1 Exploring Microbial and Functional Gene Diversity

EXPLORING MICROBIAL AND FUNCTIONAL GENE DIVERSITY IN MODERN MARINE THROMBOLITIC MAT COMMUNITIES

JENNIFER MARIE MOBBERLEY

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2013

To my parents

ACKNOWLEDGMENTS

I would like to acknowledge my advisor, Dr. Jamie Foster, for her mentorship and

guidance. Thank you for being there when I needed you and for giving me the freedom

to grow as a scientist. I am grateful to my committee members, Drs. Robert Ferl, Wayne

Nicholson, Eric Triplett, and Nian Wang for providing valuable suggestions and support.

I would like to thank Dr. Stephanie Havemann, Steven Ahrendt, Alexandra Duscher,

Artemis Louyakis, Maya Ortega, and Regine Pamphile for their technical assistance and

encouragement. I would also like to acknowledge Dr. Christina Khodadad for being

there to bounce ideas off of and helping keep me grounded.

This research was made possible by funding from the NASA Graduate Student

Researchers Project. I am also indebted to NASA Exobiology, UF Graduate Student

Council, the Department of Microbiology and Cell Science, and the Davidson fund for enabling me to present my research at international and national meetings. I would like to thank Dr. Oleksandr Moskalenko at UF’s Research Computing Center for his invaluable help with my data analysis and for giving me a little extra RAM when I needed it. Steven Ahrendt and Austin Richard-Davidson fixed my codes and helped me navigate the world of scripting.

None of this would have been possible without my parents, Belinda Mellon and

Daniel Mobberley who have provided constant love and encouragement. Thank you for believing in me even when you had no idea what I was doing. I also appreciate the support my extended family has provided over the years. Finally, I have to thank Amy

Long, David Shambaugh, Samantha Waters, and Sara Reyes for being there with unlimited support, jokes, and beverages.

Thank you all for making this dissertation possible.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 8

LIST OF FIGURES ...... 9

LIST OF ABBREVIATIONS ...... 11

ABSTRACT ...... 12

CHAPTER

1 LITERATURE REVIEW ...... 14

Microbialites ...... 14 Community Diversity in Microbialites ...... 15 The Biogeochemistry of Carbonate Biomineralization in Microbialites ...... 23 Highborne Cay Thrombolites as Model Microbialites ...... 28

2 COMPARATIVE MICROBIAL DIVERSITY ANALYSES OF MODERN MARINE THROMBOLITIC MATS BY BARCODED PYROSEQUENCING ...... 32

Introduction ...... 32 Materials and Methods ...... 34 Site Description, Sample Collection, and Microscopy ...... 34 Molecular Identification of Cyanobacterial Isolates from the Thrombolitic Mats ...... 35 Generation of Barcoded 16S Ribosomal RNA Gene Libraries ...... 37 Bioinformatic Analysis of Barcoded 16S Ribosomal RNA Gene Libraries ...... 39 Community Comparisons Between Thrombolitic Mat Types ...... 41 Results ...... 42 Morphological Characterization of Thrombolitic Mat Types ...... 42 Small Subunit rRNA Gene Diversity in Thrombolitic Mats ...... 44 Bacterial Community Composition ...... 45 Cyanobacterial Community Composition ...... 48 Archaea Community Structure ...... 50 Discussion ...... 51 Differences in Microbial Populations Between Thrombolitic Mat Types ...... 52 Cyanobacterial Diversity in Thrombolitic Mat Communities ...... 53 Low Diversity Of Archaea in Highborne Cay Thrombolitic Mats ...... 55 A Model for Microbial Succession in Thrombolitic Mat Types ...... 57

3 METABOLIC POTENTIAL OF LITHIFYING CYANOBACTERIA-DOMINATED THROMBOLITIC MATS ...... 71

Introduction ...... 71 Materials and Methods ...... 74 Thrombolitic Mat Sample Collection ...... 74 DNA Extraction and Illumina Sequencing ...... 75 Analysis of Thrombolitic Mat Metagenomes ...... 75 Comparison of Thrombolite Metagenome to Other Microbial Mat Habitats ...... 76 Metabolic Profiling with Phenotypic Microarrays ...... 77 Results ...... 78 Community Diversity Associated with Thrombolitic Mat Metagenome ...... 79 Overview of the Functional Genes of the Thrombolitic Mat Metagenome ...... 81 Comparison of Thrombolite Metagenome to Other Functional Metagenomes .. 84 Substrate Utilization Patterns Within Thrombolitic Microbial Mats ...... 85 Discussion ...... 87 Cyanobacterial Molecular Pathways Dominate the Thrombolite Metagenome ...... 88 Energy Metabolisms that Influence Carbonate Precipitation in Lithifying Microbial Mats ...... 90 Contribution of Archaea in Thrombolitic Mat Metabolic Cycling ...... 92 Gradients of Metabolic Potential Occur Within the Depth Profile of the Thrombolitic Mat ...... 94

4 METATRANSCRIPTOMIC SEQUENCING OF THROMBOLITIC MAT REVEALS DISTINCT SPATIAL GRADIENTS OF METABOLISMS ASSOCIATED WITH BIOLOGICALLY INDUCED MINERALIZATION ...... 113

Introduction ...... 113 Materials and Methods ...... 116 Sample Collection ...... 116 Microelectrode Oxygen Profiles ...... 116 Nucleic Acid Extraction and Purification ...... 117 DNA Sequencing and Read Processing ...... 118 De novo Metagenomic Assembly and Annotation ...... 119 Ribosomal RNA Transcript Analysis and Classification ...... 119 Messenger RNA Analysis ...... 120 Comparisons of Thrombolite Metagenome and Metatranscriptomes to Other Microbial Ecosystems...... 121 Results and Discussion ...... 122 Overview of Thrombolitic Mat Sequencing Libraries ...... 122 Active Members of the Thrombolitic Mat Community During Peak Photosynthetic Activity ...... 124 Functional Annotation of Gene Transcript Abundance Revealed Spatial Gradients of Gene Expression within the Thrombolitic Mat ...... 127 Conclusions ...... 134

5 SUMMARY OF RESEARCH ...... 149

LIST OF REFERENCES ...... 155

BIOGRAPHICAL SKETCH ...... 169

LIST OF TABLES

Table page

2-1 Bacterial and archaeal 16S rRNA amplicon diversity analyses of thrombolitic mats types of Highborne Cay, Bahamas ...... 61

2-2 Primers used for generating 16S rRNA barcoded libraries ...... 62

3-1 Sequence of metagenome sequencing and MG-RAST analysis ...... 97

3-2 Relative abundance of dominant metabolic pathways in thrombolitic mats ...... 98

3-3 Carbon substrate absorbance units of thrombolitic microbial mats ...... 99

3-4 Nitrogen substrate absorbance units of thrombolitic microbial mats ...... 101

3-5 Sulfur substrate absorbance units of thrombolitic microbial mats ...... 103

3-6 Phosphate substrate absorbance units of thrombolitic microbial mats ...... 104

4-1 Summary of analysis and annotation of the metagenomic and metatranscriptomic thrombolitic mat libraries ...... 136

LIST OF FIGURES

Figure page

2-1 Morphological characterization of thrombolitic mat communities ...... 63

2-2 Taxonomic diversity and community structure comparisons in thrombolitic mats ... 64

2-3 Comparisons of proteobacterial lineages in thrombolitic mats using a clustering- based approach ...... 65

2-4 Venn diagram representation of the OTU richness shared between bacterial 16S rRNA gene libraries from the four thrombolitic mat types ...... 66

2-5 Principal coordinate analysis of bacterial 16S ribosomal RNA gene libraries for the four thrombolitic mats using Fast UniFrac ...... 67

2-6 Phylogenetic tree of partial cyanobacterial 16S ribosomal RNA gene sequences from pyrosequencing libraries and clones generated from Dichothrix and Pleurocapsa isolates ...... 68

2-7 Phylogenetic tree of partial archaeal 16S ribosomal RNA gene sequences from archaeal 454 libraries ...... 69

2-8 Model for succession of Highborne Cay thrombolitic mat communities ...... 70

3-1 Modern thrombolites of Highborne Cay, Bahamas ...... 106

3-2 SEED subsystems overview of the three thrombolitic mat metagenomes ...... 107

3-3 Taxonomic distribution of the thrombolite metagenome based on MG-RAST Lowest Common Ancestor analysis ...... 108

3-4 Genus level distribution of cyanobacteria reads from the thrombolite metagenome based on MG-RAST Lowest Common Ancestor analysis ...... 109

3-5 MG-RAST functional assignment of thrombolite metagenome protein features ... 110

3-6 Comparison of the thrombolite metagenome with metagenomes of previously sequenced lithifying and non-lithifying microbial mat ecosystems ...... 111

3-7 Clustered heat map visualizing the substrate utilization patterns throughout the spatial profile of the thrombolitic mats using phenotypic microarrays ...... 112

4-1 The thrombolites of Highborne Cay, The Bahamas ...... 137

4-2 Relative abundance of dominant taxa within the thrombolitic mat metagenome and metatranscriptomes ...... 138

4-3 Relative abundance of the 20 most abundant taxa per phylogenetic domain based on expressed ribosomal RNA transcripts from the Total RNA extracts .. 139

4-4 Functional gene expression within a spatial profile of a thrombolitic mat ...... 140

4-5 Relative abundance of eukaryotic SEED subsystem reads in the assembled metagenome and metatranscriptomes ...... 141

4-6 Relative abundance of archaeal SEED subsystem reads in the assembled metagenome and metatranscriptomes ...... 142

4-7 Principle component analysis of functional gene abundances of thrombolitic mat metatranscriptomes based on SEED subsystems (Level 1) of thrombolitic metatranscriptomes and assembled metagenome ...... 143

4-8 Organisms involved in energy transformations within the thrombolitic mat during midday (12 PM) at peak photosynthesis ...... 144

4-9 Relative taxa abundance of RuBisCO reads based on KEGG pathway annotation in the assembled metagenome and metatranscriptomes ...... 145

4-10 Relative taxa abundance of nif reads based on KEGG pathway annotation in the assembled metagenome and metatranscriptomes ...... 146

4-11 Relative taxa abundance of oxidative phosphorylation reads based on the KEGG pathway in the assembled metagenome and metatranscriptomes ...... 147

4-12 Relative abundance of SEED carbohydrate subsystem reads associated with sugar metabolism in the assembled metagenome and metatranscriptomes .... 148

LIST OF ABBREVIATIONS

16S rRNA Small subunit of the bacterial and archaeal ribosome

18S rRNA Small subunit of the eukaryotic ribosome

23S rRNA Large subunit bacteria and archaeal ribosome

28S rRNA Large subunit eukaryotic ribosome

Ca2+ Calcium ion

CaCO3 Calcium carbonate

2- CO3 Carbonate ion

Contig Contiguous read

EMBL European Molecular Biology Laboratory

EPS Exopolymeric substances

GI GenBank Identifier

HBC Highborne Cay, The Bahamas

IAP Ion activity product

KEGG Koyoto Encyclopedia of Genes and Genomes

Kso Solubility product constant

LCA Last Common Ancestor

LSU rRNA Large subunit ribosomal RNA

MEGAN Metagenome Analyzer

MG-RAST Metagenomic Rapid Annotation using Subsystems Technology mRNA Messenger RNA

OTU Operational taxonomic unit

PCA Principal Component Analysis

SI Saturation Index

SSU rRNA Small subunit ribosomal

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

EXPLORING MICROBIAL AND FUNCTIONAL GENE DIVERSITY IN MODERN MARINE THROMBOLITIC MAT COMMUNITIES

Jennifer Marie Mobberley

December 2013

Chair: Jamie Foster Major: Microbiology and Cell Science

Thrombolites are unlaminated carbonate structures resulting from the trapping, binding, and mineral precipitation activities of complex microbial mat communities.

Thrombolites have a long geologic record representing one of Earth’s earliest ecosystems and harbor an unexplored area of genetic diversity associated with biologically induced carbonate precipitation. The underlying functional complexity in the thrombolitic mats of Highborne Cay, The Bahamas was explored using high-throughput sequencing techniques to determine microbial diversity, metabolic potential, and community gene expression. Bacterial, cyanobacterial, and archaeal diversity within four morphologically distinct thrombolitic mat types was assessed through 16S rRNA gene pyrosequencing. The most prominent differences between the four communities were within the cyanobacterial and archaeal populations. The “button mats” were identified as the most abundant and productive thrombolitic mat type and were characterized by an enrichment of Dichothrix spp., a cyanobacterium identified as a “hot spot” of carbonate deposition. Due to prevalence and high metabolic activity, the button thrombolitic mat type was selected for further functional analysis. The metabolic potential of these mat were delineated through metagenomic sequencing and

community-level physiological profile assays. Functional protein analysis of the metagenome indicated the presence of several key metabolic pathways known to influence carbonate mineralization including photosynthesis, aerobic heterotrophy, nitrogen and sulfur cycling. Spatial profiling of metabolite utilization suggested that the upper zone of the thrombolitic mat (0-5 mm) contained a more metabolically active community than the deeper regions of the mat. Additionally, metatranscriptomic sequencing was used to further examine spatial differences within the thrombolitic mats.

The spatial regions were delineated using oxygen microelectrode profiling and identified as oxic (0-3 mm), transitional (3-5 mm), and anoxic (5-9 mm) zones. The resulting gene expression profile revealed discrete gradients of microbial activity occurring within the unlaminated thrombolitic mat. This study showed that cyanobacterial photosynthesis is a major driver of energy production and carbon fixation in upper oxic and transitional zones, and that bacteria, eukaryotes, and archaea show spatial differences in metabolic activity. Together, this research represents the first in-depth genetic analysis of thrombolitic mat metabolisms, and provides the foundation for delineating the molecular mechanisms associated with carbonate mineralization in lithifying mat communities.

CHAPTER 1 LITERATURE REVIEW

Microbialites

Microbialites result from lithifying microbial mat communities that have a

propensity for sediment trapping, binding, and mineral precipitation of calcium

carbonates (CaCO3)(Burne, and Moore, 1987). These organosedimentary structures are classified based on their mineral fabric (mesofabric) into two primary types laminated stromatolites and unlayered, clotted thrombolites (Kennard and James,

1986). Stromatolites and thrombolites form a variety of macrostructures such as columns, platforms, domes, and conical structures that can range in length from centimeters to several meters (Riding, 2000).

Microbialites have a long fossil record, dating back 3.5 billion years, thus may represent some of Earth’s earliest microbial communities (Awramik, 1971; Kah &

Grotzinger, 1992). It has been proposed that these earliest microbial mat ecosystems contributed to the rise in oxygen and the evolution of our present biosphere (Kasting and Howard, 2001) and may have provided environments that supported the evolution of eukaryotes (Margulis, 1996). Based on their geographic and spatial distribution within the geological record, microbialites dominated the late Archaean and Proterozoic eons before their overall decline in the late Phanerozoic. That decline has been attributed to substrate competition and rise of algal and metazoan grazers (Awarmik, 1971;

Planavsky and Ginsburg, 2009). Although modern microbialites are not as prevalent, they are distributed across a range of aquatic habitats, including marine, freshwater, hypersaline, and alkaline environments (e.g. Dill et al.,1986, Laval et al., 2000; Reid et al., 2000; Burns et al., 2004; Breitbart et al., 2009; Farias et al., 2013). Considered to be

analogs to these ancient systems, modern lithifying mats provide insight into the

ecotypes and functional genes needed for the development of complex lithifying

microbial communities. In addition, since microbialites are poised at the intersection of

the biosphere and the lithosphere, these systems can be used to examine microbial

generation of biosignatures that are residuals of life that can be preserved or altered in the geologic record (Visscher & Stolz, 2005).

In lithifying microbial mats, calcium carbonate mineralization is influenced by the metabolic activity and geochemical cycling of a complex consortium of organisms. The functional groups of organisms, known as guilds, that are active in microbialite mat communities are phototrophs (i.e. oxygenic and anoxygenic), aerobic heterotrophs, sulfide-oxidizers, anaerobic heterotrophs (e.g. sulfate-reducers), and fermenters

(Visscher and Stolz, 2005; Dupraz et al., 2009). In order to understand these processes it is important to understand the underlying taxonomic and functional complexity of these communities.

Community Diversity in Microbialites

Prior to the molecular age, modern microbialites were characterized by their mesofabric or by the morphology of the dominant functional group as determined by microscopy or culturing (Riding, 2011). The advent of culture-independent methods, such as ribosomal gene studies and metagenomics, for examining lithifying community structure and diversity has enabled a more complete understanding of the organisms found in lithifying communities (Dupraz et al., 2009; Foster and Green, 2011). The main tool of these community diversity studies has been sequencing of the 16S and 18S ribosomal RNA (rRNA) small genes from Bacteria and Archaea, and Eukaryota, respectively. Due to the relatively low cost, rRNA gene surveys are often the first step in

examining the genetic complexity of lithifying mats. These phylogenetic markers have

been used to determine ecotype diversity within specific lithifying ecosystems, as well

as in comparative studies from morphologically and geographically distinct microbial

mat communities (e.g. Baumgartner et al., 2009; Goh et al., 2009; Foster and Green,

2011; Centeno et al., 2012; Farias et al., 2013).

In all lithifying microbial mats systems, bacterial metabolisms drive calcium

carbonate precipitation (Dupraz et al., 2009). The most prominent bacterial phyla, based

on 16S rRNA gene marker studies, are the Cyanobacteria and Proteobacteria, which

comprise a diverse group of phototrophs and aerobic and anaerobic heterotrophs. In

addition, other prominent phyla with a range of metabolic capabilities that are recovered

include Actinobacteria, Bacteroidetes, Chloroflexi, Firmicutes, and Planctomycetes.

While most lithifying microbial mats contain these bacteria, differences in relative

abundance of these phyla are influenced by the environmental parameters (Foster and

Green, 2011; Centeno et al., 2012).

Cyanobacteria are the driving metabolic force behind lithifying microbial mats and

are critical for accretion and early calcium carbonate precipitation (Pickney and Reid,

1997; Reid et al., 2000; Visscher and Stolz, 2005; Dupraz et al., 2009; Planavsky et al.,

2009), and comprise between 1% and 80% of the recovered 16S rRNA gene reads in lithifying systems. This broad range is likely due to the primer selectivity biases against cyanobacteria with commonly used 16S rRNA primers (Nübel et al. 1997; Foster et al.,

2009; Goh et al., 2009). As the dominant phototrophs in microbialites, cyanobacteria can be divided into two broad groups: filamentous mat builders and unicellular, coccoid cyanobacteria.

The filamentous mat builders found in microbialites include non-heterocystous

Oscilliatoriales and heterocyst-forming Nostocales orders. Oscilliatoriales are ubiquitous in microbialites, however there were differences in abundance and diversity of recovered ecotypes (Papineau et al. 2005; Goh et al., 2009; Brady et al., 2010;

Couradeau et al., 2011; Foster and Green, 2011; Nitti et al. 2012; Arp et al., 2013;

Farias et al. 2013). For example, Microcoleus spp. dominates microbialites from extreme environments such as Shark Bay hypersaline lagoon and high-altitude alkaline

Scompa Lake (Burns et al., 2004; Papineau et al., 2005; Goh et al., 2009; Farias et al.,

2013). However, in the marine stromatolites in Highborne Cay, Schizothrix spp. are

important in the trapping and binding of sediment (Reid et al., 2000; Foster et al., 2009).

A particularly interesting finding was the presence a common mat-builder Leptolyngbya

that formed clades of site-specific ecotypes in a variety of lithifying systems (Foster and

Green, 2011). Although not as ubiquitous as Oscilliatoriales, Nostocales have been

recovered from lithifying microbial mat systems including the Highborne Cay

thrombolites (Myshrall et al., 2010), and freshwater microbialites (Santos et al., 2010;

Couradeau et al. 2011; Nitti et al. 2012). In the Highborne Cay thrombolites Dichothrix

spp., which is not found in the adjacent stromatolites, had high rates of photosynthesis

and their EPS-rich sheaths were found to the initial site of lithification (Planvasky et al.,

2009; Myshrall et al., 2010). In the freshwater microbialites, Nostocales similar to

Calothrix and Nostoc were found to make up around 10% of the cyanobacterial

population, while they represented less than a percent of the population in the

Alichichica Lake microbialites (Santos et al., 2010; Couradeau et al. 2011; Nitti et al.

2012). Together these findings suggests the importance of understanding how

environmental conditions drive differences in mat-building cyanobacterial mats and how

this can effect calcium carbonate precipitation.

Chroococcales and Pleurocapsales are orders of coccoid cyanobacteria that are

found in all microbialites (Goh et al., 2009; Myshrall et al., 2009; Foster and Green,

2011; Couradeau et al., 2011; Nitti et al. 2012). Chroococcales commonly found in lithifying systems include endolithic Halothece, Cyanothece, and Solentia, the latter of which was implicated in boring and fusing ooid grains to form lithified layers in the

Highborne Cay stromatolites (MacIntyre et al., 2000; Reid et al., 2000; Foster et al.,

2009). Synechococcus and Synechocystis, which are metabolically diverse cyanobacteria, are abundant in marine microbialites (Foster et al., 2009; Myshrall et al.,

2010, Baumgartner et al., 2010). The order Pleurocapsales includes endolithic cyanobacteria that preferentially grow in low-light conditions and are UV and desiccation resistant (Krumbein et al., 1979; Dillon et al., 2002; Kremer and Kazmierczak, 2005).

This resistance explains the enrichment of Pleurocapsales ecotypes (>70%) in the microbialites found in the deeper depths of Lake Alchichica (Couradeau et al., 2011). As with the filamentous mat building cyanobacteria, more studies are needed to understand how unicellular, coccoid cyanobacteria participate in calcium carbonate precipitation.

Alphaproteobacteria, a class of Proteobacteria that encompasses a metabolically diverse group of bacteria, are the most abundant organisms recovered from 16S rRNA surveys (Foster and Green, 2011; Nitti et al., 2012, Couradeau et al., 2012). The majority of these are similar to anoxygenic photoheterotrophs and photoautotrophs belonging to the orders Rhodobacterales, Rhodospirillales, and Rhizobiales, which are

collectively called purple non-sulfur bacteria (Imhoff et al., 2005). In laboratory cultures,

anoxygenic photoautotrophs have induced calcium carbonate precipitate, although this

has yet to be confirmed in lithifying mats (Bosak et al., 2007). Additionally, Rhizobiales

also contain several nitrogen-fixing genera, which indicates these Alphaproteobacteria

contribute to nitrogen cycling in lithifying systems. In the marine and hypersaline

microbialites, Rhodobacterales and Rhizobiales are over 50% of the recovered 16S

rRNA genes (Papineau et al., 2005; Baumgarten et al., 2009; Goh et al. 2009; Myshrall

et al., 2010). Rhodospirillales have a higher abundance in freshwater lithifying systems

of Cuatros Ciénegas and Lake Socompa (Nitti et al., 2012; Farias et al., 2013).

Interestingly, in studies of spatial distribution within mats, Alphaproteobacteria tend to

be more enriched deeper in mat communities (Papineau et al., 2005; Nitti et al., 2012).

Additionally, in the Lake Socompa stromatolites, Rhodobacterales were more enriched

in the deeper microbialites while Rhizobiales had higher abundances in the shallow

microbialites (Farias et al., 2009). These results suggest that the differential

abundances of Alphaproteobacteria likely reflect niche differentiation. Although

Alphaproteobacteria are dominant in lithifying mats, these bacteria are common

constituents of the surrounding water, so their role in lithification is unresolved (Goh et

al., 2009; Baumgarten et al., 2009).

Sulfate-reducing bacteria are believed to be the major heterotrophic metabolism

contributing to carbonate precipitation in microbialites (Visscher et al., 1998; Reid et al.,

2000; Baumgarten et al., 2009; Nitti et al. 2012; Dupraz et al., 2013). Most sulfate- reducing bacteria belong to the Deltaproteobacteria class (e.g. Desulfovibrionales,

Desulfobacterales) or to Clostridiales (e.g. Desulfotomaculum, Desulfosporomusa)

although recently archaeal sulfate-reducers have been found (Barton and Fauque,

2009). Most of the studies in microbialites have focused on the Deltaproteobacterial

sulfate-reducers, which are associated with cyanobacteria as well as present in anoxic

zones of the mat in Highborne Cay and Shark Bay, and Cuatros Ciénegas microbialites

(Visscher et al., 1998; Papineau et al., 2005; Baumgartner et al. 2006; Baumgartner et al. 2009; Goh et al., 2009; Nitti et al., 2012).

In addition to the Alphaproteobacteria and sulfate-reducers, 16S rRNA gene

surveys in lithifying mats have identified a diverse number of heterotrophic phyla

including Actinobacteria, Bacteroidetes, Gammaproteobacteria, Firmicutes, and

Planctomycetes. It is likely that these bacteria modify and degrade photosynthetically

produced organic carbon (e.g. exopolymeric substance and sugars) through respiration

and fermentation (Dupraz et al. 2009). There are differences in the relative abundances

of these heterotrophs in microbialites in different environments. For example, the high

altitude lake contained the highest abundance of the UV-resistant Deinococcus-

Thermus ecotypes in any lithifying systems to date (Farias et al., 2013). Another

example is the enrichment of Bacteroidetes in deeper, anoxic zones of Shark Bay and

Cuatro Ciénegas microbialites (Papineau et al., 2005; Nitti et al., 2012). Due to their

dependence on organic carbon, it is likely that the community structure of aerobic and

anaerobic heterotrophs is greatly influenced by the specific cyanobacterial populations

present in lithifying systems.

Once thought to be confined extreme environments, microbial diversity studies

have found Archaea in nearly all environmental niches and these organisms likely

contribute to global geochemical cycles (reviewed in Offre et al., 2013). Studies on

Archaea distribution in lithifying systems have focused on microbialites found in hypersaline lagoons of Shark Bay, Australia and hypersaline lakes on the Kiritimati Atoll

(Burns et al., 2004; Papineau et al., 2005; Goh et al., 2009; Arp et al., 2012). Although

Archaea only comprised 4-10% of the total prokaryotic population in Shark Bay microbialites, between 74-100% of these sequences were similar to Halobacteria

(Euryarchaeota), which were not recovered from the surrounding water (Burns et al.,

2004; Papineau et al., 2005; Goh et al., 2009). Subsequent isolation studies and genome sequencing revealed that these Halobacteria are photoheterotrophs that are unable to reduce sulfur (Goh et al. 2006; Allen et al., 2008; Burns et al., 2012). Although not as abundant as Halobacteria, Thaumarchaeota (formerly classified as mesophilic

Crenarchaeota) similar to marine mixotrophic ammonia-oxidizers were also isolated from Shark Bay lithifying systems (Burns et al., 2004; Papineau et al. 2005).

