounding glacial moraine. Most likely formed through accumulation of more disseminated precipitates during spring snowmelt, and 4) melt pool waters - cryoconite pools filled with sulfidic water formed on glacier and the aufeis surface. Results presented in Chapter 2 show that a baseline microbial community exits within most samples regardless of sample type, typically consisting of a high abundance of an operational taxonomic unit (OTU) most closely related to the genus Flavobacterium. Organisms previously seen in polar environments include OTUs most closely related to the genera

Pseudomonas, Psychrobacter, Shewanella, Marinobacter, Loktanella, Massilia, Polaromonas, and Limnobacter, which are seen in these samples in lower relative abundance. Aqueous geochemistry results show a clear trend between the concentrations of chloride and sulfate ions between melt pools/spring discharge and aufeis/mineral precipitates, however, this trend does not appear to drive microbial community differentiation. While some acute dynamic changes in microbial community were observed in the short-term, highlighted by a large relative abundance

12 of Epsilonproteobacterial genera, these abundance values did not persist over time. More intriguing is the evidence that a base community of microorganisms exists in this unique sulfur- rich community across multiple materials, site types, and time scales.

Chapter 3 expands on the work presented in Chapter 2 and delves into the functional potential of microbial communities present in the BFP system. Through a metagenomic approach, whereby all genes from a given environmental sample are sequenced, our data show that while representative polar microorganisms are also present in the many samples and across multiple site types, a large consortium of sulfur oxidizers from the Alpha- and

Betaproteobacterial classes may also be participating heavily in sulfur oxidizing processes. These results are based on the presence of complete and almost complete sox operons (sulfur oxidizing genes), so while this only suggests that these organisms are participating in sulfur oxidation processes, it does not confirm this to be definitively the case. This is however, the next step in determining the major players in these valuable sulfur cycling processes and has shown us that, at least at BFP, sulfur oxidation may be a more widely available microbial metabolism than was previously thought. Such microbial metabolism then better informs what is possible in environmental systems, what metabolites may be produced, and whether any biologic processes are somehow reflected in the environment to serve as biosignatures.

We collected samples at BFP over multiple years (2014, 2016, and 2017) and from varied sample types (aufeis, melt pools, spring fluids, and surface mineral precipitates) for 16S rRNA gene and metagenomic whole genome sequencing (WGS). These data were assembled and binned into Metagenome Assembled Genomes (MAGs) that were queried to identify what metabolic processes are present and dominant across multiple sample types at BFP. We found that a number of sulfur metabolisms (especially those for sulfur oxidation) may be possible and

13 are coupled with carbon fixation derived from photosynthesis and chemolithoautotrophy

(presumably from the anoxic environment where spring fluids originate). This appears to occur across multiple different putative genomes as well as sample types, indicating that sulfur cycling at BFP may be aided by functional redundancy across multiple bacterial classes. This is a concept where distinctly different taxa can coexist in a system, while providing very similar energy-yielding functions (Louca et al., 2018).

Chapter 4, while not directly related to BFP, is a tangential project that details the microbial community composition of various microbial mats collected from four different locations above 80°N in the Canadian High Arctic. These mat samples are from some of the highest latitudes in published literature, and may represent how primary producers in low- temperature systems can thrive in extreme environments. Moreover, their presence in these high latitudes may contribute to the ever growing signal of climate change, with higher trophic levels forming closer to the poles as the Earth warms. We found that even in similar climates, and separated by small distances (~50 km), microbial community biogeography is highly variable.

16S rRNA gene sequencing data reveals that community diversity, while similar across some sites, is significantly different when it comes to the relative abundances inhabiting the locations.

Furthermore, metagenomes produced from one of these sites, Ice River, a rare, perennially flowing spring in the Arctic, reveals few genes associated with chemotrophy and more genes associated with diverse carbon cycling capabilities. A number of stress genes for cold shock proteins, osmotic stress, and universal stressors, were found to be abundant across mapped genomes. This characterization of High Arctic microbial mats has expanded our knowledge of layered microbial communities in an extreme low-temperature environment; however, further geochemical and metagenomic analysis of these sites would allow us to tease apart the major

14 differences in phylogeny and what is driving community shifts in these extreme locations.

1.6 Conclusions

Chapter 2 has been accepted for publication into Frontiers in Microbiology, while

Chapters 3 and 4 are slated for publication in refereed journals. These three chapters form the crux of my knowledge and skill gained at the Colorado School of Mines. Ideally this work, as a whole, will help to better inform future microbiologists and scientists about the dynamic nature of microorganisms in low-temperature environments, not limited to Arctic locales. The chapters herein can serve as the next step in piecing together how organisms in these extreme environments survive and adapt and will hopefully continue to inform astrobiologists about the importance of understanding terrestrial analog ecosystems when considering astrobiological targets.

1.7 Publications associated with CSM graduate degree

Trivedi, C. B., Lau, G. E., Grasby, S. E., Templeton, A. S., & Spear, J. R. (2018). Low- temperature sulfidic-ice microbial communities, Borup Fiord Pass, Canadian high Arctic. Frontiers in Microbiology, 9, 1622.

Bradley, J.A., Daille, L.K., Trivedi, C.B., Bojanowski, C.L., Stamps, B.W., Stevenson, B.S., Nunn, H.S., Johnson, H.A., Loyd, S.J., Berelson, W.M. and Corsetti, F.A., 2017. Carbonate- rich dendrolitic cones: insights into a modern analog for incipient microbialite formation, Little Hot Creek, Long Valley Caldera, California. NPJ biofilms and microbiomes, 3(1), p.32.

Lau, G.E., Cosmidis, J., Grasby, S.E., Trivedi, C.B., Spear, J.R. and Templeton, A.S., 2017. Low-temperature formation and stabilization of rare allotropes of cyclooctasulfur (β-S 8 and γ-S 8) in the presence of organic carbon at a sulfur-rich glacial site in the Canadian High Arctic. Geochimica et Cosmochimica Acta, 200, pp.218-231.

Stoll, Z.A., Ma, Z., Trivedi, C.B., Spear, J.R. and Xu, P., 2016. Sacrificing power for more cost- effective treatment: A techno-economic approach for engineering microbial fuel cells. Chemosphere, 161, pp.10-18.

15 1.8 References

Aguilera, A., Souza-Egipsy, V., González-Toril, E., Rendueles, O., and Amils, R. (2010). Eukaryotic microbial diversity of phototrophic microbial mats in two Icelandic geothermalhot springs. Int. Microbiol. 13, 21–32. doi:10.2436/20.1501.01.109.

Andriuzzi, W. S., Stanish, L. F., Simmons, B. L., Jaros, C., Adams, B. J., Wall, D. H., et al. (2018). Spatial and temporal patterns of microbial mats and associated invertebrates along an Antarctic stream. Polar Biol., 1–11. doi:10.1007/s00300-018-2331-4.

Barker Jøgensen, B. (1990). The sulfur cycle of freshwater sediments: Role of thiosulfate. 35, 1329–1342. Available at: https://aslopubs.onlinelibrary.wiley.com/doi/pdf/10.4319/lo.1990.35.6.1329 [Accessed May 26, 2018].

Barton, L. L., and Fauque, G. D. (2009). “Chapter 2 Biochemistry, Physiology and Biotechnology of Sulfate‐Reducing Bacteria,” in Advances in applied microbiology, 41–98. doi:10.1016/S0065-2164(09)01202-7.

Boetius, A., Anesio, A. M., Deming, J. W., Mikucki, J. A., and Rapp, J. Z. (2015). Microbial ecology of the cryosphere: sea ice and glacial habitats. Nat. Rev. Microbiol. 13, 677–690. doi:10.1038/nrmicro3522.

Bowles, M. W., Mogollón, J. M., Kasten, S., Zabel, M., and Hinrichs, K.-U. (2014). Global rates of marine sulfate reduction and implications for sub-sea-floor metabolic activities. Science 344, 889–91. doi:10.1126/science.1249213.

Bradley, J. A., Daille, L. K., Trivedi, C. B., Bojanowski, C. L., Stamps, B. W., Stevenson, B. S., et al. (2017). Carbonate-rich dendrolitic cones: insights into a modern analog for incipient microbialite formation, Little Hot Creek, Long Valley Caldera, California. npj Biofilms Microbiomes 3, 32. doi:10.1038/s41522-017-0041-2.

Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Lozupone, C. A., Turnbaugh, P. J., et al. (2011). Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc. Natl. Acad. Sci. U. S. A. 108 Suppl 1, 4516–22. doi:10.1073/pnas.1000080107.

Dahl, C., and Friedrich, C. G. (2008). Microbial Sulfur Metabolism. , eds. C. Dahl and C. G. Friedrich Berlin, Heidelberg: Springer Berlin Heidelberg doi:10.1007/978-3-540-72682-1.

Feazel, L. M., Spear, J. R., Berger, A. B., Harris, J. K., Frank, D. N., Ley, R. E., et al. (2008). Eucaryotic diversity in a hypersaline microbial mat. Appl. Environ. Microbiol. 74, 329–32. doi:10.1128/AEM.01448-07.

Finster, K. (2008). Microbiological disproportionation of inorganic sulfur compounds. J. Sulfur Chem. 29, 281–292. doi:10.1080/17415990802105770.

Gilhooly, W. P., Fike, D. A., Druschel, G. K., Kafantaris, F. C. A., Price, R. E., and Amend, J. P.

16 (2014). Sulfur and oxygen isotope insights into sulfur cycling in shallow-sea hydrothermal vents, Milos, Greece. Geochem. Trans. 15, 1–19. doi:10.1186/s12932-014-0012-y.

Gleeson, D. F., Pappalardo, R. T., Anderson, M. S., Grasby, S. E., Mielke, R. E., Wright, K. E., et al. (2012). Biosignature detection at an Arctic analog to Europa. Astrobiology 12, 135– 150. doi:10.1089/ast.2010.0579.

Gleeson, D. F., Williamson, C., Grasby, S. E., Pappalardo, R. T., Spear, J. R., and Templeton, A. S. (2011). Low temperature S0 biomineralization at a supraglacial spring system in the Canadian high Arctic. Geobiology 9, 360–375. doi:10.1111/j.1472-4669.2011.00283.x.

Grasby, S. E. (2003). Naturally precipitating vaterite (μ-CaCO3) spheres: Unusual carbonates formed in an extreme environment. Geochim. Cosmochim. Acta 67, 1659–1666. doi:10.1016/S0016-7037(00)01304-2.

Grasby, S. E., Allen, C. C., Longazo, T. G., Lisle, J. T., Griffin, D. W., and Beauchamp, B. (2003). Supraglacial sulfur springs and associated biological activity in the Canadian high Arctic—signs of life beneath the ice. Astrobiology 3, 583–596. doi:10.1089/153110703322610672.

Grasby, S. E., Beauchamp, B., and Bense, V. (2012). Sulfuric acid speleogenesis associated with a glacially driven groundwater system - Paleo-spring “pipes” at Borup Fiord Pass, Nunavut. Astrobiology 12, 19–28. doi:10.1089/ast.2011.0700.

Grasby, S. E., Proemse, B. C., and Beauchamp, B. (2014). Deep groundwater circulation through the High Arctic cryosphere forms Mars-like gullies. Geology 42, 651–654. doi:10.1130/G35599.1.

Hamady, M., Walker, J. J., Harris, J. K., Gold, N. J., and Knight, R. (2008). Error-correcting barcoded primers for pyrosequencing hundreds of samples in multiplex. Nat. Methods 5, 235–237. doi:10.1038/nmeth.1184.

Hanson, T. E., Luther, G. W., Findlay, A. J., MacDonald, D. J., and Hess, D. (2013). Phototrophic sulfide oxidation: Environmental insights and a method for kinetic analysis. Front. Microbiol. 4, 1–8. doi:10.3389/fmicb.2013.00382.

Harris, J. K., Caporaso, J. G., Walker, J. J., Spear, J. R., Gold, N. J., Robertson, C. E., et al. (2013). Phylogenetic stratigraphy in the Guerrero Negro hypersaline microbial mat. ISME J. 7, 50–60. doi:10.1038/ismej.2012.79.

Heldmann, J. L., Pollard, W. H., McKay, C. P., Andersen, D. T., and Toon, O. B. (2005). Annual Development Cycle of an Icing Deposit and Associated Perennial Spring Activity on Axel Heiberg Island, Canadian High Arctic. Arctic, Antarct. Alp. Res. 37, 127–135. doi:10.1657/1523-0430(2005)037[0127:ADCOAI]2.0.CO;2.

Hinzman, L. D., Kane, D. L., and Woo, M. (2006). “Permafrost Hydrology,” in Encyclopedia of Hydrological Sciences (John Wiley & Sons, Ltd). doi:10.1002/0470848944.hsa178.

17 Jones, D. S., Schaperdoth, I., and Macalady, J. L. (2016). Biogeography of sulfur-oxidizing Acidithiobacillus populations in extremely acidic cave biofilms. ISME J., 1–13. doi:10.1038/ismej.2016.74.

Jungblut, A. D., Hawes, I., Mackey, T. J., Krusor, M., Doran, P. T., Sumner, D. Y., et al. (2016). Microbial Mat Communities along an Oxygen Gradient in a Perennially Ice-Covered Antarctic Lake. Appl. Environ. Microbiol. 82, 620–30. doi:10.1128/AEM.02699-15.

Kiseeva, E. S., Fonseca, R. O. C., and Smythe, D. J. (2017). Chalcophile Elements and Sulfides in the Upper Mantle. Elements 13, 111–116. doi:10.2113/gselements.13.2.111.

Klotz, M. G., Bryant, D. A., and Hanson, T. E. (2011). The Microbial Sulfur Cycle. Front. Microbiol. 2. doi:10.3389/fmicb.2011.00241.

Kozich, J. J., Westcott, S. L., Baxter, N. T., Highlander, S. K., and Schloss, P. D. (2013). Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the miseq illumina sequencing platform. Appl. Environ. Microbiol. 79, 5112–5120. doi:10.1128/AEM.01043-13.

Kunin, V., Raes, J., Harris, J. K., Spear, J. R., Walker, J. J., Ivanova, N., et al. (2008). Millimeter-scale genetic gradients and community-level molecular convergence in a hypersaline microbial mat. Mol. Syst. Biol. 4, 198. doi:10.1038/msb.2008.35.

Lau, G. E., Cosmidis, J., Grasby, S. E., Trivedi, C. B., Spear, J. R., and Templeton, A. S. (2017). Low-temperature formation and stabilization of rare allotropes of cyclooctasulfur (β-S 8 and γ-S 8 ) in the presence of organic carbon at a sulfur-rich glacial site in the Canadian High Arctic. Geochim. Cosmochim. Acta 200, 218–231. doi:10.1016/j.gca.2016.11.036.

Ley, R. E., Harris, J. K., Wilcox, J., Spear, J. R., Miller, S. R., Bebout, B. M., et al. (2006). Unexpected diversity and complexity of the Guerrero Negro hypersaline microbial mat. Appl. Environ. Microbiol. 72, 3685–95. doi:10.1128/AEM.72.5.3685-3695.2006.

Louca, S., Polz, M., Mazel, F., Albright, M., Julie, H., O’Connor, M., et al. (2018). Function and functional redundancy in microbial systems. Nat. Ecol. Evol. Accepted. doi:10.1038/s41559-018-0519-1.

Luo, C., Tsementzi, D., Kyrpides, N., Read, T., and Konstantinidis, K. T. (2012). Direct comparisons of Illumina vs. Roche 454 sequencing technologies on the same microbial community DNA sample. PLoS One 7, e30087. doi:10.1371/journal.pone.0030087.

Macalady, J. L., Dattagupta, S., Schaperdoth, I., Jones, D. S., Druschel, G. K., and Eastman, D. (2008). Niche differentiation among sulfur-oxidizing bacterial populations in cave waters. ISME J. 2, 590–601. doi:10.1038/ismej.2008.25.

Mandeville, C. W. (2010). Sulfur: A Ubiquitous and Useful Tracer in Earth and Planetary Sciences. Elements 6, 75–80. doi:10.2113/gselements.6.2.75.

Mcintosh, J. C., Schlegel, M. E., and Person, M. (2012). Glacial impacts on hydrologic processes

18 in sedimentary basins: Evidence from natural tracer studies. Geofluids 12, 7–21. doi:10.1111/j.1468-8123.2011.00344.x.

Mikucki, J. A., and Priscu, J. C. (2007). Bacterial diversity associated with blood falls, a subglacial outflow from the Taylor Glacier, Antarctica. Appl. Environ. Microbiol. 73, 4029– 4039. doi:10.1128/AEM.01396-06.

Moorman, B. J. (2003). Glacier-permafrost hydrology interactions, bylot island, canada. Proc. 8th Int. Conf. Permafr., 783–788.

Niederberger, T. D., Perreault, N. N., Lawrence, J. R., Nadeau, J. L., Mielke, R. E., Greer, C. W., et al. (2009). Novel sulfur-oxidizing streamers thriving in perennial cold saline springs of the Canadian high Arctic. Environ. Microbiol. 11, 616–629. doi:10.1111/j.1462- 2920.2008.01833.x.

Parada, A. E., Needham, D. M., and Fuhrman, J. A. (2016). Every base matters: Assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol. 18, 1403–1414. doi:10.1111/1462-2920.13023.

Pepe-Ranney, C., Berelson, W. M., Corsetti, F. A., Treants, M., and Spear, J. R. (2012). Cyanobacterial construction of hot spring siliceous stromatolites in Yellowstone National Park. Environ. Microbiol. 14, 1182–1197. doi:10.1111/j.1462-2920.2012.02698.x.

Perreault, N. N., Andersen, D. T., Pollard, W. H., Greer, C. W., and Whyte, L. G. (2007). Characterization of the prokaryotic diversity in cold saline perennial springs of the Canadian high Arctic. Appl. Environ. Microbiol. 73, 1532–1543. doi:10.1128/AEM.01729- 06.

Perreault, N. N., Greer, C. W., Andersen, D. T., Tille, S., Lacrampe-Couloume, G., Lollar, B. S., et al. (2008). Heterotrophic and autotrophic microbial populations in cold perennial springs of the high arctic. Appl. Environ. Microbiol. 74, 6898–6907. doi:10.1128/AEM.00359-08.

Pollard, W. H. (2005). Icing processes associated with high Arctic perennial springs, Axel Heiberg Island, Nunavut, Canada. Permafr. Periglac. Process. 16, 51–68. doi:10.1002/ppp.515.

Pollard, W., Omelon, C., Andersen, D., and McKay, C. (1999). Perennial spring occurrence in the Expedition Fiord area of western Axel Heiberg Island, Canadian High Arctic. Can. J. Earth Sci. 36, 105–120. doi:10.1139/e98-097.

Prieto-Barajas, C. M., Alcaraz, L. D., Valencia-Cantero, E., and Santoyo, G. (2018). Life in Hot Spring Microbial Mats Located in the Trans-Mexican Volcanic Belt: A 16S/18S rRNA Gene and Metagenomic Analysis. Geomicrobiol. J., 1–9. doi:10.1080/01490451.2018.1454555.

Purcell, A. M., Mikucki, J., Alekhina, I., Achberger, A. M., Barbante, C., Brent, C., et al. (2014). Microbial sulfur transformations in sediments from Subglacial Lake Whillans. Front. Microbiol. 5, 1–28. doi:10.3389/fmicb.2014.00594.

19 Reyes, K., Gonzalez, N. I., Stewart, J., Ospino, F., Nguyen, D., Cho, D. T., et al. (2013). Surface orientation affects the direction of cone growth by Leptolyngbya sp. strain C1, a likely architect of coniform structures Octopus Spring (Yellowstone National Park). Appl. Environ. Microbiol. 79, 1302–8. doi:10.1128/AEM.03008-12.

Robertson, C. E., Spear, J. R., Harris, J. K., and Pace, N. R. (2009). Diversity and stratification of archaea in a hypersaline microbial mat. Appl. Environ. Microbiol. 75, 1801–10. doi:10.1128/AEM.01811-08.

Sapers, H. M., Ronholm, J., Raymond-Bouchard, I., Comrey, R., Osinski, G. R., and Whyte, L. G. (2017). Biological Characterization of Microenvironments in a Hypersaline Cold Spring Mars Analog. Front. Microbiol. 8, 2527. doi:10.3389/fmicb.2017.02527.

Scheidegger, J. M., Bense, V. F., and Grasby, S. E. (2012). Transient nature of Arctic spring systems driven by subglacial meltwater. Geophys. Res. Lett. 39, 1–6. doi:10.1029/2012GL051445.

Spear, J. R., Ley, R. E., Berger, A. B., and Pace, N. R. (2003). Complexity in natural microbial ecosystems: the Guerrero Negro experience. Biol. Bull. 204, 168–73. doi:10.2307/1543553.

Stibal, M., Gozdereliler, E., Cameron, K. A., Box, J. E., Stevens, I. T., Gokul, J. K., et al. (2015). Microbial abundance in surface ice on the Greenland Ice Sheet. Front. Microbiol. 6, 225. doi:10.3389/fmicb.2015.00225.

Thullner, M., Dale, A. W., and Regnier, P. (2009). Global-scale quantification of mineralization pathways in marine sediments: A reaction-transport modeling approach. Geochemistry, Geophys. Geosystems 10, n/a-n/a. doi:10.1029/2009GC002484.

Varin, T., Lovejoy, C., Jungblut, A. D., Vincent, W. F., and Corbeil, J. (2010). Metagenomic profiling of Arctic microbial mat communities as nutrient scavenging and recycling systems. Limnol. Oceanogr. 55, 1901–1911. doi:10.4319/lo.2010.55.5.1901.

Varin, T., Lovejoy, C., Jungblut, A. D., Vincent, W. F., and Corbeil, J. (2012). Metagenomic Analysis of Stress Genes in Microbial Mat Communities from Antarctica and the High Arctic. Appl. Environ. Microbiol. 78, 549–559. doi:10.1128/AEM.06354-11.

Wacey, D., Kilburn, M. R., Saunders, M., Cliff, J., and Brasier, M. D. (2011). Microfossils of sulphur-metabolizing cells in 3.4-billion-year-old rocks of Western Australia. Nat. Geosci. 4, 698–702. doi:10.1038/ngeo1238.

Wainstein, P., Moorman, B., and Whitehead, K. (2014). Glacial conditions that contribute to the regeneration of Fountain Glacier proglacial icing, Bylot Island, Canada. Hydrol. Process. 28, 2749–2760. doi:10.1002/hyp.9787.

Wasmund, K., Mußmann, M., and Loy, A. (2017). The life sulfuric: Microbial ecology of sulfur cycling in marine sediments. Environ. Microbiol. Rep. 9, 323–344. doi:10.1111/1758- 2229.12538.

20 Wedepohl, K. H. (1984). Sulfur in the Earth’s Crust, its Origin and Natural Cycle. Stud. Inorg. Chem. 5, 39–54. doi:10.1016/B978-0-444-42355-9.50007-4.

Woo, M. (2012). Permafrost Hydrology. Springer Berlin Heidelberg.

Woo, M. K., Kane, D. L., Carey, S. K., and Yang, D. (2008). Progress in permafrost hydrology in the new millennium. Permafr. Periglac. Process. 19, 237–254. doi:10.1002/ppp.613.

Wright, K. E., Williamson, C., Grasby, S. E., Spear, J. R., and Templeton, A. S. (2013). Metagenomic evidence for sulfur lithotrophy by epsilonproteobacteria as the major energy source for primary productivity in a sub-aerial arctic glacial deposit, Borup Fiord Pass. Front. Microbiol. 4, 1–20. doi:10.3389/fmicb.2013.00063.

21

CHAPTER 2

LOW-TEMPERATURE SULFIDIC-ICE MICROBIAL COMMUNITIES,

BORUP FIORD PASS, CANADIAN HIGH ARCTIC

A manuscript accepted for publication in

Frontiers in Microbiology: Extreme Microbiology1

Christopher B. Trivedi2, Graham E. Lau3, Stephen E. Grasby4, Alexis S. Templeton3, John R. Spear2

Abstract

A sulfur-dominated supraglacial spring system found at Borup Fiord Pass (BFP),

Ellesmere Island, Nunavut, Canada, is a unique sulfur-on-ice system expressed along the toe of a glacier. BFP has an intermittent flowing, subsurface-derived, glacial spring that creates a large white-yellow icing (aufeis) that extends down-valley. Over field campaigns in 2014, 2016, and

2017, numerous samples were collected and analyzed for both microbial community composition and aqueous geochemistry. Samples were collected from multiple site types: spring discharge fluid, aufeis (spring-derived ice), melt pools with sedimented cryoconite material, and mineral precipitate scrapings, to probe how microbial communities differed between site types in a

1Reprinted with permission Trivedi, C. B., Lau, G. E., Grasby, S. E., Templeton, A. S., & Spear, J. R. (2018). Low- temperature sulfidic-ice microbial communities, Borup Fiord Pass, Canadian high Arctic. Frontiers in Microbiology, 9, 1622. 2Department of Civil and Environmental Engineering, Colorado School of Mines, Golden, CO, 80401 USA 3Department of Geological Sciences, University of Colorado, Boulder, Boulder, CO, 80309 USA 4Geological Survey of Canada, Calgary, AB, T2L2A7 Canada

22 dynamic freeze/thaw sulfur-rich system. Dissolved sulfate varied between 0.07 - 11.6 mM and was correlated with chloride concentrations, where the fluids were saltiest among spring fluids.

The highest sulfate samples exhibited high dissolved sulfide values between 0.22 - 2.25 mM.

16S rRNA gene sequencing from melt pool and aufeis samples from the 2014 campaign were highly abundant in operational taxonomic units (OTUs) closely related to sulfur-oxidizing microorganisms (SOM; Sulfurimonas, Sulfurovum, and Sulfuricurvum). Subsequent sampling two weeks later had fewer SOMs and showed an increased abundance of the genus

Flavobacterium. Desulfocapsa, an organism that specializes in the disproportionation of inorganic sulfur compounds was also found. Samples from 2016 and 2017 revealed that microorganisms present were highly similar in community composition to 2014 samples, primarily echoed by the continued presence of Flavobacterium sp. Results suggest that while sulfur-cycling organisms may dominate during acute events, a basal community structure appears to dominate over time and site type. These results further enhance our knowledge of low-temperature sulfur systems on Earth, and help to guide the search for potential life on extraterrestrial worlds, such as Europa, where similar low-temperature sulfur-rich conditions may exist.

2.1. Introduction

Microorganisms that use sulfur as a source of energy are ubiquitous on Earth and are found in diverse environments (Dahl and Friedrich, 2008; Klotz et al., 2011). In marine hydrothermal fields are an abundance of organisms spanning the Proteobacteria that are involved in sulfur redox reactions (Perner et al., 2007). Sulfur oxidizing organisms are also found in deep cave systems, where chemolithoautotrophy is utilized to oxidize reduced sulfur species as electron

23 donors creating biofilms that contain up to 50% S0 by mass (Hamilton et al., 2015). Due to its ability to serve as both an electron donor and/or electron acceptor across an eight electron transfer redox state (-2 to +6), sulfur is a primary ingredient for life on Earth, and in particular to microorganisms across a multitude of environments.

One of these extreme environments is the sulfur-dominated supraglacial spring system found at Borup Fiord Pass (BFP), Ellesmere Island, Nunavut, Canada (Grasby et al., 2003). BFP is one of the few known low-temperature springs in the Canadian Arctic, along with Gypsum

Hill, Colour Peak on Axel Heiberg Island, and Ice River on Ellesmere Island (Gleeson et al.,

2011; Grasby, 2003; Grasby et al., 2012; Lau et al., 2017; Niederberger et al., 2009; Perreault et al., 2007; Pollard et al., 1999; Wright et al., 2013). The spring at BFP is unique as it discharges from a glacier and has very high concentrations of dissolved hydrogen sulfide (H2S) for a geological system (Gilhooly et al., 2014; Sievert et al., 2007), the highest of which was measured from spring fluid collected in 2009 at 3.99 mM. Biotic and/or abiotic oxidation of H2S to elemental sulfur creates visually striking yellow glacial ice, as well as yellow colored aufeis formed by spring discharge into sub-zero air temperatures. This sulfur-covered ice extends for tens to thousands of square meters down valley (Gleeson et al., 2012; Grasby et al., 2003; Lau et al., 2017). Spring water discharging from the glacier is briny, sulfide-rich, and hosts a diverse microbial community that may make use of sulfur metabolisms for growth (Gleeson et al., 2011;

Grasby et al., 2003; Wright et al., 2013). When discharging during cold winter months the spring water freezes, forming sulfidic aufeis as well as associated cryogenic minerals (Grasby, 2003;

Grasby et al., 2003; Lau et al., 2017). During the subsequent summer months surface melt pools form, releasing trapped hydrogen sulfide gas that is then oxidized to elemental sulfur that accumulates as sulfur-rich cryogenic materials (Lau et al., 2017). It remains unclear to what

24 degree, H2S oxidation is driven by microbial interaction (Gleeson et al., 2011; Grasby et al.,

2003; Wright et al., 2013), and how microbial community dynamics function in these complex sulfidic-ice environments.

The aim of this study was to use 16S rRNA gene sequencing coupled to geochemical analysis to identify and better constrain the microbial consortia that have adapted to live in this sulfur-rich, low-temperature environment. To address this, field campaigns to BFP over multiple years were undertaken to collect samples from multiple site types (spring discharge, aufeis, mineral precipitates, and melt pool waters).

2.1.1 BFP Site Background

Previous research at BFP includes data about spring geochemistry, permafrost- hydrogeology, microbiology, biomineralization, and putative sulfur metabolisms (Gleeson et al.,

2010, 2011, 2012, Grasby et al., 2003, 2012; Lau et al., 2017; Scheidegger et al., 2012; Wright et al., 2013). BFP is at the height of a north-south trending valley through the Krieger mountains at

81o 01’ N, 81o 38’ W (Figure 2.1a). The spring system is approximately 210-240 m above sea level and discharges from glacial ice near the toe of two coalesced glaciers (Grasby et al., 2003).

The spring system is thought to be perennial but discharges from different locations of the glacier from year to year (Gleeson et al., 2010; Grasby et al., 2003; Lau et al., 2017; Wright et al., 2013). Given the lack of winter observations due to 24-hour darkness, it cannot be stated with certainty that the spring is perennial; however, the annual formation of extensive sulfidic aufeis in the proglacial area indicates active discharge during the winter. The aufeis formed is characterized by extensive coverage of elemental sulfur, calcite, and gypsum, as well as rare minerals of vaterite and β- and γ-cyclooctasulfur (Grasby, 2003; Grasby et al., 2003; Lau et al.,

25 2017), all related to cryogenic processes. It is not clear, however, in what way microorganisms may play a part in their formation or utilization. A lateral fault approximately 100 m south of the toe of the glacier has been implicated in playing a role in the subsurface hydrology, in that this may be one of the areas where subsurface fluids are able to contact the surface (Grasby et al.,

2003; Scheidegger et al., 2012). Furthermore, it has been suggested that this fault may facilitate fluid flow underneath the glacier, which could deliver available organic carbon for potential subsurface microbial processes such as sulfate reduction (Grasby et al., 2003). This process would generate sulfide, which travels to the surface as dissolved aqueous phase sulfide for abiotic surface oxidation and/or biological metabolism.

2.2 Materials and Methods

2.2.1 Field Work and Sample Collection

Sample material for this study was collected during a two-week period in 2014 (21 June thru 2 July) as well as during single-day visits in 2016 (4 July) and 2017 (7 July). Samples included: 1) background control samples of glacial fluid (G1-G3); 2) fluid from an active sulfide- spring in 2016 (Spring 2016); 3) melted and filtered fluid from the aufeis (A1-A6); 4) fluid and cryoconite material collected from the melt pools on the surface of the aufeis (M1-M6); and 5) mineral precipitates that were scraped from the surface of the aufeis (two from 2014 and five from 2017; AS1-AS7).

Cryoconite sediments (melt pools and sample G1) were collected using sterile and field- washed transfer pipettes. Up to 0.5 g of this material plus water was placed into ZR

BashingBead™ lysis tubes (Zymo Research Corp., Irvine, CA, USA) containing 750 µL of

Xpedition™ Lysis/Stabilization Solution (Zymo Research Corp.) and homogenized for 45

26

Figure 2.1. Borup Fiord Pass: Sample locations and types. A) Line drawing and map showing the location of Borup Fiord Pass in the Canadian High Arctic. The inset shows the location of Borup Fiord Pass on Ellesmere Island (from Grasby et al., 2012). B) Satellite imagery of the toe of the coalescence glacier taken 2 July, 2014 during this sampling campaign. Highlighted sample sites are melt pool (gold circles), aufeis (red squares), mineral precipitates (teal square), and glacier samples (green diamonds). The location of the previous 2016 spring (black triangle) is also shown. Also highlighted are general regions: Glacial Ice (dark blue dashed line), Sulfidic Aufeis (orange dashed line), the area referred to as the Ice Blister (red oval), as well as the region of a newly emerged spring from 2016 (yellow dashed line). 2017 mineral precipitate samples are not shown, however, samples were collected within the Sulfidic Aufeis zone near the Ice Blister from 2014 (this can be seen in Figure A.3a). Also note the location of the 2014 campsite on the far east side of the figure for added scale. C) Examples of melt pool (top left), aufeis (top right), mineral precipitate (bottom left), and glacier (bottom right) sample types. Professor Alexis Templeton (top left), a five gallon rock pail (top right), Dr. Graham Lau (bottom left), and a hiking boot (bottom right) for relative scale. Satellite image courtesy of the DigitalGlobe Foundation.

27

28 seconds using the Zymo Xpedition™ portable impact drill (Zymo Research Corp.). Samples were then stored on ice until DNA extraction was performed. Mineral precipitate samples from

2014 (AS1 & AS2) were collected in 50 mL polypropylene conical tubes (VWR International,

Radnor, PA, USA). Samples were scraped from surface ice, and the resulting mixture was melted then filtered through 0.22 µm Luer-lok Sterivex™ filters (EMD Millipore; Darmstadt,

Germany), which were then capped and kept on ice in the field until returning to the laboratory where they were stored at -20 °C until extraction. Mineral precipitate scrapings from the 2017 field campaign (AS3 - 7) were collected using sterile and field-washed transfer pipettes and up to

0.5 g of material were mixed in ZR BashingBead™ lysis tubes containing 750 mL of DNA/RNA

Shield™ (Zymo Research Corp.). Samples were shaken and kept at 4 °C until returned to the laboratory, where they were then stored at -80 °C until DNA/RNA extraction could be performed. Aufeis samples were prepared by scraping approximately 10 cm of surface ice away and then digging out an area approximately half a meter wide, 30 cm tall, and 15 cm deep. The ice was placed in field-washed five gallon pails and allowed to thaw. Spring water fluid (Spring

2016) and melted aufeis (A1 - A6) samples were prepped for DNA extraction by filtration through 0.22 µm Luer-lok Sterivex™ filters (EMD Millipore), which were then capped and kept on ice in the field until returning to the laboratory where they were stored at -20 °C until extraction. Filtered water volumes are shown in Table A.1. Fluids for aqueous geochemical measurement were filtered with 0.22 µm polyethersulfone (PES; EMD Millipore, Billerica, MA,

USA) filters into 15 mL polypropylene tubes for ion chromatography (IC) and inductively- coupled plasma optical emission spectroscopy (ICP-OES) for measurement of major anions and cations, respectively. Samples for ICP-OES were acidified with approximately 0.1 mL nitric acid in the field.

29

2.2.2 DNA Extraction and 16S rRNA Gene Library Sequencing

DNA extraction of 2014 cryoconite (melt pool and glacial ice) samples was performed using the Xpedition™ Soil/Fecal DNA MiniPrep kit (Zymo Research Corp.). DNA extraction of fluids collected via 0.22 µm Luer-lok Sterivex™ filters (melt pool, spring discharge, aufeis, and

2014 mineral precipitates) was performed using the MO BIO PowerWater® Sterivex™ DNA

Isolation Kit (QIAGEN, Hilden, North Rhine-Westphalia, Germany), and DNA of 2017 mineral precipitate samples was extracted using the ZymoBIOMICS™ DNA/RNA Mini Kit (Zymo

Research Corp.). All extractions were performed according to the manufacturer’s instructions.

Extracted DNA was amplified via PCR using 16S rRNA V4 primers 515F_Y and 926R (Parada et al., 2016). The Forward primer (M13-515_Y: 5'- GTA AAA CGA CGG CCA GTC CGT

GYC AGC MGC CGC GGT AA-3') contains the M13 forward primer (bold) ligated to the 16S rRNA gene-specific sequence (underlined) to allow for barcoding in a subsequent PCR reaction

(Stamps et al., 2016). The reverse primer (926R: 5’-CCG YCA ATT YMT TTR AGT TT-3’) was described in Parada et al. (2016).

Each 25 µL PCR reaction mixture consisted of: 1X 5PRIME HOT MasterMix (Quantabio,

Beverly, MA, USA), 0.2 µM of each primer, molecular grade water, and 2 µL of extracted template DNA. Positive (ZymoBIOMICS™ Microbial Community Standard; Zymo Research

Corp.) and negative (no template) controls were also amplified along with sample template reactions. PCR included the following parameters: initial denaturation at 94 °C for 2:00 min, 30 cycles of (94 °C for 45 s, 50 °C for 45 s, and 68 °C for 1:30 min), final extension of 68 °C for 5 min, and a final 4 °C hold. A second, six cycle PCR was used to add a unique 12 bp barcode

(Hamady et al., 2008) to each previously amplified sample using a forward primer containing the

30 barcode+M13 forward sequence (5′-3′) and the 926R primer (see Appendix E, A.S1). The final barcoded PCR products were cleaned using KAPA Pure Beads (KAPA Biosystems Inc.,

Wilmington, MA, USA) at a final concentration of 0.8X v/v, quantified using the Qubit® dsDNA HS assay (Life Technologies, Carlsbad, CA, USA), pooled in equimolar amounts, and concentrated to a final volume of 80 μL using two Amicon® Ultra-0.5 mL 30K Centrifugal

Filters (EMD Millipore, Billerica, MA, USA).

The final pooled library was run on the Illumina MiSeq platform (Illumina, San Diego,

CA, USA) at the Duke Center for Genomic and Computational Biology (Duke University,

Durham, NC, USA) using PE250 V2 chemistry. After sequencing, reads were merged and de- multiplexed using QIIME (Caporaso et al., 2010b), filtered by quality, clustered into operational taxonomic units (OTUs) at 97% identity, and chimera checked using VSEARCH (Rognes et al.,

2016). OTU taxonomy was assigned using UCLUST (Edgar, 2010) and the SILVA database

(Release 123; Pruesse et al., 2007). Representative sequences were aligned using pyNAST

(Caporaso et al., 2010a) against an aligned version of the SILVA r123 database. A mapping file is included as file A.S1 (Appendix E), and the commands used to produce the final BIOM file are publicly available at https://doi.org/10.5281/zenodo.1207309. Additional diversity metrics including alpha rarefaction by observed OTUs, rank abundance, and sequencing coverage are included in the supplemental material (Figures A.5 & A.6, and file A.S2, respectively).

Estimated relative biomass for each sample was calculated based on sample amount and extracted DNA quantity as measured on the Qubit® fluorometer (Life Technologies, Carlsbad,

CA, USA). DNA concentration was multiplied by the elution volume from extraction (to get ng of DNA) and then another ~53% was added by weight to account for the mass lost during extraction. This percentage (53%) was calculated based on the volume of starting lysis buffer

31 (750 µl) prior to bead beating divided by the volume that is taken through the rest of the protocol

(400 µl). Total estimated biomass of a sample (cells mL-1 or cells mg-1, depending if the sample was extracted from a filter or sediment, respectively) was calculated based on an empirical value of 2.5 fg DNA cell-1 (Button and Robertson, 2001) from aquatic microorganisms.

2.2.3 Aqueous Geochemistry

Fluid pH (for 2014 and 2017 samples) was measured in the field with test strips (Carolina

Biological Supply Co.; Burlington, NC, USA). Temperatures were measured with a Hach

HQ40d portable field meter (Hach Company, Loveland, CO, USA). For 2014 samples, ICP-OES and IC measurements were performed on an ARL (Applied Research Laboratories) 3410+ ICP-

OES and a Dionex (Thermo Scientific; Waltham, MA, USA) 4500I IC, respectively, at the

Laboratory for Environmental and Geological Studies (LEGS) at the University of Colorado

Boulder. Total organic carbon (TOC) measurements for cryoconite material from M1 were conducted according to Lau et al. (2017)

Spring fluid collected during the summer of 2016 was analyzed at the Colorado School of

Mines. Major anions were measured using a Dionex (Thermo Scientific; Waltham, MA, USA)

ICS-90 ion chromatography system running an AS14A (4 × 250 mm) column. Major cations were measured using a Perkin-Elmer (Waltham, MA, USA) Optima 5300 DV ICP-OES. BFP-

2016 spring fluid was filtered and acidified in the same fashion as above for 2014 samples. Due to low/no-flow conditions, aqueous geochemistry data for 2017 samples was not collected.

Aqueous geochemistry was ordinated in R v3.4.3 (R Core Team, 2016) using the community ecology package “vegan” (Oksanen et al., 2016). Data was ordinated using the built-in Hellinger standardization. A principal coordinates analysis (PCoA) plot was created using the packages

32

Figure 2.2. Log-log plot of sulfate vs chloride aqueous geochemistry data. Sulfate and chloride aqueous geochemistry data appear to trend together. Note the “Control” samples that have been added for geochemical comparison, however, no 16S rRNA gene sequencing data are associated with these samples. Both sites (M4 and M5) located on top of the Ice Blister are lower in sulfate and chloride concentrations than those located just to the west (Sites M2 & M3) and south (G2). These also match closely with aqueous geochemistry data taken from spring fluids from 2009 and 2016.

33 “ggplot2” and “ggfortify” (Tang et al., 2016; Wickham, 2016) to determine which geochemical factors might influence observed differences among the different site types (Figures 2.2 and

A.4).

2.3 Results

2.3.1 General Field Observations

The 2014 field season at BFP revealed a large sulfur-covered area of ice, Sulfidic Aufeis

(Figure 2.1b). This area included an approximately 25 m diameter, sulfur-rich mound (herein referred to as the Ice Blister) at the toe of the glacier as well as what was likely spring-derived ice that was approximately 5 x 104 m2 in size and approximately 5 m thick at greatest extent (Lau et al., 2017). The majority of the aufeis had built up around the toe of the glacier while the rest followed a channel southeast down the valley for approximately 2 km (Figure 2.1b). While no flowing spring was observed during 2014, this spring-derived aufeis likely originated from within the sulfidic zone of the Ice Blister (Figure 2.1b). Ice blister formation (also known as a frost mound) is a common feature that forms in polar zones when fluid flow is present, often in the form of a spring (Pollard, 2005; Wainstein et al., 2008), and the spring was likely active sometime prior to arrival at the site in 2014.

The locations of sampled melt pools, aufeis sites, mineral precipitate sites, glacier sites, and the Ice Blister are shown in Figure 2.1b, which is a satellite image captured on 2 July 2014.

For comparison, the location of the 2016 spring discharge site is labeled to the east of the

Sulfidic Aufeis zone from 2014. Mineral precipitate samples were all collected within the

Sulfidic Aufeis zone, closer to the Ice Blister (see Figure A.3). These samples were chosen due to their variability in material type (i.e. fluid, ice, precipitate, etc.) as well their potential to

34 represent how microbial communities may or may not change in a dynamic low-temperature environment.

2.3.2 Estimated Biomass and Sample QC

Estimated biomass values (calculated from extracted DNA; Table A.1) across all samples averaged 1.30 x 104 cells mL-1 and 1.20 x 105 cells mg-1, depending on sample type (fluid or solid material). Average biomass calculated for each site type were as follows: 2.63 x 103 cells mL-1 and 4.34 x 103 cells mg-1 for melt pools; 2.70 x 103 cells mL-1 for aufeis; 6.36 x 104 cells mg-1 and 1.55 x 105 cells mL-1 for mineral precipitate samples; and 7.65 x 104 and 1.29 x 104 cells mL-1 for glacier samples. The 2016 spring fluid sample had a calculated biomass of 3.11 x

103 cells mL-1. It should be noted that biomass values are most likely an underestimate due to potential loss of DNA during extraction.

2.3.3 Melt Pools and 2016 Spring Fluid Samples

Along with the Ice Blister feature, melt pools (top left panel, Figure 2.1c) were observed across the aufeis surface (see Figure A.1). Melt pools (comprising samples M1-M6) close to the

Ice Blister exhibited a strong odor of hydrogen sulfide and had a yellow opaque layer that had formed across the surface of the water. Melt pool sizes varied but averaged 1-2 meters in width and length and were between 10 and 20 cm in depth. Many of the melt pools had cryoconite material on the bottom that was white/yellow in color and in at least one, was found to contain elemental sulfur and other sulfate minerals (Lau et al., 2017). Sites M4 and M5 are located on the south and north ends of the Ice Blister, respectively, while M2 and M3 are just to the west of the

Ice Blister (Figure A.1). Sites M1 and M6 are the furthest removed from the Ice Blister, one

35 being at the bottom of a flowing waterfall (M6) and the other being the furthest most melt pool sampled down the Sulfidic Aufeis (M1). The spring discharge sample from 2016 was to the east of the 2014 Sulfidic Aufeis and emanated higher up on the glacier (Figure 2.1b). The 2016 spring fluid is grouped with the melt pool samples based on similar community resemblance and aqueous geochemistry. Sulfide values for melt pool samples were measured for site M2 and from ice near to sites M3 and M4 (0.220, 1.00, and 2.25 mM respectively; Table 2.1). Sulfide was not measured for the 2016 spring sample. Sulfate values were measured for melt pool samples at sites M1-M5, and ranged in value from 3.28 to 11.6 mM. The 2016 spring sample contained 14.8 mM sulfate, which is the highest reported value in this study. Chloride values for melt pools and the 2016 spring ranged from 3.96 to 68.79 mM. The latter value belongs to the 2016 spring, again, the highest reported chloride value of this study.

Initial inquiry into the microbial communities of melt pool sites shows a wide range of organisms among the phylum Proteobacteria (Figure 2.3). Many of the operational taxonomic units (OTUs) most closely related to known genera (e.g. Pseudomonas, Psychrobacter,

Shewanella, Marinobacter, Loktanella, Massilia, Polaromonas, and Limnobacter) are ones that have been found in polar and glacial environments previously (Bej et al., 2009; Gleeson et al.,

2011; González-Toril et al., 2009; Van Trappen, 2004; Zhou et al., 2015). An OTU most closely related to the genus Flavobacterium was highly abundant in a number of melt pool samples (M1,

M3b, M4a-c, and M5). When queried using NCBI BLAST, the top hit (99% identity) was most closely related to Flavobacterium psychrolimnae (GenBank Accession: KY047392.1) discovered in a wetland sample. This appears to be a common theme in a majority of samples collected at

BFP, where Flavobacterium was a dominant organism across all sample types and across different years.

36 Table 2.1. Aqueous geochemistry for selected BFP sites. Aqueous geochemistry values (in mM) for the selected melt pool, aufeis, mineral precipitate, glacier, 2016 and 2009 spring fluids. Also presented are the category type and the date each sample was collected. Aqueous geochemistry values are those used to calculate the charge balance error, which is presented in Table A.1 along with microbiological data.

2016 2009 M1 M2 M3 M4 M5 A1 A2 A3 A4 A6 AS2 AS3b AS4b G2 Spring Spring Jul 7 Jul 7 Date Jun 21 Jun 21 Jun 26 Jun 27 Jun 30 Jun 25 Jun 25 Jun 28 Jun 28 Jun 30 24-Jun Jun 30 Jul 4 16 Jul 09 17 17 Type Melt Pool Aufeis Mineral Precipitate Glacier Spring fluid Sulfidea (NM) 0.2200 1.0000 2.25 (NM) 0.5200 (NM) (NM) 0.3200 (NM) (NM) (NM) (NM) (NM) (NM) 3.9900 F- 0.0127 0.0326 0.0266 (DL) 0.0128 0.0116 (DL) 0.0248 (DL) (DL) (DL) (NM) (NM) (DL) 0.0426 0.0542 Cl- 9.5815 41.9671 31.5257 3.9628 5.9468 6.8968 4.6456 23.3210 3.6215 0.1105 0.29126 (NM) (NM) 0.0128 68.7924 39.2000 Br- (DL) (DL) 0.0093 (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (NM) (NM) (DL) 0.0312 0.0330 - NO3 (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (NM) (NM) 0.0189 (DL) 0.0030 2- SO4 3.7149 11.6948 9.1208 3.8064 3.2843 1.7004 1.4300 8.6216 2.8018 0.0666 3.22504 (NM) (NM) 0.0146 14.8151 13.0000 2- S2O3 (DL) 0.2182 (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (DL) (NM) (NM) (DL) (DL) 0.1030 Si 0.0224 0.0940 0.0477 0.0096 0.0050 0.0075 0.0039 0.0459 0.0050 (DL) 3.14028 (NM) (NM) (DL) 0.0887 0.0411

MnT 0.0002 0.0004 0.0004 0.0002 (DL) 0.0002 0.0002 (DL) (DL) 0.0004 0.0032 (NM) (NM) (DL) 0.0007 0.0010

FeT (DL) (DL) (DL) (DL) 0.0005 (DL) (DL) 0.0004 0.0002 (DL) (DL) (NM) (NM) (DL) 0.0000 0.0050 Mg+ 1.5626 6.9665 3.9193 0.7427 0.8171 1.1779 0.6223 3.3213 0.8261 0.0177 0.07069 (NM) (NM) (DL) 10.1510 12.9000 Ca2+ 3.8263 11.3045 9.3785 3.5040 2.8575 1.7765 1.4410 8.2013 2.7488 0.1085 3.2365 (NM) (NM) 0.0053 14.7511 17.3000 Na+ 6.0439 27.0679 15.9823 2.8462 3.0520 4.3296 2.1968 13.6151 3.2334 0.1061 0.29406 (NM) (NM) (DL) 39.6291 51.8000 K+ 0.0470 0.1991 0.0991 (DL) (DL) 0.0414 0.0239 0.0889 0.0128 0.0128 0.01049 (NM) (NM) 0.0320 0.2619 0.3020 pH 5.5-6.8 5.5-6.0 (NM) (NM) (NM) (NM) (NM) (NM) (NM) (NM) (NM) 7.0 7.0 (NM) (NM) (NM)

Values presented as (DL) were below the detection level of the instrument.Values presented as (NM) were not measured for that - - - 2- geochemical parameter. Relevant DL values are 1.1 µM (F ), 0.6 µM (Br ), 0.8 µM (NO3 ), 1.8 µM (S2O3 ), 0.7 µM (Si), 0.02 µM + + + (MnT), 0.1 µM (FeT), 0.6 µM (Mg ), 1.9 µM (Na ), 4.0 µM (K ). aAll sulfide data were collected on 28 June 2014. bOnly pH was collected for samples during 2017.

37

Figure 2.3. BFP melt pool bar charts. Bar charts show relative abundance of 16S rRNA gene sequencing libraries, where only OTUs representing organisms more than 5% of any given sample are labeled. All other OTUs below 5% relative abundance are grouped together in the category “Genus <5%”. Labels show Class and subsequent Genus level. Sites are in order of sampling date from left to right. All melt pool sites are within the Sulfidic Aufeis zone. Sites are named such that any replicates are denoted by lowercase letters. Exact locations of all melt pool sites can be seen in Figure A.1.

Regarding sulfur cycling in the Epsilonproteobacteria, OTUs most closely related to the sulfur-oxidizing microorganisms (SOMs) Sulfurovum and Sulfurimonas are also present, and have been previously observed at BFP (Gleeson et al., 2011; Wright et al., 2013). Site M1, located southeast on the Sulfidic Aufeis, was dominated by genera commonly found in polar environments, such as Pseudomonas, Psychrobacter, Polaromonas, Limnobacter, and Loktanella

(11.5%, 33.8%, 5.7%, 8.2%, and 6.6%, respectively). Site M2 was dominated by an OTU most

38 closely related to the genus Sulfurimonas, across all four replicates (a-d), all greater than 45% relative abundance. This is also true for biological replicate M3a, but not for M3b, potentially due to the different sampling date. An OTU most closely related to Sulfurovum was found in small amounts in site M2 (<5%) and larger amounts in sites M4 (3.8 - 9.1%), M5 (15.7%), and the 2016 Spring fluid (31.1%). The presence of an OTU most closely related to the

Deltaproteobacterial genus Desulfocapsa was also found in many of the melt pool sites, with the highest abundance being seen in M2 and M3, at 11.9% and 8.1% respectively. Desulfocapsa is a relatively well-known sulfur-disproportionating microorganism which has been found mostly in subsurface sediments (Finster et al., 1998; Zeng et al., 2017), but was also shown to be present in

Antarctica’s “Blood Falls” (Mikucki and Priscu, 2007).

An OTU most closely related to the Alphaproteobacterial genus Brevundimonas was present in low abundance in sites M2 and M3, with the largest abundance seen in the 2016 spring sample at 6.4%. This organism was also found previously in Antarctic and alpine environments

(González-Toril et al., 2009) and was implicated as psychrotolerant and highly radiation resistant

(Dartnell et al., 2010). Also of note were appreciable amounts of an OTU most closely related to chloroplast, especially in sites M4 (>5%) and M6 (39.5%). When queried using NCBI BLAST, the top hit (at 97% identity) was most closely related to an “uncultured bacterial clone” from

Antarctic glacial ice (GenBank Accession: KP190124.1). It is not clear whether this chloroplast sequence belongs to a cyanobacterial or eukaryotic species.

2.3.4 Aufeis Samples

Aufeis samples, A1-A6 (top right panel, Figure 2.1c; see Figure A.2 for site locations), which are spread along a roughly 2 km area of the Sulfidic Aufeis, appeared to be layered in

39 place, indicating that multiple flow and freeze cycles had occurred during their deposition.

Sulfide was also measured for aufeis samples from sites A1 and A4 and ranged from 0.32 mM -

0.52, respectively (Table 2.1). Sulfate values for aufeis sites A1-A4 and A6 ranged in value from

0.07 - 8.62 mM. Chloride concentrations for aufeis sample with associated geochemistry ranged from 0.1105 - 23.32 mM (A6 and A3, respectively).

16S rRNA gene sequencing data for aufeis samples were obtained across six different sites with varying distances in relation to the Ice Blister (see Figures 2.1b & A.2). Figure 2.4 is a bar chart of these sites organized by sampling date from left to right. Their orientation along the

Sulfidic Aufeis can be seen in Figure A.2. Similar to the melt pools are OTUs most closely related to bacterial genera that were observed in polar environments previously, such as

Pseudomonas, Psychrobacter, Massilia, Polaromonas, and Loktanella. Similarly, we found the presence of the SOM Sulfurovum, as well as Sulfuricurvum. In previous research, Wright et al.

(2013) found these two SOM OTUs to be abundant in sediment samples, typically with

Sulfurovum being the more abundant of the two organisms. In two of the aufeis sites, Sulfurovum represents ~69% of site A1 and ~6% of site A4. However, further down the Sulfidic Aufeis in

Site A6 we observed Sulfuricurvum representing 17.2% of organisms, whereas Sulfurovum is less than 1% present. Desulfocapsa was also present in site A3 which is roughly 40 m south of the Ice Blister and the closest of all aufeis sites within the Sulfidic Aufeis zone. OTUs most closely related to the genus Flavobacterium were well represented in aufeis site samples.

Flavobacterium sp. was the most dominant organism present in sites A3, A4, A5, and A6 from

~61% all the way up to 75%. The dominant Flavobacterium OTU in melt pool samples was also present in aufeis samples. Another OTU was also present and was most closely related to

Flavobacterium aquidurense (99%) when queried against the NCBI BLAST database.

40 Additionally, an OTU most closely related to the Alphaproteobacterial genus Sphingopyxis represents ~7% in site A4. This organism was previously identified as a marine microorganism with cold-adapted proteins (Ting et al., 2010).

Figure 2.4. Aufeis bar charts show relative abundance of 16S rRNA gene sequencing libraries, where only OTUs representing organisms more than 5% of any given sample are labeled. All other OTUs below 5% relative abundance are grouped together in the category “Genus <5%”. Labels show Class and subsequent Genus level. Sites are named such that any replicates are denoted by lowercase letters. Aufeis samples are in order (from left to right) according to date sampled. Their order as they appear in the Sulfidic Aufeis zone can be seen in Figure A.2.

2.3.5 Mineral Precipitate Samples

Also littering the Sulfidic Aufeis were piles of white/yellow mineral precipitates (AS1-

AS7) that ranged from crystalline to soft, goopy piles of material most likely made up of

41 elemental sulfur, calcite, gypsum, or other sulfur minerals (Grasby, 2003; Grasby et al., 2003;

Lau et al., 2017). These samples (bottom left panel, Figure 2.1c; Figure A.3b and c) were often observed in areas that appeared to be older and less active, unlike a melt pool. Two mineral precipitate samples (AS1 & AS2) were collected from 2014 (Figure 2.1b), while all others (AS3-

AS7) were collected during 2017. All 2017 mineral precipitate samples were taken from the area that remained from the 2014 sampling effort (Figure A.3a). Because samples AS1 and AS2 were filtered, one (AS2) had enough residual fluid to be submitted for aqueous geochemistry. Sulfate values for AS2 were 3.225 mM, while chloride was measured at 0.2912 mM. No aqueous geochemistry data was captured for 2017 mineral precipitate samples (there was no liquid water).

16S rRNA gene sequencing data for mineral precipitate samples (see Figure 2.5) includes

AS1 and AS2 (from 2014, with one replicate each) and five samples collected during the 2017 field campaign (AS3-AS7, with three biological replicates for each, except for AS7 which had two). The microbial communities observed in the mineral precipitates were also present in melt pool and aufeis sites and included OTUs most closely related to the genera Massilia,

Polaromonas, Sphingopyxis, and Brevundimonas. Present in the 2017 samples were organisms not previously seen at BFP, including OTUs most closely related to the Gammaproteobacterial genus Thiomicrospira (sites AS3 and AS6), and the Alphaproteobacterial genera

Polymorphobacter (sites AS4 and AS5) and Methylobacterium (sites AS4 and AS6), all of which were seen in relatively small but consistent amounts (~2%, ~5%, and ~10% respectively). All three OTUs were previously found in Arctic and Antarctic environments (Christner et al., 2014;

Kleinteich et al., 2017; Niederberger et al., 2009). Sulfurimonas was present in small relative abundances in site A7, and Desulfocapsa, while present in very small amounts in almost all samples was most prevalent in site A6 (~6%). The most abundant OTU most closely related to

42

Figure 2.5. Mineral precipitate bar charts show relative abundance of 16S rRNA gene sequencing libraries, where only OTUs representing organisms more than 5% of any given sample are labeled. All other OTUs below 5% relative abundance are grouped together in the category “Genus <5%”. Labels show Class and subsequent Genus level. Sites are named such that any biological replicates are denoted by lowercase letters. The mineral precipitate sites are in order (from left to right) based on sampling date. Two samples are from 2014 (AS1 & AS2), while the rest are from 2017 (AS3-AS7). The 2017 sampling area is shown in Figure A.3.

Flavobacterium, comprising approximately 40-50% abundance of all samples except for site

AS7b. This is the same OTU present in melt pool sites (genus: Psychrolimnae). This site also contained a large abundance of an OTU most closely related to a Bacillus species. Also of high relative abundance were OTUs classified as chloroplast, one of which was also present in two of the melt pool samples. These OTUs were highly represented across all 2017 mineral precipitate samples (averaging above 30% abundance); with only a small percentage (~1.5%) present in

43 2014 sample AS1. The most abundant OTU in mineral precipitate samples, when queried against

NCBI BLAST resulted in a 97% identical match with an “uncultured eukaryote clone”

(GenBank Accession: EU005289.1).

2.3.6 Glacier Samples

Glacier samples (bottom right panel, Figure 2.1c) were collected outside of the Sulfidic

Aufeis zone to be used as comparisons for aqueous geochemistry (G2, Figures 2.2 & A.4) and

16S rRNA gene sequencing data (G1-G3, Figure 2.6). The locations of these sites are also displayed in Figure 2.1b, two of which are north of the Sulfidic Aufeis zone (G1 and G3) and one almost two kilometers to the east of the Sulfidic Aufeis (G2). Site G3 is similar to an aufeis site as it was processed the same way, where subsurface ice was collected and filtered. Site G1 was a melt pool with cryoconite material that was also collected and filtered. Site G2 was ice- melt/glacial runoff, and is the furthest site from the Ice Blister. Sulfate for sample G2 was 0.0146 mM, while chloride was measured at 0.0128 mM. These values are similar to those shown in

Figure 2.2 (Control; values not reported).

16S rRNA gene sequencing data for glacier sites (G1, G2, and G3) are summarized in

Figure 2.6. Surprisingly, in all sites we observed the presence of an OTU most closely related to

Thiobacillus (with the highest abundance being in G3 at ~5%), a sulfur-cycling microorganism, which we did not expect to see outside of the Sulfidic Aufeis zone. The common polar genera

Massilia and Polaromonas are present in various abundances in all glacier samples, and also seen alongside two Betaproteobacteria OTUs that are classified as “Other” and “uncultured”. An

OTU most closely related to the cyanobacterial genus Chamaesiphon, which was found previously in Arctic habitats (Lionard et al., 2012), was present and in its greatest abundance

44

Figure 2.6. Glacier bar charts show only OTUs that are represented above 5% in any given sample. Sites are ordered (from left to right) according to sampling date. G1-G3 only include those control sites where DNA was extracted for 16S rRNA sequencing. Only site G2 also has aqueous geochemistry data associated with it. Locations for glacier sites can be seen in Figure 2.1b.

(~7%) at site G2, just north of the Ice Blister. This organism was also accompanied by an uncultivated cyanobacterial OTU found in both sites G1 and G3, with its greatest abundance in site G3 at ~11%. Two OTUs most closely related to the genera Pedobacter and Ferruginibacter, both in the class Sphingobacteriia, were represented in these samples (~3% and ~10% respectively), and were previously found in Arctic environments (Peter and Sommaruga, 2016;

Zhou et al., 2012). While OTUs most closely related to the genus Flavobacterium were not as

45 abundant as observed in melt pool, aufeis, and mineral precipitate sites, it was still present in relatively large abundances especially in sites G1 and G2 (~20% and ~30% respectively). The most prevalent Flavobacterium OTU in these samples was the same as reported in the aufeis

(genus: Aquidurense). Furthermore, the presence of an OTU most closely related to the genus

Hymenobacter was seen at ~6% in site G3. This genus was previously isolated from marine sediments in the Arctic (Kim et al., 2017).

2.3.7 Additional Aqueous Geochemistry

Aqueous geochemistry data for the 2016 spring, aufeis, melt pools, mineral precipitate, and glacier sites are shown in Table 2.1. Geochemical data reported for an active spring in 2009 by

Wright et al. (2013) are included for comparison. Charge balance error (CBE) was calculated for all sites (Table A.1) to constrain the accuracy of aqueous geochemical results. The aqueous geochemistry of surveyed sites is characterized by salty, sulfide-rich fluids from both melt pool and aufeis samples, as well as those seen in spring fluids collected from previous years (Gleeson et al., 2010; Grasby et al., 2003; Wright et al., 2013). This indicates that melt pools and aufeis are not directly derived from glacial ice or glacial melt waters, but rather most likely trace back to the subsurface/sub-glacial spring source.

Aqueous geochemistry was explored via a PCoA plot (Figure A.4) of BFP sites and their associated ion concentrations. The results reveal that the analytes driving geochemical diversity

+ - 2+ 2- in the system are Na , Cl , Ca , and SO4 . Sample sites also appear to follow a trend in sulfate- chloride space (Figure 2.2). Sodium ion values ranged between below detection (G2) and 39.62 mM (2016 Spring). The highest reported sodium value however belonged to the 2009 Spring at

51.80 mM (Wright et al., 2013). Calcium ion values range from 0.0053 (G2) and 14.7 mM (2016

46 Spring) and follow the same trend as sulfate values seen in the log-log aqueous geochemical ordination of these two ions (Figure A.4). In general, 2009 and 2016 spring samples had the highest total analyte values followed closely by the melt pools and then the aufeis and mineral precipitate samples (AS1 & AS2 from 2014). The glacier sample (similar to “Control” samples) had the lowest overall values for sulfate and chloride and showed an overall negative correlation in the geochemical ordination (Figure A.4).

The pH for sites M1 and M2 was 5.5-6.8 and 5.5-6.0, respectively. pH values for aufeis sites and 2014 mineral precipitate sites were not measured. Measured pH for sites AS3 and AS4

(from 2017) was approximately 7 (Table 2.1). pH measurements were not taken for other sites.

2.4 Discussion

We used 16S rRNA gene sequencing coupled to geochemical analysis to identify and better understand the microbial communities at BFP that have adapted to live in this sulfur-rich, low-temperature environment. This was accomplished over multiple field campaigns to BFP to collect samples from multiple site types (spring discharge, aufeis, mineral precipitates, and melt pool waters). Overall, we interpret the 16S rRNA gene sequencing data presented (Figures 2.3-

2.6) across multiple site types, to show that BFP is a highly dynamic system over short time periods (weeks to months). While over longer durations (years) a basal microbial community exists and is influenced temporally by sulfur-cycling microorganisms on smaller time scales during freeze/thaw and/or spring flow events. Samples from multiple site/material types were collected in order to evaluate the potential differences/similarities between their associated microbial communities. We have included the 2016 spring 16S rRNA gene sequencing data with the melt pool data due to the presence of similar microorganisms (Figure 2.3) and fluid

47 geochemistry (Figures 2.2 & A.4) between 2014 melt pool sites and 2016 spring discharge fluid data. Aufeis sites, while different, are similar to melt pool sites in that they are the precursors to melt pools. Both melt pools and aufeis are spring-derived, with melt pools being a more active melt-thaw system derived from static in-place, frozen aufeis. It is important to note that samples for melt pool sites were taken from surface pools, while aufeis samples were from melted ice that was dug from below the surface. It should be noted however, that mixing trends between glacial runoff and melting aufeis are not wholly understood and should be further investigated in order to determine whether or not this could be influencing microbial community dynamics.

A successional microbial community trend appeared within the melt pool and aufeis sites over the course of the 2014 field sampling. One key example of this can be seen in site M3, where M3a was sampled on 26 June, 2014 along with site M2. A large, dominant community of

Sulfurimonas is seen in these samples. When site M3 was sampled for a second time (M3b; 30

June, 2014), Sulfurimonas is far less and Flavobacterium is shown to be dominant (Figure 2.3).

This can be seen with aufeis samples as well, as sites A1 and A2 (sampled on 23 June 2014) have a very low abundance of Flavobacterium, while sites A3-A6 (all sampled on or after 27

June, 2014) show this as the dominant organism (Figure 2.4). While the aufeis are more static in terms of site composition, the apparent growth of Flavobacterium in this case could be due to the speed of melting of the site thus providing a dynamic to the microbial community structure.

Aufeis samples were “slushy” and it’s possible that increased melt during this time of the season allows a fast proliferation of dominant microorganisms. The addition of 2017 mineral precipitate samples (which are remnant from the Ice Blister in 2014) shows a similar long-term trend suggesting that even over larger timescales (in this case three years), these communities, dominated by Flavobacteria and an organism with sequences classified as chloroplast (likely

48 Cyanobacteria or glacial algae) can persist (Figure 2.5). While our data show that different sulfur-oxidizing microorganisms (SOMs) do change over time and sample type/material, there are consistent organisms present (i.e. Flavobacterium sp.). The apparent succession of sample types appears to go in order of melt pool/spring fluid to aufeis, and then finally to mineral precipitates. As ices/materials get older the Flavobacterium continues to dominate while chloroplast-containing organisms (possibly Cyanobacteria, but most likely glacial algae) take hold as well. This may indicate that more dynamic site types are more amenable to SOMs, whereas more static or stable site types are preferable for Flavobacterium.

Flavobacterium is not new to BFP as reported previously (Wright et al., 2013). Wright et al. (2013) suggested that in association with SOMs, Flavobacterium, which is an aerobic organism, may be creating a ‘microoxic’ environment in surface sediments where SOMs were able to thrive at lower O2 levels (Inagaki et al., 2003, 2004). Furthermore, it has been suggested that Flavobacterium may be utilizing organic carbon present at the site via the degradation of complex molecules. It is not known however, if the detected dissolved organic carbon (DOC) previously found at BFP (3.9 - 11.4 mg L-1; Wright et al., 2013) is delivered from subsurface organic-rich shales to the surface via spring fluids directly or is being formed in situ via microbial metabolisms and/or being atmospherically delivered. Boetius et al. (2015) reported that many Arctic and Antarctic sea ice communities were found to be dominated by the bacterial classes Flavobacteriia and Gammaproteobacteria, mainly due to their ability to breakdown DOC and extracellular polymeric substances (EPS) produced by sea ice algae. It must be noted that the

Flavobacteria at BFP are commonly found in conjunction with chloroplast-containing and cyanobacterial organisms (site M4, M5, M6, all mineral precipitate sites, and G1 & G2, respectively). These findings suggest that Flavobacterium at BFP may not actually be

49 contributing directly to any type of sulfur cycling; rather they are merely part of the basal microbial community that inhabits the surrounding area. Flavobacterium species are ubiquitous on the planet and found in both aquatic and terrestrial environments with over 100 species identified (Kolton et al., 2013). Moreover, they have also been found to be associated with snowfall events in subalpine environments (Honeyman et al., 2018). It is likely in this case that

Flavobacterium found at BFP are niche species that have adapted to live in a sulfur-rich, sub- aerial environment, and are possibly acting to breakdown complex organics to be used by other microbial community members (e.g. sulfate-reducing microorganisms (SRMs)).

Desulfocapsa was found in a majority of samples including 2016 spring, melt pool, aufeis, and mineral precipitate sites, and its presence throughout BFP samples is further evidence that the reduction of some oxidized forms of sulfur (e.g. sulfite, etc.) is at least partially microbially mediated at BFP. Previous sulfur isotopic data has been interpreted to show that sulfide delivered from the subsurface is produced via biological sulfate reduction (BSR; Gleeson et al., 2012;

Grasby et al., 2003, 2012). While the organisms potentially driving such BSR have not yet been identified in unobtainable subsurface/subglacial samples, perhaps the genus Desulfocapsa present in BFP samples represents a fraction of subsurface microbiota. While Desulfocapsa is actually found in at least one sample from each site type, the majority of its presence is within melt pool sites. This further strengthens the concept that BFP Spring fluid is closely related to melt pool sites and may even actively seed surface processes via subsurface microbial communities. This is the first time that microorganisms related to sulfate reducing bacteria have been found at BFP. Moreover, the presence of organic carbon (even in a small amount of 0.20 wt.%; Lau et al., 2017) within M2 (where Desulfocapsa is present) helps to support this claim.

50 Desulfocapsa sp. have been implicated previously in the disproportionation of inorganic sulfur compounds (Janssen et al., 1996), including elemental sulfur and thiosulfate (Finster et al.,

1998). More recently it has been found that Desulfocapsa sulfexigens specializes specifically in the disproportionation of elemental sulfur, thiosulfate, and sulfite, using CO2 as its sole carbon source (Finster et al., 2013), and even more intriguingly, is unable to grow on sulfate reduction.

This may suggest then that the large presence of S0 at BFP, especially around the Ice Blister may partially be metabolized by Desulfocapsa. Furthermore, this may help to explain its presence in

2017 mineral precipitate samples (Figure 2.5) where samples were collected from brightly yellow materials. While 16S rRNA gene sequencing reads from BFP are not resolved enough to tell us the species of Desulfocapsa that are present, it may be possible that disproportionation of these inorganic sulfur species is occurring in-situ. This disproportionation process could thus be responsible for the kinds of sulfur found to be present on-site (e.g. α-, β-, γ-cyclooctasulfur; Lau et al., 2017), however, it has not been determined whether or not these rare allotropes are biogenic in origin. Moreover, it is not clear how a putatively anaerobic microorganism is actively metabolizing in a surface, oxygenic environment.

Because of the expected low biomass in the system we used extracted DNA concentrations to approximate the quantities of microbial cells in the BFP system by site type (Table A.S3). It is challenging to extract DNA from these types of systems where nucleic acids tend to adsorb strongly to mineral-rich samples (Direito et al., 2012). Lower sequencing depth is not unexpected, especially for environmental samples and low biomass environments (Boetius et al.,

2015; Stibal et al., 2015). While site values fluctuate (e.g. 4.95 x 102 - 9.80 x 103 cells mL-1 or cells mg-1 in melt pools), there are minimal differences between melt pools and aufeis sites, which both have averages between 2 - 4 x 103 cells mL-1 or cells mg-1. The only noticeable

51 difference is seen in the 2017 mineral precipitate sites that averaged 1.55 x 105 cells mL-1; however, this may be attributed to better chemistry in the ZymoBIOMICS™ DNA/RNA Mini

Kit (Zymo Research Corp.). This is further evidenced in the 16S rRNA gene sequencing data where the average number of sequence reads for melt pool sites was 7,541 reads, and 6,172 for mineral precipitate sites. While extraction kit chemistry may have differed, the average values for sequencing reads is close and does not affect the interpretation of the data.

Aqueous geochemistry across site locations follows a clear trend of higher ion concentrations found in spring fluids, followed by melt pools, aufeis, and finally glacier samples.

This logically makes sense in that melt pool sites are surface level, and most likely the most recent form of aufeis. Aufeis samples, as noted in the methods and materials, are thought to be older ice, presumably layered, and possibly diluted slightly from any glacial melting or meteoric input. Previous aqueous geochemistry from the 2009 BFP spring (Wright et al., 2013) trends very well with data collected from two melt pool sites (M2 and M3) as well as spring fluid collected during 2016 (Table 2.1). The ordination of aqueous geochemical data (Figure A.4) shows a strong correlation of the 2009 and 2016 spring fluids and saline waters. This correlation is pronounced with melt pool samples M2 and M3 as well as aufeis sample A3. The remaining samples show a negative correlation in the ordination. Graphing ions against each other (e.g.,

2- - SO4 and Cl ) is a common geochemical method to discover whether or not certain analytes are trending together. The log-log plot (Figure 2.2) generated for this purpose supports the same findings from the geochemical ordination in Figure A.4. Sulfate and chloride increase in concentration together and those associated sites are seen among melt pools, aufeis, and the 2009 and 2016 spring fluids. Due to the remote location of BFP and works done to date, knowledge of the underlying hydrology of the glacier is poorly understood. However, it is possible that

52 variability in ion concentrations, especially in spring geochemistry is governed by stochastic flow of fluids in the subsurface. Moreover, as these fluids contact the surface and begin mixing with glacial melt we would also expect more variability in ion concentrations between the different site types as well.

When comparing aqueous geochemistry to the 16S rRNA gene sequencing data there do not appear to be any general trends. Of the sites with the highest total dissolved solids, sites M2 and M3 both have high relative abundances of the OTU most closely related to Sulfurimonas; however, site A3 only has small abundances of the SOMs Sulfurovum and Sulfuricurvum.

Furthermore, the 2016 spring fluid and site A3 have high abundances of Flavobacterium, which strangely enough is not the case for 16S rRNA gene sequencing data for the 2009 Spring where an OTU most closely related to Burkholderia was the dominant organism (Wright et al., 2013).

The 2016 spring also has a large abundance of Sulfurovum, which is also seen as the primary

SOM in site A1. The only SOM seen in abundance in mineral precipitate samples is

Sulfurimonas in site AS7 which is an ice scraping seen in Figure A.3b. There isn’t any clear reason why these SOMs are in different abundances in different samples at different times, but it may very well have to do with acute events that select for given metabolisms allowing one to flourish when others do not. How aqueous geochemistry interplays with sulfur-cycling organisms at these sites remains uncertain as 16S rRNA gene sequencing only provides us with taxonomic classification, and we can only infer potential metabolic functionality. On-going metagenomic work will shed light on these questions.

While we do present sulfide data for a few of the included sites, it remains unknown how sulfide plays a role in the system other than potentially being used by SOMs as an energy source.

The link between sulfate and chloride concentration may indicate that some of the

53 microorganisms present, possibly the SOMs, have a higher tolerance to these increased ion concentrations. However, there is not a clear distinction between sites with high relative abundances of SOMs and those without where ion concentrations are high. In spring discharge fluids, melt pools, and aufeis sites, there are multiple samples where either SOMs have the highest relative abundance or Flavobacterium has the highest relative abundance. There is no clear delineation as to whether one set of organisms is more tolerant than the other. What is clear is that Flavobacterium can live at high and low concentrations of sulfide, sulfate, and chloride as evidenced in its presence in all samples along that geochemical gradient.

Microorganisms that can survive and even thrive in extreme environments on Earth have been of great interest in the study of astrobiology. One way to inform upon the types of organisms and their interactions within an environment beyond Earth is to study terrestrial environments well suited as potential analogs for extraterrestrial bodies. The Galilean moon

Europa is a target for current astrobiological investigations due to the presence of a subsurface, liquid-water ocean (Carr et al., 1998; Kivelson, 1997), as well as hydrated sulfate salts and sulfuric acid that have been detected in deposits on the icy surface (Carlson et al., 1999; McCord et al., 1999), with the ocean potentially capable of supporting life (Cockell et al., 2016; Hand et al., 2007). The presence of surface-deposited sulfate salts and sulfuric acid supports the hypothesis that sulfur-based metabolisms could be fueled by water-rock interactions within and/or below the ocean, coupled with the delivery of oxidants via the cycling of surface ice

(Hand et al., 2007; McCollom, 1999).