Interestingly, methanogens have also been recovered from hypersaline microbialites even though methanogenesis is likely inhibited due to substrate competition from sulfate-reduction (Ehrlich et al., 2002; Schimel, 2004; Burns et al. 2004; Arp et al.,

2012). As opposed to the hypersaline lithifying mats, there are very few studies on the

Archaea in marine and freshwater microbialites. A microbial diversity study and two metagenomic analyses revealed that archaea comprise less than 2% of the prokaryotic community in the Highborne Cay stromatolites and Cuatros Ciénegas microbialites

(Breitbart et al., 2008; Baumgartner et al., 2009; Khodadad and Foster, 2012). The diversity and role of archaea in lithifying mat geochemical cycling and potential influences on carbonate precipitation remains to be investigated.

Although lithifying mats are primarily driven by bacterial metabolism, Eukaryotes are also members of these mat consortia. Eukaryotes in microbialites play a wide range of roles, from contributing to mat metabolic capacity (Breitbart et al., 2008; Khodadad and Foster 2009; Couradeau et al. 2011); participating in mat stabilization and accretion

(Awarmik and Riding, 1998; Bowlin et al., 2010; Franks et al., 2010); are hypothesized to undergo bioturbation of mat and mineral structures (Bernhard et al., 2013; Edgcomb et al., 2013b; Tarhan et al., 2013). In many systems, the most dominant eukaryotes include phototrophs such as green and red algae (i.e. Chlorophytes and Rhodophytes), diatoms, and protists (Couradeau et al., 2011; Farias et al., 2012; Edgcomb et al.,

2013b). Metagenomic studies recovered photosynthetic genes from algae and diatoms, which suggest that these eukaryotes can contribute to lithifying mat photosynthesis

(Breitbart et al., 2008; Khodadad and Foster, 2012). Furthermore, diatom morphology and EPS production increased accretion in stromatolitic mats due trapping and binding of sediment, however there is no evidence that diatoms contributed directly to calcium carbonate precipitation (Paterson et al., 2010; Bowlin et al., 2010; Frank et al., 2010).

Additionally, it has been proposed that metazoans, protists, and foraminifera contribute to bioturbation through their grazing activities (Myshrall et al., 2010; Edgcomb et al.,

2013b), as well as through infiltration and alterations of the microbial mat matrix and

mesofabrics (Bernhard et al., 2013). However a survey of metazoans in a different

marine stromatolites and thrombolites found that there was no correlation between

abundance of metazoans and mesofabrics, suggesting that faunal populations have

little influence on mineralization (Tarhan et al., 2013). The impact of eukaryotes on carbonate precipitation and accretion remains to be resolved.

The Biogeochemistry of Carbonate Biomineralization in Microbialites

In the ocean, biological calcium carbonate (CaCO3) mineralization by microorganisms, including calcareous algae, cyanobacteria, and heterotrophic bacteria, is a major sink of carbon (Ehrlich, 2002; Konhauser and Riding, 2012). Biological carbonate precipitation in lithifying microbial mats occurs through two principle mechanisms: biologically influenced and biologically induced mineralization (Perry et al.,

2007; Dupraz et al., 2009). Biologically influenced is a passive mineralization process that requires organic matter (e.g. exopolymeric substances) and depends on environmental conditions instead of direct biological activity to produce a mineral precipitate (Dupraz et al., 2009). Biologically induced mineralization results from microbial metabolisms that generate favorable environmental conditions for extracellular precipitation of CaCO3 (Frankel and Bazylinksi, 2003; Visscher and Dupraz, 2005). In

both biologically influenced and induced mineralization, the resulting calcium carbonate

precipitates are indirect evidence of life (Perry et al. 2007; Dupraz et al., 2009). These

processes differ from biologically controlled carbonate mineralization in

coccolithophores, coralline algae, and mammalian skeletons, where intracellular

localization of CaCO3 crystal nucleation and growth is under direct genetic control

(Weiner and Dove, 2003; Perry et al., 2007).

Lithifying mat systems contain several functional guilds that include phototrophs, aerobic heterotrophs, anaerobic heterotrophs (i.e sulfate-reducers and methanogens), sulfide-oxidizers, and fermenters (Visscher and Stolz, 2005; Dupraz et al., 2009). These microorganisms, through their metabolic processes and cell-to-cell interactions, create light-driven biogeochemical gradients that influence carbonate precipitation (Dupraz et al., 2009). This precipitation potential is affected by two factors: the saturation index and

exopolymeric substances (Visscher and Dupraz, 2005; Dupraz et al., 2009; Decho,

2010).

2+ 2- The saturation index is the likelihood of calcium [Ca ] and carbonate [CO3 ] ions precipitating to form calcium carbonate (Morse et al., 2007). This is denoted by the ion

2+ 2- activity ion activity product (IAP; [Ca ] x [CO3 ]) and solubility product constant

-6.19 (Ksp; 10 for aragonite) of the mineral and is defined as SI = log (IAP/ Ksp). A positive

SI number indicates that calcium carbonate precipitation is thermodynamically

favorable; however, values between eight and ten-fold supersaturation (SI=0.8-1.0) are

needed for calcium carbonate precipitation to occur (Kempe and Kazmierczak, 1994;

Arp et al., 2001). The extent of calcium carbonate precipitation also depends on pH in

the microbial mat due to the carbonate equilibrium determining carbonate ion

concentration (Visscher and Stolz, 2005; Dupraz et al., 2009). In marine microbialites,

calcium is generally supersaturated; therefore metabolic processes that either increase

or decrease pH have a greater influence on carbonate precipitation (Visscher and Stolz,

2005; Dupraz et al., 2009).

In general, lithifying mat metabolic processes that increase local pH are due to

the removal of carbon dioxide via autotrophic carbon fixation (i.e. photosynthesis,

sulfate reduction, and methanogenesis) (Arp et al., 2001; Visscher et al., 1998; Visscher

and Stolz, 2005; Dupraz et al. 2013). Those metabolic processes that consume organic

carbon such as aerobic respiration and fermentation or oxidize inorganic substrates (i.e.

sulfide and ammonium) decrease the pH, therefore promoting carbonate dissolution

(Visscher and Stolz, 2005). However, it should be noted that many bacteria and

archaea are capable of performing multiple forms of metabolisms and use a range

substrates that likely influence the pH in different ways (Visscher and Stolz, 2005;

Gallagher et al., 2012). Also, as lithifying mats are dynamic ecosystems where

gradients such as pH and oxygen can dramatically change over the course of a few

hours (Visscher et al., 1998; Paerl et al. 2001), it is the overall net metabolisms that result in the formation of carbonate structures (Reid et al., 2000).

In addition to the saturation index, exopolymeric substances (EPS) also play a critical role in carbonate precipitation as it binds metal ions such at Ca2+ to key functional groups of sugars and amino acids that comprise the matrix and function as a nucleation site (Braissant et al., 2007). As an important component of the fixed carbon and nitrogen biomass, EPS serves as a substrate source for heterotrophs, as well as provides structural support to the mat in hydrodynamic systems (Decho et al., 2005;

Paterson et al., 2008). Studies in stromatolites found that while most of the EPS is produced by cyanobacteria, there was contribution from other bacteria including sulfate reducers (Kawaguchi and Decho, 2000; Braissant et al., 2007; Foster et al., 2009;

Gallagher et al., 2012). The different chemical characteristics of EPS from specific

species as well as microbial degradation can control the type and morphology of the

calcium carbonate minerals precipitated (Kawaguchi and Decho, 2002; Decho et al.,

2005; Gallagher et al., 2012). For example, the acidic residues in cyanobacterial EPS

chelate calcium ions strongly, thus inhibiting precipitation (Kawaguchi and Decho, 2002;

Guatret and Trichet, 2005), however the high EPS turnover rates (up to 60% degraded

within 24 hours) could lead to the release of the calcium ions, that under favorable pH

conditions, could result in calcium carbonate precipitation (Decho et al., 2005). Through

the saturation index and EPS cycling, the microorganisms in lithifying mat can influence

or induce calcium carbonate precipitation. The activities of these microorganisms can

be broken down in two general net effects that effect mineralization, organic carbon

production and organic carbon decomposition (Decho et al., 2010).

Cyanobacteria are the primary producers within lithifying mat systems through

their coupling of photosynthesis and carbon fixation (Paerl et al., 2001; Visscher and

Stolz; Planavsky et al., 2009; Breitbart et al., 2009; Myshrall et al., 2010). In marine and freshwater stromatolites, cyanobacteria also generate organic nitrogen through nitrogen fixation (Steppe et al., 2001; Beltran et al., 2012). High rates of cyanobacterial photosynthesis contribute to supersaturation of oxygen, high pH, and the production of organic carbon (Visscher et al., 1998; Visscher et al., 2002; Breitbart et al., 2008;

14 - Myshrall et al., 2010; Farias et al., 2012). In the Highborne Cay stromatolites, C-HCO3

tracer studies revealed that 2-4% of the cyanobacterial organic carbon was EPS

available for heterotrophic consumers (Decho et al., 2005). Although microbial diversity

surveys and metagenomic studies have found anoxygenic phototrophic bacteria to be

prevalent in lithifying systems, their role in carbon fixation and influence on pH has not

been characterized in situ (Breitbart et al., 2008; Foster and Green, 2011; Foster and

Khodadad, 2012). Additionally, although methanogenic archaea are found in lithifying

mats, methanogenesis has not been detected in marine microbialites even though it has

been reported in hypersaline mats (Robertson et al., 2009). This is likely due to sulfate-

reducing bacteria outcompeting methanogens for substrates (Ehrlich et al., 2002;

Schimel, 2004).

In microbialites, respiratory processes including aerobic and anaerobic

heterotrophy and fermentation consume a significant portion of photosynthesis-derived

biomass and can influence carbonate precipitation (Paerl et al. 2001; Visscher and

Stolz, 2005; Baumgartner et al., 2009; Myshrall et al., 2010; Gallagher et al., 2012).

EPS cycling within lithifying mats is controlled by microbial metabolisms (Decho et al.

2005; Dupraz et al., 2009). Aerobic heterotrophs and fermenters are involved in the initial processing of the EPS material. Providing there is are low oxygen microenvironments or mat anoxia, sulfate-reducing bacteria, the dominant anaerobic metabolism, use this heterotrophically processed EPS as electron donors for sulfate- reduction (Gallagher et al., 2012). The hydrogen sulfide produced by sulfate-reduction can then can be utilized by sulfide-oxidizing bacteria (Visscher and Stolz, 2005). This metabolic cycling of EPS can lead to localized calcium carbonate precipitation in a high

14 - pH environment. Microautoradiographic detection of C-HCO3 in organic matter found that calcium carbonate deposition is localized to cyanobacteria and their closely associated bacteria (Paerl et al., 2001; Planavsky et al., 2009). Further studies involving micrometer-scale mapping and radioisotope studies showed correlations between sulfate reduction and precipitation of CaCO3 in the form of a micritic crust (Visscher et

al., 1998; Visscher et al., 2000; Baumgartner et al., 2006). This was attributed to the increased alkalinity resulting from SRB activity and Ca2+ release from heterotrophic degradation of EPS (Visscher et al., 1998; Paerl et al., 2001). It is the combined metabolic activities that ultimately determine the net calcium carbonate precipitation potential within lithifying microbial mats.

Studies on the genetic regulation of metabolisms in lithifying microbial mats are lacking. To date, only a few studies have examined the metagenomes of lithifying mat communities and no studies have examined the community functional gene expression

(Desnues et al., 2008; Breitbart et al., 2009; Foster and Khodadad, 2012). Establishing

the functional complexity and underlying genetic processes that occur within

stromatolitic and thrombolitic mats will provide critical insights into metabolic processes

in lithifying mat communities.

Highborne Cay Thrombolites as Model Microbialites

Highborne Cay (HBC), on the western margin of the Exuma Sound, Bahamas, is

one of the only open ocean sites of actively forming microbialites (Reid et al, 2000). This

site contains both subtidal stromatolites and intertidal thrombolites in close proximity as

part of a 2.5 km fringing reef (Andres & Reid, 2006). This proximity is ideal for studying

different patterns of microbialite formation due to extensive monitoring of the similar

(e.g. temperature, salinity) and different environmental parameters (e.g. burial time)

each ecosystem experiences (Reid et al., 2011; Bowlin et al., 2012).

To date, the majority of research has focused on the stromatolitic mats, which have provided a basis for understanding the biogeochemical processes within microbialitic systems (reviewed in Visscher and Stolz, 2005; Dupraz et al., 2009). These biogeochemical interactions result in the formation of three major stable stromatolitic mat types characterized by distinct surface communities (Type I, Type 2, Type 3) so that each laminated subsurface represents a chronology of former mat surface (Reid et al., 2000; Bowlin et al., 2012). The Type 1 mats, the dominant stromatolitic mat type are comprised of the filamentous cyanobacteria Schizothrix spp. that produces EPS which traps and binds ooids and sediments into an unlithified grain layer (Foster et al. 2009;

Baumgartner et al., 2009). Type 2 mats represent a transitional community where heterotrophic bacterial dominated biofilms overlie a diverse population of filamentous and coccoid bacteria (Baumgartner et al., 2009; Foster et al., 2009). Calcium carbonate

precipitation occurs in the form of a micritic crust between these two layers as a result of

anaerobic heterotrophic activity dominated by sulfate-reduction (Visscher et al., 1998;

Visscher et al. 2000; Reid et al., 2000). Type 3 mats contain the most diverse bacterial community where Solentia spp., an endolithic coccoid cyanobacterium, are involved in cementing layers of micritized grains (Macintyre et al., 2000; Foster et al. 2009). Despite characterizing the biochemical and accretion processes in the stromatolitic mats, the underlying genetic processes have just recently been address by Khodadad and Foster

(2012) through metagenomic sequencing and community substrate profiling of the Type

I (nonlithifying) and Type 3 (lithifying) mats. Although they found that there was significant overlap in microbial and functional diversity, a greater number of carbon processing genes and increased organic carbon utilization was observed in the lithifying mats. This suggests that differences in lithification potential likely result from differences in metabolic activity, particularly in carbon cycling.

In contrast to the intensively studied stromatolites, relatively few investigations have been conducted on the adjacent intertidal thrombolitic microbialites. The clotted mesofabric observed in the Highborne Cay thrombolites suggests that calcium carbonate mineralization proceeds in a different manner. Microbial diversity studies have indicated that the thrombolitic mats contain diverse consortia of bacteria and eukaryotes (Myshrall et al., 2010; Edgcomb et al., 2013b; Tarhan et al., 2013).

Microscopic examination and petrographic thin sections of the primary thrombolitic mat type revealed initial calcification occurring of the EPS-rich sheaths of Dichothrix spp. filaments, a heterocyst-forming cyanobacterium that has not been found in the adjacent stromatolites (Planavsky et al., 2009; Baumgarten et al., 2009). Stable isotope data for

δ18O and δ13C suggests this calcium carbonate precipitate was induced primarily by high rates of photosynthesis with some contribution from heterotrophic metabolisms

(Planavsky et al., 2009). Although there is debate on the relative influence of microbially-driven diagenesis versus eukaryotic bioturbation in the formation of the clotted mesofabric in thrombolites; its is likely the microbial mat type is the primary driver of calcium carbonate precipitation (Planavsky et al., 2009; Planavsky and

Ginsburg, 2009; Bernhard et al., 2013; Edgcomb et al., 2013; Tarhan et al., 2013).

Detailed studies are need to understand the formation and functional of the Highborne

Cay thrombolitic ecosystem at a fundamental genetic level.

The goal of this dissertation was to increase our understanding of the metabolic processes that contribute to the formation and function of Highborne Cay thrombolites through examining the underlying genetic and functional complexity within the thrombolitic mat. My objectives were to (1) determine the bacterial and archaeal diversity within the Highborne Cay thrombolites; (2) examine the metabolic potential of these communities and to (3) examine the spatial distribution of community gene expression. A combination of techniques including microscopy, 16S rRNA gene cloning from dominant ecotypes, and 454 pyrosequencing of 16S rRNA gene amplicons were used to identify the bacterial and archaeal populations present within the thrombolitic mats. To establish the metabolic potential, metagenomic sequencing of the thrombolitic mat community was coupled to community level phenotypic microarrays. This was followed by metatranscriptomic sequencing to generate a spatial profile of thrombolitic community activity during peak photosynthesis. By determining the genetic relationship between microbial populations and metabolic processes within the modern thrombolitic

microbial mats, we gain insight into the formation of biologically complex communities on early Earth.

CHAPTER 2 COMPARATIVE MICROBIAL DIVERSITY ANALYSES OF MODERN MARINE THROMBOLITIC MATS BY BARCODED PYROSEQUENCING

Introduction

Microbialites are carbonate buildups that are derived from the trapping, binding, and mineral precipitation activities of microbial mat communities (Burne and Moore,

1987) . Microbialites are found across the globe in a wide variety of aquatic habitats

(e.g. freshwater, hypersaline, marine) and represent one of the oldest known ecosystems on the planet (Canfield and Des Marais, 1993; Grotzinger and Knoll, 1999).

Microbialites are classified by their internal micro- and mesostructure, which can range from the well-laminated stromatolites to the unlaminated, clotted thrombolites (Kennard and James, 1986). One of the few modern sites where both stromatolites and thrombolites are actively forming is the island of Highborne Cay located in the Exuma

Sound, Bahamas (Dravis,1983; Dill et al.,1986; Reid et al., 2000). Considerable progress has been made on understanding the microbiological, geological and biogeochemical processes associated with the stromatolitic microbialites of Highborne

Cay (Reid et al., 2000; Visscher et al., 2000; Baumgartner et al., 2009; Dupraz et al.,

2009; Foster et al., 2009; Foster and Green, 2011). However, far less progress has been made on understanding similar processes in the adjacent, unlaminated thrombolitic microbialites (Planavsky et al., 2009; Myshrall et al., 2010).

Initial work by Myshrall et al. (2010) provided the first assessment of the microbial communities associated with the thrombolite structures of Highborne Cay.

Reprinted with permission from Mobberley JM, Ortega MC, Foster JS. (2012). Comparative microbial diversity analyses of modern marine thrombolitic mats by barcoded pyrosequencing. Environ Microbiol 14: 82-100.

This previous study generated clone libraries to the 16S and 18S ribosomal RNA

(rRNA) genes revealing that the thrombolitic communities were predominately bacterial.

Eukaryotes comprised less than 11% of the recovered operational taxonomic units

(OTUs), with most sequences sharing similarity to grazing, bacterivorous Nematoda

(Myshrall et al., 2010). This previous study also identified two dominant thrombolitic mat

types at Highborne Cay including a highly productive, nodous mat called “button” mat

that was comprised of filamentous calcified cyanobacterial filaments (Planavsky et al.,

2009; Myshrall et al., 2010); and a “pink” encrusted mat that was found adjacent to the button thrombolitic mats (Myshrall et al., 2010). Ribosomal RNA gene analysis of the bacteria associated with these two mat types indicated that both thrombolitic mats types contained microbial communities that were distinct from the adjacent stromatolitic mats

(Myshrall et al., 2010). In this previous study, however, sequencing coverage of the microbial population, as determined by Good’s estimate, indicated that only 59% of the bacterial community was represented in the button mat clone libraries and only 49% in the pink mat type (Good, 1953; Myshrall et al., 2010).

In the present study, we expanded on this initial report by using multi-domain and deep sequencing approaches to examine and compare the microbial diversity associated with thrombolitic microbial mat types found at Highborne Cay, Bahamas. Our central goal was to examine bacterial and archaeal distribution within the thrombolites to determine whether the thrombolitic mat types represent successive stages of thrombolite formation and development. To address this goal and build on the previous study we used next-generation, barcoded pyrosequencing to improve sequencing coverage of the bacterial populations within the thrombolitic mat types. We also

generated amplicon libraries for archaeal and cyanobacterial populations to assess the

impact of these taxa on thrombolite community structure. Pyrosequencing has emerged

as a robust tool for investigating community diversity and structure in microbial

ecosystems (Edwards et al., 2006; Sogin et al., 2006; Roesch et al., 2007, 2009; Miller et al., 2009). By using this high-throughput approach to sequence the small subunit rRNA gene, we were able to more fully characterize and compare these thrombolitic mat communities in parallel to assess the microbial complexity of these thrombolite ecosystems.

Materials and Methods

Site Description, Sample Collection, and Microscopy

Thrombolitic mat samples were collected from the island of Highborne Cay located in the northern Exumas, The Bahamas (76°49ʹ W, 24°43ʹ N). The mat samples were taken from a series of intertidal thrombolites located at Site 6, as designated by

Andres and Reid (2006), in March 2008. Eight replicate samples (approximately 3 g of thrombolitic mat material) were collected from each mat type and immediately immersed in RNALater (Ambion, Austin, TX, USA) to stabilize the nucleic acids, and then frozen at

-20 °C. Frozen samples were transported to the Space Life Science Laboratory and stored at -20 °C until processing. A corresponding set of mat samples were taken for immediate microscopic examination in the field. Freshly collected thrombolitic mats were sectioned (0.5 cm) with a rock saw and examined using an Olympus SX12 stereoscope (Olympus, Center Valley, PA, USA). Mat surface morphologies were documented then cross-sections were immersed in filtered seawater and imaged using

10X, 32X, and 1000X objectives.

Molecular Identification of Cyanobacterial Isolates from the Thrombolitic Mats

To identify the morphologically dominant cyanobacterial isolates from the button and black mat types, two cells types were dissected from their respective mat types using an Olympus SZX12 stereoscope scope. Bundles of a filamentous cyanobacterium, morphologically identified as Dichothrix spp. (Planavsky et al., 2009), were dissected from the button mats and clusters of previously undescribed pigmented coccoid cells were dissected from the black mat type. Approximately 20 mg of each cell type were dissected and immediately placed into an Eppendorf tube containing DNAzol

ES reagent (Molecular Research Center, Cincinnati, OH, USA). Each tube then underwent three cycles of freeze-thawed in liquid nitrogen followed by grinding with a mortar and pestle.

Due to the presence of non-heterocystous background cyanobacteria the 16S rRNA gene from the Dichothrix spp. cell type was amplified using two unique primer sets that were designed based on alignments of full length 16S rRNA gene sequences of closely related Calothrix and Rivularia species recovered from NCBI and Rivularia- like sequences recovered from the bacterial and cyanobacterial partial 16S rRNA gene clone libraries from Myshrall et al., (2010). Primers Dicho-1F

(CTGGCTCAGGATGAACGCTG) and Dicho-1R (GCCAAACCACCTACGAACGC) produce a 523 bp amplicon of the V3-5 region. Primers Dicho-2F

(TGGGGTAAAAGCGTACCAAG) and Dicho-2R (CCACGCCTAGTATCCATCGT) produce a 600 bp amplicon of the V4-6 region. The PCR reactions contained the following final concentrations: 5x GoTaq Flexi Reaction Buffer (Promega, Madison, WI),

1.5 mM magnesium chloride, 600 µM dNTPs, 400 nM of each primer, 20 µg bovine serum albumin (BSA), 10 ng of Dichothrix spp. genomic DNA, 2.5 U of GoTaq Flexi

DNA Polymerase (Promega). The PCR conditions were 94 °C for 5 min, followed by 30 cycles of 94 °C for 1 min, 58 °C for 1 min, 72 °C for 2 min, and a final extension of 72 °C for 7 min.

Due to the presence of numerous non-cyanobacterial heterotrophic bacteria in the black cyanobacterial cell dissections, a nested primer approach was used to amplify the 16S rRNA gene. The first round used 16S universal bacterial primers 27F

(AGAGTTTGATCCTGGCTCAG) and 1525R (TAAGGAGGTGATCCAGCC; Lane, 1991) and the second round used cyanobacterial primers, Cya359F and an equimolar mixture of Cya781Ra and Cya781Rb (Nübel et al., 1997). For both rounds, the reagent concentrations were the same as in the Dichothrix PCR described above, except that in the second round 2 ng of amplicon generated in the first round of PCR was used. The conditions for the first round of PCR included 95 °C for 2 min, 30 cycles of 95 °C for 30 sec, 58 °C for 2 min, 72 °C for 2 min, and a final extension of 72 °C for 10 min. The conditions for the second round of PCR with the cyanobacterial-specific primers were the same as described for the Dichothrix PCR. The PCR products for both cell types were purified with the UltraClean PCR Kit (MoBio, Carlsbad, CA) and cloned into the pCR 2.1 vector using the Topo TA Cloning Kit (Invitrogen, Carlsbad, CA) following manufacturer’s instructions. Ten clones from each cell type were picked for sequencing with an ABI 3130 DNA sequencer at the University of Florida Interdisciplinary Center for

Biotechnology Research (UF-ICBR). Consensus sequences were generated using

Clustal W and all clone sequences were submitted to GenBank under accession numbers HQ415794 - HQ415799.

Generation of Barcoded 16S Ribosomal RNA Gene Libraries

To determine the bacterial, cyanobacterial, and archaeal community composition of the four main thrombolitic mat types barcoded 16S rRNA gene libraries were generated and sequenced with high-throughput pyrosequencing. DNA was isolated from three replicate samples for each mat type using previously described methods

(Foster et al., 2009). DNA was then PCR amplified using a fusion 454-primer that included an oligonucleotide tag (barcode) located on the 3’ end. Each barcode tag was unique for the four mat types and all sequences are listed in Table 2-1 (Hamady et al.,

2008; Roesch et al., 2009).

The PCR reactions for the bacterial 16S rRNA library, which target the V1-2 region of the gene, contained the following final concentrations: 1X Clone Pfu Reaction

Buffer (Stratagene, La Jolla, CA), 280 µM dNTPs, 2.5 µg BSA, 600 nM each primer,

0.75 ng of genomic mat DNA, 1.25 U cloned Pfu DNA Polymerase (Stratagene), and nuclease-free water (Sigma, St. Louis, MO) in a volume of 25 μl. The reactions were held at 95°C for 5 min, followed by 35 cycles of 95°C for 1 min, 64°C for 1 min, 75°C for

3 min, and a final extension of 75°C for 7 min. Negative controls were conducted and no contamination was detected. All PCR products were combined in equimolar amounts and purified using the PCR Purification kit (Qiagen, Valencia, CA).

The PCR reactions for the barcoded cyanobacterial 16S rRNA library used the

Nübel primer set, which has been shown to detect a wider range of cyanobacteria than the “universal” bacterial primer set used above (Foster et al., 2009; Wang and Qian,

2009). The primers targeted the V3-4 region of the gene, contained the following final concentrations: 1X High Fidelity PCR Buffer (Invitrogen), 2 mM magnesium sulfate, 200

µM dNTPs, 2.5 µg BSA, 400 nM 359F primer, 200 nM each of 781Ra and 781Rb

primers, 5 ng of genomic DNA, 1 U Platinum Taq High Fidelity (Invitrogen), and nuclease-free water (Sigma) to 25 µl. The PCR conditions were 94°C for 2 min, 30 cycles of 94°C for 0.5 min, 58°C for 1 min, 68°C for 2 min, and a final extension of 68°C for 10 min. PCR products were extracted from an agarose gel with the QIAquick Gel

Extraction kit (Qiagen). Negative controls were conducted for all PCR reactions and results indicated no contamination. All purified PCR products were combined in equimolar amounts.

Due to the low numbers of Archaea within the thrombolitic communities, a nested

PCR approach with two rounds of amplification was used for generating the archaeal

16S rRNA library that targeted the V3-5 region of the 16S rRNA gene. The PCR reaction concentrations were the same as the cyanobacterial library except the first round used 400 nM each of primers 23F and 958R (Barns et al., 1994; Delong, 1992) and 10 ng of genomic DNA, while the second round of PCR used 400 nM each of primers 334F and 915R (Casamayor et al., 2002) with 10 ng of round one amplicon as a template. The PCR conditions for both rounds were similar to the cyanobacterial run conditions except the annealing steps were at 55°C for 1.5 min for the first round, and

61°C for 1 min for the second round. A PCR Purification kit (Qiagen) was used to clean up the first and second rounds of PCR. Negative controls were conducted in parallel with for each round of PCR and results indicated no contamination. All amplicons from the second round were combined in equimolar amounts.

Bacterial, cyanobacterial, and archaeal 16S rRNA barcoded libraries were sequenced from the 454 Life Sciences A primer using the standard GS FLX chemistry

(454 Life Sciences, Branford, CT) in a single run performed by the UF-ICBR. The raw

sequence reads and quality files were deposited into the NCBI sequencing read archive

under project number SRP004035.

Bioinformatic Analysis of Barcoded 16S Ribosomal RNA Gene Libraries

The open-source software mothur v 1.11.0 was used for processing, cluster

analysis, and classification of the raw barcoded sequences from the 16S rRNA gene

libraries (Schloss et al., 2009). The bacterial, cyanobacterial, and archaeal libraries

were analyzed separately since they covered different variable regions of the 16S rRNA

gene. Those sequence reads that did not contain an exact match to the primer

sequences, contained any ambiguous reads, and/or had an average quality score of

less than 27 were considered to be poor quality and were excluded from the rest of the

analysis (Kunin et al., 2010). The remaining 454 reads were subjected to two additional

screening criteria. First, based on the finding that short 454 reads may be poor quality,

all reads below 125 bp for the bacterial and 200 bp for the cyanobacteria and archaeal libraries were removed (Huse et al, 2007). Secondly, based on limitations of the GS-

FLX standard chemistry reads longer than 275 bp for bacterial and 300 for cyanobacterial and archaeal sequences were removed. The corresponding mat type for each retained barcoded read was recorded and the primers were trimmed from the sequences. Since sequencing was done from the reverse primer, the reverse complements were taken and exact replicate reads were removed to speed downstream analyses.

The remaining sequences were aligned in mothur using the nearest alignment space termination (NAST) algorithm against a bacterial and archaeal SILVA 16S rRNA gene template, as this template has been shown to provide quality alignments of

variable regions (DeSantis et al., 2006; Pruesse et al., 2007; Schloss, 2009). No

sequence mask was used in this analysis as it has been previously shown to

significantly reduce the diversity observed between sequences (Schloss, 2009). The

alignment was trimmed such that all reads were aligned to the same exact region. To

account for potential pyrosequencing errors the sequences from each of the three

libraries were analyzed with the pre.cluster function in mothur, which is based on the

SLP clustering algorithm (Huse et al., 2007). Once sequencing error was minimized a pairwise distance matrix was calculated using mothur and reads were clustered into operational taxonomic units (OTUs) at 3% distance using the furthest neighbor method

(Schloss and Handelsman, 2005). SILVA alignments of representative sequences from each OTU for all three libraries were extracted from the mothur generated NAST alignments and imported into the ARB software package (Ludwig et al., 2004). The sequences were inserted into a tree composed of full-length 16S rRNA sequences from the SILVA SSU Ref v102 database using the parsimony option with domain specific position variable filters. The trees were exported to Xfig visualization software (http:

//www.xfig.org). The representative sequence from each OTU was classified using mothur with a naïve Bayesian approach at a confidence threshold of 70% (Wang et al.,

2007). The bacterial and archaeal reference databases were composed of unique, full- length sequences from the SILVA SSU Ref v102 database (http:

//www.mothur.org/wiki/Silva_reference_files) and used to classify reads from their respective barcoded amplicon libraries.

To account for the effects of different sampling depths on the alpha-diversity measurements three replicate subsets from each thrombolitic mat type, containing an

equal number of sequences, were created for each library using a random sequence selector Perl script (Paul Stothard; www.ualberta.ca/~stothard/software.html). The number of randomly sub-sampled sequences was calculated to capture 75% of the sequences in the least represented mat type (i.e. 1,248 for Bacteria, 2,548 for

Cyanobacteria, and 178 for Archaea). The sub-sampled OTUs from each mat type were used to generate alpha-diversity statistics including Chao1 non-parametric species richness estimate, Shannon indices, Shannon-based richness estimate, and Good’s coverage estimate.

Community Comparisons Between Thrombolitic Mat Types

Community analysis and comparison of the thrombolitic mat types were performed phylogenetically using Unifrac jackknifed environmental clustering and Fast

UniFrac, a web-based program designed for the comparing distances between communities in large datasets (Lozupone and Knight, 2005; Hamady et al., 2009). For the jackknifed clustering analysis 1000 permutations were run with a sampling size of

75% of the total number of sequences in the smallest library (1249 bacteria; 2349 cyanobacteria; 178 archaea). The Fast UniFrac analyses were used to determine whether Alphaproteobacteria and Cyanobacteria contributed to community structure.

Reads classified as belonging to these phyla were removed from the bacterial 16S rRNA gene library to create two artificial libraries: without Alphaproteobacteria (No-Ap) and without Cyanobacteria (No-Cy). The No-Ap and No-Cy libraries were combined with the original bacterial library and aligned within mothur. Phylogenetic trees for the bacterial (original, No-Ap, and No-Cy), cyanobacterial, and archaeal libraries for

UniFrac analysis were constructed from the NAST alignments generated above using

FastTree, a relaxed maximum-likelihood tree-building program (Price et al., 2010). Tab-

delimited sample ID mapping files were generated for each library with data collected

during mothur analysis. Principal coordinate analysis (PCA) for the three libraries was

performed on weighted and normalized data within Fast UniFrac (Hamady et al., 2009).

Results

Morphological Characterization of Thrombolitic Mat Types

Thrombolitic microbialites were localized to the intertidal zone of a fringing reef

complex along the eastern margin of Highborne Cay, Bahamas (Figure 2-1). The

distribution of the thrombolite structures extended over 1 km of the intertidal zone and

their sizes ranged up to several meters in length and width (Figure 2-1A). The surfaces

of the thrombolitic structures had four distinctive microbial mat communities that were

spatially distributed along the intertidal zone (Figure 2-1B) and were characterized by

the texture and pigmentation of the mats (Figure 2-2A-D). Those near shore

thrombolites that recently emerged from sand burial prior to collection were covered

with a dark pigmented mat referred to as “black” mat (Figure 2-2A). The black mats

exhibited a moderately lithified crust in the upper few mm of the mat. The second mat

type labeled as “beige” mats lacked any obvious surface pigmentation but exhibited

extensive encrustation of the surface (Figure 2-2B). The remaining two mat types have

been previously described (Myshrall et al., 2010) and were found on the most seaward side of the site. These mats included a smooth flat mat referred to as “pink” mats that had extensive red alga colonization (Figure 2-2C) and like the beige mat, had extensive lithification throughout the mat. The fourth mat was an irregular nodous mat referred as

“button” mats (Figure 2-2D) that contained numerous patches of vertically orientated calcified filaments. Field observations of the spatial organization of the four mat types

revealed that the button mat types were consistently superposed over the other mat

types, most commonly pink or beige mats.

Cross sections of the four mat types revealed that although there was no

laminated mesostructure in the thrombolitic build-ups, the surface microbial mat

communities did exhibit some layering (Figure 2-2E-H). Higher magnification of the four

mat types revealed that each community had a dominant ecotype that morphologically

distinguished it from the others (Figure 2-2E-P). In the black mats, the surface was

heavily colonized by clusters of a pigmented coccoid cells that ranged between 3 – 5

µm in diameter (Figure 2-2I,M). Coccoid cells dissected from the black mats had 16S rRNA gene sequences that shared 97% similarity to cyanobacterial Pleurocapsa spp. isolates from stromatolites located in Shark Bay, Australia. The beige mats were distinguished by the presence of a pronounced subsurface cyanobacterial layer rich in coccoid cells, which shared morphological similarity to Gloeocapsa spp. (Figure 2-2J,N).

The algal cells that gave the pink mats their color (Myshrall et al., 2010) were concentrated in the upper 500 µm of the mat surface (Figure 2-2K), and each cell was encased in a thick sheath (10 - 20 µm; Figure 2-2O). The red algae were previously classified as belonging to the genus Chlorophyta (Myshrall et al., 2010). The fourth mat type, referred to as the button mat types was the most widely distributed of all the mat types, and had a unique morphology in which calcified filaments formed vertical bundles throughout the surface nodes of the mat (Planavsky et al., 2009). The filaments were previously classified as belonging to the genus Dichothrix (Order Nostocales) based on the morphological characteristics of basal heterocysts and tapered apical ends

(Planavsky et al., 2009). In this study, we genetically characterized the organism using

16S rRNA gene sequencing of filaments dissected from the mats. After gene sequence

alignment, two distinct clades of organisms from the genus Dichothrix were identified,

and sequences of these clades varied up to 1.5%, and both clades shared 97%

sequence similarity to Calothrix spp. PCC7507 and Rivularia spp. PCC7116.

Small Subunit rRNA Gene Diversity in Thrombolitic Mats

Three 16S rRNA gene amplicon libraries were generated for each of the four mat

types (with eight pooled DNA extractions) and included a bacterial, cyanobacterial, and

archaeal library. Each of the 12 libraries was labeled with a unique oligonucleotide

barcode and pyrosequenced using 454 GSFLX-technology. A total of 53,456 barcoded

pyrosequences were recovered from the combined libraries with an average read length

of 260 bp. After screening for quality (see Materials and Methods) 34,027 sequences

remained with an average read length of 240 bp and were sorted by barcode, aligned

and analyzed with mothur (Schloss et al., 2009). Each mat type was represented by

between 6,000 and 11,000 sequencing reads, which were clustered into operational

taxonomic units (OTUs) at 97% sequence similarity (Table 2-1). Based on mothur

clustering, the number of OTUs in the thrombolitic mats ranged between 807 and 1193

in four thrombolitic mat types with the pink mats having the lowest number of total OTUs

and the button mat type having the highest (Table 2-1). Equalized sequence data were used to generate diversity indices for each primer set and thrombolitic mat type (see

Materials and Methods; Table 2-1). Compared to previous microbialitic mat diversity analyses, there was a 10-fold increase in the number of bacterial rRNA gene sequences recovered from these thrombolitic mats (Goh et al., 2008; Foster and Green, 2011;

Myshrall et al., 2010).

Bacterial Community Composition

Classification and community analysis of the domain Bacteria sequence libraries

were based on a defined region of the variable 2 (V2) region of the 16S rRNA gene in

all compared reads (Table 2-2). The libraries were then clustered into operational

taxonomic units (OTUs) at 97% similarity or greater. The thrombolitic bacterial

sequence libraries were composed of sequences from 13 different phyla with extensive

variation in the proportional distributions of these phyla between mat types (Figure 2-

2A,D). Proteobacteria dominated all mat types and comprised 45 - 68% of the total sequences from each library (Figure 2-3A). Within the Proteobacteria,

Alphaproteobacteria were by far the most dominant class comprising more than 60% of the recovered Proteobacteria reads in the beige, pink, and button mats and 45% of the black mats (Figure 2-3B). Bayesian analyses at a 70% confidence interval (CI) indicated that 14 - 30% of the recovered Alphaproteobacteria sequences could not be assigned to any described order. The remaining sequences shared sequence similarity (95 - 99%) to purple non-sulfur phototrophs, specifically the Rhizobiales and Rhodobacterales

families (Imhoff, 2008). Deltaproteobacterial sequences were detected in variable

abundances (3.8 - 8.6%) within the different mats. The beige mats had the highest

proportion of Deltaproteobacteria and were rich in sequences belonging to the order

Myxococcales and to metabolically versatile sulfate-reducing bacteria of the order

Desulfobacterales (Selenska-Pobell et al., 2002). The deltaproteobacterial sequences in

the pink and black mat sequence libraries were predominately Myxococcales. In the

button mats most of the recovered Deltaproteobacteria sequences were highly novel

and could not be reliably classified at deeper taxonomic levels. There were also

differences in the relative abundance of Gammaproteobacteria populations between

mat types, with the black and the beige mat types showing an enrichment of

Gammaproteobacteria, but no difference in diversity. In all four of the mat types BLAST

results of the recovered Gammaproteobacteria sequences indicated similarity (96-98%)

to methylotrophic bacteria, particularly Oceanospiralles and Methylococcales.

Jackknife environmental clustering using the domain Bacteria sequence data

indicated that the beige and the pink communities were the most similar and that the

button mat was the most distant mat type (Figure 2-2D). The clustering was

complemented by an analysis of bacterial richness using the number of shared and

unique OTUs in the four mat types (Figure 2-4). The four communities shared 59 OTUs,

which accounted for 29.5% of the total pyrosequencing reads. Thirty-three of these

OTUs were classified as Alphaproteobacteria while the rest were similar to members of

Gammaproteobacteria, Deltaproteobacteria, Actinobacteria, and Cyanobacteria. In

addition to the shared OTUs, each mat community contained OTUs that were unique to

each mat type. The highest number of unique OTUs occurred in the black (388) and

button (311) mats, with the lowest in the beige (222) and pink mat types (210). The

black mats contained several unique Pleurocapsales that were not found in other mat

types. The beige mats contained several unique Desulfobacteraceae that have been

previously associated with sulfate reduction (Miyatake et al., 2009). In the button library

there were unique sequences that shared similarity to Saprospiraceae (Bacteroidetes),

members of which have been shown to degrade complex organic compounds (Hosoya

et al., 2006; Xia et al., 2008).

The clustering analyses were complemented by using Fast UniFrac (Hamady et al., 2010) to compare the bacterial community structure in the four mat types (Figure 2-

5). Since the bacterial sequences represent the V2 partial 16S rRNA gene region, a guide tree was constructed in FastTree using only these sequences. A Principal

Coordinate Analysis (PCA) plot generated from weighted and normalized Fast UniFrac data revealed a clustering pattern of the four mat bacterial populations (black symbols) similar to the jackknife analysis (Figure 2-5); with the beige and pink mats more similar to each other than the black or button mats. To assess whether the two dominant phyla,

Alphaproteobacteria and Cyanobacteria, were significantly influencing community structure, two data subsets were created from the original Bacteria pyrosequencing data, one without the Alphaproteobacteria and one without the Cyanobacteria. The data set without the Alphaproteobacteria (gray symbols) were shifted to the right with respect to the original communities with the greatest differences detected in the beige and pink mat types. The datasets with only the Cyanobacteria removed (white symbols) resulted in a smaller shift to the left with very little change detected in the beige and button mat types. The differential clustering of these communities also illustrated how the bacterial population structure was influenced by environmental variables. Principal Component 1

(P1), which accounted for 48% of the variation in community distribution and evenness, was correlated with length or rate of burial events. The black mats, which were recently unburied at the time of collection, are often subjected to repeat sand burial due to their geographical location in the intertidal zone, whereas the button mats represent the most seaward mat type and undergoes the fewest sediment burial events. Principal

Component 2 (P2), which accounted for 20% of the variation, was correlated with the extent of lithification; the black and button mats exhibited the least amount of lithification while the beige and pink mats were highly encrusted mat types.

Cyanobacterial Community Composition

Cyanobacteria are key components of microbialitic mat communities (Reid et al.,

2000; Baumgartner et al., 2009; Foster et al., 2009; Foster and Green, 2011). To examine and compare the cyanobacterial populations an amplicon library was generated to variable region 4 (V4; Nübel et al., 1997). The relative abundance of cyanobacterial orders characterized by the V4 region of the 16S rRNA gene showed that the four thrombolitic mat communities contained complex and diverse populations of Cyanobacteria (Figure 2-2B). As was detected with in the bacterial libraries, the black mats were enriched with sequences with similarity to the non-heterocystous, coccoid order Pleurocapsales. The relative abundance of the Pleurocapsales in the four mat types decreased in the more seaward mat types (i.e., pink and buttons). In the beige and the button mats there was an enrichment of sequences with similarity to the filamentous, non-heterocystous Oscillatoriales. In both mat types around 70% of the sequences were classified as belonging to Halomicronema, which are common constituents in many hypersaline cyanobacterial mats (Abed et al., 2002; Green et al.,

2008; Allen et al., 2009). There were also novel Oscillatoriales ecotypes in the beige

(23%) and button (19%) that could not be assigned to a class using a Bayesian approach with a 70% CI. The most seaward button mats exhibited an enrichment of

Nostocales sequences that accounted for 8% of the total sequences in the button mats and shared similarity (97%) to the Calothrix spp. PCC7507 and Rivularia spp. PCC7116

(Figure 2-2B). Only a few sequences (< 0.1%) of the Nostocales sequences were found in the other three mat types. Jackknife clustering analysis of the cyanobacterial libraries

at the sequence-level revealed a different clustering pattern compared to the Bacteria

domain-specific library (Figure 2-2E) with the black mats being the most distant mat type.

The thrombolitic mat cyanobacterial diversity was further explored by constructing a phylogenetic tree with representative sequences from each OTU formed at 97% similarity from the cyanobacterial libraries (n = 1,275 OTU, 22,934 sequences) and those classified as cyanobacteria from the bacterial libraries (n = 98 OTU, 861 sequences; Figure 2-6). The cyanobacterial sequences formed sixteen distinct clusters, eleven of which were previously described (Myshrall et al., 2010). Six of the clusters contained sequences that were unique to the Highborne Cay thrombolitic mats (Figure

2-6; black-filled) and were associated with the orders Chroococcales, Oscillatoriales, and Nostocales. Nearly all of the Chroococcales sequences recovered from the pink mats formed Cluster 6, which shared similarity (99%) to the metabolically versatile

Synechocystis spp. PCC 6803 (Anderson and McIntosh, 1991). Sequences forming

Clusters 7, 10, 11, 12 and 15 were enriched in the button mats, including the two distinct Dichothrix spp. genotypes within the order Nostocales. Most of the sequences in

Cluster 11 came from cloned 16S rRNA genes from isolated Dichothrix spp. filaments, whereas Cluster 12 contained the majority of the recovered 454 pyrosequencing reads.

Eight of the cyanobacterial clusters also contained environmental clones from both Highborne Cay and Shark Bay stromatolitic mats (Figure 2-6; gray-filled; Papineau et al., 2005; Baumgartner et al., 2009; Foster et al., 2009; Goh et al., 2009). Sequences of Cyanobacteria from Cluster 1 were highly abundant in the black mats (20.7%) and contain the Pleurocapsa spp. clones that are responsible for the black surface pigmentation. Cluster 4 shared sequence similarity to Stanieria cyanosphaera and

contained between 12 to 17% of the sequences from the beige, pink, and button mats.

This cluster also contained sequences that were abundant in the Highborne Cay Type 2

and 3 stromatolite clone libraries (Baumgartner et al., 2009; Foster et al., 2009). Cluster

14, which contained 19% of the sequences from the button mats, also included

Leptolyngbya-related clones from the adjacent stromatolites (Foster and Green, 2011)

and the freshwater Ruidera Pool stromatolites from Spain (Santos et al., 2010).

Archaea Community Structure

The archaeal community was examined by generating an amplicon library that targeted the V5 region of the 16S rRNA gene using Archaea-specific primers (Delong,

1992; Barns et al.,1994). We recovered 2,362 quality sequences with a Good’s estimated coverage of between 96 and 99% in all four mat types. The recovered number of observed OTUs was low and ranged between 6 in the pink mats to 14 in the black mats (Table 2-1). The estimated diversity based on Chao1 values was also low and ranged between 17 in the pink mats and 24 in the black mats. Despite the low number of OTUs, the diversity indices of the four of the thrombolitic mats were higher than those reported in the Shark Bay stromatolites (Goh et al., 2009) but less diverse than the archaea populations reported from nonlithifying microbial communities such as the Guerrero Negro mats (Robertson et al., 2009). The recovered archaeal sequences

were classified to the order level and compared between mat types (Figure 2-2C,F).

Sequences similar to Cenarchaeales, belonging to the Thaumarchaea phylum

(Brochier-Armanet et al., 2008), were dominant in the in the black and beige mats and

decreased in abundance in the more seaward pink and button mats. The

Halobacteriales of the phylum Euryarchaeota dominated the pink mat types, but were

present at lower levels in the black, beige and button mats. The highest level of diversity

was detected in the button mat types, where the Thermoplasmatales (54%) and

unclassified Euryarchaeota (10%) sequences made up most of the button mat archaeal

community. Hierarchical clustering revealed a different relationship between the four

mat communities (Figure 2-2E). Unlike the clustering in the bacterial populations, the

archaeal populations of the near shore black and beige mats were more similar to each

other than those found in the more seaward pink and button mats.

A phylogenetic tree constructed with archaeal sequences revealed six distinct

clusters (Figure 2-7; black-filled) that were specific to the Highborne Cay thrombolites.

The closest relatives to these clusters included both lithifying (Figure 2-7; gray-filled) and non-lithifying mat (Figure 2-7; white-filled) communities (Jahnke et al., 2008; Allen et al., 2009; Goh et al., 2009; Robertson et al., 2009). The largest cluster, which included Cenarchaeum-like sequences, shared similarity to the sponge isolate

Cenarchaeum symbiosum. This group of sequences clustered separately from the

Nitrosopumilus-like clones found in the Shark Bay stromatolites (Preston et al., 1996;

Burns et al., 2004). The thrombolitic mats Cluster 3 contained Halobacteria-like sequences that were closely related to two known halophilic isolates, Haloarcula japonica and Natronomonas pharaonis (Figure 2-7). No sequences derived from known methanogens were detected in the archaeal sequence libraries.

Discussion

There has been extensive research on the microbialites of Highborne Cay, The

Bahamas, however, most of this research has been focused on the laminated stromatolitic microbialites (e.g. Pinckney and Reid, 1997; Reid et al., 2000; Visscher et al., 2000; Dupraz and Visscher, 2005; Baumgartner et al., 2009; Foster et al., 2009).

Only a handful of studies have examined the unlaminated clotted thrombolites of

Highborne Cay and these studies have targeted aspects of carbonate deposition

(Planavsky et al., 2009), phage diversity (Desnues et al., 2008) and geomicrobiology

(Myshrall et al., 2010) within the structures. The work by Myshrall et al. (2010) also

provided the first insight into the microbial ecology of the thrombolitic structures at

Highborne Cay; however, this study provided only a limited coverage of the microbial

communities associated with the thrombolites. In this study we expanded on this

previous work by characterizing two additional thrombolitic mat types and examining the

archaeal diversity within all four mats, thus providing a more comprehensive and

comparative sequencing analysis of the thrombolitic mat communities. The results of

our study provide morphological and genetic evidence that there are at least four

thrombolitic mat types at Highborne Cay and that these four mat types may represent

successive stages of thrombolitic mat development.

Differences in Microbial Populations Between Thrombolitic Mat Types

The pyrosequencing of 16S rRNA gene amplicon libraries generated to each of the four thrombolitic mat types dramatically increased the sequencing coverage of the bacterial (80 - 82%), cyanobacterial (95 - 96%), and archaeal (96 - 99%) communities

(Table 2-1) compared to the previous study by Myshrall et al., (2010), which relied on traditional clone libraries. The pyrosequencing results indicate the thrombolitic mat communities are significantly more diverse than previously measured with an estimated

807 to 1193 OTUs in each of the mat types and bacterial diversity Shannon indices ranging between 5.23 - 5.60 (based on equalized sequences; Table 2-1). These values are higher than any previously described microbialitic mat community (Baumgartner et al., 2009; Goh et al., 2009; Myshrall et al., 2010; Foster and Green, 2011) and likely reflect the 10-fold increase in sequencing coverage. However, estimates of species

richness and evenness can also be influenced by the length of the amplified region, with

shorter partial reads having higher diversity estimates than longer 16S rRNA gene

reads (Engelbrekston et al., 2010; Harris et al., 2010). These discrepancies could partially explain the higher diversity of the thrombolitic pyrosequencing libraries relative to the clone libraries from microbialitic mats (Myshrall et al., 2010; Table 2-1). To prevent such an overestimation of microbial diversity within the four thrombolitic mats we used a pair-wise alignment and similarity threshold of 97% over the same exact read length as well as a pre-clustering approach to account for potential sequencing errors

(Huse et al., 2010). These approaches have been previously shown to be effective cutoffs to reduce overestimations of microbial diversity (Huse et al., 2007; Kunin et al.,

2010; Schloss, 2010). We implemented these approaches using the mothur software

(Schloss et al., 2009) with the stringent quality control trimming resulting in 28 to 42% of our raw pyrosequences discarded during pre-processing, thereby minimizing most of the pyrosequencing errors and potential sources for diversity overestimation.

Cyanobacterial Diversity in Thrombolitic Mat Communities

Cyanobacteria have long been known to be a dominant functional group in microbial mat ecosystems (Canfield and Des Marais, 1993; Reid et al., 2000; Ley et al.,

2006; Foster and Green, 2011) and are critical for the accretion and early lithification of stromatolitic mats (Pinckney and Reid, 1997; Dupraz et al., 2004). To provide a deeper survey of the Cyanobacteria within the thrombolitic mats and overcome potential biases of using a single primer set, cyanobacterial-specific primers that targeted the V4 region

(Nübel et al., 1997) were used in addition to universal bacterial primer (Table 2-2). The cyanobacterial populations in the four mats varied significantly in their order level distribution with coccoid, non-heterocystous ecotypes abundant in the black, beige, and

pink mats, and filamentous ecotypes found in the button mats (Figure 2-2B). There were, however, eight cyanobacterial clusters detected in the thrombolitic mats that appear to be specific to microbialitic mat communities (Clusters 1 - 5, 9, 13, 14; Figure

2-6). One of these clusters, Cluster 1, contained sequences similar to the non- heterocystous coccoid Pleurocapsales and was found at a relative abundance of at least 10% in all four mat types. Previous studies have shown that these endolithic

Pleurocapsales ecotypes are resistant to extensive desiccation and UV radiation

(Krumbein and Giele, 1979; Billi et al., 2000; Dillon et al., 2002). The high relative abundance of these ecotypes in the black and beige mats relative to the pink and button mats may reflect the near shore location of these mats as these two mats are often fully exposed in the upper intertidal zone. Not all thrombolite cyanobacteria were unique to this environment, and some cyanobacterial taxa such as those in Cluster 14 (Figure 2-

6) are detected in other microbialite ecosystems. For example Cluster 14 contained sequences similar to the genus Leptolyngbya, with most sequences recovered from the button mat types. Leptolyngbya are non-heterocystous filamentous cyanobacteria that have been found in all modern stromatolitic mat systems to date, including both marine and freshwater environments (Foster et al., 2009; Goh et al., 2009; Santos, et al., 2010) and may represent a common microbialitic mat building organism.