BFP presents a system where subsurface fluids emerge at the toe of an Arctic glacier, sometimes even through conduits formed within the glacial ice itself. These supraglacial spring fluids are then introduced to a much different surface environment. For this reason, BFP has

54 been proposed as one of the key analogs for Europa on Earth (along with other spring sites such as Blood Falls and regions where ice sheets separate surface environments from cold ocean waters below). Although the surface environment at BFP does not directly reflect the surface environment at Europa (approximately −170 °C, ~10-8 torr, highly ionizing radiation, etc.; Hand and Carlson, 2015; McCord et al., 2002), the potential for interchange between Europa’s ocean and the surface via the cracks within the icy crust (the lineae) as well as the abundance of sulfur- bearing compounds on Europa (Carlson et al., 1999; McCord et al., 1999) have led to the characterization of BFP as a key analog for Europa. This study has looked at multiple site/material types and the potential changes between microbial communities among them. This is important as glacial systems such as BFP are dynamic in nature where changes to ice and surface material can occur over a relatively short time scale (weeks), but also over extended time scales as well. There is also evidence of a dynamic system on Europa of ‘chaos terrain’ where it appears that the surface has been disrupted from below (Schmidt et al., 2011). Exploring these constantly changing environments on Earth provides a good target for potential sites in the search for life on worlds such as Europa. It may be that the potential for finding life on Europa not only involves searching for organisms that utilize sulfur metabolisms (SOMs and SRMs), but also organisms that are capable of breaking down complex organic matter such as some genera of Gammaproteobacteria and Flavobacteriia (Boetius et al., 2015).

While we attempted to make our sampling efforts as exhaustive as possible, there are always limitations to what can be accomplished in the field. It should be noted that BFP is in a very remote location of the Canadian Arctic, and opportunities to sample there are limited and difficult. Due to field conditions, it was often times difficult to get instruments calibrated and this hindered our ability to collect certain types of data. While some sites had acceptable data for

55 both 16S rRNA gene sequencing and aqueous geochemistry, others did not, hence the difference in biological replicates. One example is our lack of sulfide data which may have provided a better lens through which we could judge the differentiation of SOMs in-situ. However, the 16S rRNA gene sequencing data still tells an impactful story about the composition of microbial communities between different glacial sites/materials over time. This, along with aqueous geochemistry data, have allowed us to further enhance our knowledge of microbial organisms inhabiting these low-temperature sulfur systems on Earth and possibly extraterrestrial worlds such as Europa. This information may be valuable in the preparation for sampling by a Europan lander as a difference in material type (melting and filtering ice vs preserving mineral precipitate scrapings) may require a difference in approach based on the potential microbial communities present. Moreover, based on how dynamic glacial environments can be on Earth, we recommend that a Europan lander focus on similar sites such as the ‘chaos terrain’ discussed by Schmidt et al. (2011).

2.5 Conclusion

This study used 16S rRNA gene sequencing coupled to geochemical analysis to better constrain the microbial consortia that have adapted to live in the sulfur-rich, low-temperature environment at BFP. We successfully identified new dynamics of microbial communities present at BFP, namely, how microbial assemblages are dispersed between different site/material types and locations, and which communities tend to persist over long and short-term time periods. Our results demonstrate the widespread presence of sulfur cycling organisms (i.e. Sulfurovum,

Sulfuricurvum, Sulfurimonas, and Desulfocapsa) across these site types and in varying relative abundances. Results also show that a baseline microbial community exists within most samples

56 regardless of sample type, typically consisting of a large relative abundance of an OTU most closely related to the genus Flavobacterium. This suggests that much like ‘chaos terrain’ found on Europa, dynamic environments, whether spring fluid, melted ice, or even salt-like scrapings may contain similar microbial communities, and therefore be ideal targets for astrobiological investigation. Moreover, while the focus in previous studies has been on traditional sulfur- cycling microorganisms, our data points to a potentially larger role of Flavobacterium sp. based on their ubiquity in BFP samples and their persistence over time. Even if they aren’t directly contributing to overall sulfur-cycling they may be breaking down large organic molecules for use by other microorganisms (e.g. SRMs). While it is still unclear how the metabolism of

Flavobacterium sp. influences the environment and geochemistry of BFP, metagenomic and metatranscriptomic research that is underway will shed light on these processes. For example, these data will further elucidate which of the sulfur-cycling microorganisms (Sulfurimonas,

Sulfurovum, Sulfuricurvum, and Desulfocapsa) are actively utilizing sulfur constituents, and in which part of the process they are participating or are most active.

2.6 Data Availability

After sequencing, raw reads were deposited into the NCBI sequencing read archive (SRA, https://www.ncbi.nlm.nih.gov/sra) under the accession number SRP136466. The mapping file for samples and corresponding barcodes can be found in file A.S1 (Appendix E).

2.7 Acknowledgements

Field logistics and travel to/from Borup Fiord Pass were supported through Natural

Resources Canada’s (NRC) Polar Continental Shelf Program (PCSP). JRS and SEG thank Dr.

57 Karsten Piepjohn and the Federal Institute for Geosciences and Natural Resources at BGR,

Hanover, Germany for all logistics for the 2017 field campaign of the Circum-Arctic Structural

Events-19 (CASE-19) expedition. We would like to thank the DigitalGlobe Foundation for providing us with satellite imagery. We would also like to thank Dr. Blake W. Stamps for his guidance and help with bioinformatics tools and analyses.

2.8 References

Bej, A. K., Aislabie, J., and Atlas, R. M. (2009). Polar microbiology: the ecology, biodiversity and bioremediation potential of microorganisms in extremely cold environments. Boca Raton, FL: CRC Press.

Boetius, A., Anesio, A. M., Deming, J. W., Mikucki, J. A., and Rapp, J. Z. (2015). Microbial ecology of the cryosphere: sea ice and glacial habitats. Nat. Rev. Microbiol. 13, 677–690. doi:10.1038/nrmicro3522.

Button, D. K., and Robertson, B. R. (2001). Determination of DNA content of aquatic bacteria by flow cytometry. Appl. Environ. Microbiol. 67, 1636–1645. doi:10.1128/AEM.67.4.1636- 1645.2001.

Caporaso, J. G., Bittinger, K., Bushman, F. D., Desantis, T. Z., Andersen, G. L., and Knight, R. (2010a). PyNAST: A flexible tool for aligning sequences to a template alignment. Bioinformatics 26, 266–267. doi:10.1093/bioinformatics/btp636.

Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D., Costello, E. K., et al. (2010b). QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. doi:10.1038/nmeth.f.303.

Carlson, R. W., Johnson, R. E., and Anderson, M. S. (1999). Sulfuric acid on Europa and the radiolytic sulfur cycle. Science 286, 97–99. doi:10.1126/science.286.5437.97.

Carr, M. H., Belton, M. J. S., Chapman, C. R., Davies, M. E., Geissler, P., Greenberg, R., et al. (1998). Evidence for a subsurface ocean on Europa. Nature 391, 363–365. doi:10.1038/34857.

Christner, B. C., Priscu, J. C., Achberger, A. M., Barbante, C., Carter, S. P., Christianson, K., et al. (2014). A microbial ecosystem beneath the West Antarctic ice sheet. Nature 512, 310– 313. doi:10.1038/nature13667.

Cockell, C. S., Bush, T., Bryce, C., Direito, S., Fox-Powell, M., Harrison, J. P., et al. (2016). Habitability: A Review. Astrobiology 16, 89–117. doi:10.1089/ast.2015.1295.

58 Dahl, C., and Friedrich, C. G. (2008). Preface. Microbial Sulfur Metabolism (Dahl C & Friedrich CG eds). Proceedings of the International Symposium on Microbial Sulfur Metabolism 29.06.2006–02.07.2006, pp. v-vi. Springer-Berlin, Münster, Germany.

Dartnell, L. R., Hunter, S. J., Lovell, K. V., Coates, A. J., and Ward, J. M. (2010). Low- temperature ionizing radiation resistance of Deinococcus radiodurans and Antarctic Dry Valley bacteria. Astrobiology 10, 717–732. doi:10.1089/ast.2009.0439.

Direito, S. O. L., Marees, A., and Röling, W. F. M. (2012). Sensitive life detection strategies for low-biomass environments: optimizing extraction of nucleic acids adsorbing to terrestrial and Mars analogue minerals. FEMS Microbiol. Ecol. 81, 111–123. doi:10.1111/j.1574- 6941.2012.01325.x.

Edgar, R. C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461. doi:10.1093/bioinformatics/btq461.

Finster, K., Liesack, W., and Thamdrup, B. (1998). Elemental sulfur and thiosulfate disproportionation by Desulfocapsa sulfoexigens sp. nov., a new anaerobic bacterium isolated from marine surface sediment. Appl. Environ. Microbiol. 64, 119–125.

Finster, K. W., Kjeldsen, K. U., Kube, M., Reinhardt, R., Mussmann, M., Amann, R., et al. (2013). Complete genome sequence of Desulfocapsa sulfexigens, a marine deltaproteobacterium specialized in disproportionating inorganic sulfur compounds. Stand. Genomic Sci. 8, 58–68. doi:10.4056/sigs.3777412.

Gilhooly, W. P., Fike, D. A., Druschel, G. K., Kafantaris, F. C. A., Price, R. E., and Amend, J. P. (2014). Sulfur and oxygen isotope insights into sulfur cycling in shallow-sea hydrothermal vents, Milos, Greece. Geochem. Trans. 15, 1–19. doi:10.1186/s12932-014-0012-y.

Gleeson, D. F., Pappalardo, R. T., Anderson, M. S., Grasby, S. E., Mielke, R. E., Wright, K. E., et al. (2012). Biosignature detection at an Arctic analog to Europa. Astrobiology 12, 135– 150. doi:10.1089/ast.2010.0579.

Gleeson, D. F., Pappalardo, R. T., Grasby, S. E., Anderson, M. S., Beauchamp, B., Castaño, R., et al. (2010). Characterization of a sulfur-rich Arctic spring site and field analog to Europa using hyperspectral data. Remote Sens. Environ. 114, 1297–1311. doi:10.1016/j.rse.2010.01.011.

Gleeson, D. F., Williamson, C., Grasby, S. E., Pappalardo, R. T., Spear, J. R., and Templeton, A. S. (2011). Low temperature S0 biomineralization at a supraglacial spring system in the Canadian high Arctic. Geobiology 9, 360–375. doi:10.1111/j.1472-4669.2011.00283.x.

González-Toril, E., Amils, R., Delmas, R. J., Petit, J.-R., Komárek, J., and Elster, J. (2009). Bacterial diversity of autotrophic enriched cultures from remote, glacial Antarctic, Alpine and Andean aerosol, snow and soil samples. Biogeosciences 6, 33–44. doi:10.5194/bg-6-33- 2009.

Grasby, S. E. (2003). Naturally precipitating vaterite (μ-CaCO3) spheres: Unusual carbonates

59 formed in an extreme environment. Geochim. Cosmochim. Acta 67, 1659–1666. doi:10.1016/S0016-7037(00)01304-2.

Grasby, S. E., Allen, C. C., Longazo, T. G., Lisle, J. T., Griffin, D. W., and Beauchamp, B. (2003). Supraglacial sulfur springs and associated biological activity in the Canadian high Arctic—signs of life beneath the ice. Astrobiology 3, 583–596. doi:10.1089/153110703322610672.

Grasby, S. E., Beauchamp, B., and Bense, V. (2012). Sulfuric acid speleogenesis associated with a glacially driven groundwater system - Paleo-spring “pipes” at Borup Fiord Pass, Nunavut. Astrobiology 12, 19–28. doi:10.1089/ast.2011.0700.

Hamady, M., Walker, J. J., Harris, J. K., Gold, N. J., and Knight, R. (2008). Error-correcting barcoded primers for pyrosequencing hundreds of samples in multiplex. Nat. Methods 5, 235–237. doi:10.1038/nmeth.1184.

Hamilton, T. L., Jones, D. S., Schaperdoth, I., and Macalady, J. L. (2015). Metagenomic insights into S(0) precipitation in a terrestrial subsurface lithoautotrophic ecosystem. Front. Microbiol. 5, 756. doi:10.3389/fmicb.2014.00756.

Hand, K. P., and Carlson, R. W. (2015). Europa’s surface color suggests an ocean rich with sodium chloride. Geophys. Res. Lett. 42, 3174–3178. doi:10.1002/2015GL063559.

Hand, K. P., Carlson, R. W., and Chyba, C. F. (2007). Energy, chemical disequilibrium, and geological constraints on Europa. Astrobiology 7, 1006–1022. doi:10.1089/ast.2007.0156.

Honeyman, A. S., Day, M. L., and Spear, J. R. (2018). Fresh snowfall microbiology and chemistry are driven by geography in storm-tracked events. bioRxiv. https://doi.org/10.1101/300772.

Inagaki, F., Takai, K., Kobayashi, H., Nealson, K. H., and Horikoshi, K. (2003). Sulfurimonas autrotrophica gen. nov., sp. nov., a novel sulfur-oxidizing E-proteobacterium isolated from hydrothermal sediments in the Mid-Okinawa Trough. Int. J. Syst. Evol. Microbiol. 53, 1801–1805. doi:10.1099/ijs.0.02682-0.

Inagaki, F., Takai, K., Nealson, K. H., and Horikoshi, K. (2004). Sulfurovum lithotrophicum gen. nov., sp. nov., a novel sulfur-oxidizing chemolithoautotroph within the E-Proteobacteria isolated from Okinawa Trough hydrothermal sediments. Int. J. Syst. Evol. Microbiol. 54, 1477–1482. doi:10.1099/ijs.0.03042-0.

Janssen, P. H., Schuhmann, A., Bak, F., and Liesack, W. (1996). Disproportionation of inorganic sulfur compounds by the sulfate-reducing bacterium Desulfocapsa thiozymogenes gen. nov., sp. nov. Arch. Microbiol. 166, 184–192. doi:10.1007/s002030050374.

Kim, M. C., Kim, C. M., Kang, O. C., Zhang, Y., Liu, Z., Wangmu, D., et al. (2017). Hymenobacter rutilus sp. nov., isolated from marine sediment in the Arctic. Int. J. Syst. Evol. Microbiol. 67, 856–861. doi:10.1099/ijsem.0.001685.

60 Kivelson, M. G. (1997). Europa’s magnetic signature: report from Galileo’s pass on 19 December 1996. Science (80-. ). 276, 1239–1241. doi:10.1126/science.276.5316.1239.

Kleinteich, J., Hildebrand, F., Bahram, M., Voigt, A. Y., Wood, S. A., Jungblut, A. D., et al. (2017). Pole-to-pole connections: similarities between Arctic and Antarctic microbiomes and their vulnerability to environmental change. Front. Ecol. Evol. 5, 1–11. doi:10.3389/fevo.2017.00137.

Klotz, M. G., Bryant, D. A., and Hanson, T. E. (2011). The Microbial Sulfur Cycle. Front. Microbiol. 2. doi:10.3389/fmicb.2011.00241.

Kolton, M., Sela, N., Elad, Y., and Cytryn, E. (2013). Comparative genomic analysis indicates that niche adaptation of terrestrial Flavobacteria is strongly linked to plant glycan metabolism. PLoS One 8, e76704. doi:10.1371/journal.pone.0076704.

Lau, G. E., Cosmidis, J., Grasby, S. E., Trivedi, C. B., Spear, J. R., and Templeton, A. S. (2017). Low-temperature formation and stabilization of rare allotropes of cyclooctasulfur (β-S8 and γ-S8) in the presence of organic carbon at a sulfur-rich glacial site in the Canadian High Arctic. Geochim. Cosmochim. Acta 200, 218–231. doi:10.1016/j.gca.2016.11.036.

Lionard, M., Péquin, B., Lovejoy, C., and Vincent, W. F. (2012). Benthic cyanobacterial mats in the high Arctic: multi-layer structure and fluorescence responses to osmotic stress. Front. Microbiol. 3, 140. doi:10.3389/fmicb.2012.00140.

McCollom, T. M. (1999). Methanogenesis as a potential source of chemical energy for primary biomass production by autotrophic organisms in hydrothermal systems on Europa. J. Geophys. Res. Planets 104, 30729–30742. doi:10.1029/1999JE001126.

McCord, T. B., Hansen, G. B., Matson, D. L., Johnson, T. V, Crowley, J. K., Fanale, F. P., et al. (1999). Hydrated salt minerals on Europa’s surface from the Galileo near-infrared mapping spectrometer (NIMS) investigation. J. Geophys. Res. 104, 11827–11851. doi:10.1029/1999JE900005.

McCord, T. B., Teeter, G., Hansen, G. B., Sieger, M. T., & Orlando, T. M. (2002). Brines exposed to Europa surface conditions. J. Geophys. Res. 107(E1), 4-1. doi: 10.1029/2000JE001453

Mikucki, J. A., and Priscu, J. C. (2007). Bacterial diversity associated with blood falls, a subglacial outflow from the Taylor Glacier, Antarctica. Appl. Environ. Microbiol. 73, 4029– 4039. doi:10.1128/AEM.01396-06.

Niederberger, T. D., Perreault, N. N., Lawrence, J. R., Nadeau, J. L., Mielke, R. E., Greer, C. W., et al. (2009). Novel sulfur-oxidizing streamers thriving in perennial cold saline springs of the Canadian high Arctic. Environ. Microbiol. 11, 616–629. doi:10.1111/j.1462- 2920.2008.01833.x.

Oksanen, J., Blanchet, F., Kindt, R., Legendre, P., and O’Hara, R. (2016). Vegan: community ecology package. R Packag. 2.3-3. Available at: https://cran.r-project.org/package=vegan.

61 Parada, A. E., Needham, D. M., and Fuhrman, J. A. (2016). Every base matters: Assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol. 18, 1403–1414. doi:10.1111/1462-2920.13023.

Perner, M., Seifert, R., Weber, S., Koschinsky, A., Schmidt, K., Strauss, H., et al. (2007). Microbial CO2 fixation and sulfur cycling associated with low-temperature emissions at the Lilliput hydrothermal field, southern Mid-Atlantic Ridge (9°S). Environ. Microbiol. 9, 1186–1201. doi:10.1111/j.1462-2920.2007.01241.x.

Perreault, N. N., Andersen, D. T., Pollard, W. H., Greer, C. W., and Whyte, L. G. (2007). Characterization of the prokaryotic diversity in cold saline perennial springs of the Canadian high Arctic. Appl. Environ. Microbiol. 73, 1532–1543. doi:10.1128/AEM.01729- 06.

Peter, H., and Sommaruga, R. (2016). Shifts in diversity and function of lake bacterial communities upon glacier retreat. ISME J. 10, 1545–1554. doi:10.1038/ismej.2015.245.

Pollard, W. H. (2005). Icing processes associated with high Arctic perennial springs, Axel Heiberg Island, Nunavut, Canada. Permafr. Periglac. Process. 16, 51–68. doi:10.1002/ppp.515.

Pollard, W., Omelon, C., Andersen, D., and McKay, C. (1999). Perennial spring occurrence in the Expedition Fiord area of western Axel Heiberg Island, Canadian High Arctic. Can. J. Earth Sci. 36, 105–120. doi:10.1139/e98-097.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B. M., Ludwig, W., Peplies, J., et al. (2007). SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 35, 7188–7196. doi:10.1093/nar/gkm864.

R Core Team (2016). R: A language and environment for statistical computing. R foundation for statistical Computing, Vienna, Austria; URL: https://www.R-project.org/.

Rognes, T., Flouri, T., Nichols, B., Quince, C., and Mahé, F. (2016). VSEARCH: a versatile open source tool for metagenomics. PeerJ 4, e2584. doi:10.7717/peerj.2584.

Scheidegger, J. M., Bense, V. F., and Grasby, S. E. (2012). Transient nature of Arctic spring systems driven by subglacial meltwater. Geophys. Res. Lett. 39, 1–6. doi:10.1029/2012GL051445.

Schmidt, B. E., Blankenship, D. D., Patterson, G. W., and Schenk, P. M. (2011). Active formation of ‘chaos terrain’ over shallow subsurface water on Europa. Nature 479, 502– 505. doi:10.1038/nature10608.

Sievert, S. M., Kiene, R. P., and Schulz-Vogt, H. N. (2007). The sulfur cycle. Oceanography 20, 117–123. doi:10.1017/S0016672308009336.

Stamps, B. W., Lyles, C. N., Suflita, J. M., Masoner, J. R., Cozzarelli, I. M., Kolpin, D. W., et al. (2016). Municipal solid waste landfills harbor distinct microbiomes. Front. Microbiol. 7.

62 doi:10.3389/fmicb.2016.00534.

Stibal, M., Gozdereliler, E., Cameron, K. A., Box, J. E., Stevens, I. T., Gokul, J. K., et al. (2015). Microbial abundance in surface ice on the Greenland Ice Sheet. Front. Microbiol. 6, 225. doi:10.3389/fmicb.2015.00225.

Tang, Y., Horikoshi, M., and Li, W. (2016). ggfortify: unified interface to visualize statistical results of popular R packages. R J. 8. Available at: https://journal.r- project.org/archive/2016/RJ-2016-060/RJ-2016-060.pdf.

Ting, L., Williams, T. J., Cowley, M. J., Lauro, F. M., Guilhaus, M., Raftery, M. J., et al. (2010). Cold adaptation in the marine bacterium, Sphingopyxis alaskensis, assessed using quantitative proteomics. Environ. Microbiol. 12, no-no. doi:10.1111/j.1462- 2920.2010.02235.x.

Van Trappen, S. (2004). Loktanella salsilacus gen. nov., sp. nov., Loktanella fryxellensis sp. nov. and Loktanella vestfoldensis sp. nov., new members of the Rhodobacter group, isolated from microbial mats in Antarctic lakes. Int. J. Syst. Evol. Microbiol. 54, 1263–1269. doi:10.1099/ijs.0.03006-0.

Wainstein, P., Moorman, B., and Whitehead, K. (2008). Importance of glacier–permafrost interactions in the preservation of a proglacial icing: Fountain Glacier, Bylot Island, Canada. In Kane D. L. and Hinkel K. M. eds. Proceedings of the 9th International Conference on Permafrost: extended abstracts, 29 June–3 July 2008, Fairbanks, Alaska, USA. Institute of Northern Engineering, University of Alaska Fairbanks, Fairbanks, AK.

Wickham, H. (2016). ggplot2: elegant graphics for data analysis. New York, NY: Springer- Verlag Available at: http://ggplot2.org.

Wright, K. E., Williamson, C., Grasby, S. E., Spear, J. R., and Templeton, A. S. (2013). Metagenomic evidence for sulfur lithotrophy by epsilonproteobacteria as the major energy source for primary productivity in a sub-aerial arctic glacial deposit, Borup Fiord Pass. Front. Microbiol. 4, 1–20. doi:10.3389/fmicb.2013.00063.

Zeng, Y.-X., Yu, Y., Li, H.-R., and Luo, W. (2017). Prokaryotic community composition in Arctic Kongsfjorden and sub-Arctic Northern Bering Sea sediments as revealed by 454 pyrosequencing. Front. Microbiol. 8, 2498. doi:10.3389/fmicb.2017.02498.

Zhou, Y., Liu, H., Wang, N., Jiao, N., Xu, B., Gu, Z., et al. (2015). Massilia eurypsychrophila sp. nov. a facultatively psychrophilic bacteria isolated from ice core. Int. J. Syst. Evol. Microbiol. 65, 2124–2129. doi:10.1099/ijs.0.000229.

Zhou, Z., Jiang, F., Wang, S., Peng, F., Dai, J., Li, W., et al. (2012). Pedobacter arcticus sp. nov., a facultative psychrophile isolated from Arctic soil, and emended descriptions of the genus Pedobacter, Pedobacter heparinus, Pedobacter daechungensis, Pedobacter terricola, Pedobacter glucosidilyticus and Pedobacter lentus. Int. J. Syst. Evol. Microbiol. 62, 1963– 1969. doi:10.1099/ijs.0.031104-0.

63

CHAPTER 3

METAGENOMIC INSIGHTS INTO MICROBIAL METABOLISMS OF A

SULFUR-DOMINATED GLACIAL ECOSYSTEM

A paper in preparation for submission to Applied and Environmental Microbiology

Abstract

Sulfur cycling in polar low-temperature ecosystems is an understudied phenomenon. More work is required to understand how nutrient cycling among microorganisms occurs in these environments. One low-temperature, extreme environment where sulfur cycling can be observed is at Borup Fiord Pass (BFP) in the Canadian High Arctic. Here, transient springs emerge from the toe of a glacier creating large proglacial aufeis (spring-derived ice) that are often covered in bright yellow/white sulfur and sulfate-mineral precipitates and have an associated strong odor of hydrogen sulfide. Whole genome sequencing (WGS) at this location from multiple different site and sample types has revealed the widespread presence of sulfur cycling genes, especially within the Sox complex of enzymes. Assembly and binning of WGS data produced 31 reportable metagenome assembled genomes (MAGs) that were queried for sulfur-, nitrogen- and carbon cycling/metabolism genes, which revealed common and ubiquitous metabolic processes. Genes involved sulfur oxidation via the Sox operon were found within multiple Proteobacterial classes including the Alpha-, Beta-, Gamma-, and Epsilonproteobacteria. Previous research implicated organisms within the Gamma- and Epsilonproteobacteria as the primary classes responsible for sulfur oxidation at the site. Now with the addition of Alpha- and Betaproteobacteria, a form of

64 functional redundancy appears present for more widespread sulfur oxidation. In extreme low- temperature environments, functional redundancy may be a key mechanism that microorganisms utilize to coexist whenever energy limitations are present.

3.1 Introduction

Arctic and Antarctic systems such as lakes, streams, spring-derived ice (aufeis), and saturated sediments provide insight into low-temperature terrestrial processes such as geochemical cycling and changes in microbial community dynamics. One such ecosystem in the

Arctic, Borup Fiord Pass (BFP), Ellesmere Island, Nunavut, Canada has a transient sulfidic spring system at the terminus of a coalescence glacier within a valley (Figure 3.1; Grasby et al.,

2003). Previous research at BFP has surveyed microbial activity and S(0) biomineralization

(Gleeson et al., 2011; Grasby et al., 2003), and looked at redox bioenergetics (Wright et al.,

2013). More recently, from a 2014 sampling effort, two rare allotropes of cyclooctasulfur (β and

γ) were discovered within sulfur bubbles on top of a melt pool (Lau et al., 2017). Additionally, in depth 16S rRNA gene sequencing revealed an active and diverse assemblage of autotrophic and heterotrophic microorganisms in melt pools, aufeis, spring fluid, and surface mineral deposits that persist over multiple years contributing to a basal community present in the system regardless of site or material type (Trivedi et al., 2018). Metagenomic work was conducted by

Wright et al. (2013), which constrained the bioenergetics of microbial metabolism from a mound of elemental sulfur, and demonstrated that at least in surface sediments, sulfur oxidation was likely the dominant metabolism present among Epislonproteobacteria. However, to date no work has been attempted to fully understand and characterize the metabolic capability, including sulfur metabolism, of the greater BFP system.

65

Figure 3.1. Borup Fiord Pass satellite view. This image was collected via satellite on 2 July 2014. M- and A- samples from 2014 are marked with gold circles and red squares, respectively. The 2016 Spring region is highlighted with a dashed yellow line and the location of the spring is marked with a black triangle. Samples from 2017 were collected from within the dark blue dashed area. Image courtesy of the DigitalGlobe Foundation.

Sulfur is the 10th most abundant element in the universe (Heiserman, 1992), the 6th most abundant on Earth (Mandeville, 2010), and a required element for the protein cross-linking amino acids cysteine and methionine (Brosnan and Brosnan, 2006). Prior research has shown sulfur is key for microbial metabolisms in diverse and extreme environments, such as hydrothermal vents & deep subsurface sediments (Dahl and Friedrich, 2008), sea ice (Boetius et al., 2015), glacial environments (Mikucki and Priscu, 2007; Purcell et al., 2014), and Arctic hypersaline springs (Niederberger et al., 2009; Perreault et al., 2008). One reason for this is the large range of oxidation states of known sulfur compounds, from 2- in sulfides like hydrogen

66 2- sulfide (H2S) to 6+ in sulfates (SO4 ). Two predominant forms of sulfur-utilizing microorganisms exist: sulfur-oxidizing microorganisms (SOMs), which use reduced compounds

0 (e.g. H2S and S ) as electron donors, and sulfate-reducing microorganisms (SRMs), which use

2- 2- oxidized forms of sulfur (e.g. SO4 and SO3 ) as electron acceptors. SOMs are metabolically and phylogenetically diverse (Dahl and Friedrich, 2008; Friedrich et al., 2001, 2005), and can often fix carbon dioxide (CO2) while using a variety of electron acceptors (Mattes et al., 2013).

SRMs are equally as intriguing and can utilize a range of electron donors, including organic carbon (OC), inorganic carbon, hydrogen, metallic iron (Enning et al., 2012), and even heavier metals such as uranium, where the soluble U(VI) is converted to the insoluble U(IV) under anoxic conditions (Spear et al., 1999, 2000). SRMs are particularly important in sulfate-rich marine environments where they contribute as much as 50% of total global OC oxidation

(Bowles et al., 2014; Thullner et al., 2009).

While some research has been conducted on sulfur-cycling in low-temperature environments (Gleeson et al., 2011; Grasby et al., 2012; Mikucki et al., 2009, 2016;

Niederberger et al., 2009; Purcell et al., 2014; Wright et al., 2013), which organisms participate in sulfur cycling and what metabolic pathways enable this to happen at distinct low-temperature sites across the globe is an open question. We collected samples at BFP over multiple years

(2014, 2016, and 2017; Figure 3.2) and from varied sample types (aufeis, melt pools, spring fluids, and surface mineral precipitates; Appendix B, Figure B.2) for 16S rRNA gene and metagenomic whole genome sequencing (WGS). These data were assembled and binned into

Metagenome Assembled Genomes (MAGs), which were queried to identify which metabolic processes are present and dominant across multiple sample types at BFP. We found that a number of sulfur metabolisms may be possible (especially those for sulfur oxidation) and are

67

Figure 3.2. BFP Sampling locations. A) The “Sulfidic Aufeis” from Trivedi, et al. (In review) during 2014. Note: Two samples locations are to the southwest of this picture and their approximate locations can be seen in Figure 1. B) The sulfidic aufeis on the left of the picture with the 2016 Spring (Also see Appendix B, Figure B.1) a few hundred meters to the east (indicated by the orange dashed line). C) The same location again 2017, which was the focus of sampling.

68 coupled with carbon fixation via the reductive acetyl-CoA and reverse TCA pathways. Oxygen is likely to be the terminal electron acceptor, however, metagenomics indicates that nitrate reduction may also be possible. The genes for these processes appear across multiple different putative genomes, from multiple taxa, and among different sample types, indicating that sulfur cycling at BFP may be aided by functional redundancy across multiple bacterial classes. This is a concept where distinctly different taxa can coexist in a system, while providing very similar energy-yielding functions (Louca et al., 2018).

Previous research at BFP found that sulfur cycling organisms within the Epsilon- and

Gammaproteobacterial classes were most likely the main participants in sulfur oxidation processes (Gleeson et al., 2011; Grasby et al., 2003; Wright et al., 2013). Our data show that while these organisms are also present across multiple site types, a large consortium of sulfur oxidizers from the Alpha- and Betaproteobacterial classes may also be participating in this process. These results are based on the presence of complete and almost complete sox operons.

While this suggests that these organisms may be participating in sulfur oxidation processes, it does not confirm this to be definitively the case. This is, however, the next step in determining the major players in these valuable sulfur cycling processes and has revealed that sulfur oxidation may be a more widely available microbial metabolism at BFP than was previously thought.

69 3.2 Results

3.2.1 Whole shotgun sequencing and assembly

Whole shotgun sequencing of total genomic DNA (gDNA) across nine BFP samples produced paired-end reads which were trimmed and de novo assembled, producing between 10.1

Mbp (BFP14_M2) and 638.4 Mbp (BFP17_14C) of data, and between 4,841 (BFP14_M2) and

272,035 (BFP17_14C) contigs. Co-assembled samples resulted in an assembly of 1.16 Gbp and

477,394 total contigs, with the largest being 406,785 bp long and the average being 2,440 bp long (Table B.1).

3.2.2 Binning, MAG determination, and putative taxonomic identification

Initially, 80 bins were identified via CONCOCT. After manual curation within Anvi’o, a total of 166 bins were identified. Of these, 31 were classified as MAGs (bins with greater than

50% completeness and less than 10% redundancy) after quality control within CheckM.

Reported genome statistics for assembled data include total length of MAGs (bp), contig number, N50 value, GC content, and completeness and redundancy percentages (Table 3.1). Of the 31 MAGs, nine were considered highly complete with low redundancy (90% and 10% respectively).

All MAGs were identified within the domain Bacteria. Of the 31 MAGs, 22 were identified within the phylum Proteobacteria (Table 3.1). Of the Proteobacteria, the

Betaproteobacteria and Gammaproteobacteria are the most represented, with six (MAG 29, 19,

14, 5, 20, & 27) and six (MAG 11, 23, 2, 4, 15, & 16) respectively. The class

Alphaproteobacteria contains five MAGs (MAG 10, 3, 6, 25, & 25), with the Deltaproteobacteria

(three; MAG 7, 12, & 21) and Epsilonproteobacteria (two; MAG 1 and 9) having the fewest. It

70 Table 3.1. Metadata for all reported MAGs. All MAGs shown are >50% completion with <10% redundancy via CheckM. Nearest neighbors were placed in a cladogram via CheckM/pplacer and confirmed via RAST (Figure 3.5).

MAG # Putative taxonomy Lengtha Contigsb N50a GCc Comp.c Redund.c 1 Epsilonproteobacteria; 2064274 14 214847 38.19 99.59 0 Sulfurimonas 2 Gammaproteobacteria; 4979965 67 119878 41.35 99.45 1.12 Shewanella 3 Alphaproteobacteria; 7775429 338 70375 63.24 99.15 3.72 Rhodopseudomonas 4 Gammaproteobacteria; 3581784 67 99909 43.59 98.28 2.46 Cellvibrio 5 Betaproteobacteria; 3445427 282 19477 52.80 97.39 1.13 Limnobacter 6 Alphaproteobacteria; 3009949 326 11575 58.93 93.54 1.92 Erythrobacteraceae (f) 7 Deltaproteobacteria; 2805299 284 13203 51.33 92.84 2.28 Desulfocapsa 8 Actinobacteria; 2150593 263 13565 60.87 92.5 3.33 Coriobacteriales (o) 9 Epsilonproteobacteria; 1833503 136 22712 38.75 92.21 1.84 Sulfurovum 10 Alphaproteobacteria; 3281266 483 7930 60.60 89.02 2.3 Loktanella 11 Gammaproteobacteria; 2721939 304 11951 68.73 87.86 1.12 Arenimonas 12 Deltaproteobacteria; 2954298 447 7903 50.14 84.94 2.08 Desulfocapsa 13 Cytophagia; 4253479 791 5920 41.69 80.28 9.29 Algoriphagus 14 Betaproteobacteria; 2209716 410 5958 51.10 79.43 3.13 Herminiimonas 15 Gammaproteobacteria; 3818186 748 5456 54.16 78.56 5.04 Marinobacter 16 Gammaproteobacteria; 6369279 177 60989 58.13 77.54 0.88 Pseudomonas 17 Chloroflexi; 4795723 834 6374 49.05 74.65 3.67 Chloroflexales (o) 18 Sphingobacteriia; 3604420 465 10926 33.46 73.33 5.44 Pedobacter 19 Betaproteobacteria; 2632478 497 5833 61.30 72.78 1.57 Thiobacillus 20 Betaproteobacteria; 2881094 579 5077 59.36 72.38 3.83 Rhodoferax 21 Deltaproteobacteria; 2948378 557 5659 51.99 69.53 0.65 Geopsychrobacter 22 Elusimicrobia (p); Candidatus 2483169 437 6174 58.20 69.41 2.35 Endomicrobium 23 Gammaproteobacteria; 2196521 426 5432 42.35 68.46 7.66 Thiomicrospira 24 Alphaproteobacteria; 2926576 607 4990 63.80 67.43 1.12 Sandarakinorhabdus

71 Table 3.1. Metadata for all reported MAGs continued.

25 Alphaproteobacteria; 2352194 538 4519 64.93 62.24 4.89 Sandarakinorhabdus 26 Planctomycetia; 2408829 457 5553 45.58 61.92 2.43 Isosphaera 27 Betaproteobacteria; 2910411 640 4770 62.31 61.79 0.44 Hylemonella 28 Deinococci; 3481023 613 6273 60.43 59.34 2.97 Deinococcus 29 Betaproteobacteria; 1596490 423 3649 52.68 55.17 8.31 Methylotenera 30 Mollicutes; 974526 65 28355 33.20 51.75 4.55 Acholeplasmataceae (f) 31 Flavobacteriia; 1930691 418 4726 34.69 51.45 2.36 Flavobacterium a base pairs (bp) b total number of contigs c percentage (%) should be noted that each class contains at least one highly complete, low redundancy MAG.

Trivedi et al. (2018) reported the presence of the Deltaproteobacterial genera Desulfocapsa via targeted 16S rRNA gene sequencing. Average nucleotide identity (ANI) results positively identified MAGs 7 and 12 as the genus Desulfocapsa rather than Desulfotalea in two of the three tests (Table B.2).

3.2.3 MAG recruitment by sample

The highest number of mapped reads from any given sample was from a 2017 mineral precipitate sample, BFP17 3B with 54,999,613 total reads, with the remaining samples having much lower mapping numbers (Figure 3.3a). While some samples contributed almost solely to one or two MAGs (e.g. melt pool BFP14 M2 and MAG 1) others contributed greatly to multiple

MAGs (e.g. aufeis sample BFP14 A4b and MAGs 3, 9, 10, 16, and 31). Some MAGs had contributions from many samples, however, still did not acquire great genome completion (e.g.

MAG 31; see Figure 3.3a).

MAG recruitment (Figure 3.3a) refers to how strongly a given sample contributes to the

72 overall production of a bin. For example, MAG 31 has high contributions from samples A6, 3B,

4E, and 6B, and moderate contributions from A4b and M4b. However, MAG 31 has the lowest completion while still being within the 50/10 reportable ratio. MAG 31 is most closely related to

Flavobacterium and more than one species of this genus may have been present at BFP over time, resulting in the high contribution from multiple samples, but low overall completion. Low completion even in cases of high recruitment can also be due to the differential binning method utilized by CONCOCT, where closely related sequences are not able to be effectively binned separately into high quality MAGs.