Although there were several clusters recovered from the thrombolitic mats that appear to be common to all microbialitic mats there were six clusters that were unique to the Highborne Cay thrombolitic mats and may contribute to the differences in clotted thrombolitic microfabrics compared to laminated stromatolites. The most notable of these six clusters were Clusters 11 and 12 that were enriched in the button mat types.

These two clusters were classified as the Order Nostocales and shared sequence similarity to the heterocystous, filamentous Calothrix spp. and Rivularia spp. (Rippka et al., 2001). Cluster 11 contained all of the cloned Dichothrix spp. 16S rRNA genes that were collected from the button mat type along with relatively few of the recovered pyrosequences, which may suggest some amplification or cloning bias for this ecotype.

All of the sequences associated with Cluster 12 were recovered from the pyrosequencing results and were also similar to the freshwater Calothrix spp. and to a cluster of environmental clones isolated from freshwater microbialites in Ruidera Pools in Spain (Santos et al., 2010). These results suggest that there may be two distinct populations of Dichothrix within the thrombolitic mats. However, the results may be the product of heterogeneity in the 16S rRNA gene. Extensive heterogeneity of the 16S rRNA gene is prevalent in strains of Calothrix (Berrendero et al., 2008) and previous studies have shown that a single morphotype of Calothrix in cyanobacterial mats comprised five different 16S rRNA genotypes (Hongmei et al., 2005). Further work on the Dichothrix isolates is required to resolve this issue.

Low Diversity Of Archaea in Highborne Cay Thrombolitic Mats

The 16S rRNA gene libraries generated in this study suggested that Archaea diversity was low and accounted for only 1.8% of recovered OTUs. The observed archaeal diversity indices were the lowest in the pink and beige mats (1.17 and 1.37 respectively) with the highest associated with the button mats (2.23). The results are similar to community analyses of other microbialitic mats where Archaea diversity was also low (Papineau et al., 2005; Goh et al., 2009). All six of the clusters contained sequences associated with only the thrombolitic mats (Figure 2-7). However, as no comprehensive archaeal diversity analysis has been conducted for the adjacent

stromatolites at Highborne Cay it is not clear whether these clusters are specific to the

thrombolitic mats or to the geographical location. The clusters were closely related to

other lithifying and nonlithifying microbial mat ecosystems such as Shark Bay

stromatolites (Papineau et al., 2005; Goh et al., 2009) and the hypersaline mats of

Guerrero Negro (Jahnke et al., 2008; Robertson et al., 2009). The majority of the archaeal sequences from the black and beige mats (Cluster 5) was related to a sponge symbiont, Cenarchaeum symbiosum (Hallam et al., 2006), and shared similarity to a proposed mesophilic phylum, Thaumarchaea (Brochier-Armanet et al., 2008). The

Cenarchaeum cluster was distinct, however, from Thaumarchaea found in the Shark

Bay stromatolites (Goh et al., 2009), which shared similarity to chemolithoautotrophic ammonia oxidizing Nitrosopumilus maritimus (Könneke et al., 2005). Both the C. symbiosum and N. maritimus genomes have been shown to contain similar ammonia oxidation and carbon fixation genes (Walker et al., 2010). These archaea could be contributing to nitrification and carbon cycling in the thrombolitic mats.

No archaeal ecotypes associated with methanogenesis were recovered from the thrombolitic mats nor have they been recovered from other microbialites such as the stromatolites of Shark Bay (Papineau et al., 2005; Goh et al., 2009). Although methanogenesis has been reported for other microbial mat systems such as the hypersaline mats from Guerrero Negro (Smith et al., 2008; Robertson et al., 2009) and

Solar Lake (Giani et al., 1984) methanogenesis is likely inhibited in the thrombolitic mats due to the presence of sulfate reducing bacteria, which outcompete methanogens for substrates, such as H2 and acetate (Ehrlich, 2002; Schimel, 2004). Despite the lack of methanogenic archaea recovered from the thrombolitic mats, methanotrophic bacteria

were detected in the button thrombolitic mats (e.g. Oceanospiralles spp. and

Methylococcus spp.) and may be generating methane by using noncompetitive osmoprotectants such as glycine betaine, choline and trehalose. Genes encoding these putative substrates have been found in the thrombolite metagenome (Foster unpublished). Methane production from the fermentation of complex amines including osmoprotectants has been shown to occur in a wide range of habitats including microbial mats and sediments (King, 1984; Oh et al., 2008).

A Model for Microbial Succession in Thrombolitic Mat Types

Succession in microbial communities has been well documented in microbial mat systems (Stal et al., 1985; Stolz, 2000; Grootjans et al., 1997) and is largely influenced by the changes in the environment (Mackie et al., 1999; Haruta et al., 2004; Sato et al.,

2009; Jones and Lennon, 2010). Based on the diversity analyses, we propose that the four thrombolitic mat types represent successive stages of thrombolite development in

Highborne Cay, Bahamas (Figure 2-8). Although field observations of the four distinct thrombolitic mat communities have consistently showed that button mats were superposed over the pink and beige mats, the 16S rRNA gene analysis supported these patterns of thrombolitic mat transitions.

The jackknife environmental clustering of the bacterial and cyanobacterial libraries and Fast Unifrac results indicated that the black and the button mats were the most distinctive mat communities (Figure 2-2; 2-6). We propose that the black mats, which are the most beachward mats, represent an initial surface community after thrombolites emerge from extended sand burial events. The duration of burial for the microbialites of Highborne Cay can vary from a few days to several weeks (Andres et al., 2006) and the geographical location of the black mats in the near shore intertidal

zone render these communities more susceptible to longer periods of burial, low light,

and desiccation. These conditions may impose a selection pressure on maintaining the

Pleurocapsales-dominant community in the black mats as Pleurocapsales have been

found to be resistant to low light conditions (Krumbein et al., 1979; Kremer and

Kaźmierczak, 2005) and desiccation (Krumbein and Giele, 1979). We propose that upon environmental disturbances, such as extensive sediment burial, thrombolitic mat communities shift to the black mat state (Figure 2-8) and that many of the ecotypes typically enriched in other mat types either die or enter a dormant state. Dormancy has been shown to be widespread in microbial ecosystems accounting for up to 80% of microbial populations (Cole, 1999; Cáceres and Tessier, 2003) and may facilitate an organism’s ability to survive extensive environmental perturbations (Jones and Lennon,

2010).

Once the mats become unburied due to high-energy wave action, we propose the black communities experience an environmental trigger that results in the transition to either beige or pink mat communities (black arrows; Figure 2-8B). The beta-diversity measures such as the jackknife environmental clustering (Figure 2-2D-F) and Fast

Unifrac analyses (Figure 2-5) indicated that the bacterial populations of beige and pink mats were more similar to each other than to either the black or button mats. Even when the two dominant taxa (Alphaproteobacteria and Cyanobacteria) were removed from the pyrosequencing dataset the beige and pink communities were more similar to each other, suggesting that these two mat types are closely related (Figure 2-5). The environmental factors that may initiate the transition to beige and pink may the result of light and nutrient availability. Previous studies have shown that light availability can

have a strong impact on the composition of cyanobacterial communities (Havens et al.,

1998). As photosynthetic mat communities experience increases in light availability previous studies have shown there is an increase in uptake of dissolved organic carbon

(Yannarell and Paerl, 2007). The rate of uptake is influenced by the type and composition of the cyanobacterial communities (e.g. filamentous, colony coccoid or solitary coccoid; Yannarell and Paerl, 2007). The transition to the coccoid rich communities of the beige (enriched in colony forming Pleurocapsales) or the pink

(enriched in solitary Chroococcales) mats may reflect seasonal differences in uptake or differences in the substrate-utilization patterns of the two distinct mat types.

After a transition to either beige or pink mat types, we propose there is a trend toward the succession of the button thrombolitic mat communities. Besides being the most abundant mat type, the button mats are the most productive (O2 and DIC

production) with rates higher than that of the adjacent stromatolites (Myshrall et al.,

2010). The button mats also contained the highest cyanobacterial and archaeal diversity compared to the other three thrombolitic mats (Table 2-1; Figure 2-2). It is possible that this taxonomic diversity corresponds to greater metabolic diversity, which could influence biologically controlled mineralization in the thrombolites (Dupraz and Visscher,

2005). The presence of calcium carbonate precipitation on the sheaths of Dichothrix cells is unique to the button thrombolitic mats (Planavsky et al., 2009), and the 16S rRNA genes of Dichothrix have not been recovered from the adjacent stromatolites

(Foster and Green, 2011). The transition towards the button mat type likely reflects multiple environmental cues and may include factors such as eukaryotic grazing and shifts in pH. Although eukaryotic diversity has been shown to be low in the button

thrombolitic mats numerous sequences with similarity to ecotypes associated with

bacterivorous grazing (e.g. nematodes) have been recovered (Myshrall et al., 2010).

The prevalence of filamentous cyanobacteria and bacteria within the button mats may provide resistance to grazing activity, as previous studies have shown that meiofaunal grazers selectively target smaller coccoid or rod-shaped ecotypes (Jürgens et al.,

1999).

The transitions between mat types may also reflect a shift in the pH, which may result in the differences in the patterns of biologically induced biomineralization. Shifts in pH have been shown to trigger succession events in complex microbial communities

(Haruta et al., 2004) and in the button mats there are numerous ecotypes associated with metabolisms that influence carbonate alkalinity, such as oxygenic, anoxygenic photosynthesis and sulfate reduction (Dupraz et al., 2009). Together these metabolisms

can initiate and promote carbonate precipitation (Dupraz et al., 2009), however more

work is necessary to monitor the specific changes in pH between the four mat types to

assess the impact of pH on each of the thrombolitic mat types.

Table 2-1. Bacterial and archaeal 16S rRNA amplicon diversity analyses of thrombolitic mats types of Highborne Cay, Bahamas. OTUs equal. Chao1 Shannon base sample seqs OTUsb singletsb doubletsb >1% Evenessd % cov.df Seqsa (CI)de (CI)d pairsg (Sum)bc 17 909 ± 57 5.17 ± 0.08 Bac. Black 2662 1248 652 348 115 0.84 ± 0.01 82 ± 0.01 233 (41%) (776/1096) (0.07) 13 883 ± 26 5.55 ± 0.03 Bac. Beige 1839 1248 552 295 93 0.90 82 ± 0.01 236 (22%) (769/1043) (0.06) 19 998 ± 91 5.23 ± 0.09 Bac. Pink 1801 1248 492 277 70 0.86 ± 0.01 84 ± 0.01 238 (36%) (858/1192) (0.07) 15 998 ± 91 5.47 ± 0.04 Bac. Button 1665 1248 540 321 81 0.88 81 ± 0.01 239 (35%) (858/1192) (0.07) 12 401± 44 3.07 ± 0.05 Cya. Black 7851 2548 427 220 65 0.60 96 232 (25%) (317/543) (0.07) 15 410 ± 53 3.56 ± 0.01 Cya. Beige 3398 2548 317 156 53 0.68 95 233 (77%) (344/519 (0.07) 14 350 ± 10 2.72 ± 0.01 Cya. Pink 4147 2548 277 144 44 0.54 96 232 (65%) (287/458) (0.15) 8 479 ± 47 3.80 ± 0.02 Cya. Button 8702 2548 593 292 95 0.69 95 232 (74%) (404/598) (0.06) 14 19 ± 4 1.38 ± 0.12 Arc. Black 574 178 19 7 3 0.51 ± 0.07 98 ± 0.01 250 (67%) (16/39) (0.10) 8 21± 11 1.07 ± 0.08 Arc. Beige 1258 178 20 7 1 0.43 ± 0.05 98 ± 0.02 247 (95%) (15/54) (0.08 6 11 ± 3 1.11 ± 0.12 Arc. Pink 293 178 10 5 0 0.50 ± 0.03 98 ± 0.01 251 (95%) (10/25) (0.12) 14 21 ± 7 .21 ± 0.07 Arc. Button 237 178 19 5 0 0.77 ± 0.01 99 ± 0.01 260 (97%) (18/31) (0.15) aNumber of randomized sequences used to generate the diversity measures. bThe values were calculated based on a 97% similarity threshold. cThe number of OTUs that contain sequences that make up at least 1% of the total number of sequences for each library. In parentheses are the percent of sequences within these OTUs. dMean values ± standard deviation of diversity measures using three iterations of random sequence selection to standardize libraries. eValues in parentheses represent the lower and upper 95% confidence interval associated with the Chao1 nonparametric estimator. fGood’s coverage estimator. gAverage length of quality-trimmed reads in base pairs.

Table 2-2. Primers used for generating 16S rRNA barcoded libraries. Mat Typea 454 Barcodec 16S primer Specificity Referenced primerb 27F B none AGAGTTTGATCCTGGCTCAG Bacteria Suzuki et al., 1996 Black-338R A CCAACCTT TGCTGCCTCCCGTAGGAGT universal Suzuki et al., 1996 Beige-338R A GGAATTGG TGCTGCCTCCCGTAGGAGT universal Suzuki et al., 1996 Pink-338R A AACCAACC TGCTGCCTCCCGTAGGAGT universal Suzuki et al., 1996 Button-338R A TTAAGGCC TGCTGCCTCCCGTAGGAGT universal Suzuki et al., 1996 359Fe B none GGGGAATYTTCCGCAATGGG Cyanobacteria Nübel et al., 1997 Black-781Ra A GCAACAT GACTACTGGGGTATCTAATCCCATT Cyanobacteria Nübel et al., 1997 Black-781Rb A GCAACAT GACTACAGGGGTATCTAATCCCTTT Cyanobacteria Nübel et al., 1997 Beige-781Ra A AACCGCTA GACTACTGGGGTATCTAATCCCATT Cyanobacteria Nübel et al., 1997 Beige-781Rb A AACCGCTA GACTACAGGGGTATCTAATCCCTTT Cyanobacteria Nübel et al., 1997 Pink-781Ra A CTACACC GACTACTGGGGTATCTAATCCCATT Cyanobacteria Nübel et al., 1997 Pink-781Rb A CTACACC GACTACAGGGGTATCTAATCCCTTT Cyanobacteria Nübel et al., 1997 Button-781Ra A GCAACCAT GACTACTGGGGTATCTAATCCCATT Cyanobacteria Nübel et al., 1997 Button-781Rb A GCAACCAT GACTACAGGGGTATCTAATCCCTTT Cyanobacteria Nübel et al., 1997 23Ff none none ATTCCGGTTGATCCTGC Archaea Barns et al., 1994 958Ref none none YCCGGCGTTGAMTCCATTT Archaea Delong, 1992 344Fe B none ACGGGGYGCAGCAGGCGCGA Archaea Casamayor et al., 2002 Black-915R A CCAACCAA GTGCTCCCCCGCCAATTCCT Archaea Casamayor et al., 2002 Beige-915R A CGAACCAT GTGCTCCCCCGCCAATTCCT Archaea Casamayor et al., 2002 Pink-915R A AGACAGTG GTGCTCCCCCGCCAATTCCT Archaea Casamayor et al., 2002 Button-915R A AGACACAG GTGCTCCCCCGCCAATTCCT Archaea Casamayor et al., 2002 a Number denotes the location on the Escherichia coli 16S rRNA gene. b 454 Life Sciences sequencing primer A (GCCTTGCCAGCCCGCTCAGCT) and primer B (GCCTCCCTCGCGCCATCAG) with a TA linker preceding the 16S rRNA gene primer. c Barcodes taken from Hamady et al., 2008, end with a CA linker preceding the 16S rRNA gene primer. d Reference for 16S rRNA gene primer. e Primers contain degenerate bases: Y (C,T), M (A,C). f Archaea specific 16S rRNA gene primers used in the first round of a nested PCR.

A B C D

E F G H

I J K L

eps

M N O P

Figure 2-1. Morphological characterization of thrombolitic mat communities. Surface view of thrombolitic mats: A) black; B) beige; C) pink; D) button. Bar, 1 cm. Cross-section of thrombolitic mats: E) black; F) beige; G) pink; H) button. Bar, 1 cm. 32X Magnification of thrombolitic mats indicating morphologically dominant cells in each mat: I) black; J) beige; K) pink with overlying exopolymeric substances (eps); L) button. Bar, 250 µm. Cell morphology: M) black mat Pleurocapsa spp. Bar, 10 µm; N) beige mat Gloeocapsa spp. Bar 10 µm; O) pink mat Chlorophyta. Bar, 50 µm; P) button mat Dichothrix spp. filaments with calcium carbonate precipitation (cp) on gelatinous sheath. Bar, 20 µm.

A D Mat type %DFWHULDO7D[RQ Acidobacteria Black Actinobacteria Bacteroidetes 99 &KORURÀH[L Cyanobacteria Beige Firmicutes Fusobacteria 100 Gemmatimonadetes Planctomycetes 100 Alphaproteobacteria Pink Betaproteobacteria Deltaproteobacteria Gammaproteobacteria Spirochaetes Button WS3 & TM6 unclassified bacteria 0.05 0% 20% 40% 60% 80% 100% Percent of 454 reads

B E Mat type Cyanobacterial Order E Chroococcales Black Nostocales Oscillatoriales Pleurocapsales unclassified cyanobacteria Beige 100 chloroplasts 100 non-cyanobacteria unclassified bacteria

Button 100

Pink 0.05 0% 20% 40% 60% 80% 100% Percent of 454 reads C F Mat type Archaeal Order Cenarchaeales Black Halobacteriales 100 Thermoplasmatales unclassified crenarchaea unclassified euryarchaea Beige 99 unclassified bacteria unclassified

100 Pink

Button 0.05 0% 20% 40% 60% 80% 100% Percent of 454 reads

Figure 2-2. Comparisons of thrombolitic mats using clustering-based and taxonomic diversity approaches. Dendrograms show UniFrac-based jackknife environmental clustering for A) bacteria, B) cyanobacteria, C) archaeal populations. The percent of bootstrap replicates (n= 1000) are indicated by the nodes. The histograms represent the relative abundance as the percent of the percent of the total sequences classified into a specific taxon by a Bayesian approach at a confidence interval ≥ 70% against the SILVA SSU REF v102 database with D) Bacterial libraries classified at the phyla level (class level for Proteobacterial lineages), E) Cyanobacterial libraries classified at the order level, F) Archaeal libraries classified at the order level.

64 A 100% Proteobacterial Orders 90% Caulobacterales Parvularculales 80% Rhizobiales Rhodobacterales Alpha Rhodospirillales 70% Rickettsiales Sphingomonadales 60% Burkholderiales Hydrogenophilales Beta 50% Neisseriales Nitrosomondales Bdellovibrionales 40% Desulfobacterales Delta Desulfovibrionales 30% Myxococcales Chromatiales 20% Enterobacteriales Legionellales Gamma 10% Xanthomonadales

Percent of pyrosequencing reads 0% Black Beige Pink Button Mat Type B 100% Alphaproteobacterial Families 90% Bradyrhizobiaceae Caulobacteraceae 80% Erythrobacteraceae Hyphomicrobiaceae 70% Hyphomonadaceae Parvularculaceae 60% Phyllobacteriaceae Rhodobacteraceae 50% Rhodobiaceae Rhodospirillaceae 40% Rickettsiaceae Sphingomonadaceae 30% Xanthobacteraceae

20%

Percent of pyrosequencing reads 10%

0% Black Beige Pink Button Mat Type

Figure 2-3. Comparisons of proteobacterial lineages in thrombolitic mats using a clustering-based approach. The histograms represent the relative abundance as the percent of sequences that could be assigned to a specific taxon by a Bayesian approach at a confidence interval ≥ 70% against the SILVA SSU REF v102 database, A) Proteobacteria classified at the order level, B) Alphaproteobacteria classified at the family level.

88 J. M. Mobberley, M. C. Ortega and J. S. Foster

Fig. 3. VennFigure diagram2-4. Venn representationdiagram representation of the of OTUthe OTU richness richness sharedshared between bacterialdistinct Dichothrix genotypes within the order Nostocales. between bacterial16S 16S rRNA rRNA gene genelibraries libraries from the fromfour thrombolitic the four mat types. Total observed richness was 1603 OTUs at 97% similarity. Percentages reflect those OTUsMost of the sequences in Cluster 11 came from cloned thrombolitic matunique types. to Total that mat observed type. richness was 1603 OTUs at 97% similarity. Percentages reflect those OTUs unique to that mat 16S rRNA genes from isolated Dichothrix spp. filaments, type. whereas Cluster 12 contained the majority of the recovered 454 pyrosequencing reads. Foster et al., 2009; Foster and Green, 2011). To examine Eight of the cyanobacterial clusters also contained envi- and compare the cyanobacterial populations an amplicon ronmental clones from both Highborne Cay and Shark library was generated to variable region 4 (V4; Nübel Bay stromatolitic mats (Fig. 5 grey-filled; Papineau et al., et al., 1997). The relative abundance of cyanobacterial 2005; Baumgartner et al., 2009; Foster et al., 2009; Goh orders characterized by the V4 region of the 16S rRNA et al., 2009). Sequences of Cyanobacteria from Cluster 1 gene showed that the four thrombolitic mat communities were highly abundant in the black mats (21%) and contain contained complex and diverse populations of Cyanobac- the Pleurocapsa spp. clones that are responsible for the teria (Fig. 2B). As was detected with in the bacterial librar- black surface pigmentation. Cluster 4 shared sequence ies, the black mats were enriched with sequences with similarity to the non-heterocystous, coccoid order Pleuro- 66 0.3 capsales. The relative abundance of the Pleurocapsales in the four mat types decreased in the more seaward mat types (i.e. pink and buttons). In the beige and the button 0.2 Bl:No-Cy mats there was an enrichment of sequences with similar- Bl P:No-Cy ity to the filamentous, non-heterocystous Oscillatoriales. 0.1 P In both mat types around 70% of the sequences were Bg:No-Cy classified as belonging to Halomicronema, which are Bl:No-Ap Bg common constituents in many hypersaline cyanobacterial 0 Bu:No-Cy mats (Abed et al., 2002; Green et al., 2008; Allen et al., P:No-Ap Bu 2009). There were also unclassified Oscillatoriales -0.1 ecotypes in the beige (22%) and button (23%) that could lithification increasing Bg:No-Ap not be assigned to a class using a Bayesian approach

similarity from the cyanobacterial libraries (n = 1614 OTU, Black Button 24 098 sequences; Fig. 5). The cyanobacterial 388 311 (60%) (58%) sequences formed 16 distinct clusters, 11 of which were 22 previously described (Myshrall et al., 2010). Six of the 75 37 clusters contained sequences that were unique to the Beige Pink Highborne Cay thrombolitic mats (Fig. 5; black-filled) and 222 13 14 210 (40%) (42%) were associated with the orders Chroococcales, Oscilla- 59 toriales and Nostocales. Nearly all of the Chroococcales 42 31 sequences recovered from the pink mats formed Cluster 42 50 6, which shared similarity (99%) to the metabolically versatile Synechocystis sp. PCC6803 (Anderson and McIn- 49 tosh, 1991). Sequences forming Clusters 7, 10, 11, 12 and 14 were enriched in the button mats, including the two Fig. 3. Venn diagram representation of the OTU richness shared distinct Dichothrix genotypes within the order Nostocales. between bacterial 16S rRNA gene libraries from the four Most of the sequences in Cluster 11 came from cloned thrombolitic mat types. Total observed richness was 1603 OTUs at 97% similarity. Percentages reflect those OTUs unique to that mat 16S rRNA genes from isolated Dichothrix spp. filaments, type. whereas Cluster 12 contained the majority of the recovered 454 pyrosequencing reads. Foster et al., 2009; Foster and Green, 2011). To examine Eight of the cyanobacterial clusters also contained envi- and compare the cyanobacterial populations an amplicon ronmental clones from both Highborne Cay and Shark library was generated to variable region 4 (V4; Nübel Bay stromatolitic mats (Fig. 5 grey-filled; Papineau et al., et al., 1997). The relative abundance of cyanobacterial 2005; Baumgartner et al., 2009; Foster et al., 2009; Goh orders characterized by the V4 region of the 16S rRNA et al., 2009). Sequences of Cyanobacteria from Cluster 1 gene showed that the four thrombolitic mat communities were highly abundant in the black mats (21%) and contain contained complex and diverse populations of Cyanobac- the Pleurocapsa spp. clones that are responsible for the teria (Fig. 2B). As was detected with in the bacterial librar- black surface pigmentation. Cluster 4 shared sequence ies, the black mats were enriched with sequences with similarity to the non-heterocystous, coccoid order Pleuro- 0.3 capsales. The relative abundance of the Pleurocapsales in the four mat types decreased in the more seaward mat types (i.e. pink and buttons). In the beige and the button 0.2 Bl:No-Cy mats there was an enrichment of sequences with similar- Bl P:No-Cy ity to the filamentous, non-heterocystous Oscillatoriales. 0.1 P In both mat types around 70% of the sequences were Bg:No-Cy classified as belonging to Halomicronema, which are Bl:No-Ap Bg common constituents in many hypersaline cyanobacterial 0 Bu:No-Cy mats (Abed et al., 2002; Green et al., 2008; Allen et al., P:No-Ap Bu 2009). There were also unclassified Oscillatoriales -0.1 ecotypes in the beige (22%) and button (23%) that could lithification increasing Bg:No-Ap not be assigned to a class using a Bayesian approach

P2- 24.31% of variation explained -0.2 with a 70% CI. The most seaward button mats exhibited an enrichment of Nostocales sequences that accounted Bu:No-Ap for 9% of the total sequences in the button mats and -0.3 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 shared similarity (97%) to the Calothrix sp. PCC7507 and P1- 43.42% of variation explained Rivularia sp. PCC7116 (Fig. 5). Only a few sequences decreasing sediment burial (< 0.1%) of the Nostocales sequences were found in the other three mat types. Jackknife clustering analysis of the FigureFig. 4. 2-5.Principal Principal coordinate coordinate analysis analysis of bacterialof bacterial 16S 16S rRNA ribosomal gene RNA gene libraries cyanobacterial libraries at the sequence level revealed a libraries for the four thrombolitic mats using using Fast Fast UniFrac. UniFrac. T Thehe figure shows a plot of different clustering pattern compared with the Bacteria figure showsthe first a plot two of principle the first coordinate two principal axis, coordinate which represents axis, which 43.42% (P1) and represents 43.42% (P1) and 24.31% (P2) of the variations. Each domain-specific library (Fig. 2B) with the black mats being 24.31% (P2) of the variations. Each symbol represents bacterial sequences symbol representsfrom each bacterialmat library sequences: circles (Bl, from black each mats), mat triangle library: (Bg, beige mats), the most distinctive mat type. circles (Bl,squar blacke (P, mats), pink trianglesmats), diamond (Bg, beige (Bu, mats), button squares mats). The (P, pinkshading of each The thrombolitic mat cyanobacterial diversity was mats), diamondsymbol (Bu,refers button to the mats). sequences The shading that are of present each symbol in each library: black shading refers to the sequences that are present in each library: black further explored by constructing a phylogenetic tree with represents the complete library, gray shading represent libraries with (completeAlphaproteobacteria library); grey (No-Ap, removedAlphaproteobacteria (No-Ap, Alpharemoved);proteobacteria removed); white representative sequences from each OTU formed at 97% white (No-Cy,symbolsCyanobacteria represent librariesremoved). with Cyanobacteria removed (No-Cy, Cyanobacteria removed). © 2011 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 14, 82–100

67 Figure 2-6. Phylogenetic tree of partial cyanobacterial 16S ribosomal RNA gene sequences from pyrosequencing libraries and clones generated from Dichothrix spp. and Pleurocapsa spp. isolates. Clusters containing Highborne Cay (HBC) thrombolitic sequences were assigned numbers 1-16. Clusters that also contained clone sequences from Myshrall and colleagues’ (2010) study were denoted with a P if they contained sequences associated with pink mats, and a B if they contained clones from the button mats. If clusters contained clones from both pink and button mats the cluster was labeled with asterisks(*). Clusters were also labeled based on the environments the sequences were recovered from: grey-filled, HBC thrombolitic mats and other microbialitic mat environments; black-filled, specific to HBC thrombolitic mats; stripe-filled, HBC thrombolitic mats, microbialitic mats and hypersaline mats from Guerrero Negro; white-filled clusters contained no HBC thrombolitic sequences. For each cluster the number of sequences recovered from each mat type are enclosed in parentheses (black, Bl; beige, Bg; pink, P; button, Bu). The scale bar represents 0.10 substitutions per nucleotide position.