3.2.4 Sulfur cycling genes

Of greatest interest for this study are the genes considered most important for sulfur cycling (both oxidation of reduced constituents, and the reduction of sulfate). MAGs were searched for well-known genes involved in sulfide oxidation (fcc, sqr), sulfite oxidation (sorAB), and thiosulfate oxidation (sox). A gene critical to sulfide oxidation, fcc, was found in MAGs 1, 5,

9, 19, 20, and 27, two of which (1 and 9) correspond to nearest neighbors in the

Epilonproteobacteria class (Sulfurimonas and Sulfurovum). The other five MAGs correspond to organisms within the Betaproteobacteria (Figure 3.5, see page 78). Sulfide-quinone reductase

(sqr) was only observed in four MAGs (3, 5, 6, and 25), three of which (3, 6, and 25) are

Alphaproteobacteria, with the other being most closely related to the genus Limnobacter, also containing the aforementioned fcc gene. No MAGs showed the presence of genes necessary for sulfite dehydrogenase (sorAB).

73

Figure 3.3. Anvi’o diagram and gene presence/absence table. A) This diagram shows all MAGs that are >50% completion and <10% redundancy. Higher contribution to a given MAG by a sample is indicated by darker bars, while lower contribution is indicated by lighter/no bars. B) This table shows the presence/absence of the specific genes queried in relation to the MAGs generated.

74 The Sox system is a widely studied pathway for the complete oxidation of thiosulfate

2- 2+ (S2O3 ) to sulfate (SO4 ). The metabolism is typically broken up into four different multi-gene proteins, which make up the larger Sox enzyme complex (soxAX, soxYZ, soxB, and soxCD).

Sox genes were identified in MAGs 1 (soxXA, soxYZ, and soxB) and 9 (sox XA and soxYZ), which were identified as being most closely related to the genera Sulfurimonas and Sulfurovum within the Epsilonproteobacteria, respectively. Sox genes were also identified in MAGs (5, 10,

14, 15, 19, 20, 22, 23, 27, and 29), with complete operons observed in MAGs 5, 10, 15, 20, and

23 (Figure 3.3). All but one of these MAGs are associated with Proteobacterial classes, including

Betaproteobacteria (5, 14, 19, 20, 27, and 29), Alphaproteobacteria (10), and

Gammaproteobacteria (15 and 23), with MAG 22 being nearest neighbors to the class

Candidatus Endomicrobium (Phylum Elusimicrobia). Not all MAGs had full sox gene complexes (Figure 3.3b) and some had more than one sequence copy associated with a given component (e.g. more than one soxA gene present). However, five of these MAGs (5, 10, 15, 20, and 23) appear to have genes present for all four of the sox-related proteins.

Three other sulfur-oxidizing genes that are typically co-associated are apr, sat, and qmo, which often work to oxidize sulfite to sulfate. In the green sulfur bacteria these genes are typically seen as part of the sat-aprBA-qmoABC gene operon, however, a similar aprAB- qmoABC operon has also been seen in sulfate-reducing bacteria (Dahl and Friedrich, 2008). The sat-aprBA-qmoABC was only identified in MAG 19, putatively identified within the genus

Thiobacillus (Family Thiobacillaceae). Moreover MAG 7, putatively identified as Desulfocapsa, contains both apr and qmo genes; however, this is not the case in MAG 12, also identified as

Desulfocapsa, where apr is absent, but sat is present alongside qmo. Overall, only three MAGs had qmo genes present (MAGs 7, 12, and 19) out of the 31 queried. In only one instance in the

75 BFP MAGs were apr and sat found together (MAG 5), which is most closely related to the

Betaproteobacterial genus Limnobacter. Otherwise, apr (present in 4 MAGs) was present much more sparsely that sat (present in 15 MAGs), which can be seen in Figure 3.3b. However, total homologous open reading frames (ORFs) is similar with apr having 16 and sat having 26 instances, respectively (Figure 3.4), most of which can be attributed to MAG 19 where 9 ORFs homologous to apr were present. The 3'-phosphoadenosine 5'-phosphosulfate (PAPS) gene was also identified in the majority of MAGs (18 of 31) and has shown in the past to be an important part of sulfate transfer reactions towards the formation of sulfite (Frigaard and Dahl, 2008).

Figure 3.4. Observed gene sequences within queried MAG contigs. The total number of gene sequences found when queried against all 31 reportable MAGs.Gene numbers are normalized to the total number of genes.

76 Dissimilatory sulfite-reductase (dsrAB) was also queried to identify the respiratory sulfate reduction potential of the system. dsrAB was present in three MAGs (7, 12, and 19) two of which correspond to the Deltaproteobacterial genus Desulfocapsa, while the other appears to be nearest to the Betaproteobacterial genus Thiobacillus. All three also have the DsrMKJOP gene complex present, as well as the dsr gamma subunit and some form of the assimilatory gene pathway.

3.2.5 Carbon metabolism genes

Carbon fixation potential was inferred by looking at three typical pathways: the reductive acetyl-coenzyme A pathway; the rTCA - reverse tricarboxylic acid cycle; and the Ribulose-1,5- bisphosphate carboxylase pathway (RuBisCO). ATP citrate lyase (acl), a key enzyme in the rTCA cycle (Perner et al., 2007), was only identified in two MAGs (1 and 9 respectively), while genes for rTCA were found within 17 out of 31 of the MAGs (Figure 3.3). It should be noted that the acl gene was found in both Epsilonproteobacterial MAGs (1 - Sulfurimonas and 9 -

Sulfurovum). The most commonly identified carbon fixation pathway within the queried genomes from BFP was the Wood–Ljungdahl (or reductive acetyl-CoA) pathway found in all but three MAGs (14, 22, and 31). The Calvin Benson-Bassham (RuBisCo) cycle was also identified in four MAGs with both cbbL/cbbM genes present (10, 19, 23, and 27). These results indicate that the predominant pathway for carbon fixation in the BFP system appears to be the Wood–

Ljungdahl pathway; however, the potential for use of the reductive TCA cycle for carbon fixation is possible across multiple MAGs as well.

77

Figure 3.5. Cladogram of MAGs. This cladogram of MAGs and nearest neighbors is used to help confirm putative taxonomy beyond what was determined via Centrifuge and RAST. It was created in the interactive tree of life (iTOL) using the output tree from pparser.

78 3.2.6 Nitrogen cycling genes

Genes associated with denitrification, specifically those associated with nitrate (nap and nar) and nitrite (nrf and nir) reduction were considered in MAG analysis. Genes identified as nap were more abundant than those putatively identified with homology to nar (Figure 3.4). No

MAGs had both genes, and all nap genes were found in higher quality MAGs (1, 2, 7, and 9), with the two instances of nar genes identified in lower quality MAGs (21 and 29). Nitrite reductase genes were more abundant across MAGs (nrf - 1, 2, 7, 12, 21, and 22; nir - 1, 3, 5, 8,

20, 24, 25, and 28), however, were not especially abundant in terms of the number of homologous ORFs detected. Only two MAGs appear to have the potential to completely reduce

- + nitrate (NO3 ) to ammonium (NH4 ; MAG 1 - nap, nrf, and nir; MAG 7 - nap and nrf).

3.2.7 Hydrogen, Oxygen, and Hopanoids

Due to their geobiological relevance, cytochrome c oxidase (cbb3-type) and hydrogenases hydAB and hoxU genes were also investigated. The cbb3-type cytochrome c oxidase (Cox) is an important gene in oxidative phosphorylation and was queried to look for the potential of organisms in the system to be able to utilize oxygen as a terminal electron acceptor. cbb3 was found in high abundance (18 of 31 MAGs) and is the second most abundant gene queried in the system behind those looked at for the reductive acetyl-CoA cycle. Only one instance of hyd was found in a contig associated with MAG 8. HoxU gene abundance was found in 9 out of 31

MAGs (Figure 3.3b). The gene hpnp, a hopanoid biosynthesis gene that codes for an enzyme that methylates hopanoids at the C2 position, is often used as a diagnostic for biomarker potential

(Garby et al., 2013) . However, among the BFP samples no hpnp genes were associated with any of the MAGs (data not shown).

79 3.3 Discussion

Low-temperature systems, particularly in polar environments, can provide a wealth of information about how microbial communities adapt and thrive under adverse conditions. In particular, sulfur-dominated systems such as Borup Fiord Pass (BFP) provide us with a window into microbial sulfur cycling in extreme environments. Through shotgun whole genome sequencing we have found that microbial community members at BFP have an abundance of genes related to the oxidation of reduced sulfur via the sox operon. This was predicted within known sulfur-oxidizing microorganisms (e.g. Epsilonproteobacteria); however, our results indicate that this metabolism may be functionally redundant in the system and more widespread than previously thought. The metabolic capability of microbes living in a sediment mound near the BFP 2009 spring was previously identified through metagenomic sequencing (Wright et al.,

2013); however, work presented here has expanded our understanding of the metabolic capability of low-temperature systems. The metagenome produced from Wright et al. (2013) was used to specifically address likely energetic pathways at the BFP 2009 spring based on gene availability and quantity and the most energetically favorable redox reactions. Samples for our research and another recent 16S rRNA gene survey (Trivedi et al., 2018) were gathered from multiple different site types to not only inquire about likely metabolisms but also to infer which microorganisms are participating in sulfur cycling across BFP, and to allow us to compare and contrast metabolic differences across BFP.

Sulfur cycling appears to be a crucial component for microbial energy production at BFP.

Previous work by Wright et al. (2013) from a single metagenome produced from a glacial elemental sulfur (S0) deposit (largely dominated by Epsilonproteobacteria) identified S0 oxidation as the most energetically favorable sulfur metabolism; however, questions remained as

80 to the extent of sulfur oxidation at BFP. Past BFP research identified (from spring fluid and mineral deposits) a number of the same organisms found within our MAGs: Marinobacter,

Loktanella (Gleeson et al., 2011; Grasby et al., 2003), and the sulfur oxidizers Sulfurimonas,

Sulfurovum, and Sulfuricurvum (Gleeson et al., 2011, 2012; Wright et al., 2013; Trivedi et al.,

2018). The majority of previous work was focused on sulfur cycling within the

Epsilonproteobacteria.

This study expanded on previous research, where sulfur-oxidizing genes were identified in

MAGs across BFP, and across multiple different taxonomic lineages. Our work has revealed that

MAGs belonging to previously identified genera may very well oxidize reduced sulfur compounds in addition to organisms that were not predicted to be taking part in such metabolic processes. One MAG (MAG 5), putatively identified within the genus Limnobacter, shows the presence of genes associated with the oxidation of hydrogen sulfide (fcc, sqr), as well as a complete sox pathway. This suggests it has the potential to fully oxidize hydrogen sulfide to sulfate. Chen et al. (2016) reported the full sox pathway within a Limnobacter sp. in an anaerobic methane oxidizing microbial community, while other studies have reported its ability to oxidize thiosulfate to sulfate (Kämpfer et al., 2001; Lu et al., 2011), however, we are unaware of any research linking Limnobacter sp. to environmental sulfide oxidation.

This unexpected trend of functional redundancy broadens the scope of other MAGs not previously predicted to participate in sulfur-oxidation (via sox). This includes three putatively identified within the genera Loktanella, Herminiimonas, and Rhodoferax (MAG 10, 14, and 20, respectively). Loktanella sp. were identified at BFP via 16S rRNA gene clone libraries (Gleeson et al., 2011) as well as in other low-temperature environments in the Canadian Arctic (Battler et al., 2013; Niederberger et al., 2009; Perreault et al., 2007). Loktanella was previously reported as

81 containing a soxB sequence from samples obtained at Gypsum Hill, a perennial Arctic spring on

Axel Heiberg Island (Perreault et al., 2008). This study also found that the Loktanella sp. was part of a consortium along with a Marinobacter sp. similar to what was found in previous samples at BFP (Gleeson et al., 2011; Trivedi et al., 2018). This suggests that these two organisms often coexist together in low-temperature Arctic environments.

In contrast to previous 16S rRNA gene sequencing analyses, we putatively identified the genera Herminiimonas and Rhodoferax through whole genome sequencing and binning. Despite its identification within a 120,000-year-old Greenland ice core (Loveland-Curtze et al., 2009),

Herminiimonas is rarely identified within polar environments. Moreover, the data on the capability of Herminiimonas to oxidize sulfur is equally sparse. In a recent report Koh et al.

(2017) found that Herminiimonas arsenitoxidans, which was found to oxidize arsenite, was also able to oxidize sulfur when grown on trypticase soy agar. Conversely, another

Betaproteobacteria that is more commonly found in polar environments (Lutz et al., 2015;

Madigan et al., 2000), Rhodoferax, are known as purple non-sulfur bacteria. Most are facultatively photoheterotrophic; however, some are capable of photoautotrophy when sulfide is used as an electron donor (Ghosh and Dam, 2009). A BFP MAG putatively identified as

Rhodoferax (MAG 20) also shows the presence of genes related to the reductive acetyl-CoA pathway, which could indicate that this particular organism is capable of autotrophy.

In addition to sulfur oxidation, dissimilatory sulfate reduction is also a key component in the sulfur cycle and potentially in energy generation at BFP. Out of the 31 identified MAGs, three (MAG 7, 12, and 19) contain dissimilatory sulfite reductase (dsrAB), a key step in the complete reduction of sulfate to sulfide. Two MAGs, MAG 7 and 12, are most closely related to the genus Desulfocapsa, a known sulfur disproportionator within the Deltaproteobacteria (Finster

82 et al., 1998, 2013). Another MAG most closely related to the Betaproteobacterial genus

Thiobacillus (MAG 19) also contains a complete dsr operon. Interestingly, MAG 19 is the only one that appears capable of both sulfur-oxidation via the sox pathway and sulfate-reduction via dissimilatory sulfite reductase. Thiobacillus sp. are traditionally known as SOMs and have been found in polar environments previously (Boyd et al., 2014; Harrold et al., 2016; Purcell et al.,

2014); however, it has also been suggested that certain Thiobacillus species may be able to run the dissimilatory sulfite reductase (rDsr) pathway in reverse as an alternative means to oxidize reduced sulfur compounds (Loy et al., 2009; Purcell et al., 2014). This could well be an alternative pathway that is present in the BFP system; however, it would require transcriptomic evidence to confirm its activity. It should be noted that MAG 19 appears to lack soxCD; however, the MAG is only ~73% complete, that may indicate that its full metabolic potential is not represented. So, while some genes were not detected where they may have been expected, this does not mean they are not part of that organism’s genome.

The lack of the presence of soxCD in BFP MAGs (e.g. MAGs 1, 9, and 19) from known sulfur-oxidizing microorganisms (Sulfurimonas, Sulfurovum, and Thiobacillus) is an interesting point to note. In fact this may be one possible explanation for the abundance of elemental sulfur seen in so many sites across BFP. Previous research has shown that in some organisms that lack soxCD, the Sox system is used to oxidize thiosulfate to sulfur which is then either stored inside the cell or excreted (Hensen et al., 2006). Some organisms may employ other pathways like reverse dissimilatory sulfite reductase (rDsr) to oxidize this accumulation of S0 (Hamilton et al.,

2015); however the low prevalence any sulfite reductase genes in the BFP system could point to an accumulation of elemental sulfur rather than additional biogenic oxidation.

The utilization of carbon in the BFP system is understudied and how it plays a part in

83 microbial growth dynamics is important to determine, particularly in terms of how carbon enters the system. This study focuses primarily on the fixation of carbon by Epsilonproteobacteria in the BFP system due to their known capacity for autotrophy and sulfur utilization (Campbell et al., 2006). Genes related to the reductive acetyl-CoA cycle (Wood-Ljungdahl in reverse; Carr et al., 2015) appears across all but three of the MAGs (14, 22, and 31). This is in contrast to the reductive TCA cycle where genes were present in 17 out of 31 MAGs (as well as two MAGs for acl). The reductive TCA cycle has previously been implicated as the primary means of converting inorganic carbon into biomass for organisms within the Epsilonproteobacteria (Dahl and Friedrich, 2008; Hügler et al., 2010). However, the reductive acetyl-CoA cycle, also known as the Arnon cycle (Ghosh and Dam, 2009), appear to be the predominant carbon fixation pathway throughout all of our MAG data. In a study by Momper et al. (2017), the presence of this carbon fixation pathway was suggested to likely be due to energy limitations within deep subsurface sediments, with the reductive acetyl-CoA pathway being the most energetically favorable for microorganisms compared to other carbon fixation pathways. While BFP is an extreme environment, OC data and bioenergetics calculations suggest that BFP microbes are not at the lower thermodynamic limit of life and have plenty of energy-generating pathways available to them (Lau et al., 2017; Wright et al., 2013). We should also note, the reductive acetyl-CoA pathway also requires anoxic conditions, which is contrary to the oxygen-rich surface sampling conditions at BFP. A possible explanation for this might be the establishment of anoxic zones just beneath the surface of any mineral precipitate or melt pool sediment materials. Poniecka et al. (2017) found that within cryoconite sediments an anoxic zone can re- establish at 2 to 4 mm from the surface of the sediment, as soon as an hour after disturbance.

This phenomenon may very well occur in melt pool and mineral precipitate materials at BFP

84 allowing the reductive acetyl-CoA pathway to be the predominant carbon fixation method. This is, however, in contrast to the abundance of the cbb3 cytochrome c oxidase gene, whose presence suggests that oxygen is likely the terminal electron donor for oxidative processes across BFP sites. Sulfur oxidation and carbon fixation may occur under microoxic conditions where the reductive acetyl-CoA pathway is still effective as the dominant carbon fixation process.

Cytochrome c oxidases have been shown to operate under low oxygen tension (Koh et al., 2017); however, this has not yet been established for the cbb3 cytochrome oxidase.

Based on gene presence/absence we infer that some MAGs from our system may be capable of a form of nitrate/nitrite reduction. The nap gene, or periplasmic nitrate reductase, is the most abundant identified nitrate reductase (Figure 3.4), with multiple copies identified in

MAGs 1, 2, and 9 (Sulfurimonas, Shewanella, and Sulfurovum, respectively). This suggests that the SOMs Sulfurimonas and Sulfurovum may be capable of oxidizing reduced sulfur compounds and utilizing nitrate as an electron acceptor. The nir gene, a nitrite reductase, which reduces nitrite to nitric oxide (NO) is slightly less abundant than nrf, which reduces nitrite to ammonium

+ (NH4 ). It should be noted that only MAGs which show the presence of nap also have nir. Lower levels of nitrate have been reported at BFP (0.0030 mM in 2009, and 0.0908 mM in 2014) from a spring and melt pool, respectively. While most organisms at BFP are probably utilizing oxygen as an electron acceptor, metagenomic evidence supports the idea that denitrification is a viable respiration pathway as well.

The abundance of sox genes present in BFP samples and across multiple bacterial lineages suggests that the concept of functional redundancy may play a key role in sulfur cycling in the

BFP ecosystem. Functional redundancy was recently identified as a commonplace process in environmental systems by Louca et al. (2018). These authors speculate that the degree of

85 functional redundancy in the system is largely determined by the type of environment, where large systemic forces may be in place and driving desired functions. Moreover, they call into question a previous notion that functional redundancy in a system might infer a form of ‘neutral co-existence’ between competing microorganisms (Loreau, 2004). They disagree with this sentiment, noting that it is more likely that functional redundancy is an emergent process based on environmental pressures. The response to ecological change by the ‘rare biosphere,’ i.e., the low representation of microbiota in any environment that can become dominant upon environmental condition change, thus reflects a biological manifestation of functional redundancy. The emergence of functional redundancy due to environmental pressures is likely occurring at BFP. Previous research found a core microbial community not predicted to participate in sulfur cycling exists over longer periods (Trivedi et al., 2018). Our data indicate that these core communities may have the potential to utilize sulfur compounds for growth.

Louca et al. (2018) also suggest that coexisting organisms may not always be in equal abundance as there will always be differences in functional ability based on enzyme efficiencies and growth kinetics. This also supports previous research at the BFP system, where 16S rRNA gene sequencing revealed what appeared to be acute “blooms” of activity of SOMs over short periods but would then ultimately return to a basal community (Trivedi et al., 2018). This might indicate that if there is a large influx of reduced sulfur compounds, some organisms like the

Epsilonproteobacteria are able to more efficiently use these compounds, and thus we would see an increase in their relative abundance until settling to previously identified levels. Undoubtedly, sulfur oxidation is the predominant and preferred metabolism at BFP. Our data show that this metabolism is functionally redundant and possibly able to be utilized by more microorganisms than were previously predicted.

86 3.4 Conclusion

The metabolic potential of microorganisms found across a sulfur-dominated proglacial system was characterized using high-throughput sequencing. Proteobacteria are the most dominant microorganisms present in the low-temperature BFP system, and we confirmed that sulfur oxidation is the dominant metabolism across the system. Evidence also suggests that functionally redundant sulfur-oxidation pathways (sox) are in place across organisms spanning multiple phylogenetic lineages, possible due to environmental pressures. Genes are present for dissimilatory sulfite reduction; however, it is not clear whether or not microorganisms containing these genes contributing to the production of reduced sulfur compounds in the system.

Microorganisms capable of denitrification are also present, often linked to MAGs with the ability to oxidize sulfur, suggesting these two processes are linked. Carbon metabolism clearly points to the reductive acetyl-CoA pathway being the dominant form of carbon fixation in the system, alluding to the fact that this may be occurring under anoxic or microoxic conditions. These results show that in low-temperature environments functional redundancy of key metabolisms is a worthwhile strategy for multiple bacterial lineages to coexist together. The results presented here are the potential metabolisms present within the high-quality MAGs produced from the BFP samples. Further research will begin to tease apart the differences between metabolic capability and active processes. Metatranscriptomic profiling of BFP sites across different geochemical states of the system will be invaluable in determining which of these metabolisms are active within the system, allowing us to further constrain the most energetically favorable metabolisms of core microorganisms in the BFP ecosystem.

87 3.5 Methods

3.5.1 Site Description and Sample Collection

Borup Fiord Pass is at the height of a north-south trending valley through the Krieger mountains at 81o 01’ N, 81o 38’ W (Figure 3.1). The spring system of interest is approximately

210-240 m above sea level and occurs at the toe of two coalesced glaciers (Grasby et al., 2003).

The spring system is thought to be perennial in nature but manifests in different locations along the toe of the glacier from year to year (Gleeson et al., 2010; Grasby et al., 2003; Lau et al.,

2017; Wright et al., 2013). However, given lack of winter observations it is not clear if spring flow is year-round or limited to the warmer spring and summer months. A lateral fault approximately 100 m south of the toe of the glacier has been implicated in playing a role in the subsurface hydrology, in that this may be one of the areas where subsurface fluids are able to contact the surface (Grasby et al., 2003; Scheidegger, et al. 2012). Furthermore, it has been suggested that this fault may be a source of carbon for subsurface microbial processes such as biological sulfate reduction (BSR; Lau et al., 2017)..

BFP sample material was collected during multiple days between 21 June and 2 July 2014,

4 July 2016, and 7 July 2017. Samples included: 1) Sedimented sulfur-like cryoconite material and fluid found in glacial surface melt pools (labelled M1-M6); 2) filtered fluid from thawed aufeis (labelled A1-A6); 3) Aufeis surface samples (material scraped from top of the aufeis) from

2014 and 2017 (labelled AS1-AS7); 4) Glacial control samples (G1-G3); and 5) spring fluid from 2016 (labelled Spring 2016). Cryoconite sediment was collected using sterile and field- washed transfer pipettes. Up to 0.5 g of this material plus water was placed into ZR

BashingBead™ ) lysis tubes (Zymo Research Corp., Irvine, CA, USA containing 750 µL of

Xpedition™ Lysis/Stabilization Solution (Zymo Research Corp.) and homogenized for 45

88 seconds using the Zymo Xpedition™ portable impact drill (Zymo Research Corp.). Samples were then stored on ice until DNA extraction was performed. Aufeis surface scrapings from the

2017 field campaign were collected using sterile and field-washed transfer pipettes and up to 0.5 g of material were mixed in ZR BashingBead™ lysis tubes containing 750 ml of DNA/RNA

Shield™ (Zymo Research Corp.). Samples were shaken and kept at 4 °C until returned to the laboratory, where they were then stored at -80 °C until DNA/RNA extraction could be performed. Fluid samples for DNA extraction were filtered through 0.22 µm Luer-lok Sterivex™ filters (EMD Millipore; Darmstadt, Germany), which were then capped and kept on ice in the field until returning to the laboratory where they were stored at -20 °C until extraction. Fluids for aqueous geochemical measurement were filtered with 0.22 µm polyethersulfone (PES; EMD

Millipore, Billerica, MA, USA) filters into 15 mL polypropylene tubes for ion chromatography

(IC) and inductively-coupled plasma optical emission spectroscopy (ICP-OES) for measurement of major anions and cations, respectively. Samples for ICP-OES were acidified with approximately 0.1 mL nitric acid in the field.

3.5.2 DNA extraction and metagenomic library preparation

DNA extraction of 2014 surface sediment material was performed using the Xpedition™

Soil/Fecal DNA MiniPrep kit (Zymo Research Corp.). DNA extraction of 2014 melt pool and

2016 spring fluid samples was performed using the PowerWater® Sterivex™ DNA Isolation Kit

(MO BIO Laboratories, USA), and DNA extraction of 2017 surface sediment samples was extracted using the ZymoBIOMICS™ DNA/RNA Mini Kit (Zymo Research Corp.). All extractions were performed according to the manufacturer’s instructions. Extracted DNA concentrations were determined using a QuBit 2.0 fluorometer (Thermo Fisher Scientific, Chino,

89 CA, USA) and the QuBit dsDNA HS assay (Life Technologies, Carlsbad, CA, USA).

Genomic DNA was first cleaned using KAPA Pure Beads (KAPA Biosystems Inc.,

Wilmington, MA, USA) at a final concentration of 0.8X v/v prior to library preparation. Two metagenomic sequencing runs were completed for the included sample set. The first (including samples A4b, A6, and M4b) were normalized to 1 ng of DNA in 35 µl of molecular biology grade water, and prepared using the KAPA HyperPlus Kit (KAPA Biosystems Inc., Wilmington,

MA, USA) and KAPA WGS adapters. Fragmentation and adapter ligation was performed according to the manufacturer’s instructions. Prepped samples were quality checked on an

Agilent 2100 Bioanalyzer (Agilent Genomics, Santa Clara, CA, USA) and submitted to the Duke

Center for Genomic and Computational Biology (Duke University, Durham, NC, USA). Final pooled libraries were run on an Illumina HiSeq 4000 (Illumina, San Diego, CA, USA) using

PE150 chemistry. The second sequencing run (which included samples M2, Spring 2016, and

2017 3B, 4E, 6B, and 14C) was prepped using the Nextera XT DNA Library Preparation Kit

(Illumina, San Diego, CA, USA) in conjunction with Nextera XT V2 Indexes. Libraries were hand normalized and quality checked on the Agilent 2100 Bioanalyzer (Agilent Genomics, Santa

Clara, CA, USA). Samples were submitted to the Duke Center for Genomic and Computational

Biology (Duke University, Durham, NC, USA). These libraries were sequenced on an Illumina

HiSeq 2500 Rapid Run using PE250 chemistry.

3.5.3 Metagenomic Sequencing Assembly and Analysis

Processing of metagenomic sequencing was performed on the Summit High Performance

Cluster (HPC) located at the University of Colorado, Boulder (Boulder, CO, USA). Assembly of whole-genome sequencing reads used a modified Joint Genome Institute (JGI; Walnut Creek,

90 CA, USA) metagenomics workflow. We began by using BBDuk of the BBMap/BBTools

(v37.56; Bushnell, 2014) suite to trim out Illumina adapters using the built-in BBMap adapters database. Trimmed sequences were checked for quality using FastQC v11.5 (Andrews, 2014) to ensure that adapters were removed successfully. Trimmed paired-end sequences were then joined using the PEAR package (Zhang et al., 2014) v0.9.10 using a p-value of 0.001. Adapter trimmed files, both forward (R1) and reverse (R2) were then processed using BBTools repair.sh to ensure read pairs were in the correct order. This script is designed to reorder paired reads that have become disorganized, which can often occur during trimming. The BFC (Bloom Filter

Correction) software tool was then used to error correct paired sequence reads (Li, 2015). Reads were again processed with BBTools repair.sh to ensure proper read pairing and order. Next,

Megahit v1.1.2 (Li et al., 2015) was used to generate both a co-assembly, as well as individual sample assemblies from the quality controlled reads. Post assembly, co-assembly was processed with ‘anvi-script-reformat-fasta’ for input into Anvi’o, and contigs below 2500 bp were removed. Bowtie2 v2.3.0 (Langmead and Salzberg, 2012) was then used to map individual sample sequence reads onto the co-assembly, needed for the Anvi’o pipeline. Anvi’o v4 (Eren et al., 2015) was then used to characterize, and then bin co-assembled contigs into metagenome assembled genomes (MAGs). The taxonomic classifier Centrifuge v1.0.2-beta (Kim et al., 2016) was used to provide a preliminary taxonomy of the co-assembly contigs, useful for Anvi’o binning. The NCBI Clusters of Orthologous Groups (COGs; (Galperin et al., 2015; Tatusov et al., 2000)) database was used within Anvi’o to predict COGs present within the co-assembly, and CONCOCT (Alneberg et al., 2014) was employed to create genome bins using a combination of sequence coverage and composition of our assembled contigs. Anvi’o was then used to visualize the putative bins in conjunction with other metadata to refine and identify

91 MAGs with 50% completeness and 10% redundancy. The scripts used in the processing of these data (up to the point of Anvi’o contigs database) are publicly available at https://doi.org/10.5281/zenodo.1302787.

Anvi’o refined bins were then processed using CheckM v1.1.11 (Parks et al., 2015) where a total of 31 bins were selected as MAGs (Metagenome Assembled Genomes) which followed the > 50% completeness and < 10% redundancy criteria. MAGs that met these criteria were then renamed; from 1-31 in the order of highest completeness to lowest (Table 3.1). CheckM relies on other software including pplacer v1.1.alpha17 (Matsen et al., 2010) for phylogenetic tree placement, prodigal v2.6.3 (Hyatt et al., 2012) for gene translation and initiation site prediction, and HMMER v3.1b2 (Eddy, 2011) which analyzes sequence data using hidden Markov models.

The pplacer generated tree was then visualized using the interactive Tree of Life (iTOL; (Letunic and Bork, 2016)) to search for nearest neighbors of identified MAGs (see Figure 3.5, Pg 78).

Finally, to confirm general taxonomic classification and annotate genes within our MAGs, samples were processed using the Rapid Annotations using Subsystem Technologies (RAST) server (Aziz et al., 2008). Genes of interest to this study were identified within the RAST web interface. Further taxonomic classification was done for MAGs 7 and 12 using the online tool

JSpeciesWS (Richter et al., 2016), which uses pairwise genome comparisons to measure the probability of two or more genomes. Tests within JSpeciesWS include average nucleotide identity (ANI) via BLAST and MUMmer and correlation between Tetra-nucleotide signatures.

Average nucleotide identity (ANI) was used to determine if MAGs 7 and 12 (putatively classified as Desulfotalea) were more closely related to the genus Desulfocapsa as was previously identified within BFP samples (Trivedi et al., Accepted).

92 3.6 Data availability

Raw reads are available at the NCBI sequencing read archive

(https://www.ncbi.nlm.nih.gov/sra).

3.7 References

Alneberg, J., Bjarnason, B. S., de Bruijn, I., Schirmer, M., Quick, J., Ijaz, U. Z., et al. (2014). Binning metagenomic contigs by coverage and composition. Nat. Methods 11, 1144–1146. doi:10.1038/nmeth.3103.

Andrews, S. (2014). FastQC software package. Available at: https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ [Accessed March 13, 2018].

Aziz, R. K., Bartels, D., Best, A. A., DeJongh, M., Disz, T., Edwards, R. A., et al. (2008). The RAST Server: Rapid Annotations using Subsystems Technology. BMC Genomics 9, 75. doi:10.1186/1471-2164-9-75.

Battler, M. M., Osinski, G. R., and Banerjee, N. R. (2013). Mineralogy of saline perennial cold springs on Axel Heiberg Island, Nunavut, Canada and implications for spring deposits on Mars. Icarus 224, 364–381. doi:10.1016/j.icarus.2012.08.031.

Boetius, A., Anesio, A. M., Deming, J. W., Mikucki, J. A., and Rapp, J. Z. (2015). Microbial ecology of the cryosphere: sea ice and glacial habitats. Nat. Rev. Microbiol. 13, 677–690. doi:10.1038/nrmicro3522.

Bowles, M. W., Mogollón, J. M., Kasten, S., Zabel, M., and Hinrichs, K.-U. (2014). Global rates of marine sulfate reduction and implications for sub-sea-floor metabolic activities. Science 344, 889–91. doi:10.1126/science.1249213.

Boyd, E. S., Hamilton, T. L., Havig, J. R., Skidmore, M. L., and Shock, E. L. (2014). Chemolithotrophic primary production in a subglacial ecosystem. Appl. Environ. Microbiol. 80, 6146–6153. doi:10.1128/AEM.01956-14.

Brosnan, J. T., and Brosnan, M. E. (2006). The Sulfur-Containing Amino Acids: An Overview. J. Nutr. 136, 1636S–1640S. doi:10.1093/jn/136.6.1636S.

Bushnell, B. (2014). BBMap software package. Available at: https://sourceforge.net/projects/bbmap/ [Accessed March 13, 2018].

Campbell, B. J., Engel, A. S., Porter, M. L., and Takai, K. (2006). The versatile ε-proteobacteria: Key players in sulphidic habitats. Nat. Rev. Microbiol. 4, 458–468. doi:10.1038/nrmicro1414.

Carr, S. A., Orcutt, B. N., Mandernack, K. W., and Spear, J. R. (2015). Abundant Atribacteria in

93 deep marine sediment from the Adélie Basin, Antarctica. Front. Microbiol. 6, 1–12. doi:10.3389/fmicb.2015.00872.

Chen, Y., Feng, X., He, Y., and Wang, F. (2016). Genome Analysis of a Limnobacter sp. Identified in an Anaerobic Methane-Consuming Cell Consortium. Front. Mar. Sci. 3, 257. doi:10.3389/fmars.2016.00257.

Dahl, C., and Friedrich, C. G. (2008). Microbial Sulfur Metabolism. , eds. C. Dahl and C. G. Friedrich Berlin, Heidelberg: Springer Berlin Heidelberg doi:10.1007/978-3-540-72682-1.

Eddy, S. R. (2011). Accelerated Profile HMM Searches. PLoS Comput. Biol. 7, e1002195. doi:10.1371/journal.pcbi.1002195.

Enning, D., Venzlaff, H., Garrelfs, J., Dinh, H. T., Meyer, V., Mayrhofer, K., et al. (2012). Marine sulfate-reducing bacteria cause serious corrosion of iron under electroconductive biogenic mineral crust. Environ. Microbiol. 14, 1772–1787. doi:10.1111/j.1462- 2920.2012.02778.x.

Eren, A. M., Esen, Ö. C., Quince, C., Vineis, J. H., Morrison, H. G., Sogin, M. L., et al. (2015). Anvi’o: an advanced analysis and visualization platform for ‘omics data. PeerJ 3, e1319. doi:10.7717/peerj.1319.

Finster, K., Liesack, W., and Thamdrup, B. (1998). Elemental sulfur and thiosulfate disproportionation by Desulfocapsa sulfoexigens sp. nov., a new anaerobic bacterium isolated from marine surface sediment. Appl. Environ. Microbiol. 64, 119–125.

Finster, K. W., Kjeldsen, K. U., Kube, M., Reinhardt, R., Mussmann, M., Amann, R., et al. (2013). Complete genome sequence of Desulfocapsa sulfexigens, a marine deltaproteobacterium specialized in disproportionating inorganic sulfur compounds. Stand. Genomic Sci. 8, 58–68. doi:10.4056/sigs.3777412.

Friedrich, C. G., Bardischewsky, F., Rother, D., Quentmeier, A., and Fischer, J. (2005). Prokaryotic sulfur oxidation. Curr. Opin. Microbiol. 8, 253–259. doi:10.1016/j.mib.2005.04.005.

Friedrich, C. G., Rother, D., Bardischewsky, F., Quentmeier, A., and Fischer, J. (2001). Oxidation of reduced inorganic sulfur compounds by bacteria: emergence of a common mechanism? Appl. Environ. Microbiol. 67, 2873–82. doi:10.1128/AEM.67.7.2873- 2882.2001.

Frigaard, N. U., and Dahl, C. (2008). Sulfur Metabolism in Phototrophic Sulfur Bacteria. doi:10.1016/S0065-2911(08)00002-7.

Galperin, M. Y., Makarova, K. S., Wolf, Y. I., and Koonin, E. V. (2015). Expanded microbial genome coverage and improved protein family annotation in the COG database. Nucleic Acids Res. 43, D261–D269. doi:10.1093/nar/gku1223.

Garby, T. J., Walter, M. R., Larkum, A. W. D., and Neilan, B. A. (2013). Diversity of

94 cyanobacterial biomarker genes from the stromatolites of Shark Bay, Western Australia. Environ. Microbiol. 15, 1464–1475. doi:10.1111/j.1462-2920.2012.02809.x.

Ghosh, W., and Dam, B. (2009). Biochemistry and molecular biology of lithotrophic sulfur oxidation by taxonomically and ecologically diverse bacteria and archaea. FEMS Microbiol. Rev. 33, 999–1043. doi:10.1111/j.1574-6976.2009.00187.x.

Gleeson, D. F., Pappalardo, R. T., Anderson, M. S., Grasby, S. E., Mielke, R. E., Wright, K. E., et al. (2012). Biosignature detection at an Arctic analog to Europa. Astrobiology 12, 135– 150. doi:10.1089/ast.2010.0579.

Gleeson, D. F., Pappalardo, R. T., Grasby, S. E., Anderson, M. S., Beauchamp, B., Castaño, R., et al. (2010). Characterization of a sulfur-rich Arctic spring site and field analog to Europa using hyperspectral data. Remote Sens. Environ. 114, 1297–1311. doi:10.1016/j.rse.2010.01.011.

Gleeson, D. F., Williamson, C., Grasby, S. E., Pappalardo, R. T., Spear, J. R., and Templeton, A. S. (2011). Low temperature S0 biomineralization at a supraglacial spring system in the Canadian high Arctic. Geobiology 9, 360–375. doi:10.1111/j.1472-4669.2011.00283.x.

Grasby, S. E., Allen, C. C., Longazo, T. G., Lisle, J. T., Griffin, D. W., and Beauchamp, B. (2003). Supraglacial sulfur springs and associated biological activity in the Canadian high Arctic—signs of life beneath the ice. Astrobiology 3, 583–596. doi:10.1089/153110703322610672.

Grasby, S. E., Beauchamp, B., and Bense, V. (2012). Sulfuric acid speleogenesis associated with a glacially driven groundwater system - Paleo-spring “pipes” at Borup Fiord Pass, Nunavut. Astrobiology 12, 19–28. doi:10.1089/ast.2011.0700.

Hamilton, T. L., Jones, D. S., Schaperdoth, I., and Macalady, J. L. (2015). Metagenomic insights into S(0) precipitation in a terrestrial subsurface lithoautotrophic ecosystem. Front. Microbiol. 5, 756. doi:10.3389/fmicb.2014.00756.

Harrold, Z. R., Skidmore, M. L., Hamilton, T. L., Desch, L., Amada, K., Van Gelder, W., et al. (2016). Aerobic and anaerobic thiosulfate oxidation by a cold-adapted, subglacial chemoautotroph. Appl. Environ. Microbiol. 82, 1486–1495. doi:10.1128/AEM.03398-15.

Heiserman, D. L. (1992). Exploring chemical elements and their compounds. Tab Books Available at: https://books.google.com/books/about/Exploring_Chemical_Elements_and_Their_Co.html? id=PDfHQgAACAAJ [Accessed June 14, 2018].

Hensen, D., Sperling, D., Trüper, H. G., Brune, D. C., and Dahl, C. (2006). Thiosulphate oxidation in the phototrophic sulphur bacterium Allochromatium vinosum. Mol. Microbiol. 62, 794–810. doi:10.1111/j.1365-2958.2006.05408.x.

Hügler, M., Gärtner, A., and Imhoff, J. F. (2010). Functional genes as markers for sulfur cycling and CO2 fixation in microbial communities of hydrothermal vents of the Logatchev field.

95 FEMS Microbiol. Ecol. 73, no-no. doi:10.1111/j.1574-6941.2010.00919.x.

Hyatt, D., LoCascio, P. F., Hauser, L. J., and Uberbacher, E. C. (2012). Gene and translation initiation site prediction in metagenomic sequences. Bioinformatics 28, 2223–2230. doi:10.1093/bioinformatics/bts429.

Kämpfer, P., Spring, S., and Schleifer, K. H. (2001). Limnobacter thiooxidans gen. nov., sp. nov., a novel thiosulfate-oxidizing bacterium isolated from freshwater lake sediment. Int. J. Syst. Evol. Microbiol. 51, 1463–1470. doi:10.1099/00207713-51-4-1463.

Kim, D., Song, L., Breitwieser, F. P., and Salzberg, S. L. (2016). Centrifuge: rapid and sensitive classification of metagenomic sequences. Genome Res. 26, 1721–1729. doi:10.1101/gr.210641.116.

Koh, H.-W., Hur, M., Kang, M.-S., Ku, Y.-B., Ghai, R., and Park, S.-J. (2017). Physiological and genomic insights into the lifestyle of arsenite-oxidizing Herminiimonas arsenitoxidans. Sci. Rep. 7, 15007. doi:10.1038/s41598-017-15164-4.

Langmead, B., and Salzberg, S. L. (2012). Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359. doi:10.1038/nmeth.1923.

Lau, G. E., Cosmidis, J., Grasby, S. E., Trivedi, C. B., Spear, J. R., and Templeton, A. S. (2017). Low-temperature formation and stabilization of rare allotropes of cyclooctasulfur (β-S 8 and γ-S 8 ) in the presence of organic carbon at a sulfur-rich glacial site in the Canadian High Arctic. Geochim. Cosmochim. Acta 200, 218–231. doi:10.1016/j.gca.2016.11.036.

Letunic, I., and Bork, P. (2016). Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucleic Acids Res. 44, W242–W245. doi:10.1093/nar/gkw290.

Li, D., Liu, C. M., Luo, R., Sadakane, K., and Lam, T. W. (2015). MEGAHIT: An ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics 31, 1674–1676. doi:10.1093/bioinformatics/btv033.

Li, H. (2015). BFC: Correcting Illumina sequencing errors. Bioinformatics 31, 2885–2887. doi:10.1093/bioinformatics/btv290.

Loreau, M. (2004). Does functional redundancy exist? Oikos 104, 606–611. doi:10.1111/j.0030- 1299.2004.12685.x.

Louca, S., Polz, M., Mazel, F., Albright, M., Julie, H., O’Connor, M., et al. (2018). Function and functional redundancy in microbial systems. Nat. Ecol. Evol. Accepted. doi:10.1038/s41559-018-0519-1.

Loveland-Curtze, J., Miteva, V. I., and Brenchley, J. E. (2009). Herminiimonas glaciei sp. nov., a novel ultramicrobacterium from 3042 m deep Greenland glacial ice. Int. J. Syst. Evol. Microbiol. 59, 1272–1277. doi:10.1099/ijs.0.001685-0.

96 Loy, A., Duller, S., Baranyi, C., Mussmann, M., Ott, J., Sharon, I., et al. (2009). Reverse dissimilatory sulfite reductase as phylogenetic marker for a subgroup of sulfur-oxidizing prokaryotes. Environ. Microbiol. 11, 289–99. doi:10.1111/j.1462-2920.2008.01760.x.

Lu, H., Sato, Y., Fujimura, R., Nishizawa, T., Kamijo, T., and Ohta, H. (2011). Limnobacter litoralis sp. nov., a thiosulfate-oxidizing, heterotrophic bacterium isolated from a volcanic deposit, and emended description of the genus Limnobacter. Int. J. Syst. Evol. Microbiol. 61, 404–407. doi:10.1099/ijs.0.020206-0.

Lutz, S., Anesio, A. M., Field, K., and Benning, L. G. (2015). Integrated “Omics”, Targeted Metabolite and Single-cell Analyses of Arctic Snow Algae Functionality and Adaptability. Front. Microbiol. 6, 1323. doi:10.3389/fmicb.2015.01323.

Madigan, M. T., Jung, D. O., Woese, C. R., and Achenbach, L. A. (2000). Rhodoferax antarcticus sp. nov., a moderately psychrophilic purple nonsulfur bacterium isolated from an Antarctic microbial mat. Arch. Microbiol. 173, 269–77. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10816045 [Accessed April 21, 2018].

Mandeville, C. W. (2010). Sulfur: A Ubiquitous and Useful Tracer in Earth and Planetary Sciences. Elements 6, 75–80. doi:10.2113/gselements.6.2.75.

Matsen, F. A., Kodner, R. B., and Armbrust, E. V. (2010). pplacer: linear time maximum- likelihood and Bayesian phylogenetic placement of sequences onto a fixed reference tree. BMC Bioinformatics 11, 538. doi:10.1186/1471-2105-11-538.

Mattes, T. E., Nunn, B. L., Marshall, K. T., Proskurowski, G., Kelley, D. S., Kawka, O. E., et al. (2013). Sulfur oxidizers dominate carbon fixation at a biogeochemical hot spot in the dark ocean. ISME J. 7113, 2349–2360. doi:10.1038/ismej.2013.113.

Mikucki, J. A., Lee, P. A., Ghosh, D., Purcell, A. M., Mitchell, A. C., Mankoff, K. D., et al. (2016). Subglacial Lake Whillans microbial biogeochemistry: A synthesis of current knowledge. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 374, 20140290. doi:10.1098/rsta.2014.0290.

Mikucki, J. A., Pearson, A., Johnston, D. T., Turchyn, A. V, Farquhar, J., Schrag, D. P., et al. (2009). A Contemporary Microbially Maintained Subglacial Ferrous “Ocean.” Science (80-. ). 324, 397–400. doi:10.1126/science.1167350.

Mikucki, J. A., and Priscu, J. C. (2007). Bacterial diversity associated with blood falls, a subglacial outflow from the Taylor Glacier, Antarctica. Appl. Environ. Microbiol. 73, 4029– 4039. doi:10.1128/AEM.01396-06.

Momper, L., Jungbluth, S. P., Lee, M. D., and Amend, J. P. (2017). Energy and carbon metabolisms in a deep terrestrial subsurface fluid microbial community. ISME J. 11, 2319– 2333. doi:10.1038/ismej.2017.94.

Niederberger, T. D., Perreault, N. N., Lawrence, J. R., Nadeau, J. L., Mielke, R. E., Greer, C. W., et al. (2009). Novel sulfur-oxidizing streamers thriving in perennial cold saline springs of

97 the Canadian high Arctic. Environ. Microbiol. 11, 616–629. doi:10.1111/j.1462- 2920.2008.01833.x.

Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P., and Tyson, G. W. (2015). CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–55. doi:10.1101/gr.186072.114.

Perner, M., Seifert, R., Weber, S., Koschinsky, A., Schmidt, K., Strauss, H., et al. (2007). Microbial CO2 fixation and sulfur cycling associated with low-temperature emissions at the Lilliput hydrothermal field, southern Mid-Atlantic Ridge (9°S). Environ. Microbiol. 9, 1186–1201. doi:10.1111/j.1462-2920.2007.01241.x.

Perreault, N. N., Andersen, D. T., Pollard, W. H., Greer, C. W., and Whyte, L. G. (2007). Characterization of the prokaryotic diversity in cold saline perennial springs of the Canadian high Arctic. Appl. Environ. Microbiol. 73, 1532–1543. doi:10.1128/AEM.01729- 06.

Perreault, N. N., Greer, C. W., Andersen, D. T., Tille, S., Lacrampe-Couloume, G., Lollar, B. S., et al. (2008). Heterotrophic and autotrophic microbial populations in cold perennial springs of the high arctic. Appl. Environ. Microbiol. 74, 6898–6907. doi:10.1128/AEM.00359-08.

Poniecka, E. A., Bagshaw, E. A., Tranter, M., Sass, H., Williamson, C. J., Anesio, A. M., et al. (2017). Rapid development of anoxic niches in supraglacial ecosystems. Arctic, Antarct. Alp. Res. 50, 14. doi:10.1080/15230430.2017.1420859.

Purcell, A. M., Mikucki, J., Alekhina, I., Achberger, A. M., Barbante, C., Brent, C., et al. (2014). Microbial sulfur transformations in sediments from Subglacial Lake Whillans. Front. Microbiol. 5, 1–28. doi:10.3389/fmicb.2014.00594.

Richter, M., Rosselló-Móra, R., Oliver Glöckner, F., and Peplies, J. (2016). JSpeciesWS: a web server for prokaryotic species circumscription based on pairwise genome comparison. Bioinformatics 32, 929–931. doi:10.1093/bioinformatics/btv681.

Spear, J. R., Figueroa, L. A., and Honeyman, B. D. (1999). Modeling the Removal of Uranium U(VI) from Aqueous Solutions in the Presence of Sulfate Reducing Bacteria. Environ. Sci. Technol. 33, 2667–2675. doi:10.1021/es981241y.

Spear, J. R., Figueroa, L. A., and Honeyman, B. D. (2000). Modeling reduction of uranium U(VI) under variable sulfate concentrations by sulfate-reducing bacteria. Appl. Environ. Microbiol. 66, 3711–21. doi:10.1128/AEM.66.9.3711-3721.2000.

Tatusov, R. L., Galperin, M. Y., Natale, D. A., and Koonin, E. V (2000). The COG database: a tool for genome-scale analysis of protein functions and evolution. Nucleic Acids Res. 28, 33–6. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10592175 [Accessed March 27, 2018].

Thullner, M., Dale, A. W., and Regnier, P. (2009). Global-scale quantification of mineralization pathways in marine sediments: A reaction-transport modeling approach. Geochemistry,

98 Geophys. Geosystems 10, n/a-n/a. doi:10.1029/2009GC002484.

Trivedi, C. B., Lau, G. E., Grasby, S. E., Templeton, A. S., and Spear, J. R. (2018). Low- temperature sulfidic-ice microbial communities, Borup Fiord Pass, Canadian high Arctic. Front. Microbiol. 9, 1622. doi:10.3389/FMICB.2018.01622.

Wright, K. E., Williamson, C., Grasby, S. E., Spear, J. R., and Templeton, A. S. (2013). Metagenomic evidence for sulfur lithotrophy by epsilonproteobacteria as the major energy source for primary productivity in a sub-aerial arctic glacial deposit, Borup Fiord Pass. Front. Microbiol. 4, 1–20. doi:10.3389/fmicb.2013.00063.

Zhang, J., Kobert, K., Flouri, T., and Stamatakis, A. (2014). PEAR: A fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics 30, 614–620. doi:10.1093/bioinformatics/btt593.

99

CHAPTER 4

CHARACTERIZATION OF FOUR MICROBIAL MATS ABOVE 80° N

IN THE CANADIAN HIGH ARCTIC

A paper in preparation for submission to Frontiers in Microbiology

Abstract

Photosynthetic microbial mats are found across the world. Microorganisms from all three domains of life interact in consortia within microbial mats, often under extreme conditions. Mats are found in extreme environments such as hydrothermal systems, hypersaline environments oligotrophic lakes, and across cold polar regions. While microbial mats in Antarctica are well known and characterized, less is known about mats from the far north Arctic regions. This study characterized and compared microbial mat samples gathered from four sites in the Canadian

High Arctic: Cape Evans, Derberville, Ice River, and Moss Valley, all of which are above 80° N on Ellesmere Island, Nunavut. 16S rRNA gene sequencing data revealed that sites significantly differed in their microbial community composition. Abundances of microbial genera varied between sites, namely the Bacteriodetes genus Ferrginibacter in Cape Evans and Derberville, high abundances of Flavobacterium in Moss Valley and Ice River, and large differences in abundances of operational taxonomic units (OTUs) most closely related to chloroplast sequences between all sites. Mat samples from Ice River in 2014 and 2017 were queried via metagenomics and revealed some potential for sulfur oxidation and nitrate/nitrite reduction. The dominant carbon cycling mechanisms among metagenome assembled genomes (MAGs) were related to the

100 reductive acetyl-CoA cycle and the reverse TCA cycle. Also found were genes that allow microorganisms to adapt to adverse conditions in extreme environments. Microbial mats appear to provide primary carbon and nitrogen fixation for later ecological succession in the north and appear to be the ‘first responders’ of ecosystem change. This study expands our knowledge of microbial mats in low-temperature polar environments and describes how they can be barometers of environmental change.

4.1 Introduction

Microbial mats are found in extreme environments around the world such as geothermal hot springs in the United States (Bradley et al., 2017; Pepe-Ranney et al., 2012; Reyes et al.,

2013); Iceland (Aguilera et al., 2010); Mexico (Prieto-Barajas et al., 2018); and hypersaline environments such as Guerrero Negro, Baja California Sur, Mexico (Feazel et al., 2008; Harris et al., 2013; Kunin et al., 2008; Ley et al., 2006; Robertson et al., 2009; Spear et al., 2003). For low-temperature polar regions microbial mats have been described in Antarctica (Andriuzzi et al., 2018; Jungblut et al., 2016; Varin et al., 2010, 2012) and Axel Heiberg Island (Sapers et al.,

2017) and Ellesmere Island (Mueller et al., 2005; Varin et al., 2010, 2012) in the Canadian

Arctic. While an abundance of information highlights their formation, biogeography, and microbiological structure, little is known about their presence in extreme polar locales, such as their continued ability to adapt to extreme environments. Furthermore, for microbial mats that have been discovered in low-temperature polar environments, little is known about their ability to adapt to changes in solar conditions (summer vs. winter) and how their presence and migration in high latitudes might signal continued changes on our planet due to climate change effects.

The presence of layered microbial mats at high latitudes may indicate a succession of

101 microbial communities due to climate change via a shift in Earth’s natural cycles, possibly due to anthropogenic influence. Primary production is an ecological concept where organisms are able to synthesize organic compounds by carbon fixation via chemolithotrophy and/or photosynthesis.

As such, the oxygen byproduct of microbial photosynthesis led to the Great Oxidation Event

(GOE) that brought oxygen to the atmosphere of the Earth and allows for aerobic life today

(Buick, 2008; Crowe et al., 2013; Holland, 2006; Planavsky et al., 2014). While plants and higher-order fauna are responsible for primary production in terrestrial landscapes, microorganisms are responsible for this process in the world’s oceans (Kirchman, 2012).

Additionally, in extreme environments, such as those at high latitudes, microorganisms are able to flourish where plants cannot, including on and in rocks (endolith; de la Torre et al., 2003;

Walker and Pace, 2007). The presence of microbial mats at high latitudes could indicate an successional role towards higher trophic levels resulting from an overall increase in polar temperatures and glacial retreat (Garcia-Lopez and Cid, 2017). If the flow of energy continues towards higher trophic levels then the emergence of terrestrial microbial mats could be an important indicator of climate change in Arctic regions.

We present here a microbiological characterization of glacier-associated microbial mats obtained for four sites (Cape Evans, Derberville, Ice River, and Moss Valley) in northern areas of Ellesmere Island in the Canadian High Arctic. To our knowledge this is one of only a few surveys of low-temperature, microbial mats at latitudes above 80° N. Previous studies looked at microbial mats associated with Arctic ice shelves (Mueller et al., 2005; Varin et al., 2010, 2012), whereas the microbial mats surveyed in this study are glacial melt/spring-associated. We found that even in similar climates and site types, and separated by smaller distances (~50 km); microbial community composition is highly variable. 16S rRNA gene sequencing data reveals

102 that community diversity, while similar across some sites, is significantly different in terms of relative abundances of microorganisms inhabiting these locations, particularly in terms of the phyla Bacteroidetes and Cyanobacteria. Furthermore, shotgun whole genome sequencing of Ice

River samples reveals fewer genes associated with chemotrophy and more genes associated with diverse carbon cycling capabilities. A number of stress genes for cold shock proteins, osmotic stress, and universal stress proteins, were found to be abundant across mapped metagenome assembled genomes (MAGs). This characterization of High Arctic microbial mats has expanded our knowledge of layered microbial communities in an extreme low-temperature environment; however, further geochemical and metagenomic/metatranscriptomic analysis of these sites would allow us to tease apart the major differences in phylogeny and what is driving community shifts in these extreme locations. Moreover, microbial succession to higher trophic levels, via the presence of microbial mats in these high latitudes, may be an indicator of continued climate change in the Arctic.

4.2 Materials and Methods

4.2.1 Site Description and Sample Collection

Mat samples were collected from four different Arctic sites named by the field research team (Cape Evans, Derberville, Ice River, and Moss Valley) during the 2017 Arctic field season

(Figure 4.1). Two samples from Ice River were also collected during the 2014 Arctic field season. The exact location of each site is available in Table 4.1. Mat sites contained a variety of different sample types including: green mat, brown mat, and in dry or wet variations. More description of each sample can be found in Table 4.2 (see page 106). Representative pictures of

Ice River and Moss Valley can be seen in Figure 4.1c-e. Microbial mat samples from 2014 were

103 collected and field processed as found in Trivedi et al. (Accepted). Mat samples from the 2017 field campaign were collected using sterile spatulas and approximately 0.5 g of material were mixed in ZR BashingBead™ (Zymo Research Corp., Irvine, CA, USA) lysis tubes containing

750 ml of DNA/RNA Shield™ (Zymo Research Corp.). Samples were shaken and kept at 4 °C until returned to the laboratory, where they were then stored at -80 °C until DNA/RNA extraction could be performed. Extra sample material was collected into 15 mL polypropylene tubes, covered in 2-3 mL of DNA/RNA Shield™ (Zymo Research Corp.) and frozen on ice until transported back to the lab and stored at -80 °C.

Table 4.1. Location data of microbial mat sites. Latitude and longitude coordinates are provided in decimal format.

Site Latitude Longitude Cape Evans 82.6226 -82.3400 Derberville 80.6072 -79.5905 Ice River 81.1880 -86.7466 Moss Valley 81.8622 -82.5263

4.2.2 DNA extraction and 16S rRNA gene library preparation

DNA extraction of 2014 Ice River samples (1A and 3.1; Table 4.2) was performed using

Xpedition™ Soil/Fecal DNA MiniPrep kit (Zymo Research Corp.) and DNA extraction of all

2017 samples was performed using the ZymoBIOMICS™ DNA/RNA Mini Kit (Zymo Research

Corp.). All extractions were performed according to the manufacturer’s instructions. Extracted

DNA was amplified via PCR using 16S rRNA V4 primers 515F_Y and 926R (Parada et al.,

2016). The Forward primer (M13-515_Y: 5'- GTA AAA CGA CGG CCA GTC CGT GYC

AGC MGC CGC GGT AA-3') contains the M13 forward primer (bold) ligated to the 16S rRNA

104

Figure 4.1. Microbial mat locations and sites. A) Line drawing of Canada (inset) with a zoom in of Ellesmere Island, Nunavut. B) Locations are plotted on a composite image (IBCAO and Landsat/Copernicus) obtained from Google Earth and marked with orange dots. Cape Evans is the northern most site, with Derberville being the southern-most. Examples of sample sites for Ice River from 2014 (C) and 2017 (D) are shown. Additionally, the Moss Valley site location can be seen in E. Special thanks to © OpenStreetMap contributors for access to the line drawing map.

105

Table 4.2. SSU rRNA Gene Library Summary. This table contains visual description of each sample as well as total sequences obtained post QC, and the year of sampling.

Sample Site Description Sequences Year 32A Cape Evans Mat on till - dry 1969 2017 32B Cape Evans Mat on till - dry 2354 2017 33A Cape Evans Mat on till - stream 4089 2017 33B Cape Evans Mat on till - stream 793 2017 34A Cape Evans Leather-like mat 1639 2017 34B Cape Evans Leather-like mat 2576 2017 38A Derberville Leather-like mat 1000 2017 38B Derberville Leather-like mat 1938 2017 11A Ice River Top mat - green/brown 6651 2017 11B Ice River Top mat - green/brown 5584 2017 11C Ice River Top mat - green/brown 9240 2017 12A Ice River Lower mat - leather-like 2182 2017 12B Ice River Lower mat - leather-like 9130 2017 13A Ice River Brown mat - gelatinous 8110 2017 13B Ice River Brown mat - gelatinous 10749 2017 13C Ice River Brown mat - gelatinous 8461 2017 26B Ice River Matt plus moss 1022 2017 1A Ice River Lower mat - in stream 26821 2014 3.1 Ice River Upper mat - in stream 19283 2014 16A Moss Valley Lower mat - brown 12300 2017 16B Moss Valley Lower mat - brown 10611 2017 18A Moss Valley Brown mat 10232 2017 18B Moss Valley Brown mat 14387 2017

106 gene-specific sequence (underlined) to allow for barcoding in a subsequent PCR reaction

(Stamps et al., 2016). The reverse primer (926R: 5’-CCG YCA ATT YMT TTR AGT TT-3’) was described in Parada et al. (2016). Ice River and Moss Valley samples were amplified using

1X 5PRIME HOT MasterMix (Quanta BioSciences Co., Beverly, MA, USA), while Cape Evans and Derberville samples were amplified using 1X ZymoBIOMICS™ PCR PreMix (Zymo

Research Corp.). PCR cycling conditions can be found in Table C.1. The final pooled library was run on the Illumina MiSeq platform (Illumina, San Diego, CA, USA) at the Duke Center for

Genomic and Computational Biology (Duke University, Durham, NC, USA) using PE250 V2 chemistry.

4.2.3 16S rRNA gene sequencing analysis

After sequencing, reads were merged and de-multiplexed using QIIME (Caporaso et al.,

2010b), filtered by quality, clustered into operational taxonomic units (OTUs), and chimera checked using VSEARCH (Rognes et al., 2016). OTU taxonomy was assigned using UCLUST

(Edgar, 2010) and the SILVA database (Release 123; Pruesse et al. 2007). Representative sequences were aligned using pyNAST (Caporaso et al., 2010a) against an aligned version of the

SILVA r123 database. A mapping file is included as file C.S1 (Appendix E), and the commands used to produce the final BIOM file are publicly available at https://doi.org/10.5281/zenodo.1302423.

SSU rRNA gene sequencing analyses were performed in R (R Core Team, 2016). Samples were imported into Phyloseq v1.23.1 (McMurdie and Holmes, 2013) where alpha diversity, ordination via non-metric multidimensional scaling (NMDS), principle coordinates analysis

(PCoA), and statistical analyses (ADONIS) were performed. The Ampvis2 package v2.3.7

107 (Albertsen et al., 2015) was used to create heat maps, and the stacked bar charts were created using ggplot2 (Wickham, 2016) and the R package Fantaxtic (Teunisse, 2018). All R code is contained within an R markdown notebook which can be accessed at https://doi.org/10.5281/zenodo.1302423.

4.2.4 Metagenomic sequencing and analysis

Processing of metagenomic sequencing data was performed on the Summit High

Performance Cluster (HPC) located at the University of Colorado, Boulder (Boulder, CO, USA).

We began by using BBDuk of the BBMap/BBTools (v37.56; Bushnell, 2014) suite to trim out

Illumina adapters using the built-in BBMap adapters database. Trimmed sequences were checked for quality using FastQC v11.5 (Andrews, 2014) to ensure that adapters were removed successfully. Trimmed paired-end sequences were then joined using the PEAR package (Zhang et al., 2014) v0.9.10 using a p-value of 0.001. Adapter trimmed files, both forward (R1) and reverse (R2) were then processed using BBTools ‘repair.sh’ to ensure read pairs were in the correct order. This script is designed to reorder paired reads that have become disorganized, which can often occur during trimming. The BFC (Bloom Filter Correction) software tool was then used to error correct paired sequence reads (Li, 2015). Reads were again processed with

BBTools repair.sh to ensure proper read pairing and order. Next, Megahit v1.1.2 (Li et al., 2015) was used to generate both a co-assembly, as well as individual sample assemblies from the quality controlled reads. Post assembly, co-assembly was processed with ‘anvi-script-reformat- fasta’ for input into Anvi’o, and contigs below 2500 bp were removed. Bowtie2 v2.3.0

(Langmead and Salzberg, 2012) was then used to map individual sample sequence reads onto the co-assembly, needed for the Anvi’o pipeline. Anvi’o v4 (Eren et al., 2015) was then used to

108 characterize, and then bin co-assembled contigs into metagenome assembled genomes (MAGs).

The taxonomic classifier Centrifuge v1.0.2-beta (Kim et al., 2016) was used to provide a preliminary taxonomy of the co-assembly contigs, useful for Anvi’o binning. The NCBI Clusters of Orthologous Groups (COGs; (Galperin et al., 2015; Tatusov et al., 2000)) database was used within Anvi’o to predict COGs present within the co-assembly, and CONCOCT (Alneberg et al.,

2014) was employed to create genome bins using a combination of sequence coverage and composition of our assembled contigs. Anvi’o was then used to visualize the putative bins in conjunction with other metadata to refine and identify MAGs with 50% completeness and 10% redundancy. The scripts used in the processing of these data (up to the point of Anvi’o contigs database) are publicly available at https://doi.org/10.5281/zenodo.1302423.

Anvi’o refined bins were then processed using CheckM v1.1.11 (Parks et al., 2015) where a total of 14 bins were selected as MAGs (Metagenome Assembled Genomes; Table C.3) which followed the > 50% completeness and < 10% redundancy criteria. CheckM relies on other software including pplacer v1.1.alpha17 (Matsen et al., 2010) for phylogenetic tree placement, prodigal v2.6.3 (Hyatt et al., 2012) for gene translation and initiation site prediction, and

HMMER v3.1b2 (Eddy, 2011) which analyzes sequence data using hidden Markov models.

Genes of interest to this study were identified and queried via the bioinformatics tool Prokka v1.13 (Seemann, 2014). The scripts used in the processing of these data (up to the point of

Anvi’o contigs database) are publicly available at https://doi.org/10.5281/zenodo.1302423.

4.3 Results

4.3.1 Mat microbial communities contain distinct assemblages

16S rRNA gene sequencing revealed that mat samples between the four different sites are

109 diverse and microbial communities are different from one another. This is evidenced by significant clustering (Table C.2) of site samples via NMDS analysis (Figure 4.2). Furthermore, when the top 25 OTUs were queried by site, these trends were further observed across varying taxonomic abundances (Figure C.2).

NMDS of sample sites revealed significant clustering between all compared sites except for Derberville and Moss Valley (see Table C.2). R-squared values are all below 0.5 in terms of strength of the ADONIS test run against the Bray-Curtis distance matrix. These statistical results corroborate with the clustering shown in the NMDS ordination (Figure 4.2). Alpha diversity of sites shows similar amounts of OTUs per site except for Derberville which had the fewest OTUs of the four sites (Figure C.1). A heat map of the top 12 Phyla (Figure C.2) indicates that all four sites are inhabited generally by the same Phylum-level organisms (e.g. Cyanobacteria,

Proteobacteria, and Bacteroidetes). This is echoed as well in Figure 4.3 where the top 100 OTUs were queried and then agglomerated on the OTU level and then broken out by individual sample type. The further breakdown of each site reveals how microbial communities across site types are highly variable on lower taxonomic levels.

4.3.2 Cape Evans

Cape Evans, near the top of Ellesmere Island, is a site with a retreating glacier that used to flow into the Arctic Ocean that has now left bare several hundred meters of predominately rocky cobble outcrop with some soil and few plants. Surface waters run off /under the glacier toward the ocean to the north. Samples were taken in duplicate from three locations at the site. The most abundant taxa of Cape Evans samples are from the Phylum Proteobacteria, on average comprising 38.6% from the top 12 Phyla represented. Next most abundant are the Bacteroidetes

110

Figure 4.2. NMDS of Arctic mat sites. Sites cluster independently aside from some outlying samples from Ice River and Moss Valley. These results are further supported by community diversity 16S rRNA gene sequencing data and ADONIS tests.

at 23.3%, followed by the Phyla Cyanobacteria at 15.4%. Also present among the top OTUs on the Phylum level are Planctomycetes, Verrucomicrobia, Acidobacteria, and Actinobacteria.

These Phyla are all present in greater than 2% relative abundance (see Figure C.2). The

Cyanobacterial overall average in Figure C.2 is higher due to samples 32A and 32B (see Figure

4.3), which have close to five times more Cyanobacterial OTUs represented than samples 33A/B and 34A/B. The reason for this, by taxonomic classification, is that breaking down the top 15

OTUs by genus level, site 32 has greater than 16% relative abundance of an OTU most closely

111

Figure 4.3. SSU rRNA Gene Bar Chart Organized by Location. Bars are representative of the top 100 OTUs agglomerated by Phyla within each sample and grouped by site. Any OTUs that were not part of the top 100 are grouped into the gray “Other” category. Bars are normalized to percent relative abundance.

related to the Family I Family of Cyanobacteria (Figure 4.4), while sites 33 and 34 both have below 2% relative abundance for that OTU. All sites contained between 4.4 and 12.2% of an

OTU most closely related the genus Ferruginibacter within the Phylum Bacteroidetes.

Remaining top taxonomy belonged to an OTU most closely related to Sphingomonas (4.1-6.5%),

Polaromonas (1.3-4.2%), and two Proteobacterial OTUs most closely related to the Family

Comamonadaceae (2-4.2% & 2.4-6.2%), which also houses the genus Polaromonas. It should be

112 noted that of the three sample sets only site 32 has those top 100 overall OTUs (Figure 4.3) expressed above 50% of the total relative abundance of the sample. Site 34 has over 50% of relative abundance located in “Other” OTUs.

Figure 4.4. Top 15 OTUs by Site broken down by Phylum, Class, and Genus. The top 15 OTUs across all samples as expressed in a heat map. Phylum, Class, and Genus are shown from right to left.

113 4.3.3 Derberville

Derberville at the southern end of our study area, though still in northern Ellesmere, has mats represented by one site with duplicate samples. The mats were green/brown, had a leather- like consistency and were under 10 cm of cold, flowing water emanating from the summer seasonal melt of an upstream glacier. At the Phylum level the most abundant OTUs are almost exclusively within the Cyanobacteria (43.8%), Proteobacteria (20.3), and Bacteroidetes (25.3%).

The only other Phylum level OTU above 2% is Verrucomicrobia at 2.9% (Figure C.2). Much like Cape Evans site 32, the Derberville replicates (38A/B) have greater than 9% relative abundance of the OTU most closely related to the Family Family I (Figure 4.4). The highest

Bacteroidetes OTU represented within the Derberville samples belongs to Ferruginibacter at

12.2% and 11.1% (38A and 38B, respectively). This is similar to relative abundances found in

Cape Evans Mat samples. The highest relative abundance OTU of the top 15 queried was most closely related to the Cyanobacteria genus Phormidium at 20.4% for sample 38A and 14.3% for

38B. The majority of other top OTUs were all below 3% except for the Bacteroidetes genus

Flavobacterium (2.8 and 4.4%), and another Cyanobacteria genus Leptolyngbya at 3.1% and

4.3% (38A and 38B, respectively; Figure 4.4).

4.3.4 Ice River

Ice River is thought to be a perennially flowing spring, possibly the highest such spring in the Arctic, but has been difficult to fully characterize due to its location and lack of winter access. The source-spring for the river environment comes from a few sources, and as the river flows and further cools, it acquires more and more ice on its surface that can be colored with accumulating iron oxides of unknown origin (i.e., they could be products of geobiological

114 microbial activity and/or surface deposition from upwind sources). Ice River contains the most extensive set of samples taken for this study including two that were taken during the summer of

2014 (1A and 3.1) and eight in 2017. Two sets of samples are in triplicate (11A-C and 13A-C), one is in duplicate (12A/B), and the remaining are single samples (26B, 1A, and 3.1). According to the bar charts created from DNA sequence analysis of the top 100 OTUs across all samples

(Figure 4.3), Ice River microbial communities are similar to those found at the other site locations, with the majority of OTUs belonging to Cyanobacteria, Bacteroidetes, and

Proteobacteria. This holds true for all samples except the duplicate samples of site 12 which are different in that they have very few Cyanobacterial OTUs and small relative abundance increases of Acidobacteria and Plancomycetes.

The majority of Ice River sites (11, 13, 1A, and 3.1) have high relative abundances associated with OTUs most closely related to the Class chloroplast. The first OTU is associated with relative abundances between 15.7% (11A) and 59.3% (3.1), while the second OTU spans between 14.7% (11A) and 33.4% (3.1). With the exception of site 1A which has 5.9% relative abundance of an OTU most closely related to Leptolyngbya, and site 26B which has 8.7% relative abundance of the OTU related to the family Family I, the majority of Cyanobacterial

OTUs for Ice River samples relate to Chloroplast sequences, which are most likely eukaryotic sequences. All samples except for sites 12 and 3.1 show at least 3% relative abundance with an

OTU most closely related to the genus Flavobacterium, with sites 26B and 1A having the highest relative abundance of the genus (17.3 and 14.4%, respectively). Site 12 also contains approximately 12% of an Alphaproteobacteria OTU most closely related to the Family A0839.

115 4.3.5 Moss Valley

In the middle of an Arctic desert on Ellesmere Island lies a large outcrop with a dominant south to north notch running through it with a source spring on the upper, south side, creating a series of pools that flow north characterized by various mosses, small flowering plants and microbial mats. Termed “Moss Valley” because of its green appearance visible from afar by helicopter, the site presents as a watering hole, an oasis location, for Arctic plants and animals.

Moss Valley has two sampled sites each with one biological replicate (16A/B and 18A/B). These samples are also among the least diverse in terms of the majority of the top 100 OTUs that belong to the top 12 Phyla (Figure 4.3) in this study. This is similar to the 2014 Ice River samples that also have below 25% “Other” in terms of relative abundance. Moss Valley samples are largely dominated by Cyanobacteria and Bacteroidetes, and have overall fewer OTUs associated with Proteobacteria than the other Arctic mat sites. The most abundant OTU among both sites is most closely related to the Family I group of organisms as seen in Cape Evans and

Derberville sites. In Moss Valley samples this OTU is by far the most abundant across all samples and sites (37.8-65.7%). Conversely, OTUs related with Chloroplast sequences, like those seen in many of the Ice River samples are all below 1%. Moss Valley also has the highest relative abundance of the OTU most closely related to Flavobacterium (between 16-29.3%).

Also above 3% across all replicates is the OTU most closely related to the Proteobacteria genus

Polaromonas. Of note are the Bacteroidetes genera Ferruginibacter (>3.3% at site 18B) and

Spirosoma (>7.6% at site 16). See Figure 4.4.

4.3.6 Ice River metagenomes

Assembly and binning of Ice River metagenomes (from three samples: 1A, 11B, and 13A)

116 resulted in 64 bins produced by CONCOCT prior to hand refinement. Bin refinement as well as further quality control through CheckM produced 14 reportable (>50% completion & <10% redundancy) metagenome assembled genomes (MAGs) (Figure 4.5). MAG length ranged from

1.80 Mbp (MAG 16) to 6.06 Mbp (MAG 2). A summary of MAG statistics is available in supplemental table C.3. The majority of mapped reads for a given MAG were recruited from the

2014 Ice River sample (Sample 1A, Figure 4.5). This is most likely due to a greater number of overall sequencing being associated with sample 1A (26,821) than 11B and 13A (5,584 and

8,110, respectively; Table 4.2). All MAGs were queried for a variety of functional genes including those for sulfur, nitrogen, and carbon cycling, as well as those found in various stress responses. Genes responsible for sulfur cycling, flavocytochrome c (fcc) and sulfide-quinone reductase (sqr) were only present in 5 of 14 reported MAGs (MAG 9, 4, and 12 & 6 and 18, respectively). Only the gene that codes for subunit A of the SoxAX operon was identified in

MAGs 9, 4, and 12. Sulfate adenylyl transferase (sat) was present in all but four (MAG 3, 7, 4, and 16) genomes. Adenylyl sulfate reductase (apr) was not observed in any MAG.

Nitrogen cycling genes were queried including those for nitrate reduction (nap and nar), nitrite reduction (nrf and nir), nitric oxide reduction (nor), nitrous oxide reduction (nos), and the fixation of atmospheric nitrogen (nif). There were no genes present for nap; however, MAGs 1,

5, 7, 9, 4, and 12 had genes for nar. Nrf genes were also absent from all MAGs, with only MAGs

7 and 4 having genes present for nir. Only MAG 12 had a gene present for nitric oxide reductase

(nor), and only MAG 6 had a gene present for nitrous oxide reductase (nos). There were no genes present for nitrogen fixation. Carbon cycling genes queried included those responsible for photosynthesis (Photosystem II, (PSII)), the reductive acetyl-CoA cycle (rCoA), the reverse tricarboxylic acid cycle (rTCA), RuBisCo (cbb), as well as genes for fermentation, Acetyl-CoA

117

fcc - flavocytochrome c nif - nitrogen fixation sqr - sulfide quinone reductase PSII- photosystem II sox - sulfur oxidation rCoA- reductive acetyl-CoA sat - sulfate adenylyl transferase rTCA- reverse tricarboxylic acid apr - adenylyl sulfate reductase cbb - RuBisCo nap - periplasmic nitrate reductase csp - cold shock protein nar - nitrate reductase ots - trehalose-6-phosphate nrf - nitrite reductase usp - universal stress nir - nitrite reductase nor - nitric oxide reductase nos - nitric oxide synthase

Figure 4.5. Summary of Ice River metagenomes. Summary statistics of metagenome assembled genomes (MAGs). Total MAG length as well as completeness and redundancy percentages are shown. Mean coverage by sample is a heat map indicating which samples contributed the most reads to a given MAG (dark blue is the highest, with grey shades being lower). Queried genes are shown as a presence (black) absence plot (white).