Figure 2-7. Phylogenetic tree of partial archaeal 16S ribosomal RNA gene sequences from archaeal 454 libraries. The tree was constructed in ARB by inserting sequences using a parsimony method with archaea specific position variable filters into a tree containing 16S rRNA gene sequences from SILVA SSU Ref v102 database. Clusters containing Highborne Cay (HBC) thrombolitic sequences were assigned numbers 1-10. Clusters were labeled based on the environments the sequences were recovered from: white filled, non-lithifying hypersaline mats from Guerrero Negro and Shark Bay; black-filled, HBC thrombolitic mats; grey-filled, only lithifying Shark Bay stromatolites. For each cluster the number of HBC thrombolitic sequences recovered from each mat are enclosed in parentheses (black, Bl; beige, Bg; pink, P; button, Bu). The scale bar represents 0.10 substitutions per nucleotide position.

69 A B

Figure 2-8. Model for succession of Highborne Cay thrombolitic mat communities. A) Spatial distribution of thrombolitic mat types in situ: black (bl), beige (bg), pink (p), and button (bu). Bar 0.25 m, B) Representation of the potential succession of thrombolitic mats. Black arrows indicate pathways associated with the gradual succession of thrombolitic mat types. Grey arrows indicate pathways that may occur in response to a sediment burial event.

CHAPTER 3 METABOLIC POTENTIAL OF LITHIFYING CYANOBACTERIA-DOMINATED THROMBOLITIC MATS

Introduction

Microbialites are carbonate-depositing ecosystems that are the products of multiple microbial metabolisms and complex biogeochemical cycling. Microbialites have a long fossil record dating back billions of years and are generally thought to be one of the oldest known ecosystems on Earth (Grotzinger and Knoll, 1999). Microbialites are distinguished by their carbonate macrostructure. Those microbialites with a laminated macrostructure are referred to as stromatolites, whereas microbialites with unlaminated clotted fabrics are known as thrombolites. The laminated stromatolites are the result of iterative microbial growth that accrete through the precipitation of calcium carbonate as well as the trapping and binding of inorganic sediment (Reid et al., 2000; Macintyre et al., 2000; Paerl et al., 2001; Andres et al., 2006). In contrast to the stromatolites, the thrombolites show a high degree of heterogeneity and variability in carbonate precipitation; however, the underlying microbial processes that form these thrombolitic clotted structures are not well understood.

To explore this issue further the thrombolites of Highborne Cay, an island located in the Exuma Sound, The Bahamas were targeted. At Highborne Cay, both stromatolites and thrombolites actively grow in the subtidal and intertidal zones, respectively. The thrombolites of Highborne Cay form large, unlaminated carbonate structures ranging in size from a few centimeters to several meters wide (Figure 3-1A).

Overlaying these thrombolitic carbonate structures is a thick (∼ 1 cm) microbial mat

Reprinted with permission from Mobberley JM, Khodadad CLM, Foster JS. (2013). Metabolic potential of lithifying cyanobacteria-dominated thrombolitic mats. Photosynth Res 118: 125-140.

(Figure 3-1B) comprised of a diverse microbial consortium, which is thought to drive the

precipitation of extracellular carbonate through their various metabolic activities

(Planavsky et al., 2009; Myshrall et al., 2010; Mobberley et al., 2012). There have been

several recent studies that have characterized the taphonomy of the thrombolite

carbonate structures (Planavsky and Ginsburg 2009), the surrounding oolitic sand

grains (Edgcomb et al., 2013a), as well as the biogeochemistry and microbial diversity

of the thrombolite-forming microbial mats, from here on referred to as thrombolitic mats

(Desnues et al., 2008; Myshrall et al., 2010; Mobberley et al., 2012). These previous

analyses have revealed several key functional groups of organisms associated with the

mats and include: phototrophs, aerobic heterotrophic bacteria, sulfate-reducing bacteria,

sulfide-oxidizing bacteria, and fermentative bacteria. All of these guilds have been

shown to be critical for carbonate precipitation and/or dissolution in other lithifying

microbial mat ecosystems such as stromatolites (e.g. Dupraz and Visscher, 2005;

Visscher and Stolz, 2005; Baumgartner et al., 2009). Together, these groups of

organisms generate steep vertical chemical gradients within the thrombolites with large

diel fluctuations. For example, O2 levels throughout the diel cycle can go from undetectable to several-fold supersaturated in the top 1-5 mm of the lithifying mat surface (Myshrall et al., 2010). The dominant organisms responsible for this O2-rich

layer are photosynthetic cyanobacteria, which have been shown to be a driving force in

the biogeochemical cycling in thrombolites (Myshrall et al., 2010).

In the thrombolitic mats of Highborne Cay, one of the prominent cyanobacteria

that comprises the upper few millimeters of the thrombolitic mats are calcified

filamentous Dichothrix spp. (Figure 3-1C; Planavsky et al., 2009; Mobberley et al.,

2012), which are characterized by their basal heterocysts and tapered apical ends

(Planavsky et al., 2009). These bundles of vertically-orientated filaments produce

copious amounts of exopolymeric substances (EPS) and appear to be “hot spots” for

carbonate precipitation in the thrombolitic mats. The carbonate deposition is thought to

be due to the elevated rates of photosynthesis, which increases the pH and the

carbonate saturation state within the mats and facilitates carbonate precipitation

(Visscher et al., 2002).

The production of EPS material has been shown to be a critical factor associate

with carbonate precipitation in modern microbialites. The EPS matrix within microbial

mats not only serves as a structural support for the growing community (Decho 1990)

but it can also chelate cations such as Ca2+ and function as nucleation sites for calcium carbonate nucleation (Decho 2000; Dupraz and Visscher, 2005). The EPS associated with the adjacent stromatolites of Highborne Cay are predominantly generated by cyanobacteria and sulfate reducing bacteria (Decho 2000; Braissant et al., 2007). The cyanobacterial EPS material contains approximately 50% carbohydrates, such as mannose, xylose and fucose, whereas the remaining material is a mixture of amino acids, uronic acids and glycans (Kawaguchi and Decho, 2000). Through the heterotrophic degradation of the EPS material in the stromatolites, the Ca2+ binding capacity of the EPS can be altered resulting in the release of Ca2+ ions, which can increase the saturation index of the microenvironment and promote localized areas of carbonate precipitation within the mat community (Dupraz and Visscher, 2005).

Despite recent studies regarding the viral and microbial diversity, sedimentology, and biogeochemistry of the Bahamian thrombolitic mats (Desnues et al., 2008;

Planavsky and Ginsburg, 2009; Planavsky et al., 2009; Myshrall et al., 2010; and

Mobberley et al., 2012; Edgcomb et al., 2013b) there remains an overall lack of understanding of the molecular mechanisms that underlie and regulate the metabolisms associated with these communities. In this study, we survey the metabolic potential of thrombolitic mats by sequencing the metagenome using massively parallel DNA sequencing. The metagenomic sequencing was coupled with phenotypic microarray analysis, which assessed vertical metabolite utilization patterns throughout the thrombolitic mats, thereby generating a spatial profile of metabolic activities. These two approaches help to increase our understanding of the underlying molecular pathways and overall genomic complexity associated with these unlaminated lithifying communities.

Materials and Methods

Thrombolitic Mat Sample Collection

Thrombolitic mats were collected from the island of Highborne Cay, The

Bahamas (76°49’ W, 24°43’N) in February 2010 from a single intertidal thrombolite platform located at Site 5 (Figure 3-1; Andres and Reid, 2006). The thrombolitic mats were cut into three 9-mm vertical sections and immediately placed in RNAlater (Life

Technologies, Inc, Grand Island, NY) to preserve the nucleic acids for later metagenomic analysis. Corresponding thrombolitic mat sections for metabolic phenotypic microarray studies were collected and placed into a sterile Nalgene container and transported to the Space Life Sciences Lab at the Kennedy Space

Center, FL. The mats were incubated in a walk-in growth chamber maintained at

Highborne Cay ambient conditions (24°C; 2000 µE/m2/s; 12 h light/dark cycle) for two

weeks until they were processed for the phenotypic microarray analysis as described

below.

DNA Extraction and Illumina Sequencing

Total community DNA was extracted from 1.2 g of each of the RNAlater

preserved samples using a previously described modified xanthogenate method

(Mobberley et al., 2012). The DNA was concentrated for sequencing with the Wizard

Genomic DNA Purification kit (Promega, Madison, WI) following manufacturers protocol.

Thrombolite community DNA was sequenced using an Illumina GAIIx platform

generating three distinct libraries of non-overlapping paired-end reads (Table 3-1). All

raw sequencing data was deposited into GenBank NCBI short read archive

SRP021141.

Analysis of Thrombolitic Mat Metagenomes

Each of the three metagenomic libraries (Thr-A, Thr-B, and Thr-C) were analyzed

and annotated with the MetaGenome Rapid Annotation using Subsystem Technology

(MG-RAST) version 3.3 pipeline and are publically available under project 3438 (Meyer

et al., 2008). Within the MG-RAST pipeline, the sequences in each library were quality

screened using the pre-processing program DynamicTrim (Cox et al., 2010) in which the phred score was set to 20 and only 5 bases or less below this threshold was allowed. Sequences of low quality were trimmed from each database resulting in 11.5,

10.7, 10.1 % of the raw sequences being removed from the Thr-A, Thr-B, and Thr-C libraries, respectively. Additionally, the libraries were screened for artificially replicated sequences as previously described (Gomez-Alvarez et al., 2009) with 1.9, 2.0, and 1.9

% of the sequences from the Thr-A, Thr-B, and Thr-C libraries removed, respectively.

The MG-RAST analysis of the replicate metagenomic libraries consisted of BLAT comparisons of the sequences against the SILVA SSU and LSU databases (version

104) for ribosomal gene reads and the multiple sourced non-redundant M5NR database

(version 1) for the protein encoding reads (Wilke et al., 2012). Ribosomal reads were annotated to the two SILVA databases at 97% identity with bit scores above 50. Protein encoding reads were assigned to SEED subsystems and KEGG pathways using only matches of > 60% similarity and > 60 base pairs that had an E-value of ≤ 10-5. To

acquire the broadest representation, taxonomic assignment of reads was performed

using the MG-RAST Lowest Common Ancestor (LCA) approach with phylogenetic trees

using the NCBI taxonomy created in MEGAN (version 4.70.4; Huson et al., 2011).

Deeper taxonomic exploration of metabolic genes in SEED subsystems and KEGG

pathways were carried out in MEGAN through LCA assignment using default

parameters of the 10 best NCBI nr database BLASTX hits for reads of interest.

Comparison of Thrombolite Metagenome to Other Microbial Mat Habitats

To explore functional differences between different lithifying and non-lithifying

mat systems, functional comparisons of subsystems distributions on whole

metagenome datasets publically available in MG-RAST were carried out by with the on

level 1 SEED subsystems with the following parameters: minimum E-value of 10-5 and minimum identity of 60%. The following datasets were used in this analysis (study reference and MG-RAST IDs given): thrombolitic mat replicates (this study), Type 1 and

3 Highborne Cay stromatolitic mats (Khododad and Foster, 2012; 4449590.3,

4449591.3), Cuatro Ciénegas microbialites (Breitbart et al., 2009; 44440060.4,

4440067.3), Cuatro Ciénegas non-lithifying mats (Peimbert et al., 2012; 4442466.3,

4441363.3), Guerrero Negro hypersaline mats (Kunin et al., 2008; 4440964.3-

4440972.3 were combined into one dataset for this analysis), Octopus Springs mat

(Bhaya et al., 2007; 4443749.3), and a Global Ocean Sampling Sargasso Sea water column sample (Rusch et al., 2007; 4441570.3). Multivariate analysis of the metagenomes functional subsystems was performed within the R statistical program (R

Development Core Team, 2010) using the randomForest (version 4.6-2; Liaw and

Wiener, 2002) and bpca (version 1.0-10; Faria and Demetrio, 2012) libraries described in Dinsdale et al., 2013. To account for differences in sampling depth, the subsystem abundance counts were normalized to the total number of reads that were able to be assigned a subsystem for each sample. To predict the most important subsystems

(response variables) for separating out the datasets, a supervised Random Forest was carried out the normalized subsystem counts for each environmental dataset using

5000 bootstrap iterations. Principle Component Analysis (PCA) for all metagenomes was performed and the directionality of the top six most important subsystems variables were calculated and overlain as a biplot.

Metabolic Profiling with Phenotypic Microarrays

Following acclimation in the growth chamber, the living thrombolitic mats transported from Highborne Cay were sectioned into three discreet zones that corresponded to known in situ oxygen depth profiles that occur at the peak of photosynthetic activity (Myshrall et al., 2010). The three zones included the upper 0 – 3 mm (Zone 1; oxic), the middle 3 – 5 mm (Zone 2; transition) and the lower 5 – 9 mm

(Zone 3; anoxic) region of the mats (Figure 3-1B). The metabolic capacity of each of the thrombolitic mat zones was characterized by diluting 250 mg of the sections into 1.7 mL of filter-sterilized artificial seawater. The samples were then homogenized by vortexing for 15 min and centrifuged to remove the larger sand particles. The cell density within

the homogenates was examined spectrophotometrically (A590nm; Genesys 20, Thermo

Fisher Scientific, Waltham, MA) and normalized to 1 x 106 cells per mL of seawater.

Aliquots (100 µl) of the mat slurries were then placed into Phenotypic Microarray (PM)

plates that contained either carbon (PM1), nitrogen (PM3B), phosphorus (PM4A) or

sulfur (PM4A) substrates (Biolog Inc., Hayward, CA). With the exception of the carbon

plates, all other plates were supplemented with 20 mM sodium succinate and 2µM ferric

citrate as described in the manufacturer’s protocol. The inoculated plates were

incubated for 24 h at 30°C and then examined every 15 min at A590nm using an Omnilog reader, which measures cellular respiration of each substrate (Biolog, Inc., Hayward,

CA). Only those substrates with absorbance readings above the set threshold were considered utilized by the mat community. The utilization threshold was considered to be 20% of the highest recorded absorbance on the PM plates. All samples were examined in triplicate and the differences between the different thrombolitic mat zones were statistically compared using ANOVA with a significance of p ≤ 0.05. Heatmap analyses were carried out Euclidean complete linkage clustering using the pheatmap package (version 0.7.4; Kolde, 2013) for the statistical program R (R Development Core

Team, 2010).

Results

Three replicate metagenome libraries representing the entire 9-mm thrombolitic mat profile were sequenced independently resulting in a combined 103,897,831 high quality sequencing reads derived from the thrombolitic mats and were highly reproducible (Table 3-1; Figure 3-2). An ANOVA analysis of the three independent libraries showed no statistical differences between replicates (p ≥ 0.986), and from here

on will be described as a single data set. The average guanine-cytosine (GC) content of the thrombolitic mat metagenome was 46% and most of the recovered reads were protein encoding genes representing between 73 – 74% of the recovered sequences with less than 0.1% of the recovered metagenome sequences corresponding to rRNA genes (Table 3-1). Although most of the reads recovered were protein encoding only

22% of the reads could be specifically annotated and only 13% were assigned to a

SEED subsystem or KEGG functional category.

Community Diversity Associated with Thrombolitic Mat Metagenome

Analysis of the annotated protein encoding and ribosomal features revealed that

91.7% were assigned to the domain Bacteria, whereas 7.4% were assigned to

Eukaryota and only 0.8% to Archaea (Figure 3-3). Within the domain Bacteria a total of

16 phyla were represented in the metagenomic libraries with extensive differences in the distribution of each of phyla (Figure 3-3A). The dominant phyla represented in the metagenomic library were the Cyanobacteria, which comprised 32.8% of the total annotated sequences. These results differed from previous clone and barcoded 16S rRNA gene libraries where the Proteobacteria were the dominant recovered taxa

(Myshrall et al., 2010; Mobberley et al., 2012). In the metagenomic libraries

Proteobacteria comprised 21.7% of the total number of bacterial sequences while

Bacteroidetes (10.4%) and Planctomycetes (1.1%) were also abundant. Five additional phyla were detected in the metagenomic libraries not previous identified through a barcoded 16S rRNA gene-pyrosequencing approach (Mobberley et al., 2012), although they comprised less than 1% of the total sequencing reads. These phyla included the

Chlamydiae, Chlorobi, Syngeristetes, Tenericutes, and Nitrospirae.

Although the Cyanobacteria were the dominant bacterial taxa, 70% of the reads assigned to the Cyanobacteria were unable to be designated to a taxon beyond the phylum level (Figure 3-4). Those sequences that could be further classified 47.6% were assigned to the order Chroococcales, with 39.4% of the classifiable Chroococcales sequences associating with the genus Cyanothece. The Nostocales were also prevalent, comprising 43.8% of the cyanobacterial sequences, with most sharing similarity to the genes typically associated with the genera Nostoc and Anabena (Figure

3-4). The Oscillatoriales were also present (7.7%) although 95% of the sequences

associated with this order could not be assigned beyond the order level.

In the archaeal population, most of the recovered sequences were assigned to

the phyla Thaumarchaeota (92.1%) with most assigned to the ammonia-oxidizing genus

Nitrosopumilus (63.5%; Figure 3-3B). Metagenomic sequencing also revealed

sequences associated with methanogenic archaea, which had not been previously

detected with 16S rRNA gene analysis (Mobberley et al., 2012). Specifically, sequences

with similarity to the Euryarchaeota families Methanosarcinaceae and

Methanobacteriaceae were recovered (Figure 3-3B). Other archaea phyla, such as

Crenarchaeota, comprised less than 0.2% of the archaeal taxa associated with the

thrombolitic mat metagenome.

The recovered reads assigned to Eukaryota taxa were diverse containing more

than ten superphyla and phyla within the thrombolitic mat metagenome (Figure 3-3C).

Of those Eukaryota sequences that could be classified, Metazoans accounted for 9.6%

of the total eukaryotic reads with the majority of the reads associated with Arthropoda

(6.8%), Chordata (1.2%), Annelida (0.8%) and Nematoda (0.4%). Bacillariophyta, or

diatoms, were also prevalent within eukaryotic population comprising 8.6% of the

protein encoding and ribosomal genes. Other prevalent eukaryotic taxa included the

Viridiplantae, specifically the Streptophyta (4.7%), Ascomycota (3.4%), Alveolata

(3.0%), Rhodophyta (2.1%), and the Foraminifera (0.5%).

Overview of the Functional Genes of the Thrombolitic Mat Metagenome

The protein encoding genes within the thrombolitic mat metagenome were compared to the M5NR database using the MG-RAST platform and the KEGG database using BLAT. Approximately 73.7% of the recovered sequences were identified as protein encoding genes, with 26.2% of the sequences unable to be characterized

(Table 3-1). Subsystem-level analysis of the thrombolitic mat metagenome indicated most of the protein encoding genes were of unknown function (75.5%), however, those sequences that were annotated were assigned to 28 different SEED subsystems

(Figure 3-5A). The two subsystems with the highest relative abundance of annotated sequencing reads were Protein Metabolism and RNA Metabolism subsystems representing a wide range of housekeeping genes involved in for the thrombolitic mats

(Figure 3-5A). Another subsystem with a high relative abundance of sequencing reads was the Carbohydrate Metabolism subsystem, which contained numerous genes associated with central carbon metabolism and carbon dioxide uptake. There was also a high representation of genes associated with mono-, di- and oligosaccharide metabolism and biosynthesis, specifically xylose, mannose, fructose, and trehalose

(data not shown). Screening the metagenome for genes associated with various energy metabolisms showed a diverse metabolic potential within the thrombolitic mats (Figure

3-5B). Genes associated with oxidative phosphorylation and photosynthesis were

abundant in annotated metagenome as were hydrogenases and genes associated with

fermentation, methane metabolism, sulfur oxidation, and sulfide reduction (Figure 3-5B).

A similar trend was observed when the mat metagenome was compared to the

KEGG database. There was an increase in the relative abundance of genes associated with the functional categories Carbohydrate and Energy Metabolism (Table 3-2).

Specifically, in the Carbohydrate Metabolism genes associated with the core carbon metabolisms, such as glycolysis and gluconeogenesis, the citric acid cycle, and pentose phosphate pathway were enriched. The dominant taxa associated with these pathways were primarily Cyanobacteria, with most reads sharing similarity to the orders

Chroococcales and Nostocales with bacterial orders Rhodobacterales, Rhizobiales, and

Planctomycetales also represented (Table 3-2). Interestingly, there were also several high-frequency matches to order Nitrosopumilales of the phylum Thaumarchaeota in the functional categories associated with core carbon metabolisms and in fructose and mannose metabolisms. The majority of the gene reads associated with glyoxylate and dicarboxylate metabolism were similar to the Ribulose-1,5-bisphosphate carboxylase oxygenase (RuBisCO, Form III) genes from the methanogenic Methanosarcinales.

Within the Energy Metabolism category there were five dominant pathways including oxidative phosphorylation, photosynthesis, methane, nitrogen, and sulfur metabolism (Table 3-2). The dominant taxa associated each metabolism varied, however, the Cyanobacteria, specifically the order Chroococcales, had the highest number of matches in each category. These included the majority of the reads for genes encoding ATP synthase complexes, cytochromes, and Photosystems I and II, although there were smaller contributions from other bacterial phyla as well as diatoms, plants,

and cyanophage (psbA). In methane metabolism, genes associated with the oxidation of methane (e.g. fbaA, metF), had a high frequency of matches to the Chroococcales, specifically there was similarity to the genus Cyanothece spp.. Surprisingly, most of the hits for coenzyme F420 hydrogenase (frhA) were more similar to [NiFe]-hydrogenase homologs found in cyanobacteria instead of methanogens. No methyl-coenzyme reducatase genes were detected at the cutoffs of this analysis, however, there was a low abundance of archaeal genes associated with methanogenesis such as tetrahydromethanopterin S-methyltransferase (mtr) and formylmethanofuran dehydrogenase (fwd) found in the metagenome (Table 3-2). Soluble methane monooxygenase genes were also recovered suggesting that methane oxidation could be occurring, although it should be noted that these genes shared a high sequence similarity to ammonia monooxygenase genes.

In nitrogen metabolism, the vast majority of nitrogen fixation genes were associated with Cyanobacteria (i.e., Chroococcales and Nostocales) and

Chloroflexales. With regard to other nitrogen metabolisms, genes involved in assimilatory nitrate reduction were predominately associated with Cyanobacteria, while dissimilatory reduction genes were found to be similar to those in Alphaproteobacteria,

Gammaproteobacteria and Planctomycetes. Denitrification genes (e.g. nor) were associated with Flavobacteriales, Rhodobacterales, and Nitrosopumilales. There were very few nitrification genes detected within the thrombolite metagenomes, those that were recovered appeared similar to ammonia monooxygenase genes from the archaea

Nitrosopumilus maritimus.

In sulfur metabolism, most of the recovered genes were associated with assimilatory sulfate reduction, such as sulfate adenylyltransferases, and shared similarity to Chroococcales (met3, cysC), Rhodobacterales (met3, cysD), and

Nitrosopumilales (cysD). Although pathways associated with assimilatory sulfate reduction had the highest relative abundance in the thrombolitic mat metagenome, genes associated dissimilatory sulfate reduction pathways (e.g. dsrB, aprAB) were also detected and were associated with the Deltaproteobacteria, specifically the orders

Desulfobacterales and Desulfovibrionales. Thrombolitic mat metagenome reads associated with sulfur oxidation were most similar to the sox genes found in

Rhodobacterales and Rhizobiales. In addition, there were reads similar to genes from

Rhodobacterales, Cytophagales, and Chroococcales that were involved in the transport and utilization of organic sulfur compounds including sulfonates and DMSP.

Comparison of Thrombolite Metagenome to Other Functional Metagenomes

The three replicate thrombolite metagenomes of this study were compared to

nine publically available metagenomes derived from both lithifying and non-lithifying

microbial mat habitats using MG-RAST (Figure 3-6). These environments included: the

non-lithifying hypersaline mats of Guerrero Negro, Mexico; nonlithifying and lithifying

freshwater mats of Cuatros Ciénegas, Mexico; and the marine stromatolites of

Highborne Cay, The Bahamas. A PCA plot of the functional metagenomes is visualized

in Figure 3-6A. Overlaying the PCA plot of the different metagenomes were six SEED

subsystems that appeared to drive several differences between the communities. The

results indicated that the Bahamian thrombolitic mats were distinct from the other nine

microbial mat metagenomes with the analysis explaining 92.85 % of the variance.