118 carboxylase (acc) and Propionyl-CoA carboxylase (pcc). All queried PSII genes were found associated with MAG 1 via annotation with the Prodigal package. This MAG was putatively identified within the phylum Cyanobacteria. Reductive CoA and rTCA were both well represented across all MAGs and only absent in 3 (rCoA) and 2 (rTCA) genomes, respectively

(Figure 4.5). Surprisingly, RuBisCo genes were only present in MAGs 1 and 4, but because they are MAGs, not all genes that may be in a genome are necessarily in the sequence coverage zone of the assembled MAG.

Lastly, stress response genes were queried for the Ice River metagenomes including changes in temperature (csp), osmotic pressure/oxygen (ots), and a general stress response (usp) gene. All but four MAGs (1, 9, 12, and 18) showed the presence of the cold shock protein (csp).

Seven MAGs had the Trehalose-6-phosphate (ots) gene present (Figure 4.5). Three MAGs (8, 3, and 7) had the universal stress gene (usp) that was queried for present within their genomes.

4.4 Discussion

The presence of terrestrial layered microbial mats in extreme locales is interesting due to their potential to adapt to adverse conditions. Moreover, finding such structures in high latitudes may be able to act as an indicator for primary production and climate change effects. As the

Earth anthropogenically warms, low-latitude forests have the potential for desertification; higher latitudes have tree-line migration as has been described for the North American Arctic treeline

(Pearson et al., 2013); and places where there was heavy glaciation, northern Ellesmere, upon melt will have exposed rock where microbes will build mats and soils to then become covered in small plants to produce tundra followed by ecological succession. Beyond this, what can these mats tell us about microbial community structure in low-temperature systems where full light is

119 only available during half of the year? This study set out to characterize microbial mats that, to our knowledge, are some of the most northern ever identified and may indicate how climate change is playing a role in warming our polar environments and allowing for the growth of microbial assemblages not previously able to proliferate in these ecosystems.

Based on the hundreds of kilometer-scale distances these mats are separated by, we expected differences in community composition despite the shared environment of low temperature, a no-light to high-light seasonal environment and high latitude. Of the four sites studied, Cape Evans and Derberville are believed to receive fluid flow predominantly from glacial melt, though groundwater and spring influences are also possible, while Ice River and

Moss Valley are most likely spring-fed. Communities were significantly different overall by an

ADONIS test, yet Derberville and Moss Valley were more similar to one another than Ice River and Cape Evans, which contained distinct microbial communities.

Of all the sample sites, Cape Evans was the furthest north (Figure 4.1). The sample site was costal with glacial melt as its primary water source. Correspondingly, some of the more abundant taxa identified were most closely related to the Betaproteobacterial genus

Polaromonas. This often heterotrophic genus (Hamilton and Havig, 2017) is associated with glacial environments (Boetius et al., 2015; Gleeson et al., 2011; Raymond-Bouchard et al.,

2018), thus their presence is not unexpected. One of the most abundant OTUs at Cape Evans

(higher in samples 32 and 33; Figure 4.4) was identified as being most closely related to the

Bacteroidetes genus Ferruginibacter. This genus was previously isolated from freshwater lake sediments in South Korea (Lim et al., 2009), and has also been shown to live in association with snow algae (Hamilton and Havig, 2017). Not all samples were identical at Cape Evans, and some microbial mats with a “brown elephant skin” texture had greater relative abundances of

120 OTUs related to chloroplast sequences. Other differences in samples (notably 32 and 33), are likely due to location, where sample 32 was a mat growing atop unsubmerged glacial till, whereas sample 33 was submerged.

Interestingly, two sites, Derberville and Moss Valley, were similar in community membership despite being several hundred kilometers apart (Figure 4.1). The Derberville site is the furthest south of the four sites, and is only characterized by one set of samples with duplicates (site 38); however, it is distinct from all other mat sites. This mat was described as leathery with likely photosynthetically produced bubbles on top. While it does share some similarities to Cape Evans samples, especially with abundant OTUs most closely related to the class Family I as well as Ferruginibacter, its highest relative abundant OTU is most closely related to the Cyanobacterial genus Phormidum. In all other mat samples this OTU is represented less than 3.2% (Figure 4.4). This OTU is ubiquitous, and has been found in sites across the globe including Antarctica (Jungblut et al., 2016), the Arctic (Bottos et al., 2008; Bradley et al., 2016), endolith in the United States Rocky Mountains (Walker and Pace, 2007), and even as part of dendrolitic microbialite cones in hot springs (Bradley et al., 2017). Phormidium sp. (along with

Leptolyngbya, also present in Derberville samples) are important “engineer” Cyanobacteria in glacial cryoconite hole development, releasing extracellular polymeric substances (EPS), which can assist in greater community survival at the bottom of these surface-glacial water pockets in harsh polar environments (Anesio et al., 2017).

Moss Valley is the second northern most site behind Cape Evans. It is characterized as a brown microbial mat, and the most dominant organisms present are most closely related to OTUs matching the Family I class within the Cyanobacteria (Figure 4.4). This was also observed in both Derberville samples as well as sample 32 of Cape Evans, however, not to such a great

121 extent (over 37% in all Moss Valley samples). The Bacteroidetes genus Flavobacterium is also well represented in these samples, all above 16% relative abundance. Previously, Varin et al.

(2012b) observed Flavobacterium sp. within microbial mat metagenomes from polar ecosystems on the northern coast of Ellesmere Island and the McMurdo Ice Shelf in Antarctica. It is a ubiquitous organism that has been found in Antarctic surface snow (Lopatina et al., 2016), freshwater sediments (Tamaki et al., 2003), and a High Arctic sulfur-dominated glacial system

(Gleeson et al., 2011; Trivedi et al., In review; Wright et al., 2013). The genus Polaromonas is also present in abundances below 5%, which is similar to results found at Cape Evans. Lastly, in sample 16 from Moss Valley an OTU mostly closely related to the Bacteroidetes genus

Spirosoma is observed above 7.6% (Figure 4.4). Tytgat et al. (2014) found that when looking at bacterial diversity of Antarctic terrestrial vs aquatic microbial mats it was found in much higher abundance in the terrestrial mats. This is similar to Moss Valley samples where mats were located on top of tundra rather than submerged under water, though in zones of water saturation.

By opportunity, Ice River was the most sampled site in 2017 (four samples; 11, 12, 13, and

26), and two from 2014 (1A and 3.1) with metagenomes obtained from three of these samples

(1A from 2014, and 11B and 13A from 2017). Samples 11 and 12 are from the same location; however, represent different portions of the mat. Sample 11 (with replicates A, B, and C) are from the top, green part of the mat, whereas Sample 12 (with replicates A and B) are from the bottom, brown part of the mat. Dissimilarity is evidenced in the mat top OTUs as seen in observed relative abundance (Figure 4.4). Sample 11 is largely dominated by OTUs most closely related to chloroplast sequences, likely due to eukaryotic organisms like algae and/or diatoms; however 18S rRNA genes were not queried in this study for further follow-up. Sample 12 has greater than 12% relative abundance of an OTU most closely related to the Alphaproteobacterial

122 family A0839 represented in both sample replicates. This genus is part of the Order Rhizobiales, which Preisner et al. (2016) found present in microbial mats in conjunction with photosynthetic organisms from samples on San Salvador Island, The Bahamas. Sample 13 (with replicates A, B, and C) was described as “brown and gelatinous”; however, in community composition highly resembles sample 11 rather than the “brown” sample 12. Sample 26B is a “Matt + Moss” sample that, in microbial community composition, highly resembles Moss Valley samples (Figure 4.4).

Samples from 2014 (1A and 3.1) are very similar to samples 11 and 13 from 2017 in their high relative abundance of OTUs related to Chloroplast sequences. Also of note is the higher relative abundance of OTUs most closely related to Flavobacterium (14.4%) and Leptolyngbya (5.7%) in sample 1A as compared to sample 3.1 (1.8% and 0%, respectively).

The near-total lack of sunlight during winter months at Ice River, and the possibility of dry, or non-flowing conditions as well as a potential lack of a clear organic carbon source at each sample location suggests that the microbial mats undergo periods of intense stress and/or possible dormancy. To answer these questions and to better understand the metabolic function of such far-northern microbial mats metagenomic sequencing was carried out at Ice River. A single

MAG (MAG 1) was identified with genes for photosynthesis. Specifically, genes associated with

Photosystem II are highly abundant. This could very well be attributed to the high relative abundance of sequences related to Chloroplast OTUs. Genes representing the reductive acetyl

Coenzyme-A cycle (rCoA) and the reverse tricarboxcylic acid cycle (rTCA) are well represented across all of the MAGs, suggesting that heterotrophy in the system has multiple potential pathways and is likely ubiquitous. RuBisCo genes were only found in two MAGs (1 and 4), corresponding to putative taxonomies of Cyanobacteria and Proteobacteria. Oxidative respiration coupled to partial sulfur oxidation appears common in the sequenced mat samples. MAGs with

123 genes present for the oxidation of reduced sulfur compounds (fcc, sqr, and sox) all belong to the

Proteobacteria (MAGs 9, 4, 6, and 12), with the exception of MAG 18 which is putatively placed in the domain Bacteria. No MAG was found to have a full Sox operon, however. Sulfate adenylyltransferase (sat) was present in all but four MAGs, and there was no evidence of adenylyl sulfate reductase (apr) which typically work together in the oxidation of sulfite in green sulfur bacteria (Dahl and Friedrich, 2008).

A gene responsible for response to cold shock, specifically csp, was present in all but four

MAGs (Figure 4.5). Genes associated with cold shock response are often present in microorganisms in low-temperature systems, and aide in protein folding (Varin et al., 2012).

Another gene associated with cold shock, ots, or trehalose phosphate synthase, is a common stress response gene that encodes for trehalose synthesis, a solute that can aide in lowering the freezing temperature within a cell (Liljeqvist et al., 2015). Trehalose phosphate synthase genes were found in seven of the MAGs and typically associated with the Proteobacteria and

Bacteroidetes phyla (Figure 4.5). Given the environment and this finding of an abundance of the ots gene, it appears that the majority of microbiota in this low temperature location likely survive and adapt to adversity through common genetic adaptations. Further screening of ots variance and other stress-related genes in Ice River mats could likely inform on evolutionary trends for the adaptations to low temperature.

Finally usp, a known universal stress protein gene was identified in three MAGs (8, 3, and

7). This gene was previously identified in Arctic permafrost bacteria and can be activated by multiple environmental stressors (Raymond-Bouchard et al., 2018) such as starvation for key nutrients and temperature exposure (Kvint et al., 2003). The ubiquity of cold-shock associated genes suggests that microorganisms in microbial mats of the far north are well adapted to deal

124 with long periods of intense cold during dark winter months. Nitrogen fixation within the mats appears lacking by our sequencing effort, but this is likely a result of sequence coverage, or lack thereof. Denitrification appears possible with nitrate reductase (nar), nitrite reductase (nir), nitric oxide reductase (nor), and nitrous oxide reductase (nos) all identified indicating possible

+ reduction to N2 gas or even ammonium (NH4 ). Atmospheric nitrogen fixation via nap, nrf, or nif was absent although additional work is required to confirm their absence in the total microbial community. Further sampling and more complete analysis of Ice River mats will likely inform both on the endemic nature of the community as well as how that community is responding to environmental pressures.

Open questions remain for future investigation. Differences in community composition were noted, but additional work is required to determine if this is a physical or temporal difference. Community dynamics at such high latitudes could be extreme as endless sunlight turns to total darkness in the winter. Correspondingly shifts in relative abundance could very likely be due to changes in aqueous geochemistry and flow dynamics. Broadly, the existence of microbial mats at such a high latitude should be a continuing topic of discussion. Is the presence of these mats due to increasing temperatures globally, or are photosynthetic microbial mats more resilient to long periods of stress than previously imagined? Either way, the microbial mats of the

Arctic serve as bellwethers for life in the North and how that life may change due to variance in the cycles of the Earth.

Microbial mats are ubiquitous across the planet, representing evidence of early life on

Earth and oftentimes found at extremes of temperature, pH and salinity to name a few. This study characterized microbial mats found in the Canadian High Arctic at some of the highest spring-associated latitudes ever reported. While all sites were within a few hundred kilometers of

125 each other, we report that individual sites vary significantly based on their 16S rRNA gene sequencing community composition. Furthermore, metagenomic sequencing of one site (Ice

River) revealed a small diversity of carbon metabolisms and some potential for the oxidation of sulfur and the reduction of nitrogen. Queried environmental stress genes were abundant and expected in these low-temperature systems. This work further extends our knowledge of mats in extreme conditions and may hint at how microbial life responds to the changing climate of our world.

4.5 Data availability

After sequencing, raw reads were deposited into the NCBI sequencing read archive (SRA) for 16S rRNA raw data & metagenome raw data (https://www.ncbi.nlm.nih.gov/sra). The mapping file for samples and corresponding barcodes can be found in file C.S1 (Appendix E).

4.6 References

Aguilera, A., Souza-Egipsy, V., González-Toril, E., Rendueles, O., and Amils, R. (2010). Eukaryotic microbial diversity of phototrophic microbial mats in two Icelandic geothermalhot springs. Int. Microbiol. 13, 21–32. doi:10.2436/20.1501.01.109.

Albertsen, M., Karst, S. M., Ziegler, A. S., Kirkegaard, R. H., and Nielsen, P. H. (2015). Back to Basics – The Influence of DNA Extraction and Primer Choice on Phylogenetic Analysis of Activated Sludge Communities. PLoS One 10, e0132783. doi:10.1371/journal.pone.0132783.

Alneberg, J., Bjarnason, B. S., de Bruijn, I., Schirmer, M., Quick, J., Ijaz, U. Z., et al. (2014). Binning metagenomic contigs by coverage and composition. Nat. Methods 11, 1144–1146. doi:10.1038/nmeth.3103.

Andrews, S. (2014). FastQC software package. Available at: https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ [Accessed March 13, 2018].

Andriuzzi, W. S., Stanish, L. F., Simmons, B. L., Jaros, C., Adams, B. J., Wall, D. H., et al. (2018). Spatial and temporal patterns of microbial mats and associated invertebrates along an Antarctic stream. Polar Biol., 1–11. doi:10.1007/s00300-018-2331-4.

126 Anesio, A. M., Lutz, S., Chrismas, N. A. M., and Benning, L. G. (2017). The microbiome of glaciers and ice sheets. npj Biofilms Microbiomes 3, 10. doi:10.1038/s41522-017-0019-0.

Boetius, A., Anesio, A. M., Deming, J. W., Mikucki, J. A., and Rapp, J. Z. (2015). Microbial ecology of the cryosphere: sea ice and glacial habitats. Nat. Rev. Microbiol. 13, 677–690. doi:10.1038/nrmicro3522.

Bottos, E. M., Vincent, W. F., Greer, C. W., and Whyte, L. G. (2008). Prokaryotic diversity of arctic ice shelf microbial mats. Environ. Microbiol. 10, 950–966. doi:10.1111/j.1462- 2920.2007.01516.x.

Bradley, J. A., Arndt, S., Šabacká, M., Benning, L. G., Barker, G. L., Blacker, J. J., et al. (2016). Microbial dynamics in a High Arctic glacier forefield: A combined field, laboratory, and modelling approach. Biogeosciences 13, 5677–5696. doi:10.5194/bg-13-5677-2016.

Bradley, J. A., Daille, L. K., Trivedi, C. B., Bojanowski, C. L., Stamps, B. W., Stevenson, B. S., et al. (2017). Carbonate-rich dendrolitic cones: insights into a modern analog for incipient microbialite formation, Little Hot Creek, Long Valley Caldera, California. npj Biofilms Microbiomes 3, 32. doi:10.1038/s41522-017-0041-2.

Buick, R. (2008). When did oxygenic photosynthesis evolve? Philos. Trans. R. Soc. Lond. B. Biol. Sci. 363, 2731–43. doi:10.1098/rstb.2008.0041.

Bushnell, B. (2014). BBMap software package. Available at: https://sourceforge.net/projects/bbmap/ [Accessed March 13, 2018].

Caporaso, J. G., Bittinger, K., Bushman, F. D., Desantis, T. Z., Andersen, G. L., and Knight, R. (2010a). PyNAST: A flexible tool for aligning sequences to a template alignment. Bioinformatics 26, 266–267. doi:10.1093/bioinformatics/btp636.

Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D., Costello, E. K., et al. (2010b). QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. doi:10.1038/nmeth.f.303.

Crowe, S. A., Døssing, L. N., Beukes, N. J., Bau, M., Kruger, S. J., Frei, R., et al. (2013). Atmospheric oxygenation three billion years ago. Nature 501, 535–538. doi:10.1038/nature12426.

Dahl, C., and Friedrich, C. G. (2008). Microbial Sulfur Metabolism. , eds. C. Dahl and C. G. Friedrich Berlin, Heidelberg: Springer Berlin Heidelberg doi:10.1007/978-3-540-72682-1. de la Torre, J. R., Goebel, B. M., Friedmann, E. I., and Pace, N. R. (2003). Microbial diversity of cryptoendolithic communities from the McMurdo Dry Valleys, Antarctica. Appl. Environ. Microbiol. 69, 3858–67. Available at: http://www.ncbi.nlm.nih.gov/pubmed/12839754 [Accessed June 30, 2018].

Eddy, S. R. (2011). Accelerated Profile HMM Searches. PLoS Comput. Biol. 7, e1002195. doi:10.1371/journal.pcbi.1002195.

127 Edgar, R. C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461. doi:10.1093/bioinformatics/btq461.

Eren, A. M., Esen, Ö. C., Quince, C., Vineis, J. H., Morrison, H. G., Sogin, M. L., et al. (2015). Anvi’o: an advanced analysis and visualization platform for ‘omics data. PeerJ 3, e1319. doi:10.7717/peerj.1319.

Feazel, L. M., Spear, J. R., Berger, A. B., Harris, J. K., Frank, D. N., Ley, R. E., et al. (2008). Eucaryotic diversity in a hypersaline microbial mat. Appl. Environ. Microbiol. 74, 329–32. doi:10.1128/AEM.01448-07.

Galperin, M. Y., Makarova, K. S., Wolf, Y. I., and Koonin, E. V. (2015). Expanded microbial genome coverage and improved protein family annotation in the COG database. Nucleic Acids Res. 43, D261–D269. doi:10.1093/nar/gku1223.

Garcia-Lopez, E., and Cid, C. (2017). “The Role of Microbial Ecology in Glacier Retreat,” in Glaciers Evolution in a Changing World (InTech). doi:10.5772/intechopen.69097.

Gleeson, D. F., Williamson, C., Grasby, S. E., Pappalardo, R. T., Spear, J. R., and Templeton, A. S. (2011). Low temperature S0 biomineralization at a supraglacial spring system in the Canadian high Arctic. Geobiology 9, 360–375. doi:10.1111/j.1472-4669.2011.00283.x.

Hamilton, T. L., and Havig, | J (2017). Primary productivity of snow algae communities on stratovolcanoes of the Pacific Northwest. 15, 280–295. doi:10.1111/gbi.12219.

Harris, J. K., Caporaso, J. G., Walker, J. J., Spear, J. R., Gold, N. J., Robertson, C. E., et al. (2013). Phylogenetic stratigraphy in the Guerrero Negro hypersaline microbial mat. ISME J. 7, 50–60. doi:10.1038/ismej.2012.79.

Holland, H. D. (2006). The oxygenation of the atmosphere and oceans. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 361, 903–15. doi:10.1098/rstb.2006.1838.

Hyatt, D., LoCascio, P. F., Hauser, L. J., and Uberbacher, E. C. (2012). Gene and translation initiation site prediction in metagenomic sequences. Bioinformatics 28, 2223–2230. doi:10.1093/bioinformatics/bts429.

Jungblut, A. D., Hawes, I., Mackey, T. J., Krusor, M., Doran, P. T., Sumner, D. Y., et al. (2016). Microbial Mat Communities along an Oxygen Gradient in a Perennially Ice-Covered Antarctic Lake. Appl. Environ. Microbiol. 82, 620–30. doi:10.1128/AEM.02699-15.

Kim, D., Song, L., Breitwieser, F. P., and Salzberg, S. L. (2016). Centrifuge: rapid and sensitive classification of metagenomic sequences. Genome Res. 26, 1721–1729. doi:10.1101/gr.210641.116.

Kirchman, D. L. (2012). Processes in Microbial Ecology. Oxford University Press Available at: https://books.google.com/books/about/Processes_in_Microbial_Ecology.html?id=U6I5vNc 4g14C [Accessed June 13, 2018].

128 Kunin, V., Raes, J., Harris, J. K., Spear, J. R., Walker, J. J., Ivanova, N., et al. (2008). Millimeter-scale genetic gradients and community-level molecular convergence in a hypersaline microbial mat. Mol. Syst. Biol. 4, 198. doi:10.1038/msb.2008.35.

Kvint, K., Nachin, L., Diez, A., and Nyström, T. (2003). The bacterial universal stress protein: function and regulation. Curr. Opin. Microbiol. 6, 140–145. doi:10.1016/S1369- 5274(03)00025-0.

Langmead, B., and Salzberg, S. L. (2012). Fast gapped-read alignment with Bowtie 2. Nat. Methods 9, 357–359. doi:10.1038/nmeth.1923.

Ley, R. E., Harris, J. K., Wilcox, J., Spear, J. R., Miller, S. R., Bebout, B. M., et al. (2006). Unexpected diversity and complexity of the Guerrero Negro hypersaline microbial mat. Appl. Environ. Microbiol. 72, 3685–95. doi:10.1128/AEM.72.5.3685-3695.2006.

Li, D., Liu, C. M., Luo, R., Sadakane, K., and Lam, T. W. (2015). MEGAHIT: An ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics 31, 1674–1676. doi:10.1093/bioinformatics/btv033.

Li, H. (2015). BFC: Correcting Illumina sequencing errors. Bioinformatics 31, 2885–2887. doi:10.1093/bioinformatics/btv290.

Liljeqvist, M., Ossandon, F. J., González, C., Rajan, S., Stell, A., Valdes, J., et al. (2015). Metagenomic analysis reveals adaptations to a cold-adapted lifestyle in a low-temperature acid mine drainage stream. FEMS Microbiol. Ecol. 91, 1–9. doi:10.1093/femsec/fiv011.

Lim, J. H., Baek, S. H., and Lee, S. T. (2009). Ferruginibacter alkalilentus gen. nov., sp. nov. and ferruginibacter lapsinanis sp. nov., novel members of the family “chitinophagaceae” in the phylum Bacteroidetes, isolated from freshwater sediment. Int. J. Syst. Evol. Microbiol. 59, 2394–2399. doi:10.1099/ijs.0.009480-0.

Lopatina, A., Medvedeva, S., Shmakov, S., Logacheva, M. D., Krylenkov, V., and Severinov, K. (2016). Metagenomic Analysis of Bacterial Communities of Antarctic Surface Snow. Front. Microbiol. 7, 398. doi:10.3389/fmicb.2016.00398.

Matsen, F. A., Kodner, R. B., and Armbrust, E. V. (2010). pplacer: linear time maximum- likelihood and Bayesian phylogenetic placement of sequences onto a fixed reference tree. BMC Bioinformatics 11, 538. doi:10.1186/1471-2105-11-538.

McMurdie, P. J., and Holmes, S. (2013). phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS One 8, e61217. doi:10.1371/journal.pone.0061217.

Mueller, D. R., Vincent, W. F., Bonilla, S., and Laurion, I. (2005). Extremotrophs, extremophiles and broadband pigmentation strategies in a high arctic ice shelf ecosystem. FEMS Microbiol. Ecol. 53, 73–87. doi:10.1016/j.femsec.2004.11.001.

Parada, A. E., Needham, D. M., and Fuhrman, J. A. (2016). Every base matters: Assessing small

129 subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol. 18, 1403–1414. doi:10.1111/1462-2920.13023.

Parks, D. H., Imelfort, M., Skennerton, C. T., Hugenholtz, P., and Tyson, G. W. (2015). CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res. 25, 1043–55. doi:10.1101/gr.186072.114.

Pearson, R. G., Phillips, S. J., Loranty, M. M., Beck, P. S. A., Damoulas, T., Knight, S. J., et al. (2013). Shifts in Arctic vegetation and associated feedbacks under climate change. Nat. Clim. Chang. 3, 673–677. doi:10.1038/nclimate1858.

Pepe-Ranney, C., Berelson, W. M., Corsetti, F. A., Treants, M., and Spear, J. R. (2012). Cyanobacterial construction of hot spring siliceous stromatolites in Yellowstone National Park. Environ. Microbiol. 14, 1182–1197. doi:10.1111/j.1462-2920.2012.02698.x.

Planavsky, N. J., Asael, D., Hofmann, A., Reinhard, C. T., Lalonde, S. V., Knudsen, A., et al. (2014). Evidence for oxygenic photosynthesis half a billion years before the Great Oxidation Event. Nat. Geosci. 7, 283–286. doi:10.1038/ngeo2122.

Preisner, E. C., Fichot, E. B., and Norman, R. S. (2016). Microbial Mat Compositional and Functional Sensitivity to Environmental Disturbance. Front. Microbiol. 7, 1632. doi:10.3389/fmicb.2016.01632.

Prieto-Barajas, C. M., Alcaraz, L. D., Valencia-Cantero, E., and Santoyo, G. (2018). Life in Hot Spring Microbial Mats Located in the Trans-Mexican Volcanic Belt: A 16S/18S rRNA Gene and Metagenomic Analysis. Geomicrobiol. J., 1–9. doi:10.1080/01490451.2018.1454555.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B. M., Ludwig, W., Peplies, J., et al. (2007). SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 35, 7188–7196. doi:10.1093/nar/gkm864.

R Core Team (2016). R: A language and environment for statistical computing. R foundation for statistical Computing, Vienna, Austria; URL: https://www.R-project.org/.

Raymond-Bouchard, I., Goordial, J., Zolotarov, Y., Ronholm, J., Stromvik, M., Bakermans, C., et al. (2018). Conserved genomic and amino acid traits of cold adaptation in subzero- growing Arctic permafrost bacteria. FEMS Microbiol. Ecol. 94, 1–13. doi:10.1093/femsec/fiy023.

Reyes, K., Gonzalez, N. I., Stewart, J., Ospino, F., Nguyen, D., Cho, D. T., et al. (2013). Surface orientation affects the direction of cone growth by Leptolyngbya sp. strain C1, a likely architect of coniform structures Octopus Spring (Yellowstone National Park). Appl. Environ. Microbiol. 79, 1302–8. doi:10.1128/AEM.03008-12.

Robertson, C. E., Spear, J. R., Harris, J. K., and Pace, N. R. (2009). Diversity and stratification of archaea in a hypersaline microbial mat. Appl. Environ. Microbiol. 75, 1801–10. doi:10.1128/AEM.01811-08.

130 Rognes, T., Flouri, T., Nichols, B., Quince, C., and Mahé, F. (2016). VSEARCH: a versatile open source tool for metagenomics. PeerJ 4, e2584. doi:10.7717/peerj.2584.

Sapers, H. M., Ronholm, J., Raymond-Bouchard, I., Comrey, R., Osinski, G. R., and Whyte, L. G. (2017). Biological Characterization of Microenvironments in a Hypersaline Cold Spring Mars Analog. Front. Microbiol. 8, 2527. doi:10.3389/fmicb.2017.02527.

Seemann, T. (2014). Prokka: rapid prokaryotic genome annotation. Bioinformatics 30, 2068– 2069. doi:10.1093/bioinformatics/btu153.

Spear, J. R., Ley, R. E., Berger, A. B., and Pace, N. R. (2003). Complexity in natural microbial ecosystems: the Guerrero Negro experience. Biol. Bull. 204, 168–73. doi:10.2307/1543553.

Stamps, B. W., Lyles, C. N., Suflita, J. M., Masoner, J. R., Cozzarelli, I. M., Kolpin, D. W., et al. (2016). Municipal solid waste landfills harbor distinct microbiomes. Front. Microbiol. 7. doi:10.3389/fmicb.2016.00534.

Tamaki, H., Hanada, S., Kamagata, Y., Nakamura, K., Nomura, N., Nakano, K., et al. (2003). Flavobacterium limicola sp. nov., a psychrophilic, organic-polymer-degrading bacterium isolated from freshwater sediments. Int. J. Syst. Evol. Microbiol. 53, 519–526. doi:10.1099/ijs.0.02369-0.

Tatusov, R. L., Galperin, M. Y., Natale, D. A., and Koonin, E. V (2000). The COG database: a tool for genome-scale analysis of protein functions and evolution. Nucleic Acids Res. 28, 33–6. Available at: http://www.ncbi.nlm.nih.gov/pubmed/10592175 [Accessed March 27, 2018].

Teunisse, G. M. (2018). Fantaxtic. Available at: https://github.com/gmteunisse/Fantaxtic.

Tytgat, B., Verleyen, E., Obbels, D., Peeters, K., De Wever, A., D’hondt, S., et al. (2014). Bacterial Diversity Assessment in Antarctic Terrestrial and Aquatic Microbial Mats: A Comparison between Bidirectional Pyrosequencing and Cultivation. PLoS One 9, e97564. doi:10.1371/journal.pone.0097564.

Varin, T., Lovejoy, C., Jungblut, A. D., Vincent, W. F., and Corbeil, J. (2010). Metagenomic profiling of Arctic microbial mat communities as nutrient scavenging and recycling systems. Limnol. Oceanogr. 55, 1901–1911. doi:10.4319/lo.2010.55.5.1901.

Varin, T., Lovejoy, C., Jungblut, A. D., Vincent, W. F., and Corbeil, J. (2012). Metagenomic Analysis of Stress Genes in Microbial Mat Communities from Antarctica and the High Arctic. Appl. Environ. Microbiol. 78, 549–559. doi:10.1128/AEM.06354-11.

Walker, J. J., and Pace, N. R. (2007). Phylogenetic composition of Rocky Mountain endolithic microbial ecosystems. Appl. Environ. Microbiol. 73, 3497–504. doi:10.1128/AEM.02656- 06.

Wickham, H. (2016). ggplot2: elegant graphics for data analysis. New York, NY: Springer- Verlag Available at: http://ggplot2.org.

131 Wright, K. E., Williamson, C., Grasby, S. E., Spear, J. R., and Templeton, A. S. (2013). Metagenomic evidence for sulfur lithotrophy by epsilonproteobacteria as the major energy source for primary productivity in a sub-aerial arctic glacial deposit, Borup Fiord Pass. Front. Microbiol. 4, 1–20. doi:10.3389/fmicb.2013.00063.

Zhang, J., Kobert, K., Flouri, T., and Stamatakis, A. (2014). PEAR: A fast and accurate Illumina Paired-End reAd mergeR. Bioinformatics 30, 614–620. doi:10.1093/bioinformatics/btt593.

132

CHAPTER 5

CONCLUDING REMARKS

5.1 Conclusions

The research detailed in this dissertation expands our knowledge of microbial communities in polar environments, specifically sulfur-impacted ecosystems, and in exploring the variability in community composition of microbial mats of the Canadian High Arctic. Polar ecosystems are incredibly important, providing humanity a window through which to study life processes under adverse conditions such as temperature extremes, high salinity, and solar radiation (e.g. polar night and midnight sun). Polar systems, among others, are also often used as terrestrial analog sites to study extraterrestrial worlds such as Jupiter’s moon Europa (Gleeson et al., 2011; Lorenz et al., 2011). They are also used as a barometer to measure how climate change is affecting our world, which can be seen in metrics like glacial mass loss (Noël et al., 2018), and an increase in glacial melt due to microbial growth on the surface of ice (Ganey et al., 2017). Using 16S rRNA gene sequencing linked to aqueous geochemistry I investigated the sulfur-rich, low-temperature environment at Borup Fiord Pass, Ellesmere Island, Nunavut, Canada shining a light on the variability and temporal dynamics of microbial communities in the High Arctic. Microbial assemblages differed between different site/material types during short time periods (days and weeks), while Flavobacterium sp. community members primarily persisted over longer time periods (years). Sulfur-utilizing organisms across the phylum Proteobacteria (i.e. Sulfurovum,

Sulfuricurvum, Sulfurimonas, and Desulfocapsa) were widespread across multiple site types and in varying relative abundances. A basal microbial community existed regardless of sample type,

133 consisting of a large relative abundance of OTUs most closely related to the genus

Flavobacterium across BFP. While the focus in previous studies has been on traditional sulfur- cycling microorganisms, my data point to a larger role of Flavobacterium sp. based on their ubiquity in BFP samples and their persistence over time. While these heterotrophs may not directly contribute to overall sulfur-cycling in the BFP system they may break down large organic molecules into smaller more easily metabolized carbon substrates for use by other microorganisms (e.g. sulfate-reducing microorganisms). While 16S rRNA gene sequencing was essential in establishing the presence of Flavobacterium sp., a higher resolution approach such as shotgun whole genome sequencing, or metagenomic sequencing, was needed to determine the overall metabolic potential of BFP.

Metagenomic sequencing was used to elucidate the potential metabolisms of the microbial communities of BFP. Confirming the work presented in chapter two, chapter three detailed that organisms from the Alpha-, Beta-, Gamma-, Delta-, and Epsilonproteobacteria were well represented via whole genome sequencing. The Proteobacteria at BFP appear capable of a diverse swath of metabolic functions, confirming previous work where sulfur oxidation is the most likely chemolithotrophic metabolism and is employed across BFP (Wright et al., 2013).

The process of sulfur oxidation appears functionally redundant across multiple genomes within and between samples, most likely due to environmental pressures and selection. Functional redundancy is a concept wherein taxonomically distinct organisms coexist in a system despite sharing metabolic, typically energy-producing, functionality. While a tantalizing finding, it does not prove that all organisms are capable of growing and conserving energy via sulfur oxidation.

Rather, some organisms may only have these genes to help mediate the oxidation of sulfur for assimilatory (integration of sulfur compounds into cellular components), rather than

134 dissimilatory, processes. Conversely, sulfate reduction genes (dsr) were found linked to organisms capable of sulfur disproportionation, a process by which sulfur compounds (such as elemental sulfur, thiosulfate, and sulfite) can serve as both electron donor and acceptor,

2- generating both reduced (H2S) and oxidized (SO4 ) forms (Finster, 2008). It remains unclear, however, whether the reduced sulfur compounds in the system are primarily biogenic or abiogenic in origin.

While sulfur cycling was the focus of this dissertation, nitrogen metabolism is also an important consideration in the High Arctic. Nitrate and nitrite reduction genes were also present at BFP, with the majority contained in metagenome assembled genomes (MAGs) that also contain an abundance of sulfur-oxidation genes, suggesting these two processes are more than likely linked. Finally, functional inquiry into carbon metabolism clearly points to the reductive acetyl-CoA pathway being the dominant form of carbon fixation in the system, alluding to the fact that this may occur under anoxic or microoxic conditions, even within melt pools, aufeis, and mineral precipitates at the surface of the glacier. Microoxic niches may allow for this in many cases. Metagenomics only provides a view into the potential metabolisms present in the system whether microorganisms are alive or dead, and active or not. Overall the results of chapter 3 reveal that metabolic functional redundancy in low-temperature environments is a crucial strategy for multiple bacterial lineages to coexist.

From sulfur-impacted ice my thesis moved to another unexpected feature of the Canadian

High Arctic, microbial mats found across Ellesmere Island. These microbial mats were found across streams whose flow is generated via glacial melt or by subsurface spring flow. Similar to my previous two chapters, the purpose of this research was to characterize these sites via use of

16S rRNA gene sequencing, and additionally for one site, Ice River, metagenomic sequencing.

135 The studied microbial mats are located at, to my knowledge, the highest latitudes ever reported

(above 80° N). While all sites were within a few hundred kilometers of each other, my results showed that individual sites vary greatly based on their 16S rRNA gene sequencing community composition. Within each site the phyla Cyanobacteria, Bacteroidetes, and Proteobacteria were present; however, their relative abundances varied explaining the significant differences in community composition and diversity. Likely driving these shifts in microbial community composition are both the environmental conditions and nutrient provision of geochemically distinct source waters. Furthermore, metagenomic sequencing of Ice River revealed a small diversity of carbon metabolisms and some potential for the cycling of sulfur and nitrogen. Genes associated with stress or environmental shock were abundant- not unexpected in an environment in which extremes in temperature, solar exposure, or water content are common. This work contributes to the body of knowledge in microbial mat systems of polar climates, and suggests that regardless environmental conditions are the main driving force behind taxonomic differences and variability in adaptive stress genes. Moreover, the implications for the impact of climate on the development of these sites should be noted. No research has previously been conducted to look at how primary productivity in microbial mats is affected by latitude. This study could indicate that due to climate change more growth of layered assemblages like microbial mats can be expected at higher latitudes in the coming years. The carbon fixed by laminated microbial mats serves as the primary node in ecosystem development, whereby a microbial mat forms, grows, cycles, becomes a basis for ‘soil’ where small plants and mosses colonize, and subsequently higher plants and trees. In the Arctic, this is how the ecosystem develops from nothing but rock, ice and water to rock, ice, “tundra” and water to rock, ice,

“tundra”, “forest” and water. Watching these shifting patterns as an ecological consequence of

136 climate change could be in our very near future.

5.2 Future Directions

The presented research is the sum of 4.5 years of concentrated effort across multiple kilometers of the high Arctic providing a high-resolution view of the metabolic potential of sulfur-impacted glacial systems here on Earth and potentially beyond. Even so, my work has led to numerous outstanding questions. Through my research and several authors before me, Borup

Fiord Pass has been extensively but not exhaustively studied. Additional work should be conducted to help constrain and characterize the system to a greater degree including hydrogeologic investigations of the site, sulfur geochemistry, metatranscriptomics, and the impact of climate change over time.

One drawback to my work was the lack of a well characterized hydrologic model of the flow beneath the glacier at BFP. Concerted efforts to develop a model for the hydrogeology of

BFP have been conducted in the past (Scheidegger et al., 2012). This was previously challenging due to logistical efforts and the cost of transporting heavy scientific equipment. These efforts can hopefully be lessened now due to advancements in technology leading to more portability in systems like ice penetrating radar (Ye et al., 2011).

Questions also remain in considering the fate of sulfur both biological and abiotic in the

BFP system. Lau et al. (2017) discovered rare forms of cyclooctasulfur in one of the melt pools from 2014 samples. Although unable to determine whether or not these rare allotropes (β- and γ- cyclooctasulfur) were biogenic in origin or if their formation was biogenically mediated in any way, it stands to reason that this is a real possibility. A targeted effort to recover more β- and γ- cyclooctasulfur from sites at BFP would be a priority task for a future field campaign to the site.