Protein encoding genes associated with the subsystem Respiration appeared to be the

most important subsystem driving the differences between the Highborne Cay

thrombolites and other metagenomes. Additionally, an independent multivariate analysis

of 12 metagenomes was conducted using the random forest package in R (Figure 3-6B;

Liaw and Weiner, 2002; Dinsdale et al., 2013). These results also indicated that the genes associated with Respiration were the most important variable that distinguished the thrombolite metagenomes from other lithifying and non-lithifying mat communities.

Genes associated with subsystems such as Photosynthesis, Nitrogen Metabolism,

Cofactors and Pigments all appear to be conserved between the 12 metagenomes

(Figure 3-6B).

Substrate Utilization Patterns Within Thrombolitic Microbial Mats

A spatial profile of the metabolic activity within thrombolitic mats was generated using phenotypic microarray analysis, which assesses the ability of the mat communities to grow on a wide range of carbon (C), nitrogen (N), phosphorus (P) and sulfur (S) substrates. Live mat samples were sectioned into three discrete zones ranging from 0 – 3 mm (Zone 1); 3 – 5 mm (Zone 2) and 5 – 9 mm (Zone 3; Figure 3-

1B) and homogenized into slurries that were normalized for cell density. An overview of the substrate utilization patterns for each of the zones within the thrombolitic mat is visualized in Figure 3-7. The full lists of the substrates, their usage in the three mat zones and significance values are given (Table 3-3, 3-4, 3-5, 3-6). The upper 3 mm of the thrombolitic mat utilized the greatest number of tested C, N, P, S substrates (n=149) compared to the 3-5 mm (n=131) and 5-9 mm (n=60) zones under our experimental conditions. However, when each substrate category was examined individually there was a pronounced difference in the extent of the substrate utilization within the different zones.

Although only 31% of the carbon substrates were metabolized by the thrombolite mat slurries most were utilized in Zones 1 (n=21) and 2 (n=24; Figure 3-7A). Zone 2 had

the highest number of exclusive carbon sources (n=9) with most containing an amine

group, while Zone 1 (n=4) exclusively used D-fructose-6-phosphate and a few carboxylic acids. Several of the substrates were utilized at significantly different levels between the zones. For example, the substrates adenosine and D-mannose were

utilized at a significantly higher level in Zones 1 and 2, where as Zone 3 utilized alpha-

ketoglutaric acid at a higher level than the other two zones (Figure 3-7A; Table 3-3).

The C sources metabolized by all three zones were mostly substrates and intermediates in energy generating processes (e.g. TCA cycle) such as pyruvic acid and citric acid (Figure 3-7A; Table 3-3). Of the nitrogen substrates, more than half (53%) of them were utilized as sole nitrogen sources by the thrombolitic mat zones (Figure 3-7A;

Table 3-4). Zone 1 metabolized the most N substrates (51%), and had high levels of utilization of cytidine, alanine and glycine dipeptides, and most amino acids compared to Zones 2 and 3. Although Zone 2 utilized the least number of substrates, it used adenosine at a higher rate than any of the other zones. Complete linkage cluster analysis of total C and N substrate usage shows that the overall mat metabolism for

Zone 1 is distinct from those found in Zones 2 and 3 (Figure 3-7A).

The thrombolitic mat slurries metabolized P (94%) and S (71%) substrates at an overall higher rate compared to the C (31%) and N (53%; Figure 3-7B). For P substrate utilization, Zones 1 and 2 had the highest utilization rate (94%) while Zone 3 slurries were only able to metabolize nine P substrates by 24 h (Table 3-5). The P usage pattern for Zones 1 and 2 were similar and exhibited high rates of utilization for

adenosine-containing nucleotides and thiophosphates. All three thrombolitic mat zones

metabolized cyclic nucleotides, pyrophosphate, and glycerol phosphate. As with P

substrate utilization, S metabolism patterns in Zones 1 and 2 were similar and included

high rates of growth on L-djenkolic acid and L-cysteine (Figure 3-7B; Table 3-6).

Additionally these two thrombolitic mat zones were able to use inorganic sulfate,

thiosulfate, and thiophosphates. Interestingly, the organisms in Zone 3 were able to

metabolize two sulfone-containing substrates as their sole sulfur source with L-

methionine sulfone being exclusively used. Since Zone 3 was not metabolizing most P

and S substrates at 24 h, we examined utilization at 48 hours and found that more of

the P and S substrates (all on PM4 plates) were used suggesting delayed utilization

under the experimental testing conditions (data not shown). Cluster analysis revealed

that overall P and S utilization was most similar between Zones 1 and 2 with little or

slower utilization rates by the organisms in Zone 3.

Discussion

This study represents the first metagenomic survey of the functional gene complexity associated with Bahamian thrombolites and provides a spatial profile of the substrate utilization patterns within the thrombolitic mat community. The results of this study suggest that 1) cyanobacterial molecular pathways dominate the thrombolite metagenome; 2) all microbial energy metabolisms that have been proposed to promote carbonate precipitation are present within the metagenome; 3) archaea, although low in diversity, contribute to metabolic cycling within the mats; and 4) discrete gradients of metabolite utilization occur within the depth profile of the thrombolitic mat.

Cyanobacterial Molecular Pathways Dominate the Thrombolite Metagenome

Cyanobacteria have long been known to dominate microbialitic mat communities

(e.g. Reid et al., 2000; Breitbart et al., 2009; Myshrall et al., 2010; Santos et al., 2010;

Foster and Green, 2011; Khodadad and Foster, 2012). The metagenomic sequencing of the thrombolitic mats reinforces this concept and shows that the majority of the recovered reads from the Bahamian thrombolitic mats were derived from cyanobacteria.

The thrombolite metagenome was enriched in genes with sequence similarity to nitrogen fixing organisms such as Cyanothece, Nostoc, and Anabaena, whereas

common microbialite mat builders such as Schizothrix spp. and Dichothrix spp. were not

well represented. The metagenome results differed from previous 16S rRNA gene

amplicon studies where Oscillatoriales and Pleurocapsales were highly prevalent in the

thrombolites (Myshrall et al., 2010; Mobberley et al., 2012). This discrepancy between

gene centered surveys and whole metagenomic sequencing may result from biases in

ribosomal primer design, variation in copies of the 16S rRNA genes between

organisms, as well as database differences and limitations (Liu et al., 2007; Stevens et

al., 2012). In general, most complete cyanobacterial genomes found in genomic

databases, including those used by MG-RAST, are from ubiquitous aquatic and marine

ecotypes such as Nostoc spp. and Anabaena spp., whereas trapping and binding

filamentous Oscillatoriales and endolithic Pleurocapsales typically associated with

microbialite communities are significantly underrepresented. Despite these caveats

regarding the assignment of specific genes and pathways to individual taxa below the

order and family-level the metagenome does provide important insight into the specific

metabolisms associated with the cyanobacterial community.

It has been previously established, based on biogeochemical profiling and stable isotope analyses, that cyanobacteria play a major role in energy generation and carbonate mineralization in Highborne Cay thrombolitic mats (Planavsky et al., 2009;

Myshrall et al., 2010). These findings are supported by the recovered metagenome

results with most of the cyanobacterial genes being similar to well characterized

Photosystems I and II predominately from the orders Chroococcales and Nostocales.

Besides converting light energy into usable biological energy, the thrombolitic mat

cyanobacteria are involved in processing of fixed organic carbon through a variety of

anabolic (e.g. gluconeogenesis, sugar biosynthesis) and catabolic (e.g. glycolysis,

oxidative pentose phosphate, fermentation) pathways to create and consume complex

carbohydrates. For example, the presence of many genes involved in the conversion of

mannose and fructose to molecules such as rhamnose, alginate, xylose, and trehalose

indicates that a portion of thrombolitic mat organic carbon is utilized in the formation of

the EPS matrix of the thrombolitic mats. The EPS matrix is integral to biologically

influence mineralization via the binding of calcium cations (Kawaguchi and Decho,

2000; Dupraz et al., 2009).

In addition to energy metabolism and carbohydrate synthesis, cyanobacterial

genes associated with nitrogen fixation were abundant in the metagenome. Nitrogen is

often a limiting nutrient in marine microbialitic mat systems (Pinckney et al., 1995) and

biological nitrogen fixation may represent an important metabolism for the growth and

maintenance of the thrombolitic mats. Numerous cyanobacterial nitrogenases (e.g.

nifDH) and associated genes were recovered from the thrombolitic mats, as were other

diazotrophic organisms (Table 3-2). Nitrogenase activity has been measured in the

adjacent stromatolites of Highborne Cay and has been shown to peak at night, with the

activity strongly influenced by photosynthesis and the availability of reduced organic

carbon (Steppe et al., 2001). The thrombolites, however, are dominated by

heterocystous-forming cyanobacteria such as Dichothrix spp. (Mobberley et al., 2012) and are not readily found in the adjacent stromatolites. These population differences in the thrombolitic mats may result in different temporal and spatial patterns of nitrogen fixation compared to the stromatolites. Nitrogen fixation is also of interest as it is closely associated with the evolution of H2. Genes encoding other H2 producing enzymes such

as [NiFe]-hydrogenases were prevalent in the thrombolite metagenome. Overall genes

encoding hydrogenases, including both uptake (e.g. hya, hyb) and bidirectional

hydrogenases (hox operon) comprised 5% of the energy metabolism subsystem.

Although in most systems the microbial communities are quite effective in sequestering

the evolved H2 (Peters et al., 2013), the rates of H2 production in these lithifying

microbial mats remains to be examined. The novelty of many of the cyanobacteria

within the thrombolitic mats and their various hydrogenases may help to improve the

understanding of H2 evolution in these lithifying ecosystems, and has the potential to be exploited for renewable energy carrier production (Bothe et al., 2010; Peters et al.,

2013).

Energy Metabolisms that Influence Carbonate Precipitation in Lithifying Microbial Mats

The thrombolitic mat metagenome contained genes associated with all of the proposed functional guilds associated with the precipitation and dissolution of carbonate including phototrophs, sulfate-reducing bacteria, sulfide-oxidizing bacteria, aerobic heterotrophs, and fermentative bacteria (Visscher and Stolz, 2005; Dupraz et al., 2009).

As discussed above, cyanobacteria were the major oxygenic phototrophs detected

within the thrombolitic mats, however, eukaryotic algae (i.e. Streptophyta; Chlorophyta)

and diatoms (i.e. Bacillariophyta) were also present at low levels. Genes associated

with anoxygenic photosynthetic photopigments such as bacteriochlorophylls and

proteorhodopsins were also recovered from the thrombolitic metagenome suggesting a

diverse phototrophic community in the thrombolitic mat photosynthesis. However,

comparison to other microbial mat metagenomes indicates that genes associated with

photosynthesis and pigment synthesis are highly conserved between both lithifying and

non-lithifying metagenomes, whereas genes associated with the subsystem Respiration

was an important variable in distinguishing the Bahamian thrombolites from other known

microbial mat metagenomes (Figure 3-6A).

In microbialite forming mats the role of anaerobic respiration, specifically sulfate

reduction, in promoting carbonate precipitation through remineralization has been well

documented (Visscher et al., 1998; Visscher et al., 2000; Paerl et al., 2001;

Baumgartner et al., 2006; Nitti et al., 2012; Gallagher et al., 2013). Amplicon libraries of the 16S rRNA gene amplicon studies have previously shown the presence of several sulfate-reducing bacteria (SRB) mostly associated with the Desulfovibrionales in the thrombolitic mats (Mobberley et al., 2012; Myshrall et al., 2010, Planavsky et al., 2009).

The metagenomic survey of the top 9 mm of the thrombolitic mats, however, showed that the relative abundance of dissimilatory sulfate reduction genes was low compared to metagenomic studies of other microbialitic mat communities (Breitbart et al., 2009;

Khodadad and Foster, 2012). Additionally, most of the recovered genes in thrombolitic mats were similar to Desulfobacterales, although a few genes associated with

Desulfovibrionales were detected (Table 3-2). The low relative abundance may suggest

that the sulfate reducing metabolisms are located deeper in the thrombolitic mat

communities compared to the adjacent stromatolites. Future assessment of the

biogeochemical profiles of SRBs activity of these genes through biogeochemical

profiling (e.g., HS-) as well as measuring the activity of the genes with either metatranscriptomics or quantitative PCR will be required to fully assess this important metabolism within the thrombolitic mat communities.

The thrombolitic metagenome also contained many bacterial organic substrate degradation pathways that have also been recovered from other microbialitic mat systems (Breitbart et al., 2009; Khodadad and Foster, 2012) and were primarily derived from phyla Proteobacteria, Bacteroidetes, and Planctomycetes. Additionally, genes were recovered from lactate and mixed acid fermentation pathways from a variety of bacteria with most reads sharing similarity to cyanobacteria. Although not well documented in microbialitic mats, cyanobacterial fermentation under anoxic conditions in non-lithifying mats leads to the production of organic acids, H2, and CO2 (van der

Meer et al., 2007; Burow et al., 2012). Together these heterotrophic metabolisms

represented in the metagenome may play an important role in organic matter cycling

within the thrombolitic mat. For example, the degradation and alteration of

cyanobacterial EPS by heterotrophic metabolisms can lead to the release of calcium

cations from the polymers, which can subsequently be used for carbonate precipitation

(Decho, 2000).

Contribution of Archaea in Thrombolitic Mat Metabolic Cycling

Microbial diversity studies have detected the presence of archaea in lithifying

microbialitic mats ecosystems (Burns et al., 2004; Goh et al., 2009; Coreadeau et al.,

2011; Arp et al., 2012; Khodadad and Foster, 2012; Mobberley et al., 2012), but little

work has been done investigating the contribution of these organisms to microbialite

physiology. Although archaea comprised only a small portion of the thrombolitic mat

metagenome, the recovery of reads in major carbohydrate and energy KEGG functional

categories (Table 3-2) suggest that these organisms are not insignificant contributors to

thrombolitic mat metabolic cycling. The overabundance of thrombolitic metagenomic

reads similar to the ammonia-oxidizing chemolithoautotroph Nitrosopumilus maritimus is

different from other metagenomic analyses of microbialitic mat ecosystems, such as the

adjacent stromatolites of Highborne Cay (Khodadad and Foster, 2012) and freshwater

microbialites of Cuatros Ciénegas (Breitbart et al., 2009). In these systems most of the

recovered archaeal protein encoding genes were assigned to the Euryarchaeota.

Genomic sequencing of cultured isolates of Nitrosopumilus maritimus has identified several metabolic pathways for ammonia oxidation and carbon fixation through a modified hydroxypropionate-hydroxybutylate cycle and an incomplete TCA cycle

(Walker et al., 2010), both of which are detected in the thrombolitic mat metagenome.

Additionally, the presence of Nitrosopumilales-like genes in glycolysis/gluconeogenesis and sugar metabolisms indicates that archaeal carbohydrate processing may occur in the thrombolite. These results suggest that Nitrosopumilus maritimus-like archaea may be significant contributors to nitrification and carbon cycling in the thrombolitic mats thus potentially influence carbonate precipitation.

The metagenomics results also suggest that archaea may also be contributing to the thrombolitic mat carbon cycling through methanogenesis. To date, the role of methanogenesis, an anaerobic respiration metabolic pathway, in modern marine

microbialitic mats is unclear since only a few methanogenic ecotypes have been

recovered from microbial diversity studies of the hypersaline stromatolites from Shark

Bay and the Kiritimati Atoll (Burns et al., 2004; Arp et al., 2012). Methanogenic ecotypes have not been previously identified in the Bahamian thrombolitic mats based on 16S rRNA gene analyses (Myshrall et al., 2010; Mobberley et al., 2012), nor have they been

identified in the metagenome of the adjacent stromatolite (Khodadad and Foster, 2012).

Analyses of the thrombolitic mat metagenome in this study did not recover any methyl-

coenzyme reducatase genes; however, genes encoding accessory enzymes that are

typically associated with methanogenesis, such as tetrahydromethanopterin S-

methyltransferase (mtr) were present and appear to be derived from the

Methanosarcinaceae. Members of this family undergo methanogenesis with CO2,

acetate, and methylated one-carbon compounds (Feist et al., 2006), suggesting that

methanogenesis may be occurring in Bahamian thrombolitic mats; however its role in

carbonate mineralization, if any, needs to be explored further.

Gradients of Metabolic Potential Occur Within the Depth Profile of the Thrombolitic Mat

The phenotypic microarray measured the respiration of carbon, nitrogen,

phosphorous, and sulfate substrates within the thrombolitic mats thereby serving as

proxies of heterotrophic metabolism. The overall substrate utilization patterns in the

upper 5 mm of the thrombolitic mat, as represented by Zones 1 and 2, contained the

most active communities. This result combined with the wide range of heterotrophic

respiratory and degradation pathways found in the metagenome (Figure 3-5; Table 3-2),

suggests that the top two zones of the thrombolitic mat support a more robust and

adaptable community of heterotrophic metabolisms. Although it is likely that the full

metabolic capacity of the lower zone of the mat was not captured, as it typically remains

anoxic throughout the day (Myshrall et al., 2010).

The differential usage of carboxylic acids and saccharides by the three discrete

zones was correlated to the diverse array of carbohydrate metabolisms observed in the

metagenome, such as decarboxylases, and sugar transporters and processing pathways. For example, Zone 1 of the thrombolite mats used a broader range of

carboxylic acids than the other zones of the thrombolitic mats suggesting that the

microbes in this upper portion of the mat more readily processed organic acids.

Preferential degradation of acidic residues within the thrombolitic mat EPS could

liberate Ca2+ cations that could serve as nucleation sites for carbonate precipitation

(Decho, 2000). The Zone 2 microarray results indicated that these microbial

communities utilized a wide range of sugars particularly those with amine groups, which

also suggests a diverse, but different range of EPS degradation capabilities in this zone

of the thrombolitic mat. Together, these examples indicate discrete spatial differences in

carbohydrate metabolism by the microbial communities within the thrombolitic mats,

suggesting differences in the potential for carbonate precipitation throughout the vertical

profile of the thrombolitic mat.

With regard to N substrate utilization, distinct differences were also observed

throughout the vertical profile of the mats. The varied uptake and assimilation of

different nitrogen sources between the thrombolitic mat zones suggested the potential

partitioning of nitrogen metabolisms. As with carbohydrate metabolisms, the greater

utilization of dipeptides and amino acids by the heterotrophic community in Zone 1

could also promote carbonate precipitation through alteration of the EPS chemistry

during degradation (Dupraz et al. 2009). In contrast to the C and N patterns, the S and

P utilization were comparable in all three zones of the thrombolitic mats, although in

Zone 3 the microbial communities required an additional 48 h for utilization of many of the various substrates (data not shown). This delay likely reflects the difficulty these anaerobic communities may have had growing under the aerobic experimental conditions.

Taken together, the results of the phenotypic microarrays suggest the presence of distinct vertical gradients of metabolic activity within the clotted, unlaminated thrombolitic mats. Similar trends have also been reported in the freshwater microbialites of Cuatros Ciénegas, Mexico (Nitti et al. 2012). Results from these studies, which used a combined genomic, lipid and stable isotope analyses, also showed distinct spatial differences within the lithifying communities although no lithified layers were present

(Nitti et al. 2012) reinforcing the idea that the microbial communities can form discrete metabolic zones within unlaminated microbialite forming communities. Although the metagenomic survey correlated many of the specific pathways associated with the use of these different substrates, future work is required to delineate the specific expression of these molecular pathways to assess how the various metabolic activities are coordinated and regulated throughout the vertical profile of these lithifying microbial communities.

Table 3-1. Sequence of metagenome sequencing and MG-RAST analysis. Sample Sequence rRNA Protein Annotated Proteins readsa gene encoding proteinsd (%) assigned readsb (%) readsc (%) functione (%) Thr-A 34234300 38674 25055197 7581161 4638130 (0.1) (73.2) (22.1) (13.6) Thr-B 34958742 39324 25742370 7824431 4791538 (0.1) (73.6) (22.4) (13.7) Thr-C 34704789 39283 25792847 7741612 4782100 (0.1) (74.3) (22.3) (13.8) aTotal number of high-quality sequences post processing and de-replication. bNumber of reads annotated against the SILVA SSU and LSU databases (ver. 104) at 97% identity with bit scores above 50. cNumber of reads predicted by FragGeneScan to encode protein. dAnnotation of putative proteins against the M5NR database, 60% identity, 60 bp alignment, e-value >10-3. eAnnotated proteins assigned to a SEED Subsystem and or KEGG ontology functional category.

Table 3-2. Relative abundance of dominant metabolic pathways in thrombolitic mats. a KEGG Protein or gene name with Total d Functional Category ID high relative abundancebc Reads Dominant taxa (No. Reads) Carbohydrate metabolism 1110 Glycolysis, 10 ALDO, fbaAB, GAPDHS, 19212 Chroococcales (503) Gluconeogenesis gap2, cel, porAB Rhodobacterales (320) (aceE, pfkA, pckA) Nitrosopumilales (288) Nostocales (175) Planctomycetales (136) Bacillariophyta (73)

Citric Acid (TCA cycle) 20 acnB, fumAB, sdhA, 14449 Chroococcales (533) SDH1, frdA Nostocales (166) (gltA, sucC, korA) Rhodobacterales (163) Planctomycetales (149)

Pentose phosphate 30 Phosphoketolase, pgl, 644 Nostocales (602) pathway kguK (UGDH, rbsK) Rhizobiales (11)

Glyoxylate, dicarboxylate 630 Formamidase, purU, 1871 Methanosarcinales (180) metabolism rbcSL, Chroococcales (178) (phbB, glcD, ttdB) Rhizobiales (57)

Fructose, mannose 51 manB, rhaB, GMPP, fucK, 3939 Chroococcales (553) metabolism gmd, fcl(rhaA, gutB) Nostocales (324) Chloroflexales (30) Nitrosopumilales (18) Energy metabolism 1120 Oxidative phosphorylation 190 ATPF1, ATPeF1, ATPV, 23872 Chroococcales (3134) ccoN, coxC, ndhACEJL Nostocales (1276) (MQCRB, hoxF) Planctomycetales (339) Rhodobacterales (266) Archaea (101) Viridiplante (12) Photosynthesis 195 psaABEX, psbABCDF28, 20942 Chroococcales (2411) petACEF Nostocales (908) Oscillatoriales (541) Chlorophyta (97) Bacillariophyta (74)

Methane metabolism 680 fbaA, frhA, fwbB/fmdB, 5410 Chroococcales (235) mtd, mtrF,hdr Nostocales (75) (coxSL, fwdDCFH) Bacillariophyta (46) Rhodobacterales (6)

Nitrogen metabolism 910 nifHDNV 1743 Chloroflexales (60) (napACE, narVW, nosZ, Chroococcales (46) norDF, nirAB, narB, amoC) Nostocales (31)

Sulfur metabolism 920 met3, aprAB, cysC, cysD 6212 Chroococcales (411) (cysQHJ, dsrB, soxBYZ) Rhodobacterales (260) Nitrosopumilales (61) Desulfobacterales (37) Desulfovibrionales (11) acategories based on KEGG classification system. bgenes that were highly represented within the thrombolitic mat metagenomes. c genes similarities found from 10-3 to 10-5 in parentheses dreads assigned based on LCA algorithm in MEGAN.

Table 3-3. Carbon substrate absorbance units of thrombolitic microbial mats. Carbon Substrates (n = 95)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value 1,2-Propanediol 22.00 ± 3.00 24.33 ± 4.67 24.67 ± 3.18 0.860 2-aminoethanol 56.00 ± 11.50 40.33 ± 12.84 51.00 ± 17.50 0.739 2-Deoxy-Adenosine 47.00 ± 6.24 32.33 ± 9.84 20.33 ± 1.20 0.082 α-D-Glucose 31.00 ± 7.21 19.33 ± 3.33 22.00 ± 6.66 0.408 α-D-Lactose 23.67 ± 1.20 21.00 ± 0.58 19.33 ± 0.67 0.032 α-Hydroxy Butyric Acid 7.67 ± 3.38 5.67 ± 3.84 2.67 ± 1.76 0.556 c α-Hydroxy GAGL 19.67 ± 9.87 22.00 ± 3.06 15.67 ± 1.33 0.763 α-keto-Butyric Acid 11.00 ± 1.73 7.00 ± 2.08 9.00 ± 0.58 0.284 α-keto-Glutaric Acid 145.67 ± 23.25 26.33 ± 14.08 68.67 ± 25.30 0.020 α-Methyl-D-Galactoside 13.33 ± 1.67 13.33 ± 0.67 12.00 ± 2.00 0.789 Acetic Acid 59.67 ± 1.86 61.67 ± 1.67 56.33 ± 0.33 0.101 Acetoacetic acid 63.67 ± 1.20 64.33 ± 1.76 64.00 ± 1.73 0.957 Adenosine 59.00 ± 0.00 42.67 ± 5.78 35.67 ± 4.63 0.021 Adonitol 15.00 ± 2.08 16.33 ± 1.20 15.00 ± 2.00 0.838 β-Methyl-D-Glucoside 23.00 ± 3.61 19.00 ± 1.73 20.00 ± 2.08 0.560 Bromo Succinic Acid 102.67 ± 14.90 116.3 ± 54.27 52.33 ± 6.39 0.403 Citric Acid 175.67 ± 45.67 137.0 ± 22.28 279.33 ± 10.53 0.037 D-Alanine 10.00 ± 3.51 15.33 ± 5.55 13.67 ± 5.78 0.757 D-Aspartic Acid 4.00 ± 3.06 3.33 ± 0.88 9.67 ± 2.73 0.207 D-Celiobiose 27.33 ± 2.03 31.00 ± 2.00 18.00 ± 1.00 0.005 D-Fructose 31.00 ± 3.51 28.33 ± 3.48 18.67 ± 0.67 0.050 D-Fructose-6-Phosphate 57.33 ± 34.83 30.67 ± 6.33 24.67 ± 2.60 0.525 D-GAGLc 4.33 ± 3.84 5.33 ± 0.33 1.33 ± 1.33 0.500 D-Galactose 71.33 ± 16.34 32.67 ± 5.21 43.33 ± 9.26 0.116 D-Galacturonic Acid 43.67 ± 3.33 37.00 ± 6.43 42.33 ± 4.41 0.619 D-Gluconic Acid 99.33 ± 13.57 88.33 ± 17.23 41.00 ± 19.35 0.104 D-Glucosaminic Acid 20.33 ± 2.03 20.67 ± 2.03 25.67 ± 3.71 0.360 D-Glucose-1-Phosphate 18.00 ± 2.08 20.67 ± 0.33 20.00 ± 1.00 0.403 D-Glucose-6-Phosphate 43.00 ± 3.06 34.00 ± 3.61 33.67 ± 1.86 0.110 D-Glucuronic Acid 122.67 ± 54.57 27.67 ± 10.67 39.67 ± 6.96 0.157 D-Malic Acid 28.67 ± 6.12 20.00 ± 3.06 21.67 ± 7.51 0.572 D-Mannitol 39.00 ± 3.21 36.33 ± 2.91 25.67 ± 1.67 0.028 D-Mannose 62.00 ± 2.31 43.33 ± 3.93 40.67 ± 8.19 0.050 D-Melibiose 29.67 ± 1.20 27.00 ± 1.00 30.33 ± 2.03 0.312 D-Psicose 23.00 ± 3.79 20.33 ± 5.24 22.33 ± 4.41 0.911 D-Ribose 71.67 ± 2.33 71.67 ± 3.18 67.67 ± 2.33 0.507 D-Saccharic Acid 23.67 ± 3.71 24.33 ± 6.64 21.33 ± 4.91 0.914 D-Serine 26.33 ± 1.33 27.00 ± 2.08 24.67 ± 2.60 0.727 D-Sorbitol 9.33 ± 5.90 13.00 ± 5.00 9.67 ± 4.18 0.856 D-Threonine 20.00 ± 2.52 17.67 ± 0.33 19.67 ± 1.45 0.599 D-Trehalose 34.33 ± 9.17 32.00 ± 5.13 25.00 ± 3.00 0.582 D-Xylose 60.00 ± 5.20 54.33 ± 2.73 53.67 ± 1.76 0.432 D,L-α-Glycerol Phosphate 13.00 ± 3.79 9.67 ± 2.03 3.00 ± 1.53 0.088 D,L-Malic Acid 177.67 ± 64.24 108.0 ± 18.19 194.67 ± 61.75 0.506 Dulcitol 58.00 ± 9.87 67.67 ± 6.89 54.00 ± 13.05 0.646 Formic Acid 3.00 ± 3.00 6.33 ± 3.18 2.33 ± 2.33 0.598 Fumaric Acid 67.00 ± 10.12 24.00 ± 18.77 29.00 ± 14.01 0.158 Glucuronomide 47.33 ± 4.81 38.67 ± 6.98 40.00 ± 6.08 0.579 Glycerol 35.67 ± 6.36 17.67 ± 2.85 21.67 ± 9.28 0.217 Glycolic Acid 13.00 ± 1.00 7.67 ± 2.03 9.67 ± 4.18 0.433 Glycyl-L-Aspartic Acid 19.33 ± 2.03 18.33 ± 3.28 16.00 ± 3.21 0.719 Glycyl-L-Glutamic Acid 13.00 ± 1.00 8.67 ± 3.93 9.33 ± 2.03 0.494

Table 3-3. Continued. Carbon Substrates (n = 95)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value Glycyl-L-Proline 19.00 ± 2.65 14.33 ± 0.67 21.00 ± 2.31 0.142 Glyoxylic Acid 31.00 ± 1.00 25.33 ± 6.23 28.67 ± 1.20 0.583 Inosine 54.67 ± 2.85 46.67 ± 3.76 43.33 ± 5.9.3 0.248 L-Alanine 17.33 ± 2.91 16.00 ± 4.51 15.67 ± 2.85 0.939 L-Alanyl-Glycine 21.33 ± 2.19 10.67 ± 4.70 14.33 ± 6.36 0.338 L-Arabinose 49.33 ± 3.71 47.33 ± 3.38 47.33 ± 4.91 0.923 L-Asparagine 19.33 ± 3.33 20.33 ± 2.40 18.00 ± 1.73 0.819 L-Aspartic Acid 14.67 ± 6.69 19.00 ± 4.36 20.00 ± 5.13 0.774 L-Fucose 39.33 ± 5.55 40.00 ± 5.29 35.00 ± 8.33 0.846 L-GAGLc 19.33 ± 0.88 9.67 ± 5.36 14.67 ± 3.84 0.281 L-Glutamic Acid 51.00 ± 19.30 56.00 ± 7.51 37.67 ± 9.67 0.621 L-Glutamine 13.67 ± 2.60 19.00 ± 4.04 19.00 ± 0.58 0.361 L-Lactic Acid 65.33 ± 21.40 34.00 ± 3.51 21.33 ± 4.33 0.116 L-Lyxose 95.00 ± 1.53 92.67 ± 1.33 93.33 ± 1.86 0.592 L-Malic Acid 142.33 ± 27.42 77.67 ± 41.45 110.67 ± 56.05 0.599 L-Proline 18.00 ± 7.09 7.33 ± 2.03 12.00 ± 4.62 0.383 L-Rhamnose 35.67 ± 3.71 34.33 ± 1.76 37.67 ± 1.33 0.656 L-Serine 18.67 ± 0.88 20.00 ± 1.53 19.33 ± 2.19 0.848 L-Threonine 33.33 ± 9.28 23.67 ± 8.29 23.33 ± 5.61 0.619 Lactulose 18.67 ± 0.88 18.00 ± 3.51 16.33 ± 2.40 0.802 m-Hydroxy Phenyl Acetic Acid 30.00 ± 2.08 27.67 ± 5.55 21.33 ± 3.18 0.332 M-Inositol 26.00 ± 1.00 25.33 ± 1.86 29.00 ± 3.79 0.574 M-Tartaric Acid 3.00 ± 2.00 6.00 ± 1.53 2.33 ± 1.45 0.327 Maltose 24.33 ± 4.63 13.00 ± 2.52 15.33 ± 4.33 0.181 Maltotriose 22.67 ± 2.96 21.33 ± 2.60 20.00 ± 0.58 0.727 Methyl pyruvate 149.00 ± 24.11 140.0 ± 32.75 136.33 ± 21.67 0.942 Mono Methyl Succinate 5.00 ± 0.58 6.00 ± 1.00 5.00 ± 2.08 0.842 Mucic Acid 12.00 ± 2.89 7.00 ± 0.58 9.67 ± 1.33 0.244 N-Acetyl-β-D-Mannosamine 28.33 ± 2.33 22.33 ± 3.18 25.00 ± 2.65 0.364 N-Acetyl-D-Glucosamine 35.33 ± 1.20 29.33 ± 0.67 27.00 ± 2.52 0.029 p-Hydroxy-Phenyl Acetic Acid 20.00 ± 3.06 8.00 ± 3.46 19.67 ± 2.60 0.053 Phenylethylamine 30.33 ± 2.60 27.33 ± 4.84 29.33 ± 4.18 0.866 Propionic Acid 26.33 ± 8.35 25.00 ± 3.21 15.33 ± 2.67 0.354 Pyruvic Acid 254.67 ± 28.85 129.0 ± 67.99 184.67 ± 29.69 0.233 Succininc Acid 7.33 ± 3.76 4.00 ± 2.00 6.33 ± 6.33 0.863 Sucrose 30.67 ± 6.89 27.67 ± 8.21 27.33 ± 10.40 0.956 Thymidine 52.33 ± 4.91 54.33 ± 4.18 54.00 ± 5.29 0.952 Tricarballylic Acid 22.00 ± 6.51 8.00 ± 1.53 9.00 ± 5.20 0.158 Tween 20 38.33 ± 9.84 40.00 ± 11.50 23.33 ± 3.18 0.404 Tween 40 16.00 ± 1.15 28.33 ± 8.84 19.00 ± 1.53 0.293 Tween 80 23.67 ± 0.88 30.67 ± 6.17 24.00 ± 1.73 0.387 Tyramine 39.00 ± 4.58 27.00 ± 4.58 33.67 ± 3.33 0.211 Uridine 47.00 ± 1.73 43.33 ± 1.86 43.67 ± 7.06 0.809 aSubstrates were considered utilized if absorbance readings were above threshold of 50 units. bValues represent mean absorbance unit for three replicate phenotypic microarrays. c GAGL;Glutaric Acid-γ-Lactone

100

Table 3-4. Nitrogen substrate absorbance units of thrombolitic microbial mats. Nitrogen Substrates (n = 95)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value δ-Amino-N-Valeric Acid 11.75 ± 4.57 12 ± 4.20 16.25 ± 15.27 0.932 α-Amino-N-Valeric Acid 51 ± 17.20 36.25 ± 18.15 20 ± 9.81 0.405 Acetamine 35.75 ± 10.27 17 ± 5.67 21 ± 2.27 0.185 Adenine 59 ± 15.14 4 ± 3.67 17.25 ± 10.21 0.014 Adenosine 220.75 ± 26.80 253.25 ± 5.11 73.5 ± 54.55 0.012 Agmatine 197.25 ± 36.47 27.5 ± 9.54 74.75 ± 41.82 0.013 Ala-Asp 296.5 ± 22.51 145 ± 54.83 71.7 ± 57.73 0.025 Ala-Gln 247.25 ± 19.72 139.5 ± 49.82 87.5 ± 58.88 0.092 Ala-Glu 152.25 ± 39.55 32.75 ± 8.61 77.75 ± 18.30 0.027 Ala-Gly 252.5 ± 32.01 63 ± 20.19 91.25 ± 43.02 0.006 Ala-His 125.25 ± 10.77 23.75 ± 9.63 57.75 ± 20.38 0.002 Ala-Leu 136.75 ± 21.72 35.75 ± 12.78 58.75 ± 23.58 0.014 Ala-Thr 219.75 ± 24.17 196 ± 51.05 37.25 ± 19.07 0.009 Allantoin 22.75 ± 2.87 15 ± 5.35 17.75 ± 6.17 0.560 Alloxan 45.25 ± 12.26 45.75 ± 13.68 29.5 ± 2.33 0.501 Ammonia 78.5 ± 31.83 95 ± 56.81 8.25 ± 4.64 0.275 β-Phenylethylamine 7.5 ± 5.19 4.5 ± 4.5 0.5 ± 0.29 0.486 Biuret 0.5 ± 0.50 11.5 ± 11.50 0.75 ± 0.75 0.444 Cytidine 265 ± 12.23 204.5 ± 61.08 90 ± 69.14 0.118 Cytosine 7.75 ± 4.27 2.5 ± 1.85 8.5 ± 4.63 0.502 D-Alanine 163.75 ± 35.41 146 ± 52.78 78.5 ± 28.36 0.331 D-Asparagine 5 ± 1.96 10 ± 4.55 15.5 ± 4.13 0.192 D-Aspartic Acid 0.75 ± 0.75 4 ± 1.41 5.5 ± 2.90 0.252 D-Galactosamine 57 ± 33.80 31.25 ± 29.26 4 ± 1.78 0.388 D-Glucosamine 174 ± 43.19 126.5 ± 40.74 52 ± 48.06 0.199 D-Lysine 1.5 ± 1.50 0 ± 0.00 5.25 ± 5.25 0.506 D-Mannosamine 83.25 ± 39.92 87.25 ± 30.0 56.5 ± 2.63 0.724 D-Serine 73.75 ± 21.27 11.25 ± 9.07 46.75 ± 25.36 0.137 D-Valine 10 ± 10.0 5.75 ± 5.42 0 ± 0.0 0.577 D,L-a-Amino-N-Butyric Acid 6 ± 5.05 13.25 ± 5.41 6.25 ± 6.25 0.600 D,L-a-Amino-N-Caprylic Acid 7.5 ± 4.33 12.5 ± 7.01 11 ± 6.36 0.836 D,L-Lactamide 6.25 ± 4.25 17.25 ± 2.81 2.5 ± 2.5 0.028 ε-Amino-N-Caproic Acid 2.5 ± 2.18 6.5 ± 6.50 13 ± 7.36 0.466 Ethanolamine 0 ± 0.00 0.75 ± 0.75 7.5 ± 5.19 0.210 Ethylamine 10.5 ± 5.87 12.75 ± 7.36 3.5 ± 3.5 0.525 Ethylenediamine 1.25 ± 0.95 2.75 ± 1.70 1 ± 1.0 0.589 Formamide 10.25 ± 6.38 2.75 ± 1.38 7.75 ± 4.4 0.519 γ-Amino-N-Butyric Acid 74.25 ± 69.61 16.75 ± 10.58 9.25 ± 6.85 0.496 G-Glutamic Acid 9.5 ± 3.75 30.75 ± 30.75 5.5 ± 2.33 0.583 Glucuronamide 46.75 ± 11.12 14.5 ± 7.03 15.5 ± 6.49 0.040 Gly-Asn 173 ± 49.07 122.5 ± 69.45 139.5 ± 49.57 0.819 Gly-Gln 229.75 ± 41.67 138.25 ± 58.72 170.25 ± 61.46 0.512 Gly-Glu 78.5 ± 28.19 35.75 ± 18.84 9.5 ± 2.87 0.092 Gly-Met 189.25 ± 22.69 81 ± 33.80 86.25 ± 74.30 0.263 Glycine 208 ± 36.48 63.25 ± 45.88 71.5 ± 47.07 0.075 Guanine 21.5 ± 11.51 2.5 ± 2.50 10.25 ± 6.01 0.261 Guanosine 205.5 ± 21.53 77.75 ± 21.85 92.5 ± 54.03 0.064 Histamine 5 ± 2.35 5.75 ± 2.53 7.25 ± 2.87 0.826 Hydroxylamine 59 ± 23.89 26 ± 28.86 9.75 ± 4.33 0.191 Inosine 87 ± 21.64 50.5 ± 20.20 72.75 ± 44.94 0.714 L-Alanine 138.75 ± 31.71 69.75 ± 29.32 39 ± 29.15 0.108 L-Arginine 189.25 ± 21.53 171.25 ± 22.94 99.5 ± 44.58 0.159 L-Asparagine 94.5 ± 18.43 32.75 ± 20.91 6.25 ± 4.73 0.011

101

Table 3-4. Continued. Nitrogen Substrates (n = 95)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value L-Aspartic Acid 192.75 ± 37.53 102.25 ± 43.71 92.75±62.19 0.326 L-Citrulline 115 ± 30.83 14.25 ± 11.74 29.7 ± 517.92 0.020 L-Cysteine 235 ± 0.71 113 ± 44.49 199.5 ± 18.85 0.034 L-Glutamic Acid 178.5 ± 33.53 144.5 ± 46.47 125.5 ± 56.73 0.725 L-Glutamine 221.75 ± 18.99 68.5 ± 37.10 56.25 ± 53.92 0.028 L-Histidine 46.25 ± 28.72 13.5 ± 5.68 12 ± 2.35 0.319 L-Homoserine 3.25 ± 1.70 8 ± 6.10 0.5 ± 0.5 0.383 L-Isoleucine 17 ± 12.61 8.25 ± 4.25 12.5 ± 4.27 0.752 L-Leucine 89.25 ± 39.20 10 ± 9.03 18.5 ± 10.96 0.086 L-Lysine 3.5 ± 2.02 13.25 ± 7.65 2.25 ± 2.25 0.253 L-Methionine 46.5 ± 20.10 1.75 ± 1.75 23.5 ± 16.09 0.161 L-Ornithine 165.5 ± 40.08 104 ± 32.30 64.75 ± 31.97 0.178 L-Phenylalanine 40 ± 23.27 24.5 ± 11.98 16.75 ± 7.99 0.589 L-Proline 87.5 ± 40.51 73.75 ± 23.13 71.5 ± 4.99 0.904 L-Pyroglutamic Acid 12 ± 3.54 10 ± 5.79 39.5 ± 18.07 0.168 L-Serine 77 ± 44.06 8.25 ± 4.64 55.5 ± 54.84 0.501 L-Threonine 190.75 ± 19.78 70.25 ± 24.49 99.5 ± 51.67 0.087 L-Tryptophan 128.25 ± 32.89 52.25 ± 7.80 41 ± 15.95 0.037 L-Tyrosine 56.75 ± 3.35 63.25 ± 5.28 54 ± 8.21 0.552 L-Valine 20.5 ± 5.24 2.25 ± 1.93 5.5 ± 4.01 0.022 Met-Ala 237.25 ± 32.01 153.75 ± 44.17 147.5 ± 50.95 0.307 Methylamine 11.25 ± 5.95 5 ± 5.0 3 ± 3.0 0.479 N-Acetyl-D-Galactosamine 65.5 ± 38.40 58.75 ± 26.39 59.75 ± 47.80 0.991 N-Acetyl-D-Glucosamine 89 ± 9.60 101.5 ± 29.07 59.5 ± 22.41 0.416 N-Acetyl-D-Mannosamine 1.75 ± 1.18 5 ± 1.96 4 ± 2.61 0.527 N-Acetyl-D,L-Glutamic Acid 3.75 ± 2.17 5 ± 2.92 4 ± 1.35 0.917 N-Amylamine 13 ± 9.70 15.25 ± 13.95 1 ± 1.0 0.566 N-Butylamine 8.75 ± 4.61 5 ± 4.06 5 ± 4.36 0.786 N-Phthaloyl-L-Glutamic Acid 73 ± 45.35 18.25 ± 12.51 0 ± 0.0 0.197 Nitrate 66.5 ± 24.17 16.25 ± 2.50 56.25 ± 28.76 0.276 Nitrite 23 ± 8.93 11.5 ± 4.29 22 ± 7.99 0.499 Parabanic Acid 69.5 ± 52.18 29 ± 18.67 16.5 ± 11.11 0.512 Putrescine 190.25 ± 35.41 38 ± 37.34 0.25 ± 0.25 0.003 Thymidine 12.25 ± 3.82 7.25 ± 4.42 7 ± 3.34 0.579 Thymine 3.5 ± 1.19 11.5 ± 8.02 3.75 ± 2.06 0.445 Tyramine 7.5 ± 4.05 10 ± 4.81 12.25 ± 5.74 0.796 Uracil 5.5 ± 1.66 4 ± 2.61 0 ± 0.0 0.134 Urea 117.25 ± 42.53 64.5 ± 53.77 149.5 ± 60.83 0.541 Uric Acid 78 ± 15.72 53.5 ± 14.89 60.75 ± 10.70 0.473 Uridine 7.25 ± 4.23 17.75 ± 12.03 2 ± 1.68 0.355 Xanthine 130 ± 39.06 104.25 ± 38.22 96.75 ± 43.59 0.833 Xanthosine 13 ± 6.86 4.75 ± 4.75 4.75 ± 4.75 0.505 aSubstrates were considered utilized if absorbance readings were above threshold of 50 units. bValues represent mean absorbance unit for three replicate phenotypic microarrays.

102

Table 3-5. Sulfur substrate absorbance units of thrombolitic microbial mats Sulfur Substrates (n = 35)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value 1-Thio-b-D-Glucose 60.50 ± 2.75 55.25 ± 1.18 2.25 ± 0.85 0.001 2-Hydroxyethane Sulfonic Acid 36.75 ± 4.01 29 ± 4.18 6.75 ± 2.39 0.001 Butane Sulfonic Acid 35.50 ± 2.25 28.25 ± 2.72 2.00 ± 0.91 0.001 Cystathionine 102.50 ± 2.10 99.25 ± 4.19 3.00 ± 2.38 0.001 Cysteamine 106.50 ± 3.01 87 ± 2.27 0.00 ± 0.0 0.001 D-Cysteine 65.50 ± 1.76 63.75 ± 2.72 6.75 ± 6.75 0.001 D-Methionine 88.25 ± 2.14 75.5 ± 5.24 7.50 ± 7.5 0.001 D,L-Ethionine 36.75 ± 2.06 31 ± 2.35 0.00 ± 0.0 0.001 D,L-Lipomide 186.50 ± 2.10 164.5 ± 2.99 17.00 ± 14.16 0.001 Dithiophosphate 97.00 ± 5.12 110 ± 5.85 14.50 ± 14.50 0.001 Glutathione 112.00 ± 5.64 106 ± 1.68 17.00 ± 17.0 0.001 Glycyl-L-Methionine 130.25 ± 4.82 116.25 ± 5.62 24.75 ± 23.76 0.001 Hypotaurine 41.50 ± 0.96 33 ± 1.41 2.00 ± 0.41 0.001 L-Cysteic Acid 25.75 ± 2.29 23 ± 2.27 0.00 ± 0.0 0.001 L-Cysteine 233.50 ± 4.94 223 ± 5.34 2.25 ± 2.25 0.001 L-Cysteine Sulfinic Acid 107.00 ± 9.05 106.5 ± 6.64 33.75 ± 14.88 0.001 L-Cysteinyl-Glycine 112.75 ± 0.95 111.75 ± 1.38 33.00 ± 33.0 0.025 L-Djenkolic Acid 290.50 ± 3.69 286 ± 3.49 1.75 ± 1.18 0.001 L-Methionine 145.75 ± 2.10 129 ± 4.42 0.00 ± 0.0 0.001 L-Methionine Sulfone 45.00 ± 8.50 50.75 ± 10.9 42.50 ± 16.3 0.890 L-Methionine Sulfoxide 123.75 ± 1.80 113.5 ± 4.63 39.25 ± 19.68 0.001 Lanthionine 58.25 ± 4.77 46.75 ± 0.85 32.50 ± 9.31 0.044 Methane Sulfonic Acid 37.75 ± 1.70 32 ± 2.35 11.75 ± 7.85 0.010 N-Acetyl-D,L-Methionine 90.50 ± 1.19 86 ± 1.78 32.75 ± 15.37 0.002 N-Acetyl-L-Cysteine 37.50 ± 1.76 29.5 ± 0.96 0.00 ± 0.0 0.001 p-Amino Benzene Sulfonic Acid 36.75 ± 1.75 28.25 ± 1.25 0.00 ± 0.0 0.001 S-Methyl-L-Cysteine 75.75 ± 2.46 72.75 ± 1.25 0.00 ± 0.0 0.001 Sulfate 152.00 ± 3.49 140.5 ± 13.52 0.00 ± 0.0 0.001 Taurine 37.75 ± 2.50 30.75 ± 3.40 0.75 ± 0.48 0.001 Taurocholic Acid 39.00 ± 1.63 37.25 ± 1.65 0.00 ± 0.0 0.001 Tetramethlene Sulfone 58.50 ± 5.95 56 ± 8.13 41.75 ± 16.33 0.538 Tetrathionate 131.00 ± 6.18 97.5 ± 20.6 8.75 ± 5.28 0.001 Thiophosphate 74.25 ± 5.19 57.75 ± 0.75 13.00 ± 13.0 0.001 Thiosulfate 140.00 ± 1.78 123.5 ± 14.40 1.75 ± 1.18 0.001 Thiourea 85.25 ± 3.99 83.25 ± 7.34 12.75 ± 12.75 0.001 aSubstrates were considered utilized if absorbance readings were above threshold of 50 units. bValues represent mean absorbance unit for three replicate phenotypic microarrays.

103

Table 3-6. Phosphate substrate absorbance units of thrombolitic microbial mats. Phosphate Substrates (n = 58)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value 2-Aminoethyl Phosphonic Acid 101.25 ± 10.90 123.00 ± 3.03 5.50 ± 3.28 0.001 2-Deoxy-D-Gluc-6-Phosphate 94.75 ± 21.51 122.25 ± 4.87 0.25 ± 0.25 0.001 6-Phospho-Gluconic Acid 135.00 ± 6.20 123.50 ± 8.27 3.25 ± 1.31 0.001 Adenosine 2,3'-cmpc 256.75 ± 8.04 267.75 ± 2.87 54.75 ± 51.8 0.001 Adenosine 3,5'-cmpc 226.00 ± 9.51 240.00 ± 12.19 77.25 ± 32.98 0.001 Adenosine-2'-monophosphate 188.50 ± 7.84 197.75 ± 8.66 0.50 ± 0.50 0.001 Adenosine-3'-monophosphate 251.75 ± 13.20 262.00 ± 6.94 1.25 ± 1.25 0.001 Adenosine-5'-monophosphate 181.00 ± 3.16 180.00 ± 5.12 2.50 ± 1.66 0.001 b-Glycerol Phosphate 125.75 ± 13.34 135.75 ± 2.87 10.75 ± 2.98 0.001 Carbamyl phosphate 97.75 ± 7.79 118.00 ± 6.16 2.50 ± 2.50 0.001 Cysteamine-S-Phosphate 218.75 ± 69.92 290.50 ± 2.18 18.50 ± 2.66 0.003 Cytidine 2,3'-cmpc 144.00 ± 41.71 208.25 ± 8.53 7.50 ± 5.25 0.001 Cytidine 3,5'-cmpc 142.50 ± 4.17 154.25 ± 4.50 47.75 ± 2.53 0.001 Cytidine-2'-Monophosphate 167.50 ± 13.68 179.00 ± 7.56 2.00 ± 2.0 0.001 Cytidine-3'-Monophosphate 160.00 ± 3.70 176.25 ± 9.50 1.50 ± 1.50 0.001 Cytidine-5'-Monophosphate 149.50 ± 6.46 160.50 ± 10.69 4.50 ± 2.87 0.001 D-2-PhosphoGlyceric Acid 99.25 ± 13.37 130.75 ± 3.52 2.00 ± 2.0 0.001 D-3-PhosphoGlyceric Acid 109.75 ± 22.31 109.75 ± 8.52 12.50 ± 9.67 0.002 D-Glucosamine-6-Phosphate 137.00 ± 15.93 135.50 ± 7.58 15.75 ± 15.75 0.001 D-Glucose-1-Phosphate 154.25 ± 9.87 149.50 ± 4.44 4.50 ± 0.65 0.001 D-Glucose-6-Phosphate 157.50 ± 6.51 163.75 ± 7.03 16.25 ± 3.64 0.001 D-Mannose-1-Phosphate 134.00 ± 2.27 146.25 ± 6.01 6.50 ± 3.23 0.001 D-Mannose-6-Phosphate 133.25 ± 5.98 131.75 ± 11.81 2.50 ± 1.66 0.001 D,L-a-Glycerol Phosphate 136.75 ± 3.30 121.00 ± 8.58 37.50 ± 16.48 0.001 Dithiophosphate 295.00 ± 13.39 278.00 ± 36.18 11.00 ± 3.89 0.001 Guanosine 2,3'-cmpc 186.00 ± 5.02 197.50 ± 3.33 2.75 ± 1.70 0.001 Guanosine 3,5'-cmpc 108.25 ± 10.50 113.50 ± 8.14 33.75 ± 12.45 0.001 Guanosine-2'-monophosphate 167.25 ± 3.64 180.00 ± 4.64 2.75 ± 1.60 0.001 Guanosine-3'-monophosphate 132.50 ± 41.68 180.75 ± 12.97 6.25 ± 5.01 0.002 Guanosine-5'-monophosphate 139.00 ± 7.38 140.25 ± 1.93 1.50 ± 0.96 0.001 Hypophosphite 17.00 ± 2.61 16.75 ± 2.39 1.00 ± 1.0 0.001 Inositol Hexaphosphate 48.75 ± 3.75 70.75 ± 9.15 6.75 ± 3.90 0.001 Methylene Diphosphonic Acid 14.75 ± 2.95 15.75 ± 0.85 4.25 ± 4.25 0.046 O-Phospho-D-Serine 96.50 ± 6.59 105.50 ± 6.25 2.75 ± 2.75 0.001 O-Phospho-D-Tyrosine 123.75 ± 3.97 93.50 ± 16.83 5.00 ± 2.27 0.001 O-Phospho-L-Serine 118.75 ± 8.31 141.50 ± 5.52 1.00 ± 1.0 0.001 O-Phospho-L-Threonine 116.50 ± 3.48 101.75 ± 2.72 8.00 ± 4.90 0.001 O-Phospho-L-Tyrosine 112.75 ± 4.44 119.00 ± 3.03 1.50 ± 0.96 0.001 O-Phosphyl-Ehtanolamine 83.00 ± 3.94 107.50 ± 14.62 3.50 ± 318 0.001 Phosphate 118.00 ± 8.09 109.25 ± 8.63 10.50 ± 5.55 0.001 Phospho Glycolic Acid 103.50 ± 11.55 120.75 ± 3.73 8.50 ± 8.17 0.001 Phospho-L-Arginine 123.50 ± 25.99 154.50 ± 6.28 19.00 ± 7.62 0.001 Phosphocreatine 126.25 ± 7.32 103.75 ± 4.96 16.75 ± 4.37 0.001 Phosphoenol pyruvate 98.75 ± 19.52 137.50 ± 5.56 8.75 ± 4.70 0.001 Phosphono Acetic Acid 29.75 ± 1.18 31.50 ± 2.87 0.00 ± 0.0 0.001 PhosphorylCholine 118.00 ± 4.51 114.00 ± 8.95 19.50 ± 7.10 0.001 Pyrophosphate 97.25 ± 17.88 123.25 ± 11.52 36.75 ± 5.15 0.003 Thiophosphate 262.25 ± 25.05 245.75 ± 11.18 3.50 ± 2.36 0.001 Thymidine 3,5'-cmpc 111.00 ± 6.79 120.25 ± 2.95 26.00 ± 9.06 0.001 Thymidine-3'-monophosphate 129.00 ± 10.72 113.25 ± 8.44 0.00 ± 0.0 0.001 Thymidine-5'-monophosphate 89.75 ± 30.01 132.50 ± 1.55 0.00 ± 0.0 0.001 Triethylphosphate 16.00 ± 6.58 24.50 ± 6.09 0.00 ± 0.0 0.024

104

Table 3-6. Continued. Phosphate Substrates (n = 58)a Zone 1b ± SEM Zone 2 ± SEM Zone 3 ± SEM P-value Tripolyphosphate 77.75 ± 22.65 100.00 ± 2.35 0.00 ± 0.00 0.001 Uridine 2,3'-cmpc 149.25 ± 7.73 175.00 ± 2.12 0.00 ± 0.0 0.001 Uridine 3,5'-cmpc 110.75 ± 5.36 121.00 ± 3.72 27.50 ± 8.10 0.001 Uridine-2'-monophosphate 125.75 ± 5.91 131.50 ± 8.91 15.50 ± 9.67 0.001 Uridine-3'-monophosphate 134.00 ± 7.77 116.00 ± 25.54 3.25 ± 3.25 0.001 Uridine-5'-monophosphate 149.50 ± 6.28 155.50 ± 6.06 2.75 ± 2.75 0.001 aSubstrates were considered utilized if absorbance readings were above threshold of 50 units. bValues represent mean absorbance unit for three replicate phenotypic microarrays. c cmp: cyclic monophosphate.