137 This effort would also include targeted cultivation efforts to better determine if microorganisms are playing a role in allotrope formation and utilization. If that were the case then it would be possible that these allotropes could possibly serve as a biomarker of sorts in these polar low- temperature environments to indicate microbial activity. This would be of keen interest to scientists studying astrobiological targets such as Europa and Enceladus. Moreover, it would further expand our knowledge of these polar ecosystems and what microorganisms are capable of when subjected to adverse environmental conditions.

While I was able to answer questions towards the metabolic capability of microorganisms across BFP and potential implications towards sulfur cycling, specific activity of microbial communities is crucial to explore to fully answer if the predicted processes are occurring. One approach that will help answer questions about microbial activity is metatranscriptomics, or the sequencing of environmental RNA to further interpret the functionality of keystone species at

BFP. One primary target for metatranscriptomic study is Flavobacterium sp. found across multiple BFP sites, sample types, and years. Some of this is already underway, but is beyond the scope of this thesis. Furthermore, genes detected by metatranscriptomic profiling could be further linked to other sensitive measures of microbial activity in the future such as environmental proteomics, or protein sequencing, or in situ measures of sulfur oxidation.

Comparing these data to the metagenomic data will further enlighten us about the differences between potential metabolic processes and active metabolic processes occurring at BFP.

Polar environments remain as one of the final, largely unexplored frontiers across planet

Earth. These frigid environments provide valuable insight into low-temperature geochemical and biological processes across our world and worlds beyond. Polar environments truly act as a barometer to a changing planet, responding to human influence. If humanity is not careful, these

138 polar terrestrial laboratories will melt away before we are fully able to realize their scientific value and potential.

5.3 References

Finster, K. (2008). Microbiological disproportionation of inorganic sulfur compounds. J. Sulfur Chem. 29, 281–292. doi:10.1080/17415990802105770.

Ganey, G. Q., Loso, M. G., Burgess, A. B., and Dial, R. J. (2017). The role of microbes in snowmelt and radiative forcing on an Alaskan icefield. Nat. Geosci. 10, 754–759. doi:10.1038/ngeo3027.

Gleeson, D. F., Williamson, C., Grasby, S. E., Pappalardo, R. T., Spear, J. R., and Templeton, A. S. (2011). Low temperature S0 biomineralization at a supraglacial spring system in the Canadian High Arctic. Geobiology 9, 360–375. doi:10.1111/j.1472-4669.2011.00283.x.

Lau, G. E., Cosmidis, J., Grasby, S. E., Trivedi, C. B., Spear, J. R., and Templeton, A. S. (2017). Low-temperature formation and stabilization of rare allotropes of cyclooctasulfur (β-S 8 and γ-S 8 ) in the presence of organic carbon at a sulfur-rich glacial site in the Canadian High Arctic. Geochim. Cosmochim. Acta 200, 218–231. doi:10.1016/j.gca.2016.11.036.

Lorenz, R. D., Gleeson, D., Prieto-Ballesteros, O., Gomez, F., Hand, K., and Bulat, S. (2011). Analog environments for a Europa lander mission. Adv. Sp. Res. 48, 689–696. doi:10.1016/j.asr.2010.05.006.

Noël, B., van de Berg, W. J., Lhermitte, S., Wouters, B., Schaffer, N., and van den Broeke, M. R. (2018). Six decades of glacial mass loss in the Canadian Arctic Archipelago. J. Geophys. Res. Earth Surf. doi:10.1029/2017JF004304.

Scheidegger, J. M., Bense, V. F., and Grasby, S. E. (2012). Transient nature of Arctic spring systems driven by subglacial meltwater. Geophys. Res. Lett. 39, 1–6. doi:10.1029/2012GL051445.

Wright, K. E., Williamson, C., Grasby, S. E., Spear, J. R., and Templeton, A. S. (2013). Metagenomic evidence for sulfur lithotrophy by epsilonproteobacteria as the major energy source for primary productivity in a sub-aerial arctic glacial deposit, borup fiord pass. Front. Microbiol. 4, 1–20. doi:10.3389/fmicb.2013.00063.

Ye, S., Zhou, B., Wu, B., Zhao, B., and Fang, G. (2011). An improved transient-type ice- penetrating radar. J. Glaciol. 57, 295–301. doi:https://doi.org/10.3189/002214311796405861.

139

APPENDIX A

SUPPORTING INFORMATION FOR CHAPTER 2

Two supplemental files with additional information are included for Chapter 2. The first is a mapping file and the corresponding barcodes. This is included as a Microsoft Excel file in

Appendix E (A.S1 - Ch2_Mapping.xlsx). The second file is a sequencing coverage summary as the output file from the script ‘biom summarize-table’. This is included as a Microsoft Excel file in Appendix E (A.S2 - Ch2_Seq_coverage.xlsx).

Table A.1. Sample information table including sequencing counts, charge balance error, and calculated biomass. Site names are shown with included replicates (a, b, c, etc.). Samples are grouped based on sample type category (A - Aufeis; AS - Mineral precipitates; M - Melt Pool; G - Glacier) with included approximate charge balance error (CBE) shown. Biomass values were calculated using DNA concentration and initial sample amount (mass or volume).

DNA conc. Sample Date Site Type CBE Seq. read # Sample amount Biomass (cells/ml) (ng/μl) A1 23-Jun-14 <15% 6209 7570 ml BD ND A2 23-Jun-14 <15% 12922 5678 ml 0.425 1.15E+03 cells/ml A3 27-Jun-14 <15% 3269 5678 ml BD ND A4a 27-Jun-14 <5% 13564 2840 ml 0.828 4.46E+03 cells/ml Aufeis A4b 27-Jun-14 <5% 19106 2840 ml 0.837 4.51E+03 cells/ml A5a 27-Jun-14 >15% 5936 3785 ml 0.517 2.09E+03 cells/ml A5b 27-Jun-14 >15% 14203 3785 ml 0.318 1.29E+03 cells/ml A6 28-Jun-14 <15% 5615 Unknown 4.0 ND AS1 22-Jun-14 >15% 1243 100 ml 0.4 6.14E+04 cells/ml AS2 24-Jun-14 <15% 10031 100 ml 0.43 6.58E+04 cells/ml AS3a 7-Jul-17 ND 8493 500 mg 6.88 2.11E+05 cells/mg AS3b 7-Jul-17 ND 11095 500 mg 11 3.37E+05 cells/mg AS3c 7-Jul-17 ND 5592 500 mg 2.73 8.35E+04 cells/mg AS4a 7-Jul-17 ND 5965 500 mg 7.34 2.25E+05 cells/mg AS4b 7-Jul-17 ND 5246 500 mg 5.37 1.64E+05 cells/mg AS4c 7-Jul-17 ND 7422 500 mg 6.76 2.07E+05 cells/mg AS4d 7-Jul-17 Mineral ND 6980 500 mg 3.56 1.09E+05 cells/mg AS4e 7-Jul-17 Precip. ND 1468 500 mg 4.03 1.23E+05 cells/mg AS4f 7-Jul-17 ND 4420 500 mg 0.117 3.58E+03 cells/mg AS5a 7-Jul-17 ND 1474 500 mg 4.86 1.49E+05 cells/mg AS5b 7-Jul-17 ND 5130 500 mg 5.4 1.65E+05 cells/mg AS5c 7-Jul-17 ND 9243 500 mg 5.89 1.80E+05 cells/mg AS6a 7-Jul-17 ND 8376 500 mg 4.34 1.33E+05 cells/mg AS6b 7-Jul-17 ND 1769 500 mg 9.14 2.80E+05 cells/mg AS6c 7-Jul-17 ND 9201 500 mg 3.21 9.82E+04 cells/mg AS7a 7-Jul-17 ND 2821 500 mg BD ND

140 Table A.1 BFP Sample information table (continued).

AS7b 7-Jul-17 ND 10235 500 mg 0.141 4.31E+03 cells/mg G1a 21-Jun-14 Glacier ND 17103 500 mg 3.91 1.20E+05 cells/mg G1b 21-Jun-14 ND 10156 500 mg 1.08 3.30E+04 cells/mg G2 27-Jun-14 ND 4798 360 ml 0.275 1.17E+04 cells/ml G3a 28-Jun-14 <5% 2997 360 ml 0.26 1.11E+04 cells/ml G3b 28-Jun-14 <5% 11377 360 ml 0.375 1.59E+04 cells/ml M1 21-Jun-14 <5% 5177 300 ml 0.0097 4.95E+02 cells/ml M2a 21-Jun-14 <15% 13889 500 mg 0.3201 9.80E+03 cells/mg M2b 21-Jun-14 <15% 13681 500 mg 0.1679 5.14E+03 cells/mg M2c 21-Jun-14 <15% 6032 360 ml BD ND M2d 21-Jun-14 <15% 7510 500 mg 0.0401 1.23E+03 cells/mg M3a 26-Jun-14 <15% 6007 500 mg 0.0394 1.21E+03 cells/mg Melt Pool M3b 30-Jun-14 <15% 4794 360 ml BD ND M4a 27-Jun-14 <15% 2478 360 ml 0.0224 9.52E+02 cells/ml M4b 27-Jun-14 <15% 10240 360 ml 0.074 3.15E+03 cells/ml M4c 27-Jun-14 <15% 7648 500 mg BD ND M5 30-Jun-14 <15% 8242 360 ml 0.179 7.61E+03 cells/ml M6 30-Jun-14 >15% 4795 240 ml BD ND 2016 Spring 4-Jul-16 Spring <15% 18139 2000 ml 0.406 3.11E+03 cells/ml

Figure A.1. Melt Pool site locations. Melt pool locations M2-M6 are shown using light blue arrows. The location of M1 is further down the Sulfidic Aufeis and not shown in this figure, but rather Figure A.2. An indicator of the direction of M1 is included. Note the location of sites M4 and M5 directly on top of the Ice Blister (not explicitly highlighted).

141

Figure A.2. Aufeis site locations. Aufeis site locations are shown as they continue down the Sulfidic Aufeis zone. Note the location of M1 which is farther down the aufeis and not able to be shown in Figure A.1. Site A3 is the closest to the location of the Ice Blister and the majority of melt pool sites seen in Figure 2.1b and Figure A.1.

142

Figure A.3. BFP17 Site location. The general location of the BFP 2014 site as seen on July 7, 2017 (A). The yellow staining in A is the approximate site of the Ice Blister from 2014. Panels B and C show examples of samples collected from 2017 and represented as mineral precipitate samples for comparison to 2014 data. The scale bar in panels B and C is 7 cm, and approximately 8 cm for the total scale card.

143

Figure A.4. Principal Component Analysis (PCoA) of BFP aqueous geochemistry. Samples are separated into categories and ordinated according to geochemical values. The longer the eigenvector (blue arrow) for each value the stronger its effect on the shown diversity of the system. Sites that are in the same direction of a given eigenvector are considered strongly correlated with that parameter and vice versa for those sites opposite of a given eigenvector. Analytes with eigenvectors of 0 were removed for clarity of the figure. Glacier/Control samples are denoted by the same blue square. Sample G2 has associated 16S rRNA gene sequencing data whereas “Control” samples were used only to compare aqueous geochemistry of the system.

144

Figure A.5. Alpha rarefaction plot of BFP site samples. Alpha rarefaction plot is shown based on the number of observed OTUs per sample. The upper limit of rarefaction is the median sequence value (6962) which can also be seen in Table A.1.

145

Figure A.6. Rank abundance plot of BFP site types. Rank abundance is shown by sample type. The y-axis is in log form whereas the x-axis is not. A rank abundance plot is used to show similar diversity across site types in terms of relative abundance vs the number of species (OTUs) averaged across the samples for that site type.

146

APPENDIX B

SUPPORTING INFORMATION FOR CHAPTER 3

Table B.1. Co-assembly metadata. These data represent raw values from Megahit prior to mapping through Bowtie2 and final processing via Anvi’o.

Sample Total bp Contigs Max contig (bp) Avg (bp) N50 (bp) BFP14_A4b 85484280 30017 380944 2848 3571 BFP14_A6 82535183 33359 128624 2474 2871 BFP14_M2 10144305 4841 418178 2095 1977 BFP14_M4b 1.59E+08 71593 111481 2215 2359 BFP16_Spring 60894340 18996 206795 3206 4748 BFP17_14C 6.38E+08 272035 297076 2347 2611 BFP17_3B 1.98E+08 78690 78857 2513 3028 BFP17_4E 89868307 27395 44699 3280 4468 BFP17_6B 1E+08 34932 60991 2866 4186

Co-Assembly 1.16E+09 477394 406785 2440 2778

147

Table B.2. JSpeciesWS results. MAGs 7 and 12 were compared to database genomes for Desulfocapsa sulfexigens and Desulfotalea psychrophila LSv54. Average nucleotide identity was calculated for BLAST and MUMmer as well as Tetra-nucleotide comparison.

#ANI BLAST and aligned percentage Desulfotalea Desulfocapsa MAG 12 MAG 7 psychrophila LSv54 sulfexigens DSM 10523 Desulfotalea psychrophila * 66.86 [20.05] 67.07 [14.44] 67.13 [14.59] LSv54 Desulfocapsa sulfexigens 66.75 [18.50] * 68.41 [25.07] 68.16 [24.91] DSM 10523 MAG 12 67.28 [20.09] 68.74 [35.44] * 81.71 [44.41] MAG 7 67.17 [20.68] 68.48 [36.64] 82.11 [44.96] *

#ANI MUMmer and aligned percentage Desulfotalea Desulfocapsa MAG 12 MAG 7 psychrophila LSv54 sulfexigens DSM 10523 Desulfotalea psychrophila * 87.49 [1.04] 87.97 [0.04] 85.18 [0.14] LSv54 Desulfocapsa sulfexigens 87.74 [0.47] * 82.75 [0.55] 82.87 [0.26] DSM 10523 MAG 12 87.97 [0.02] 82.75 [0.75] * 85.30 [33.12] MAG 7 85.18 [0.17] 82.84 [0.41] 85.30 [34.60] *

#Tetra results Desulfotalea Desulfocapsa MAG 12 MAG 7 psychrophila LSv54 sulfexigens DSM 10523 Desulfotalea psychrophila * 0.90901 0.86371 0.84073 LSv54 Desulfocapsa sulfexigens 0.90901 * 0.92621 0.90605 DSM 10523 MAG 12 0.86371 0.92621 * 0.98997 MAG 7 0.84073 0.90605 0.98997 *

148

Figure B.1. BFP16 Spring location. Shown is the location of the 2016 spring. It’s exact location is at the northern-most portion of the yellow-colored area, approximately in the center of the picture. Also shown is the area known as the “Sulfidic Aufeis” from Trivedi, et al. (in prep) towards the right edge of the picture. This is the location of sampling from 2014 and 2017 field campaigns.

149

Figure B.2. Examples of sample types used for metagenomes. A) Melt pool; B) Aufeis sample, where surface ice was removed, a pit dug and ice from this pit was melted and filtered; C) Mineral precipitates from the 2017 BFP site.

150

APPENDIX C

SUPPORTING INFORMATION FOR CHAPTER 4

An Arctic mats mapping file is included as supplemental information in Appendix E. This is a mapping file with associated Illumina barcodes for demultiplexing of 16S rRNA gene sequencing data uploaded to NCBI. This is included as a Microsoft Excel file in Appendix E

(C.S1 - Ch4_Mapping.xlsx).

Table C.1. Mat sample PCR amplification parameters.

Temperature Duration Notes 94 °C 2 minutes Initial Denaturation 94 °C 30 seconds 1X 5PRIME HOT 50 °C 30 seconds 30 cycle amplification 68 °C 1:30 minutes 68 °C 5 minutes Final Extension 4 °C Max 24 hours Hold until transitioned into storage

Temperature Duration Notes 95 °C 10 minutes Initial Denaturation 1X 94 °C 30 seconds ZymoBIOMICS™ 52°C 20 seconds 30 cycle amplification PCR PreMix 72 °C 30 seconds 72 °C 5 minutes Final Extension 4 °C Max 24 hours Hold until transitioned into storage

151

Table C.2. ADONIS Results between mat sites. Samples were subset and an ADONIS test was run to check significance between each site based on microbial community diversity. P-values are reported with R2 values shown in parentheses.

Adonis Cape Evans Derberville Ice River Moss Valley Cape Evans .043 (.305) .002 (.254) .008 (.499) Derberville .016 (.236) .06 (.807) Ice River .002 (.349) Moss Valley

Table C.3. Ice River MAG putative taxonomy. Taxonomy was generated via CheckM.

Marker lineage Bin Length (bp) Comp. Redund. Name c__Gammaproteobacteria (UID4761) MAG 8 4637852 98.45 1.62 p__Cyanobacteria (UID2182) MAG 1 4161358 98.35 0.94 s__algicola (UID2847) MAG 3 3399685 95.75 0.19 c__Alphaproteobacteria (UID3422) MAG 5 3954031 93.04 2.28 s__algicola (UID2846) MAG 7 4212278 92.81 0.66 o__Burkholderiales (UID4000) MAG 9 3425651 90.68 1.62 o__Sphingomonadales (UID3310) MAG 13 2815611 88.59 0.75 f__Rhodobacteraceae (UID3356) MAG 4 2902300 88.36 0.5 p__Bacteroidetes (UID2591) MAG 2 6068462 87.47 2.08 p__Bacteroidetes (UID2591) MAG 6 4529778 85.24 0.51 o__Burkholderiales (UID4000) MAG 12 2804090 72.6 0.24 o__Sphingomonadales (UID3310) MAG 16 1807025 65.79 4.8 o__Cytophagales (UID2936) MAG 11 3460703 62.81 8.72 k__Bacteria (UID203) MAG 18 5117701 56.03 2.59

152

Figure C.1. Alpha diversity of Arctic mat sites. Box and whisker plots show similar numbers of observed OTUs across all sites. Ice River has by far the widest range, but also has the most samples. Cape Evans and Moss Valley are similar while Derberville has the fewest number of observed OTUs, but also the fewest number of samples as well.

153

Figure C.2. Heatmap of top 12 phyla by site. Heat map of the top 25 OTUs of each site displayed on the Phylum level.

154

APPENDIX D

CARBONATE-RICH DENDROLITIC CONES: INSIGHTS INTO A MODERN ANALOGUE

FOR INCIPIENT MICROBIALITE FORMATION, LITTLE HOT CREEK,

LONG VALLEY CALDERA, CALIFORNIA

A paper published in NPJ Biofilms and Microbiomes1

Bradley, James A.2*; Daille, Leslie K.3*; Trivedi, Christopher B.4*; Bojanowski, Caitlin L.5,6; Stamps, Blake W.4; Stevenson, Bradley S.7, Nunn, Heather S.7; Johnson, Hope A.8; Loyd, Sean J.9; Berelson, William M.2; Corsetti, Frank A.2; Spear, John R.4^

Abstract

Ancient putative microbial structures that appear in the rock record commonly serve as evidence of early life on Earth, but the details of their formation remain unclear. The study of modern microbial mat structures can help inform the properties of their ancient counterparts, but

1Reprinted with permission Bradley, J. A., Daille, L. K., Trivedi, C. B., Bojanowski, C. L., Stamps, B. W., Stevenson, B. S., Nunn, H. S., Johnson, H.A., Loyd, S. J., Berelson, W. M., Corsetti, F. A., & Spear, J. R. (2017). Carbonate-rich dendrolitic cones: insights into a modern analog for incipient microbialite formation, Little Hot Creek, Long Valley Caldera, California. NPJ biofilms and microbiomes, 3(1), 32. 2Department of Earth Sciences, University of Southern California, Los Angeles, USA. 3Facultad de Ciencias Biológicas, Pontificia Universidad Católica de Chile, Santiago, Chile. 4Department of Civil and Environmental Engineering, Colorado School of Mines, Golden, CO, USA. 5Soft Matter Materials Branch, Materials and Manufacturing Directorate, Air Force Research Laboratory, Wright-Patterson AFB, OH 45433 6UES, Inc., Dayton, OH 7Department of Microbiology and Plant Biology, University of Oklahoma, Norman, Oklahoma, USA. 8Department of Biological Science, California State University, Fullerton, Fullerton, California, USA. 9Department of Geological Sciences, California State University, Fullerton, Fullerton, California, USA.

155 modern mineralizing mat systems with morphological similarity to ancient structures are rare.

Here, we characterize partially lithified microbial mats containing cm-scale dendrolitic coniform structures from a geothermal pool (“Cone Pool”) at Little Hot Creek, California, that if fully lithified, would resemble ancient dendrolitic structures known from the rock record. Light and electron microscopy revealed that the cm-scale ‘dendrolitic cones’ were comprised of intertwined microbial filaments and grains of calcium carbonate. The degree of mineralization

(carbonate content) increased with depth in the dendrolitic cones. Sequencing of 16S rRNA gene libraries revealed that the dendrolitic cone tips were enriched in OTUs most closely related to the genera Phormidium, Leptolyngbya, and Leptospira, whereas mats at the base and adjacent to the dendrolitic cones were enriched in Synechococcus. We hypothesize that the consumption of nutrients during autotrophic and heterotrophic growth may promote movement of microbes along diffusive nutrient gradients, and thus microbialite growth. Hour-glass shaped filamentous structures present in the dendrolitic cones may have formed around photosynthetically-produced oxygen bubbles – suggesting that mineralization occurs rapidly and on timescales of the lifetime of a bubble. The dendrolitic-conical structures in Cone Pool constitute a modern analogue of incipient microbialite formation by filamentous microbiota that are morphologically distinct from any structure described previously. Thus, we provide a new model system to address how microbial mats may be preserved over geological timescales.

D.1 Introduction

Ancient fossilized putative microbial structures appear in the rock record as morphologically distinctive indicators of early life on Earth (Allwood et al., 2006; Flannery and

Walter, 2012). Tufts in fossilized Archaean structures have been interpreted as Cyanobacteria and thus evidence for oxygenic photosynthesis (Buick, 1992). However, as structures that have

156 undergone lithification and post-depositional diagenetic alteration, most ancient microbialites lack preserved microfossils or unobscured information concerning their formation and the chemistry of their environments (Bosak et al., 2013; Grotzinger and Knoll, 1999). An understanding of the processes that control the formation of microbialites and determining the geochemistry of these structures will help to more accurately interpret the ambient ancient environmental, geochemical and physical conditions.

Conical microbialite forms are well known from the rock record and occur at various scales (millimeter to decimeter), but modern examples, especially from carbonate-precipitating environments, are rare (Berelson et al., 2011; Lindsay et al., 2017; Mata et al., 2012; Pepe-

Ranney et al., 2012). Some examples of modern conical microbialite structures have previously been described in geothermal springs (Bosak et al., 2012; Petroff et al., 2010) and the Antarctic lakes Untersee (Andersen et al., 2011) and Vanda (Sumner et al., 2016). It is useful to characterize these and other modern conical analogs of ancient microbialites to support the development of a formation model and to better understand their preservation potential in rock over geological timescales.

Here, we present a detailed physical, geochemical and genomic characterization of semi- lithified carbonate, dendrolitic conical microbial structures (i.e. dendrolitic cones) from a geothermal pool (Cone Pool) at Little Hot Creek (LHC) spring, in the Long Valley Caldera,

California, USA (Figure D.1). We describe the dendrolitic cone structure as a cm-scale central coniform structure on which smaller scale dendrolitic structures root, and as such, is a unique structure that has some similarity with both coniform and dendrolitic microbialites in the rock record. The dendrolitic cones, surrounding microbial mats, and water, were examined to

157

Figure D.1. Cone Pool at Little Hot Creek. (A) Satellite view of Little Hot Creek (LHC) geothermal spring in California, USA. White squares indicate current study sampling sites. White asterisks indicate sampling sites of Vick et al. (2010). (B) View of Cone Pool from the West rim. (C) Dendrolitic cones in situ at Cone Pool East. (D) Extracted microbial mat showing dendrolitic cone assemblage and size differences. (E) Photograph of extracted dendrolitic cone indicating macrostructure. Cone Tip, Cone Middle and Cone Base labelled.

determine their morphological structure, microbial and geochemical composition, and degree of lithification. The microbial composition of the dendrolitic cones, underlying microbial mat, and the layered microbial mat from an adjacent outflow pool (LHC Outflow pool) were determined by sequencing amplified libraries of small subunit ribosomal 16S rRNA genes. The autotrophic growth rates of the community were measured by 13C-bicarbonate uptake from dissected samples of the tips of dendrolitic cones. The data presented here provide a basis to better understand the structure and generation mechanism of LHC dendrolitic cones. These findings provide insight into dendrolitic cone microbialite formation and preservation in modern geothermal springs.

They also provide an analogue to ancient microbialites.

158 D.2 Results

D.2.1 Site Characterization

The Cone Pool at LHC is a small geothermal pool that measures approximately 6m by 4m and is ~0.75m deep in the center. The pool is fed by a subsurface spring via an inlet vent along its northern rim (Figure D.1A). The geothermal spring upwelling that originates in Cone Pool trickles downstream to a second larger pool (Outflow Pool). Cone Pool and the Outflow Pool contain microbial mats, yet, uniquely, Cone Pool contains partially lithified dendrolitic cones that emerge from an underlying, laminated microbial mat foundation. Physical and chemical characteristics of Cone Pool and the Outflow Pool spring water are presented in Table D.1. At the time of sampling, Cone Pool was substantially warmer (45.6°C) than the downstream LHC

Outflow Pool (34.1°C). Cone Pool and the Outflow Pool exhibited similar slightly alkaline pH

(8.08 and 8.29 respectively), and both were super-saturated with respect to calcite (ΩCa = 4.08 and 3.40 respectively) (Figure D.6). Concentrations of dissolved total CO2 (TCO2), dissolved oxygen, and Ca2+, K+, Mg2+, and Na+ were similar between the Cone Pool and Outflow Pool

(Table 1). Based on visits to Cone Pool prior to and after the sampling trip of the present study

Table D.1. Physical and chemical properties of stream water at Cone Pool and Outflow Pool.

a b Dissolved b d b b Site Temperature Ω TCO2 pH c Ca K Mg Na oxygen Cone 45.6 4.08 11.98 8.08 130.3 0.43 0.70 29.42 17.49 Pool Outflow 34.1 3.40 11.52 8.29 141.4 0.26 0.75 35.76 18.23 Pool a °C; b Millimolar (mM); c Percent relative to vapor saturated air (100% DO); d Micromolar (µM)

159 (June 2015), dendrolitic cones may be an ephemeral feature of the microbial mat: dendrolitic cones were visible from the first visit (December 2013) until September 2015, however, were not visible on return to the site in June 2016.

D.2.2 Dendrolitic Cone Description

The dendrolitic cones ranged in height between 0.5 – 2.0 cm (Figure D.1C) and were orange to brown in color, with an upright, fractal, branching appearance; that is, the cm-scale dendrolitic cones had mm-scale protrusions upon them, creating an ‘arborescent’ appearance.

The mm-scale protrusions were somewhat rounded at the tip, and oriented at an upward angle from the surface of the dendrolitic cone. The dendrolitic cones were rigid, did not deform when removed from the pool, and remained firm to touch upon collection and transport (Figure D.1D-

E). Geochemical analyses revealed that the dendrolitic cones were predominantly comprised of

CaCO3 (76.1% dry weight). Organic carbon comprised 5.9% of the dry weight of dendrolitic cone material. One representative dendrolitic cone, measuring 1.2 cm across the base and 2 cm high, was subdivided into Tip, Middle, and Base for further analysis (Figure D.1E).

Scanning Electron Microscopy (SEM) imaging revealed that the Cone Tip, Middle, and

Base contained morphologically distinct microstructures. The dendrolitic cone tip was covered with the mm-scale protrusions comprised of domed pinnacles (0.5-1mm length) (Figure D.2A).

The pinnacle elements consisted of intertwined microbial filaments that formed matrices and bridging structures. Voids in the dendrolitic cone on the scale of >600 μm diameter occurred between densely packed filamentous bridging structures (Figure D.2B). These filaments appeared to be intertwined around calcium carbonate grains ranging in size from ~1-50 μm in diameter, which were slightly rounded but contained multiple facets. The Cone Middle

160

Figure D.2. Microscopic characterization of dendrolitic cones. Scanning electron microscopy images of (A) dendrolitic cone macrostructure and (B) microscale structures. Notable features of dendrolitic cone micro-structure indicated by numbers: (1) major voids, (2) bridging structure, (3) CaCO3 grains, (4) small ‘bubble-like’ voids. (C) Confocal microscopy images of dendrolitic cone filament sections, at a magnification of 400X. Photoautotrophic cells are shown in red.

161 exhibited a web of intertwined filaments with increasingly abundant mineral grains (white) compared to the Cone Tip (Figure D.2A). Examination of the Cone Base revealed an even higher degree of mineralization: grains in the Cone Base were largest and most common (Figure D.2A,

Base). The high degree of mineralization in the Cone Base likely occluded the microbial filament structures that characterized the upper parts of the dendrolitic cone. A notable, but uncommon feature from SEM imaging, was the appearance of hour-glass shaped void structures (Figure

D.2B) that, at ~50 μm diameter, were substantially smaller than the larger voids (>600 μm) mentioned above. These smaller voids seemed to cluster together in spatially discrete locations, where filaments were less tightly bundled. Confocal analyses showed that the auto-fluorescence in the dendrolitic cone increases from the Cone Base to the Cone Tip, suggesting a greater abundance of photoautotrophic microorganisms in upper sections of the dendrolitic cone (Figure

D.2C). SEM images of a dendrolitic cone sample were taken before and after the addition of 1 M hydrochloric acid (HCl), which resulted in the dissolution of the calcium carbonate grains, leaving voids (~50 μm) in the sample (Figure D.3). In addition, the sample lost its rigidity.

D.2.3 Growth of Microorganisms in the Cone Tip

The uptake of 13C-labeled bicarbonate was measured in samples from the Cone Tip and the microbial mat from the outflow pool. Fixation of bicarbonate in samples from the Cone Tip was confirmed with measured rates of autotrophic growth of 0.15 % d-1 (± 0.01; 1 SD). Autotrophic growth in the microbial mat from the Outflow Pool was 0.17 % d-1 (± 0.05; 1SD). Thus, measured growth relative to killed controls showed that resident autotrophs were actively fixing carbon in the dendrolitic cone tips and in the microbial mat in the Outflow Pool.

162

Figure D.3. Analysis of structure of dendrolitic cone filament. (A) Scanning electron microscopy images from an individual dendrolitic cone filament, pre- (left) and post- (right) treatment with one drop of 1M HCl to remove calcium carbonate. (B)-(C) Grains intertwined between filamentous microorganisms, prior to treatment with 1M HCl. (D)-(E) Post-treatment observation of voids.

D.2.4 Microbial community assemblage in mats at Little Hot Creek

Several Operational Taxonomic Units OTUs were more abundant in the Cone Tip compared to the Cone Middle. These included genera most closely related to the Cyanobacteria

Leptolyngbya (1 to 6%), and Phormidium (~13 to 26%), an uncultured Planctomycetes (~1 to

5%), and the Spirochaetae genus Leptospira (~3 to 11%). Lower relative abundances in members of the Chloroflexi (~16 to 5%), and Bacteriodetes (~3 to 0.2%) were noted in the Cone Tip compared to the Cone Middle, along with a marked decrease in members of the Cyanobacteria genus Synechococcus (~43 to 30%). Comparative community analyses based on 16S rRNA library sequencing of dendrolitic cone subsections and the Adjacent Mat (top layer of mat adjacent to growing dendrolitic cones) showed remarkably similar microbial communities

(Figure D.4A). Although no statistically significant difference was found between samples from

163

Figure D.4. Bacterial community composition from Cone Pool East. (A) Relative abundance plots of different sections of the dendrolitic cones, including the surrounding bacterial communities present in the Adjacent Mat and Water. Analysis was performed using 16S rRNA gene sequencing. (B) Cluster dendrogram between communities from Cone Pool East, using Bray-Curtis dissimilarity and SIMPROF test. Significant clustering (p < 0.05) is indicated with coloured branches.

164 dendrolitic cone subsections and the Adjacent Mat using -diversity analyses (data not shown), a dissimilarity analysis was performed to determine if communities훽 were different across the dendrolitic cone structure. Cluster dendogram representation showed that samples from Cone Tip were different from surrounding water, Cone Middle and Adjacent Mat (Figure D.4B). Branches of the same color in the analysis represent significant clustering evaluated with the SIMPROF test. This distinctive shift in community composition of Cone Tip suggests that microbial populations with higher abundance of photosynthetic, phototrophic and filamentous bacteria may play a role in formation or stabilization of the dendrolitic cone structure.

Analysis of 16S rRNA libraries showed that members of the Phyla Bacteroidetes and

Chloroflexi, and all Cyanobacteria were more abundant in the microbial communities of LHC mat samples (Figure D.5). Similar relative abundances of these Phyla (~19%, ~10%, and ~40% respectively) were found in the top layer of the Outflow Pool microbial mat and the top layer of mat from the west side of Cone Pool (Cone Pool West). When compared to the Adjacent Mat, the relative abundance of Bacteroidetes decreased (~-17% decrease to ~2%), but the Chloroflexi

(~+10% to ~20%) and Cyanobacteria (~+10% to ~50%) increased (Figure D.5A). Values presented here are averages of relative abundance percentages from triplicate samples.

The microbial communities from each sample type were compared via -diversity analyses and visualized using principal coordinate analysis (PCoA) (Figure D.5B). While훽 the communities between the Outflow Pool, Cone Pool West, and Adjacent Mat appear similar in composition (Figure D.5A), Adjacent Mat samples showed a distinct increase in Cyanobacteria

(the genus Synechococcus specifically; ~+30% shift) and therefore, these samples were separated from the rest in the ordination (Figure D.5B).

165

Figure D.5. Bacterial community composition of mats from Little Hot Creek. (A) Relative abundance plots of mats based on 16S rRNA gene sequencing from the Outflow Pool, Cone Pool West and Cone Pool East adjacent to the dendrolitic cones (Adjacent Mat). (B) Principal coordinate analysis (PCoA) of variance between communities from mats in LHC.

166 D.3 Discussion

The structure of the dendrolitic cones present in Cone Pool are morphologically distinct from any other modern-day conical or dendrolitic microbialite structures previously described in the literature. Notably, the dendrolitic cones lacked laminations, distinguishing them from laminated coniform stromatolites described at Antarctic lakes Untersee and Vanda (Andersen et al., 2011; Sumner et al., 2016), and Yellowstone National Park geothermal spring-associated mats and stromatolites (Berelson et al., 2011; Bosak et al., 2009, 2012; Reyes et al., 2013). The dendrolitic cone structures present in Cone Pool were also visually and morphologically different from preserved Archean and Proterozoic conical stromatolites (Allwood et al., 2006), which are laminated and do not show fractal-like branching fabrics. However, the distinctly arborescent appearance of the dendrolitic cones resembled a hybrid of modern-day pinnacles (Andersen et al., 2011) and ancient dendrolites/shrubs (Fraiser and Corsetti, 2003; Shapiro and Rigby, 2004).

A particularly striking difference between the dendrolitic cones present in Cone Pool and those found elsewhere in modern settings was that they were partially lithified, appearing firm and crunchy (Figure D.2) rather than soft and spongy (Andersen et al., 2011; Bosak et al., 2009;

Sumner et al., 2016). The crunchy nature of the dendrolitic cones, comprised predominantly of

CaCO3 (76.1% dry weight), is relevant to the potential preservation of these structures and thus use as a paleoenvironmental analogue. The high degree of mineralization may have permitted the apparent preservation of macro- and micro-structures following sampling, transportation and storage. However, SEM images presented here are not of sufficient magnification to determine the degree of lithification of individual microbial filaments. Higher magnification images of individual microbial filaments may enable this supposition to be reconciled, and this should be a priority in future studies.

167 SEM images clearly showed that the dendrolitic cones were comprised of microbial filaments. These filaments, in combination with calcium carbonate minerals, formed what appeared to resemble bridging structures, which likely provide structural support for the dendrolitic cone form (Figure D.2B). These bridging structures are not unlike what has been seen in the microbial component of certain speleothems described in a geothermal mine adit (Spear et al., 2007). The microbial filaments comprising the dendrolitic cones of Cone Pool had no well- defined, consistent spatial orientation. Instead, the microbial filaments appeared to be intertwined around calcium carbonate grains (Figures D.2 and D.3). Intertwined filamentous dendrolitic cones at LHC were remarkably different from modern siliceous stromatolites from Obsidian Pool

Prime, Yellowstone (Berelson et al., 2011; Mata et al., 2012; Pepe-Ranney et al., 2012; Spear and Corsetti, 2013), where filaments within the same lamination were oriented in a uniform direction. In both cases, however, the interlocking organization of filaments at both Cone Pool and Obsidian Pool Prime may give rise to firm, microbiotically built, physically stable macrostructures. This contrasts with the soft, non-lithified, pinnacle-like structures identified previously in both Yellowstone (Bosak et al., 2009) and Antarctica (Andersen et al., 2011;

Sumner et al., 2016). It has been suggested that physical entanglement of filaments generally increases mat cohesion and thus aids the development of such complex microbialite structures

(Gerbersdorf and Wieprecht, 2015; Walter et al., 1976) and perhaps leads to increased preservation potential.

D.3.1 Hourglass structures

SEM imaging revealed hour-glass shaped structures approximately ~50 μm in diameter defined by bundles of filaments surrounding voids (Figure D.2B). These structures are

168 remarkably similar in size and morphology to structures preserved in a laminated siliceous stromatolite from a Yellowstone geothermal spring (Mata et al., 2012), which have been interpreted as relics of oxygen-rich bubbles generated during microbial photosynthesis. The high relative abundance of members of the phylum Cyanobacteria combined with an increased abundance of members of the phylum Planctomycetes in the Cone Tip could be related to the hour-glass structures. It has been found that the genus Isosphaera, a strict aerobic organism, can harbor gas vesicles that may help to maintain the buoyancy of cells in warm water conditions

(Youssef and Elshahed, 2014). This has also been described for benthic Cyanobacteria, which have been shown to form photosynthetic bubbles (Knapp et al., 1983). These bubbles may be preserved within mats upon CaCO3 precipitation (Figure D.2B), similar to the model described by Bosak et al. (2010) for photosynthetic mats in Yellowstone. The preservation of hour-glass structures in dendrolitic cones at LHC (and similar to stromatolites at Yellowstone) suggests that the rate of mineralization of bacterial filaments is on the timescale of the lifespan of a bubble, perhaps as short as a diel cycle. Oxygen production and bubble-forming potential is highest during daylight hours, while gas dissolution/utilization occurs during nighttime hours. In the

Obsidian Pool Prime example of stromatolites in Yellowstone (Mata et al., 2012), rapid silicification of filaments occurred as hydrothermal fluids rich in dissolved silica cooled near the pool margins, preserving the hourglass structures likely surrounding oxygen bubbles. Similarly, the Cone Pool hourglass structures are preserved, but with calcium carbonate. Fresh geochemical constituents are continuously delivered from the spring source.

169 D.3.2 Carbonate Precipitation and Dendrolitic Cone Lithification

The Cone Pool water was supersaturated with calcium carbonate (ΩCa = 4.08). Images prior to and after acidification with HCl show how a loss of structural integrity was caused by the dissolution of CaCO3 grains that were previously bound between inter-twined microbial filaments (Figure D.3B, C, D and E). This indicates a direct link between the presence of carbonate grains and the structural integrity of dendrolitic cones, and implies that mineralization of the structure is responsible for the resistance to deformation upon removal from the pool. It was difficult to determine unambiguously if the microbial community played a major role in the precipitation of minerals, for example through alteration of the micro-environment via metabolic processes, or enzymatic activation. However, mineralization increased with depth in the dendrolitic cones (Figure D.2A), suggesting progressive precipitation of CaCO3. Photosynthetic, chemosynthetic, thermophilic and heterotrophic microorganisms can induce biofilm calcification by actively mediating the formation of calcium carbonate in aquatic environments (Arp et al.,

2001; Pentecost and Viles, 1994), and may be critical in mediating the formation of dendrolitic conical microbialite structures (Sim et al., 2012). Results from 13C incorporation experiments suggested that photo-autotrophy was an active metabolism for microbes in Cone Pool. We detected an increase in organic matter carbon by 0.15 % per day through uptake of inorganic bicarbonate. The relative higher abundance of auto-fluorescent pigments in the Cone Tip and

Middle compared to the Cone Base (Figure D.2C) suggested an abundance of autotrophic microorganisms in the upper sections of the pinnacles. Photosynthetic activity in the Cone Tip may promote carbonate precipitation by removing CO2 from the spring water and increasing alkalinity (Equation 1):

170 - 2- + 106 CO2 + 16 NO3 + HPO4 + 122 H2O + 18 H → C106N16H263O110P + 138 O2 [1]

Although this equation is a gross simplification of photosynthetic carbon fixation reactions occurring in the mat, the saturation state of CaCO3 in the spring water is highly sensitive to changes in CO2 concentration and alkalinity (Figure D.6). Thus, photosynthetic activity in the

Figure D.6. Calcium carbonate saturation in water of Cone Pool. Alkalinity and TCO2 concentrations are represented on a fan diagram with the respective omega value for water adjacent to the dendrolitic cones (ΩCa).

171 dendrolitic cone may drive localized concentration gradients in water chemistry and encourage carbonate precipitation, driving increases in dendrolitic cone firmness. Microbial filaments and their extra-cellular secretions can also mediate crystal nucleation in geothermal springs by acting as nuclei upon which mineral growth may proceed (Fouke, 2001, 2011). Quantitative characterization of extracellular polymeric substances (EPS) in dendrolitic cone sections would provide additional insight on the role of EPS as nucleation points for carbonate precipitation, and we suggest this as a focus of future investigations into carbonate-based microbialite formations.

Further, the accumulation of calcium carbonate in the base of the dendrolitic cones could also result from secondary processes associated with the heterotrophic metabolisms of the microbial community in the dendrolitic cone base. Calcium carbonate precipitation could be favored in environments rich in organic matter, by ammonification, dissimilatory nitrate reduction,

2- - degradation of urea or uric acid and sulfate reduction. Production of CO3 , HCO3 , and ammonia

(for nitrogen metabolisms) or hydrogen sulfide (for sulfate reduction) may cause a pH change that induces carbonate precipitation (Stolz, 2000). The relative importance of individual microbial metabolisms on dendrolitic cone mineralization warrants further attention.

D.3.3 Model of Dendrolitic Cone Formation

Our δ13C data from the growth experiment shows that microorganisms in the Cone Tip are actively fixing dissolved inorganic carbon. The similarity in autotrophic growth rate between the

Cone Tip (0.15 % d-1) and the microbial mat in the Outflow Pool (0.174 % d-1) indicated that growth differences are not driving dendrolitic cone formation. However, in a low-flow environment, such as between filaments within the dendrolitic cones, the development of diffusive gradients on a micro-scale may promote the movement and migration of microbes.

172 Experimental evidence has suggested that slow diffusive transport in stagnant fluids in-between microbial filaments causes limited exchange of nutrients and promotes a growth response in modern cone-forming microbial communities (Petroff et al., 2010).

The distinct morphological features of the dendrolitic cones present in Cone Pool suggest that biological activity may exert a control on the formation of these structures. The observed differences between the microbial communities found in the Cone Tip and other locations, based on 16S rRNA gene sequence analysis, support the hypothesis that biological activity may be a controlling factor on dendrolitic cone formation and lithification. Our data show that filamentous organisms with the potential for photo- and chemo- taxis are more abundant in the tip of the dendrolitic cone than in the Adjacent Mat. OTUs most closely related to the Cyanobacterial genus Leptolyngbya were more abundant in the Cone Tip relative to the Cone Middle and

Adjacent Mat. Members of this same genus were abundant in coniform mat structures in

Yellowstone geothermal springs and other incipient microbialite structures in aqueous environments (Andersen et al., 2011; Bosak et al., 2012; Lau et al., 2005; Reyes et al., 2013;

Walter et al., 1976). Furthermore, an OTU in the Cone Pool (between 1-10% relative abundance) most closely related to Gloeomargarita lithophora (Figure D.7, OTU 325) was identified.

Previous research has shown it to be capable of internal biomineralization of carbonate grains

(Couradeau et al., 2012). More recently Benzerara et al. (2014) found that internal biomineralization of carbonate was much more common in cyanobacteria than previously thought, including some strains of Synechococcus. While we have no evidence of this type of biomineralization in the Cone Pool system, the omnipresence of Synechococcus may allude to this type of process. The potential role of internal mineralization on microbialite formation warrants further investigation in future studies. Moreover, members of the Planctomycetes genus

173

Figure D.7. Phylogenetic tree of the top 25 OTUs represented across LHC sample types. known to be able to degrade high molecular weight organic matter and biopolymers (Cottrell and Kirchman, 2000; Ruvindy et al., 2016) and are often found, as they are at Cone Pool, in close relation to Cyanobacteria and other primary producers (Warden et al., 2016; Wong et al., 2015). The possible symbiotic relationship between primary producers such as Cyanobacteria and heterotrophic Bacteroidetes at Cone Pool warrants further exploration, since these groups may contribute to biologically induced lithification (Stolz, 2000).

174

175 Isosphaera, a non-Cyanobacterial organism with phototactic capabilities (Giovannoni et al.,

1987), was found in low abundance (file D.S2 in Appendix E, OTU 114). Bacteroidetes, which are present in 16S rRNA sequences throughout LHC, are present in several other lithifying microbialite systems (Warden et al., 2016; Wong et al., 2015). These organisms are

D.3.4 Relevance to Ancient Microbialites

Unlike the conical incipient stromatolites found in Lake Untersee, Antarctica (Andersen et al., 2011), and modern stromatolites from Obsidian Pool Prime, Yellowstone (Berelson et al.,

2011; Mata et al., 2012; Pepe-Ranney et al., 2012; Spear and Corsetti, 2013), the dendrolitic cones in Cone Pool are not laminated, and thus comparisons with the most common conical stromatolites (e.g., Conophyton) is not warranted. However, the surface expression of the cm- scale dendrolitic cones with mm-scale protrusions does appear to resemble fractal-like microbial fabrics from the geologic record, e.g. carbonate shrubs and/or dendrolites (Fraiser and Corsetti,

2003; Shapiro and Rigby, 2004) (Figure D.1E). Interestingly, the formation of carbonate shrubs/dendrolites is commonly attributed to calcimicrobes of unknown affinity, or coccoidal bacteria of some kind (Chafetz and Guidry, 1999). The Cone Pool examples demonstrate that shrub-like morphologies can originate from filamentous microbes as well. These incipient microbialite dendrolitic cones may constitute a modern analogue to address how complex microbialite structures are formed, become part of the rock record, and also inform subsequent interpretation of paleoenvironmental conditions and biological evolution through fossilized stromatolites and microbialites.

176 D.4 Methods

D.4.1 Study site

The Cone Pool is part of the LHC geothermal spring system located within the Long

Valley Caldera, Mono County, California, USA (37°41'N, 118°50'W). The site has been described previously in Vick et al (2010). LHC is in a mixed geological setting of Holocene alluvium silts, sands and gravel, lacustrine sediments, and Pleistocene sandstone and conglomerates. The Cone Pool is slightly northeast of the accompanying geothermal spring outlet in that same area.

D.4.2 Sampling

All samples and measurements were taken on 26th June 2015. Samples of each dendrolitic cone were collected for molecular and microscopic analyses using sterile scalpels (Sklar Surgical

Instruments, West Chester, PA, USA). Each dendrolitic cone was sectioned into the Cone Tip,

Cone Middle, Cone Base, and Adjacent Mat (Figure D.1). Each sub-sample was bifurcated, where half were preserved for microscopy, and half were used for DNA extraction. Samples used for DNA extraction were placed in ZR BashingBead™ (Zymo Research Corp., Irvine, CA, USA) lysis tubes containing 750 µL of Xpedition™ (Zymo Research Corp., Irvine, CA, USA)

Lysis/Stabilization Solution, and homogenized for 45 seconds in the field. For molecular analyses of the surrounding water community, approximately 2.4 L of spring water was filtered through triplicate 25 mm 0.22 µm polyethersulfone (PES) filters (Merck Millipore Corp.,

Darmstadt, Germany). Each filter was removed aseptically and transferred to ZR BashingBead™ lysis tubes and preserved as above.

Fluid samples for water chemistry were taken in triplicate by drawing up approximately 12

177 mL of spring water from the center of the stream closest to the station marker, and then filtering it through a 25 mm 0.45 µm PES filter (Merck Millipore Corp., Darmstadt, Germany) into 15 mL polypropylene conical tubes (Corning, Inc., Corning, NY, USA). A volume of 4 ml of filtered spring water was added to pre-weighed, evacuated Exetainer® (Labco Limited,

Lampeter, Ceredigion, UK) vials for TCO2, and 2 dram vials (Fisher Scientific, Waltham, MA,

USA) were overfilled and capped for cation analyses. All samples were stored on ice and transported for further lab analyses.

Samples for microscopic analyses were transported on ice, and refrigerated for 4 days before confocal microscopy analysis. Samples were then imaged by SEM 10 days later (a total of

14 days after sampling).

D.4.3 Water chemistry

Water temperature, pH and oxygen concentration were measured using a SevenGo Duo proTM (Mettler-Toledo, LLC, Columbus, OH, USA). Exetainer® vials containing water samples for TCO2 concentration analyses were weighed and the vial masses subtracted to determine water volume. Fluids were then acidified with 10% phosphoric acid using an Automate®

Carbonate Preparation device (Automate FX, Inc., Bushnell, FL, USA) in order to liberate all inorganic carbon as gaseous CO2. Produced CO2 was passed into a Picarro® G-2121i Cavity

Ringdown Spectrometer (CRDS) (Picarro Inc., Santa Clara, CA, USA) and the TCO2 content determined by comparison with NBS 915b pure carbonate standards. The saturation state of stream water from Cone Pool and the Outflow Pool with respect to calcium carbonate (ΩCa) was calculated using measured calcium ion concentrations, temperature, a saturation product (Ksp) of

9.81x10-9, and PHREEQC modelling software (Parkhurst, 1995).

178 D.4.4 Inorganic and organic carbon, and 13C bicarbonate uptake

Samples of individual dendrolitic cones with the underlying mat from Cone Pool, and the microbial mat in the Outflow Pool, were cored with a 10 mm plastic drinking straw and placed into individual test tubes (~25 ml). Inorganic C content of dendrolitic cones was determined by first drying portions of the dendrolitic cone, grinding and weighing 20-40 mg into an Exetainer® vial with a septum cap. Tubes were evacuated with vacuum, acidified with 10% phosphoric acid and the resultant CO2 analyzed using Picarro® G-2121i CRDS (Picarro Inc., Santa Clara, CA,

USA); specific methods are described in Subhas et al. (2015). Reagent grade CaCO3 served as a standard. Organic C content of dendrolitic cones was determined by the difference between Total

C and Inorganic C (weight %). Total C was determined by weighing 2-10 mg of dried powdered dendrolitic cone material into an aluminum-foil cup and burning with O2 in a Costech Elemental

Analyzer. The CO2 produced in combustion was determined quantitatively by the Picarro method (above). An internal standard of San Pedro Channel mud (San Pedro, California) was used to calibrate the % C.

For growth rate experiments, 0.22 µm filtered spring water from Cone Pool was amended with 13C labeled sodium bicarbonate to a final concentration of 4-5 mM and δ13C of 1800 per mil. The spring water was then added to 10 ml test tubes to fill completely without headspace and closed with a rubber stopper. Six different dendrolitic cones were incubated with and without the addition of 100 µL saturated mercury chloride + 300 mM sodium azide as an abiotic control. The vials were incubated at 40° C for 24 hours in a 12 hr light-dark cycle. Following the incubation, the top one third of the dendrolitic cone (excluding any mat material) was removed and acidified with 1M HCl to dissolve calcium carbonate. The sample was then centrifuged at

3000 x g for 5 min in 15 ml tubes and washed with PBS buffer three times to remove all

179 inorganic carbon. The resulting biological pellets were dried overnight at 60°C, ground in an agate mortar and pestle, weighed to 2-3 mg, and then packaged into aluminum foil cups. As above, the 13C/12C ratio of control and experimental samples were determined by Picarro analysis. An isotope mass balance (using 13C/12C ratios) was constructed to establish the amount of spiked bicarbonate uptake into the biomass. Replicate measurements provide confidence of precision to ± 0.01 per mil (1 SD). In all cases, the poisoned material was of unchanged isotopic value and the non-poisoned organic carbon was isotopically heavier. In all analyses, standards were run that relate the amount of organic carbon to the ppm of CO2 determined by the Picarro®

2 G-2121i CRDS and the standard curve was linear over a range of 8000 ppm CO2 with R >

0.999. The calibration of carbon isotopes, standardized using U.S. Geological Survey (USGS) 40 standard material, yielded an uncertainty of <0.1 per mil through a range of 1500 to 7500 ppm

13 CO2. All samples delivered CO2 within the range of standards. Rates of C uptake were converted to an estimated rate of autotrophic growth.

D.4.5 Microscopy and morphological characterization

For confocal microscopy, a dendrolitic cone dissected according to Figure D.1E was used.

Wet samples were mounted onto clean glass slides, covered with square cover slips, and sealed with nail polish. Samples were observed at 400x magnification on a Leica TCS SP2 Inverted

Scanning Confocal Microscope (Leica Microsystems Heidelberg GmbH, Heidelberg, Germany).

SEM microscopy was performed using both a fully dissected dendrolitic cone to determine structure and a full dendrolitic cone filament to analyze the contribution of minerals to the rigid structure and appearance of the dendrolitic cone. Samples were mounted onto an aluminum stage and observed with a Hitachi TM-1000 environmental SEM (Hitachi Ltd., Japan). Imaging was

180 performed prior to and post acidification with 1M HCl.

D.4.6 DNA extraction and 16S rRNA gene library sequencing

DNA extraction was performed using an Xpedition™ Soil/Fecal DNA MiniPrep kit according to manufacturer’s instructions (Zymo Research Corp., Irvine, CA, USA). Extracted

DNA was amplified using primers that spanned the V4 region of the 16S rRNA gene between positions 519 and 802 (Escherichia coli numbering), producing a product of approximately 266 bp. The primer pair represents a broad distribution of both the Bacteria and Archaea (Klindworth et al., 2013). The forward primer (M13L-519F: 5′- GTA AAA CGA CGG CCA GCA CMG

CCG CGG TAA -3′) contains the M13 forward primer (in bold), followed by the 16S rRNA gene-specific sequence (underlined) to allow for barcoding of each sample in a separate reaction(Stamps et al., 2016). The reverse primer (785R: 5′-TAC NVG GGT ATC TAA TCC-3′) was taken directly from the reverse primer “S-D-Bact07850b-A-18” in Klindworth et al. (2013).

Each 50 µL reaction mixture consisted of: 1X 5 PRIME HOT master mix (5 PRIME Inc.,

Gaithersburg, MD), 0.2 µM of each primer, and molecular grade water. A volume of 4 µL of extracted template DNA was added to each reaction. Polymerase chain reaction (PCR) cycling was carried out as previously described(Stamps et al., 2016). Positive (E. coli) and negative (no template) controls were also amplified along with sample template reactions. The amplified

DNA molecules were then purified using AmpureXP paramagnetic beads (Beckman Coulter

Inc., Indianapolis, IN, USA) at a final concentration of 0.8 x v/v. A second, six cycle PCR was used to add a unique 12 bp barcode(Hamady et al., 2008) to each previously amplified sample using a forward primer containing the barcode+M13 forward sequence (5′-3′) and the 785R primer (See Appendix E, D.S1). The final barcoded PCR products were again cleaned using

181 AmpureXP paramagnetic beads at a final concentration of 0.8X, quantified using the QuBit dsDNA HS assay (Life Technologies, Carlsbad, CA, USA), pooled in equimolar amounts, and concentrated to a final volume of 80 μL using two Amicon® Ultra-0.5 mL 30K Centrifugal

Filters (EMD Millipore, Billerica, MA, USA).

The final pooled library was run on the Illumina MiSeq platform (Illumina, San Diego,

CA, USA) using PE250 V2 chemistry. After sequencing, reads were merged and de-multiplexed using QIIME (Caporaso et al., 2010b), filtered by quality, clustered into operation taxonomic units (OTUs), and chimera checked using VSEARCH (Rognes et al., 2016). OTU taxonomy was assigned using UCLUST(Edgar, 2010) and the SILVA database (Release 123; Pruesse et al.,

2007)). Representative sequences were aligned using pyNAST(Caporaso et al., 2010a) against an aligned version of the SILVA r123 database. A phylogenetic tree was created using FastTree

(Price et al., 2010) for use in community composition analyses. Differences in community composition were estimated using weighted UniFrac indices (Lozupone and Knight, 2005).

Sample libraries were subsampled to 12,500 reads to generate a weighted UniFrac distance matrix in order to compare microbial diversity between sites. A mapping file is included in

Appendix E as D.S1, and the commands used to produce the final BIOM file are publicly available at https://doi.org/10.5281/zenodo.582679. Beta diversity PCoA plots were generated using the R package phyloseq (McMurdie and Holmes, 2013), and analysis of multivariate homogeneity of group variances was tested using the PERMDISP2 test (Anderson et al., 2006) through QIIME using the R package “vegan” (Dixon, 2003). Hierarchical clustering

(dendrogram) was performed using the R package “clustig” (Whitaker and Christman, 2014).

Cluster significance was also determined within this package using a similarity profile analysis

(SIMPROF; Clarke et al., 2008).

182 Secondary alignment was produced using SILVA SINA. Post alignment OTUs were added into the r123 SILVA database using ARB (Westram et al., 2011). After OTUs were added into the tree, near neighbors were selected, including closely related cultivated representatives, if possible. The selected sequences were exported in FASTA format, and imported into MEGA v7.0.16 (Kumar et al., 2016). The alignment was filtered to only include the sequence region represented in OTU sequences. A maximum likelihood tree was generated using the Tamura-Nei method (Tamura and Nei, 1993), and 500 bootstrap replicates. Changes in relative abundance across sample types were statistically tested using a multivariate of homogeneity of group variances test (PERMDISP2; Anderson et al., 2006).

D.5 Data availability

After sequencing, raw reads were deposited into the NCBI sequencing read archive (SRA) under the accession number SRX2830741 (https://www.ncbi.nlm.nih.gov/sra). The mapping file for samples and corresponding barcodes can be found in Appendix E, D.S1.

Acknowledgements

We would like to thank student participants and instructors from the 2015 International

Geobiology Course, who assisted with sample collection and laboratory analyses, and insightful discussion of the results: Emma Bertran, Victoria Petryshyn, Olivia Piazza, Russell Shapiro, Joy

Boungiorno, Luoth Chou, Sharon Grim, Megan Krusor, Gabriela Libanori, Katie Rempfert,

Danielle Santiago, Lennart van Maldegem, Dylan Wilmeth, Feifei Zhang, Laura Zinke, Amber

Brown and Ann Close. Ron Ormeland originally suggested Cone Pool as a field site for the

International Geobiology Course. Stephen Karl assisted with microscopy. Nick Rollins

183 conducted the Picarro measurements and data reduction. Benjamin Glasner assisted with bioinformatic analysis. DNA sequencing services were provided by the Oklahoma Medical

Research Foundation, and data were archived by the OU Supercomputing Center for Education

& Research (OSCER). A research permit was granted to J.R.S. from the U.S. Forest Service

(Permit #MLD15053) Inyo District, Bishop, California, to conduct fieldwork and sample the

Cone Pool / LHC system.

D.6 References

Allwood, A. C., Walter, M. R., Kamber, B. S., Marshall, C. P., and Burch, I. W. (2006). Stromatolite reef from the Early Archaean era of Australia. Nature 441, 714–718. doi:10.1038/nature04764.

Andersen, D. T., Sumner, D. Y., Hawes, I., Webster-Brown, J., and Mckay, C. P. (2011). Discovery of large conical stromatolites in Lake Untersee, Antarctica. Geobiology 9, 280– 293. doi:10.1111/j.1472-4669.2011.00279.x.

Anderson, M. J., Ellingsen, K. E., and McArdle, B. H. (2006). Multivariate dispersion as a measure of beta diversity. Ecol. Lett. 9, 683–693. doi:10.1111/j.1461-0248.2006.00926.x.

Arp, G., Reimer, a, and Reitner, J. (2001). Photosynthesis-induced biofilm calcification and calcium concentrations in Phanerozoic oceans. Science 292, 1701–1704. doi:10.1126/science.1057204.

Benzerara, K., Skouri-Panet, F., Li, J., Férard, C., Gugger, M., Laurent, T., et al. (2014). Intracellular Ca-carbonate biomineralization is widespread in cyanobacteria. Proc. Natl. Acad. Sci. 111, 10933–10938. doi:10.1073/pnas.1403510111.

Berelson, W. M., Corsetti, F. A., Pepe-Ranney, C., Hammond, D. E., Beaumont, W., and Spear, J. R. (2011). Hot spring siliceous stromatolites from Yellowstone National Park: Assessing growth rate and laminae formation. Geobiology 9, 411–424. doi:10.1111/j.1472- 4669.2011.00288.x.

Bosak, T., Bush, J. W. M., Flynn, M. R., Liang, B., Ono, S., Petroff, A. P., et al. (2010). Formation and stability of oxygen-rich bubbles that shape photosynthetic mats. Geobiology 8, 45–55. doi:10.1111/j.1472-4669.2009.00227.x.

Bosak, T., Knoll, A. H., and Petroff, A. P. (2013). The Meaning of Stromatolites. Annu. Rev. Earth Planet. Sci. 41, 21–44. doi:10.1146/annurev-earth-042711-105327.

184

Bosak, T., Liang, B., Sim, M. S., and Petroff, A. P. (2009). Morphological record of oxygenic photosynthesis in conical stromatolites. Proc. Natl. Acad. Sci. U. S. A. 106, 10939–10943. doi:10.1073/pnas.0900885106.

Bosak, T., Liang, B., Wu, T. D., Templer, S. P., Evans, A., Vali, H., et al. (2012). Cyanobacterial diversity and activity in modern conical microbialites. Geobiology 10, 384–401. doi:10.1111/j.1472-4669.2012.00334.x.

Buick, R. (1992). The antiquity of oxygenic photosynthesis: evidence from stromatolites in sulphate-deficient Archaean lakes. Science 255, 74–7. doi:10.1126/science.11536492.

Caporaso, J. G., Bittinger, K., Bushman, F. D., Desantis, T. Z., Andersen, G. L., and Knight, R. (2010a). PyNAST: A flexible tool for aligning sequences to a template alignment. Bioinformatics 26, 266–267. doi:10.1093/bioinformatics/btp636.

Caporaso, J. G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F. D., Costello, E. K., et al. (2010b). QIIME allows analysis of high-throughput community sequencing data. Nat. Methods 7, 335–336. doi:10.1038/nmeth.f.303.

Chafetz, H. S., and Guidry, S. A. (1999). Bacterial shrubs, crystal shrubs, and ray-crystal shrubs: bacterial vs. abiotic precipitation. Sediment. Geol. 126, 57–74. doi:http://dx.doi.org/10.1016/S0037-0738(99)00032-9.

Clarke, K. R., Somerfield, P. J., and Gorley, R. N. (2008). Testing of null hypotheses in exploratory community analyses: similarity profiles and biota-environment linkage. J. Exp. Mar. Bio. Ecol. 366, 56–69. doi:http://dx.doi.org/10.1016/j.jembe.2008.07.009.

Cottrell, M. T., and Kirchman, D. L. (2000). Natural assemblages of marine proteobacteria and members of the Cytophaga-flavobacter cluster consuming low- and high-molecular-weight dissolved organic matter. Appl. Environ. Microbiol. 66, 1692–1697. doi:10.1128/AEM.66.4.1692-1697.2000.

Couradeau, E., Benzerara, K., Gérard, E., Moreira, D., Bernard, S., Brown, G. E., et al. (2012). An Early-Branching Microbialite Cyanobacterium Forms Intracellular Carbonates. Science (80-. ). 336, 459 LP-462.

Dixon, P. (2003). VEGAN, a package of R functions for community ecology. J. Veg. Sci. 14, 927–930. doi:10.1111/j.1654-1103.2003.tb02228.x.

Edgar, R. C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461. doi:10.1093/bioinformatics/btq461.

Flannery, D. T., and Walter, M. R. (2012). Archean tufted microbial mats and the Great Oxidation Event: new insights into an ancient problem. Aust. J. Earth Sci. 59, 1–11. doi:10.9774/GLEAF.978-1-909493-77-3_30.

185

Fouke, B. W. (2001). Depositional Facies and Aqueous-Solid Geochemistry of Travertine- Depositing Hot Springs (Angel Terrace, Mammoth Hot Springs, Yellowstone National Park, U.S.A.): Reply. J. Sediment. Res. 71, 497–500. doi:10.1306/2DC40959-0E47-11D7- 8643000102C1865D.

Fouke, B. W. (2011). Hot-spring Systems Geobiology: Abiotic and biotic influences on travertine formation at Mammoth Hot Springs, Yellowstone National Park, USA. Sedimentology 58, 170–219. doi:10.1111/j.1365-3091.2010.01209.x.

Fraiser, M. L., and Corsetti, F. A. (2003). Neoproterozoic Carbonate Shrubs: Interplay of Microbial Activity and Unusual Environmental Conditions in Post-Snowball Earth Oceans. Palaios 18, 378 LP-387.

Gerbersdorf, S. U., and Wieprecht, S. (2015). Biostabilization of cohesive sediments: Revisiting the role of abiotic conditions, physiology and diversity of microbes, polymeric secretion, and biofilm architecture. Geobiology 13, 68–97. doi:10.1111/gbi.12115.

Giovannoni, S. J., Schabtach, E., and Castenholz, R. W. (1987). Isosphaera pallida, gen. and comb. nov., a gliding, budding eubacterium from hot springs. Arch. Microbiol. 147, 276– 284. doi:10.1007/BF00463488.

Grotzinger, J. P., and Knoll, A. H. (1999). Stromatolites in Precambrian carbonates: Evolutionary mileposts or environmental dipsticks? Annu. Rev. Earth Planet. Sci. 27, 313– 358. doi:DOI 10.1146/annurev.earth.27.1.313.

Hamady, M., Walker, J. J., Harris, J. K., Gold, N. J., and Knight, R. (2008). Error-correcting barcoded primers for pyrosequencing hundreds of samples in multiplex. Nat. Methods 5, 235–237. doi:10.1038/nmeth.1184.

Klindworth, A., Pruesse, E., Schweer, T., Peplies, J., Quast, C., Horn, M., et al. (2013). Evaluation of general 16S ribosomal RNA gene PCR primers for classical and next- generation sequencing-based diversity studies. Nucleic Acids Res. 41, e1–e1. doi:10.1093/nar/gks808.

Knapp, E. B., Elliott, L. F., and Campbell, G. S. (1983). Carbon, Nitrogen and Microbial Biomass Interrelationships during the Decomposition of Wheat Straw - a Mechanistic Simulation-Model. Soil Biol. Biochem. 15, 455–461. doi:Doi 10.1016/0038-0717(83)90011- 1.

Kumar, S., Stecher, G., and Tamura, K. (2016). MEGA7: Molecular Evolutionary Genetics Analysis Version 7.0 for Bigger Datasets. Mol. Biol. Evol. 33, 1870–1874. doi:10.1093/molbev/msw054.

Lau, E., Nash, C. Z., Vogler, D. R., and Cullings, K. W. (2005). Molecular diversity of cyanobacteria inhabiting coniform structures and surrounding mat in a Yellowstone hot

186 spring. Astrobiology 5, 83–92. doi:10.1089/ast.2005.5.83.

Lindsay, M. R., Anderson, C., Fox, N., Scofield, G., Allen, J., Anderson, E., et al. (2017). Microbialite response to an anthropogenic salinity gradient in Great Salt Lake, Utah. Geobiology 15, 131–145. doi:10.1111/gbi.12201.

Lozupone, C., and Knight, R. (2005). UniFrac: A new phylogenetic method for comparing microbial communities. Appl. Environ. Microbiol. 71, 8228–8235. doi:10.1128/AEM.71.12.8228-8235.2005.

Mata, S. A., Harwood, C. L., Corsetti, F. A., Stork, N. J., Eilers, K., Berelson, W. M., et al. (2012). Influence of gas production and filament orientation on stromatolite microfabric. Palaios 27, 206–219. doi:10.2110/palo.2011.p11-088r.

McMurdie, P. J., and Holmes, S. (2013). phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS One 8, e61217. doi:10.1371/journal.pone.0061217.

Parkhurst, D. L. (1995). User's guide to PHREEQC : a computer program for speciation, reaction-path, advective-transport, and inverse geochemical calculations. Lakewood, Colo. : U.S. Dept. of the Interior, U.S. Geological Survey ; Denver, CO : Earth Science Information Center, Open-File Reports Section [distributor], 1995.

Pentecost, A., and Viles, H. (1994). a Review and Reassessment of Travertine Classification. Géographie Phys. el Quat. 48, 305–314. doi:10.7202/033011ar.

Pepe-Ranney, C., Berelson, W. M., Corsetti, F. A., Treants, M., and Spear, J. R. (2012). Cyanobacterial construction of hot spring siliceous stromatolites in Yellowstone National Park. Environ. Microbiol. 14, 1182–1197. doi:10.1111/j.1462-2920.2012.02698.x.

Petroff, A. P., Sim, M. S., Maslov, A., Krupenin, M., Rothman, D. H., and Bosak, T. (2010). Biophysical basis for the geometry of conical stromatolites. Proc. Natl. Acad. Sci. U. S. A. 107, 9956–9961. doi:10.1073/pnas.1001973107.

Price, M. N., Dehal, P. S., and Arkin, A. P. (2010). FastTree 2 - Approximately maximum- likelihood trees for large alignments. PLoS One 5. doi:10.1371/journal.pone.0009490.

Pruesse, E., Quast, C., Knittel, K., Fuchs, B. M., Ludwig, W., Peplies, J., et al. (2007). SILVA: A comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 35, 7188–7196. doi:10.1093/nar/gkm864.

Reyes, K., Gonzalez, N. I., Stewart, J., Ospino, F., Nguyen, D., Cho, D. T., et al. (2013). Surface orientation affects the direction of cone growth by Leptolyngbya sp. strain C1, a likely architect of coniform structures Octopus Spring (Yellowstone National Park). Appl. Environ. Microbiol. 79, 1302–8. doi:10.1128/AEM.03008-12.

187 Rognes, T., Flouri, T., Nichols, B., Quince, C., and Mahé, F. (2016). VSEARCH: a versatile open source tool for metagenomics. PeerJ 4, e2584. doi:10.7717/peerj.2584.

Ruvindy, R., White, R. A., Neilan, B. A., and Burns, B. P. (2016). Unravelling core microbial metabolisms in the hypersaline microbial mats of Shark Bay using high-throughput metagenomics. ISME J. 10, 183–96. doi:10.1038/ismej.2015.87.

Shapiro, R. S., and Rigby, J. K. (2004). First occurrence of an in situ Anthaspidellid sponge in a dendrolite mound (Upper Cambrian; Great Basin, USA). J. Paleontol. 78, 645–650. doi:DOI: 10.1666/0022-3360(2004)078<0645:FOOAIS>2.0.CO;2.

Sim, M. S., Liang, B., Petroff, A. P., Evans, A., Klepac-Ceraj, V., Flannery, D. T., et al. (2012). Oxygen-Dependent Morphogenesis of Modern Clumped Photosynthetic Mats and Implications for the Archean Stromatolite Record. Geosciences 2, 235–259. doi:10.3390/geosciences2040235.

Spear, J. R., Barton, H. A., Robertson, C. E., Francis, C. A., and Pace, N. R. (2007). Microbial community biofabrics in a geothermal mine adit. Appl. Environ. Microbiol. 73, 6172–6180. doi:10.1128/AEM.00393-07.

Spear, J. R., and Corsetti, F. . (2013). The Evolution of Geobiology in the Context of Living Stromatolites. Web Geol. Sci. Adv. Impacts Interact. Geol. Soc. Am. Spec. Pap. #500. doi:doi: 10.1130/2013.2500(17).

Stamps, B. W., Lyles, C. N., Suflita, J. M., Masoner, J. R., Cozzarelli, I. M., Kolpin, D. W., et al. (2016). Municipal solid waste landfills harbor distinct microbiomes. Front. Microbiol. 7. doi:10.3389/fmicb.2016.00534.

Stolz, J. F. (2000). “Structure of Microbial Mats and Biofilms,” in Microbial Sediments, eds. R. E. Riding and S. M. Awramik (Berlin, Germany: Springer), 1–8. doi:10.1007/978-3-662- 04036-2.

Subhas, A. V., Rollins, N. E., Berelson, W. M., Dong, S., Erez, J., and Adkins, J. F. (2015). A novel determination of calcite dissolution kinetics in seawater. Geochim. Cosmochim. Acta 170, 51–68. doi:10.1016/j.gca.2015.08.011.

Sumner, D. Y., Jungblut, A. D., Hawes, I., Andersen, D. J., Mackey, T. J., and Wall, K. (2016). Growth of elaborate microbial pinnacles in Lake Vanda, Antarctica. Geobiology 14, 556– 574. doi:10.1111/gbi.12188.

Tamura, K., and Nei, M. (1993). Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Mol. Biol. Evol. 10, 512–526.

Vick, T. J., Dodsworth, J. A., Costa, K. C., Shock, E. L., and Hedlund, B. P. (2010). Microbiology and geochemistry of Little Hot Creek, a hot spring environment in the Long

188 Valley Caldera. Geobiology 8, 140–154. doi:10.1111/j.1472-4669.2009.00228.x.

Walter, M. R., Bauld, J., and Brock, T. D. (1976). Chapter 6.2 Microbiology and Morphogenesis of Columnar Stromatolites (Conophyton, Vacerrilla) from Hot Springs in Yellowstone National Park. Dev. Sedimentol. 20, 273–310. doi:10.1016/S0070-4571(08)71140-3.

Warden, J. G., Casaburi, G., Omelon, C. R., Bennett, P. C., Breecker, D. O., and Foster, J. S. (2016). Characterization of Microbial Mat Microbiomes in the Modern Thrombolite Ecosystem of Lake Clifton, Western Australia Using Shotgun Metagenomics. Front. Microbiol. 7, 1064. doi:10.3389/fmicb.2016.01064.

Westram, R., Bader, K., Prüsse, E., Kumar, Y., Meier, H., Glöckner, F. O., et al. (2011). “ARB: A Software Environment for Sequence Data,” in Handbook of Molecular Microbial Ecology I (John Wiley & Sons, Inc.), 399–406. doi:10.1002/9781118010518.ch46.

Whitaker, D., and Christman, M. (2014). clustsig: Significant Cluster Analysis. R package version 1.1.

Wong, H. L., Smith, D. L., Visscher, P. T., and Burns, B. P. (2015). Niche differentiation of bacterial communities at a millimeter scale in Shark Bay microbial mats. Sci. Rep. 5, 15607. doi:10.1038/srep15607.

Youssef, N. H., and Elshahed, M. S. (2014). “The Phylum Planctomycetes,” in Prokaryotes Other Major Lineages Bact. Archaea, eds. E. Rosenberg, E. F. DeLong, S. Lory, E. Stackebrandt, and F. Thompson (Berlin: Springer), 759–810.

189

APPENDIX E

SUPPLEMENTAL ELECTRONIC FILES

The table below includes auxiliary files from thesis chapters that are too large to be put into in- line tables. It also contains electronic versions (PDF) of co-author papers.

Table E.1. Table of electronic files associated with Chapters and Appendices.

File Name Chapter Description of file Mapping file for 16S rRNA sequencing data published in A.S1 - Ch2_Mapping.xlsx Chapter 2 Chapter 2. This includes all relevant sequencing indexes for demultiplexing. Sequencing coverage excel file. This includes average A.S2 - sequencing statistics as well as Chapter 2 Ch2_Seq_coverage.xlsx overall sequencing read numbers for each sample reported in Chapter 2. Mapping file for 16S rRNA sequencing data published in C.S1 - Ch4_Mapping.xlsx Chapter 4 Chapter 4. This includes all relevant sequencing indexes for demultiplexing. PDF copy of co-author manuscript describing differences in anode type for Stoll_Manuscript.pdf NA microbial fuel cells and the potential for more cost effective solutions. PDF copy of co-author manuscript describing the discovery of unexpected and Lau_Manuscript.pdf NA rare cyclooctasulfur allotropes in the Canadian High Arctic at Borup Fiord Pass.

190 Table E.1 Table of electronic files associated with Chapters and Appendices (continued).

Mapping file for 16S rRNA sequencing data associated D.S1 - with the publication in Appendix D Appendix_D_Mapping.xlsx Appendix D. This includes all relevant sequencing indexes for demultiplexing. Excel table of absolute OTU abundances detected within D.S2 - Appendix D each sample from the Appendix_D_OTUs.xlsx published work in Appendix D.

191

APPENDIX F

COPYRIGHT FOR PUBLISHED WORKS

F.1 Open Access copyright permission for published article in Appendix D. https://www.nature.com/articles/s41522-017-0041-2#rightslink

192 F.2 Permission from all authors to reprint article from Appendix D for this dissertation.

193 F.2 Permission from all authors to reprint article from Appendix D for this dissertation (cont’d).

194 F.2 Permission from all authors to reprint article from Appendix D for this dissertation (cont’d).

195 F.2 Permission from all authors to reprint article from Appendix D for this dissertation (cont’d).

196 F.2 Permission from all authors to reprint article from Appendix D for this dissertation (cont’d).

197