105

ABC

1 2 3

Figure 3-1. Modern thrombolites of Highborne Cay, Bahamas. A) Thrombolitic build-ups along the intertidal zone. Bar = 30 cm. B) cross section of thrombolitic microbial mat demarking three zones within the mat. Zone 1 comprised the upper 3 mm of the mat whereas Zone 2 consisted of 3 – 5 mm beneath the surface of the mat, and Zone 3 contained the lower portion of the mat between 5 – 9 mm. Bar = 1 cm. C) within the upper Zone 1 there is an increase in the relative abundance of the filamentous cyanobacterium Dichothix spp. (arrows) which appear as hot spots of carbonate deposition within the thrombolitic mats. Bar = 0.5 mm.

106

100

80 Cofactors, Vitamins, PG, Pigments

40 Fatty Acids, Lipids, Isoprenoids

20 Relative Abundance (percentage) Relative

0 Thb A Thb B Thb C

Figure 3-2. SEED subsystems overview of the three thrombolitic mat metagenomes.

107

A B Actinobacteria (2388) Apicomplexa (854) Alveolata Bacteroidetes (729981) Ciliophora (10495) Chlorobi (2431) Dinophyceae (5631) Chlamydiae (3452) Verrucomicrobia (25115) Cryptophyta (72) Chloroflexi (24501) Diplonemida (45) Eugelenozoa Cyanobacteria (2503977) Euglenida (36) Acidobacteria (205) Firmicutes (13571) Kinetoplastida (966) Fusobacteria (309) Glaucocystophyceae (136) Nitrospirae (369) Planctomycetes (78276) Haptophyceae (209) ĮSURWHR Caulobacterales (158) Choanoflagellida (71) bacteria Rhizobiales (50363) Chrytridiomycota (15) (387350) Rhodobacterales (216750) Rhodospirillales (13292) Ascomycota (19260) ȕSURWHR Rickettsiales (6451) Fungi (4) Basidomycota (138) bacteria Sphingomonadales (11986) Fungi incertae sedis (19) (10765) Burkholderiales (87) Methylophilales (21) Glomeromycota (15) Bacteria įSURWHR Neisseriales (685) Microsporidia (73) (2085035) bacteria Bdellovibrionales (510) Platyhelminthes (703) Desulfobacterales (2498) (23668) Desulfovibrionales (1978) Chordata (6633) Desulfuromonadales (2947) Echinodermata (52) Myxococcales (460) Annelida (4531) Syntrophobacterales (163) Campylobacterales (1354) Echiura (54) Proteo Acidithiobacillales (123) Brachiopoda (13) bacteria Aeromonadales (1128) Bryozoa (39) (516733) Alteromonadales (52996) Cardiobacteriales (31) Mollusca (657) Chromatiales (3242) Arthropoda (38460) Enterobacteriales (531) Tardigrada (31) Legionellales (3504) Priapulida (15) ȖSURWHR Methylococcales (116) Oceanospirillales (1789) Sipuncula (115) bacteria Pasteurellales (13) Nemotoda (2013)

(183315) Pseudomonadales (2405) Eukaryota (373216) Thiotrichales (931) Rotifera (10) Vibrionales (17430) Cnidaria (597) Xanthomonadales (990) Ctenophora (28) Spirochaetes (4040) Syngeristetes (382) Porifera (99) Tenericutes (378) Opisthokonta (25) Archaea (61814) Cercozoa (258) Eukaryota (562361) Rhizaria viruses (3075) Foraminifera (2616) Rhodophyta (12048) Crenarchaeota (99) Bacillariophyta (48591) C Thermoproteales (2) Bicosoecida (75) Archaeoglobaceae (24) Chrysophyceae (28) Archaeoglobus (2) Dictyochophyceae (22) Halobacteriaceae (3) Labyrinthulida (386) Methanobacteriaceae (29) Archaea Methanothermobacter (9) stramenopiles Oomycetes (189) (3759) Pelagophyceae (532) Methanospirillum (1) Pinguiophyceae (191) Methanosaeta (2) Phaeophyceae (349) Xanthophyceae (151) Methanosarcinaceae (574) Raphidophyceae (435) Ferroplasma (2) Synurophyceae (37) Thaumarchaeota (16719) Cenarcheum (1338) Telonemida (29) Viridiplantae Chlorophyta (4357) Nitrosopumilus (39250) Streptophyta (26639)

Figure 3-3. Taxonomic distribution of the thrombolite metagenome based on MG-RAST Lowest Common Ancestor analysis, A) overview of reads assigned to Bacterial Phyla (>100 reads) with Proteobacteria assigned to order level, B) Higher resolution of Archaea reads, C) Phyla level (>10 reads) resolution of Eukaryota reads. All trees were created in MEGAN using MG-RAST annotation server read abundance data for each respective taxonomic level is given in parentheses.

108

Acaryochloris (4253) Chamaesiphon (3) Crocosphaera (11622) Cyanobacterium (44) Cyanothece (143190) Chroococcales (203491) Dactylococcopsis (7) Halothece (28) Microcystis (198) Snowella (2) Synechoccocus (97) Synechocystis (21) Thermosynechococcus (9) Microchaetaceae (1322) Microchaete (415) Tolypothrix (47) Anabaena (1311) Aphanizomenon (23)

Cylindrospermopsis (3443) Nostocales Nostocaceae (6943) (309804) Cylindrospermum (4) Dolichospermum (212) Nodularia (5) Nostoc (11841) Sphaerosperopsis (1) Calothrix (234) Scytonema (24) Anthrospira (3) Cyanobacteria (1742466) Geitlerinima (109) Jaaginema (8) Leptolyngbya (136) Lyngbya (2094) Oscillatoriales Microcoleus (46) (55790) Oscillatoria (90) Phormidium (25) Planktothrix (1) Pseudoanabaena (112) Spirulina (167) Symploca (7) Trichodesmium (227) Chroococcidiopsis (49) Pleurocapsales (48) Pleurocapsa (59) Staneria (122) Xenococcus (87) Prochlorales (4) Prochlorococcus (3478) Prochlorothrix (88) Stigonematales (39) Fischerella (83) Mastigocladus (45)

Figure 3-4. Genus level distribution of cyanobacteria reads from the thrombolite metagenome based on MG-RAST Lowest Common Ancestor analysis. Tree created in MEGAN using MG-RAST annotation server read abundance data for each respective taxonomic level is given in parentheses

109

A B

Figure 3-5. MG-RAST functional assignment of thrombolite metagenome protein features. A) Overview of SEED subsystems (Level 1). B) SEED subsystems (Level 3) associated with energy metabolism. Protein feature abundances are given in parentheses.

110

A Guerrero Negro Rio Mesquites Oncolite Type 3 Stromatolite Los Venados Red Mat Pozas Azules Green Mat Thr-A (this study) Octopus Spring, YNP Pozas Azules Thrombolite Thr-B (this study) Sargasso Sea 7\SH6WURPDWROLWH Thr-C (this study) 2

Regulation & Cell Signaling Potassium

Cell Wall & Capsule

0 Cell Division & Cell Cycle Respiration Nucleosides & Nucleotides ï incipal Component 2 (20.66%) ï P r

ï ï ï 0 2 3 4 Principal Component 1 (72.19%)

B Respiration DNA Metabolism Mobile elements Dormancy and Sporulation Iron metabolism Miscellaneous Cell Wall and Capsule Amino Acids Metabolism Stress Response Phosphorus Metabolism ClusterLQJïEDVHGVXEV\VWHPV Carbohydrates Virulence, Disease and Defense Potassium metabolism Aromatic Compound Metabolism RNA Metabolism Nucleosides and Nucleotides Secondary Metabolism Sulfur Metabolism Fatty Acids, Lipids, and Isoprenoids Protein Metabolism Motility and Chemotaxis Membrane Transport Cofactors, Vitamins, Pigments Nitrogen Metabolism Cell Division and Cell Cycle Photosynthesis Regulation and Cell signaling 2 3 4 5 6 7 Mean Decrease in Accuracy

Figure 3-6. Comparison of the thrombolite metagenome with metagenomes of previously sequenced lithifying and non-lithifying microbial mat ecosystems. A) Principle Component Analysis of SEED Subsystems (Level 1) derived from several distinct habitats. B) SEED subsystem (i.e. variable) importance determined by random forest analysis of lithifying and non-lithifying microbial mat metagenomes.

111

A B

Figure 3-7. Clustered heat map visualizing the substrate utilization patterns throughout the spatial profile of the thrombolitic mats using phenotypic microarrays. Dendograms generated based on Euclidean complete linkage clustering. Color scale reflects the intensity of usage with Biolog absorbance units ranging from 0 to 250. White boxes indicate that absorbance readings were below the threshold level. A) Carbon (C) and nitrogen (N) substrate utilization within the three zones of the thrombolitic mats, B) Phosphate (P) and Sulfur (S) substrate utilization within the three zones of the thrombolitic mat.

112

CHAPTER 4 METATRANSCRIPTOMIC SEQUENCING OF THROMBOLITIC MAT REVEALS DISTINCT SPATIAL GRADIENTS OF METABOLISMS ASSOCIATED WITH BIOLOGICALLY INDUCED MINERALIZATION

Introduction

Microbialites are carbonate build-ups produced through the trapping, binding, and mineral precipitation activity of cyanobacterial enriched microbial mats. These photosynthetic ecosystems are considered to be analogs to the earliest ecosystems on

Earth (Burne and Moore, 1987; Grotzinger and Knoll, 1999) and may have contributed to the rise of atmospheric oxygen, thus enabling the evolution of more complex communities and higher organisms (Kasting and Howard, 2006; Papineau et al., 2013).

Microbialites form at the interface of the biosphere and lithosphere rendering these ecosystems ideal models for understanding the development and formation of lithifying microbial ecosystems.

Previously, much of the research on microbialites has focused on determining the carbonate microstructure (Reid et al., 2000; Riding 2011; Planavsky and Ginsburg,

2009), biogeochemistry (Visscher et al., 2000; Paerl et al. 2001; Myshrall et al., 2010;

Nitti et al., 2012), and microbial diversity (Burns et al., 2004; Baumgartner et al., 2008;

Myshrall et al., 2010; Mobberley et al., 2012) of these ecosystems. These earlier studies showed that the microbial activity through metabolic energy transformations and cell-to- cell interactions create both spatial and temporal biogeochemical gradients that influence calcium carbonate precipitation within microbialite systems. Several function guilds have been detected in the microbialites that that are thought to influence carbonate mineralization and include: oxygenic and anoxygenic phototrophs, aerobic heterotrophs, sulfate-reducers, sulfide-oxidizers, and fermenters (Visscher et al., 1998;

113

Visscher et al., 2000; Steppe et al., 2001; Baumgartner et al., 2009; Myshrall et al.,

2010; Mobberley et al., 2012; Mobberley et al., 2013). Together, these guilds influence the net precipitation potential of calcium carbonate within lithifying mats by impacting the saturation index and cycling of exopolymeric substances (EPS; reviewed in Dupraz et al., 2009). The saturation index is determined by the availability of free calcium ion and the carbonate alkalinity. Carbonate alkalinity is influenced by microbial metabolic processes that either increase the pH (e.g. photosynthesis; sulfate-reduction) to promote precipitation or decrease the pH (e.g. aerobic respiration; sulfide oxidation), thus promoting carbonate dissolution (Visscher and Stolz, 2005). It is the overall net balance of these various metabolisms that determines the location and extent of carbonate precipitation within lithifying mat communities.

Another key factor in the biologically induced precipitation of carbonate in modern microbialites is through the production of EPS. The EPS matrix in lithifying mats can serve as a substrate for heterotrophic growth (Decho et al., 2005), provide structural stability to the community in physically dynamic systems (Paterson et al.,

2008), and function as a nucleation sites for calcium carbonate precipitation through the binding and release of calcium cations (Decho 2000; Kawaguchi and Decho, 2000;

Braissant et al., 2007). Studies in stromatolites found that EPS produced by cyanobacteria (Kawaguchi and Decho, 2000; Foster et al., 2009) and sulfate-reducing bacteria (Braissant et al., 2007; Gallagher et al., 2012) have differences in chemical composition and cycling that result in varying carbonate precipitation potentials

(Visscher et al., 2009; Kawaguchi and Decho, 2002; Dupraz et al., 2009; Gallagher et al., 2012). For instance, sulfate reduction has been linked to calcium carbonate

114

precipitation due to the increased alkalinity and release of calcium cations from

heterotrophic degradation of EPS (Visscher et al., 1998; Visscher et al., 2000; Paerl et

al., 2001; Baumgartner et al., 2006). Calcium carbonate precipitation associated with

EPS-rich cyanobacterial sheaths is primarily attributed to photosynthetic activity

(Planavsky et al., 2009). However, the genetic processes underlying these metabolic

controls of carbonate precipitation within lithifying mats is unresolved.

Although microbialites are found globally in a wide range of freshwater, marine

and hypersaline environments (e.g. Laval et al., 2000; Burns et al., 2004; Breitbart et al.,

2009; Bosak et al., 2012; Farias et al., 2013), the microbialites at Highborne Cay, The

Bahamas are relatively unique in that the two dominate types of microbialite, laminated stromatolites and clotted thrombolites, co-exist within this single habitat (Myshrall et al.,

2010). The well-laminated, subtidal stromatolites result from iterative microbial growth through cycles of sediment trapping and binding as well as accretion (Reid et al., 2000;

Macintyre et al., 2000; Paerl et al., 2001; Bowlin et al., 2012), however, the specific processes that form the clotted thrombolites structures are unknown (Figure 4-1A,B). As in their stromatolite counterparts thrombolite platforms are overlain by a centimeter thick microbial mat (Figure 4-1C,D) containing a taxonomically and metabolically diverse community of organisms capable of carbonate precipitation (Desnues et al., 2008;

Planavsky et al., 2009; Myshrall et al., 2010; Mobberley et al., 2012; Edgcomb et al.,

2013b; Taharan et al. 2013). Despite these geochemical and microbial diversity characterizations, the distribution of functional genetic pathways linking the underlying genetics to metabolic gradients is just beginning to be studied (Mobberley et al., 2013).

A recent metagenomic survey of thrombolite forming mat from Highborne Cay detected

115

the presence of several microbial guilds and functional genetic pathways associated

with carbonate precipitation, suggesting there were spatial differences in metabolite

utilization (Mobberley et al. 2013). However, the molecular controls and expression of

the functional genes within these lithifying ecosystems remains unknown. In this study,

we use metatranscriptomic sequencing to examine the expression of these functional

pathways within the thrombolite forming mats during peak times of photosynthesis. The

sequencing effort was combined with microelectrode monitoring of oxygen productivity

to generate a spatial profile of those dominant metabolisms associated with these

lithifying microbial communities.

Materials and Methods

Sample Collection

Thrombolitic mats were collected from the island of Highborne Cay, The

Bahamas (76°49’ W, 24°43’N) in February and March 2010. All samples were collected

at 12 pm in triplicate from a single intertidal thrombolitic platform from Site 5, as

designated by Andres and Reid (2006)(Figure 4-1B,C). The thrombolitic mats used for

metatranscriptome analysis were sectioned into three zones based on the O2 profile generated as described below and immediately placed into RNALater (Life

Technologies, Inc, Grand Island, NY)(Figure 4-1D,E). Corresponding samples were also collected in triplicate for metagenomic analysis (Mobberley et al., 2013). All samples were transported to Space Life Sciences Lab, Kennedy Space Center, FL and stored at

-80°C until processing.

Microelectrode Oxygen Profiles

Prior to sample collection, biochemical depth profiles for O2 were measured

within the thrombolitic mat. Mats were probed in situ using polarimetric Clark needle

116

electrodes with an outer diameter of 0.4 mm and a sensing tip of 10-20 µm (Visscher et al., 1991). Measurements were taken in 0.2 mm increments with the aid of a motorized ministage MM-3M-EX-2.0 and a programmable Servo 3000 MC4B controller box

(National Aperture, Salem, NH).

Nucleic Acid Extraction and Purification

Total RNA was extracted from each mat zone in triplicate by a modified RNAzol protocol (MRC, Cincinnati, OH). Briefly, 1 g of thrombolitic mat was ground to a fine powder in a mortar under liquid N2 then divided into 80-100 mg aliquots that were each added to 1mL RNAzol RT reagent. These mixtures were vortexed for 15 s and centrifuged for 5 min at 23°C. The supernatants were transferred to fresh tubes and

DEPC-treated water was added to precipitate the DNA and proteins. After mixing through inversion, the samples were incubated for 15 min at 23°C, then centrifuged for

15 min at 4°C to recover the RNA containing supernatant. To increase recovery of total

RNA, 1 µl of RNA precipitate carrier (MRC,Cincinnati, OH) was added prior to precipitation with 1 mL 75% ethanol. These samples were incubated at 23°C for 10 min followed by centrifugation for 8 min at 4°C. The resulting total RNA pellets were washed twice with 75% ethanol, air dried for 2 min, then resuspended in molecular grade water.

The additional RNA purification described below was carried out with materials and protocols derived from Life Technologies, Inc., Grand Island, NY unless otherwise noted. All RNA samples were treated with Turbo-DNase to remove any remaining DNA and re-concentrated with the MegaClear kit. Samples were further concentrated by centrifugation under vacuum at 4°C. Any additional inhibitors were removed with a lithium chloride precipitation step. To generate the mRNA-enriched fractions, total RNA

117

(2 µg) was enriched for mRNA using MicrobExpress, which removes 16S and 23S

rRNAs. This enrichment was followed by purification with MegaClear.

In each thrombolitic mat zone RNA was amplified from total RNA (40 ng) and

mRNA-enriched fractions (20 ng) in triplicate using the MessageAMP II-Bacterial RNA

kit (Life Technologies, Inc, Grand Island, NY). The in vitro transcription reaction was

incubated for 14 h at 37°C and the resultant antisense RNA (aRNA) was purified and

concentrated. cDNA was synthesized from the 1 µg aRNA reaction product in triplicate

using Promega Universal RiboClone cDNA Synthesis System (Promega, Madison, WI)

with random hexamer primers and 1.25 U GoScript enzyme (Promega, Madison, WI).

The second strand synthesis step was performed with T4 polymerase using

manufacturer’s protocol. The double stranded cDNA was purified with the Wizard

Genomic DNA Purification kit (Promega, Madison, WI) and quantified with a Qubit 2.0

fluorometer.

To complement the metatranscriptomic libraries, metagenomic DNA was

extracted from corresponding whole thrombolitic mat sections using a previously

described xanthogenate method that has been previously optimized for microbialites

(Foster et al., 2009, Mobberley et al., 2012).

DNA Sequencing and Read Processing

cDNA and DNA were sequenced using the Illumina GAIIx platform at the

University of Florida’s Interdisciplinary Center for Biotechnology Research. Single-end

reads were generated for total RNA and mRNA-enriched fractions from the 0 - 3 mm zone. Paired end reads were generated for total RNA from the 3-5 mm and 5-9 mm zone, the mRNA-enriched fraction from the 3-5 mm zone, and the metagenomic DNA.

118

All Illumina libraries were trimmed and quality filtered with the command-line

version of Prinseq to remove reads with ambiguous positions, quality scores below 20,

and low complexity sequences (Schmieder and Edwards, 2011). Quality-filtered reads

were analyzed as described below.

De novo Metagenomic Assembly and Annotation

Assembly of the metagenomic reads was performed using a hybrid assembly

approach as described in Luo et al. (2011). Briefly, the metagenomic reads from each

Illumina flow cell underwent a series of de novo assemblies using different kmer (K)

sizes with the programs MetaVelvet (K= 21, 25, 29) and SOAP de novo (K= 23, 27, 31).

The resulting contigs were merged into one file and assembled with Newbler (version

2.7; Roche, Branford, CT) using the following parameters: -ml 100 -mi 95 –a 200 –l 500.

Ribosomal RNA (rRNA) genes were identified and analyzed from the assembled contigs

as described below. Protein-encoding genes in the contigs were identified using the

program MetaGeneMark (Zhu et al., 2010) and analyzed with the messenger RNA as

described below.

Ribosomal RNA Transcript Analysis and Classification

Ribosomal rRNA transcripts were identified and removed from the

metatranscriptomic libraries by alignment against the combined SILVA small subunit

(SSU) and large subunit (LSU) databases (V-111) using blastn with a maximum score

of e-5. Following annotation, read counts were parsed using SILVA mapping files with

SILVA taxonomy assignments. For the total RNA library, Phyloseq (McMurdie and

Holmes, 2013) was used to for comparative analyses of the different thrombolitic mat zones with the top 20 most abundant organisms from each taxonomic domain selected

119

based on the number of hits to a specific organism. Bubble charts were generated using a perl script developed by Charles Howes (Zaikova et al., 2009).

Messenger RNA Analysis

After removal of rRNA transcripts, protein-encoding genes in each of the metatranscriptomes were identified with FragGeneScan assuming an error rate of one percent (Rho et al., 2010). To assign a putative function, protein-encoded genes were annotated via BLAT against the non-redundant database protein database (nr release

February 02, 2013; minimum identity 30), with the best RefSeq hit retained (maximum score e-3) to represent each annotated gene. The BLAT results were imported into

MEGAN5 (version 5.0.78-beta; Huson et al., 2011) for further functional analysis. To obtain a broad functional overview of each zone within the thrombolitic mat annotated gene reads were compared to the SEED subsystem database (July 2010). For more in- depth analyses of energy metabolisms, the KEGG pathway database (June 2011) was used. To account for differences in sampling depth, the resulting abundance data was log(n+1) transformed in R prior to heatmap generation with Euclidean complete linkage clustering using with pheatmap package (version 0.7.4; Kolde 2013). Principal component analyses were carried out using the R package bpca (Faria and Demetrio

2008) as described by Dinsdale et al. (2013). The five most important subsystems based on highest variance are displayed as a biplot for each PCA analysis. To account for potential annotation deficiencies in the SEED and KEGG subsystems, the best blat hit tables from each library were searched against GenBank (GI number) and Refseq

(Ref number) records created on Sept 1st 2013 for the following genes and/or pathways: sulfate reduction (dsrAB), sulfur oxidation (sox), ammonia oxidation (amo),

120

methanogenesis (formylmethanofuran dehydrogenase), and cyanophage photosystem

genes (psbD).

Comparisons of Thrombolite Metagenome and Metatranscriptomes to Other Microbial Ecosystems

The assembled metagenome contigs and raw metatranscriptomic libraries were

also submitted and analyzed through the MG-RAST standard pipeline (MG-RAST

version 3.3.6). Dereplication was not performed to maintain gene expression

abundance levels in the metatranscriptomes. Previously published microbial mat

metagenome datasets were mined from MG-RAST and compared to the results of the

current study (Project number mgp6163). These datasets (study reference and MG-

RAST IDs given) included: unassembled marine thrombolitic mat metagenome from

Highborne Cay, The Bahamas (Mobberley et al., 2013; 4513715.3, 4513716.3,

4513717.3); non-lithifying (Type 1) and lithifying (Type 3) marine stromatolitic mats from

Highborne Cay, The Bahamas (Khodadad and Foster, 2012; 4449590.3, 4449591.3);

freshwater microbialites from Cuatros Ciénegas, Mexico (Breitbart et al., 2009;

44440060.4, 4440067.3), hypersaline non-lithifying mats from Guerrero Negro, Mexico

(Kunin et al. 2008; 4440964.3-4440972.3 were combined into one library). All samples

were analyzed at a max e-value cutoff of 10-5 identity > 60%, and minimum alignment of

45 peptides. To account for differences in sampling sizes SEED subsystem (Level 1) abundance counts were normalized to total number of SEED annotated reads for each sample. To maximize functional differences between the environments analyzed,

“Clustering-based subsystems” and “Miscellaneous” were excluded. Principal component analyses were carried out using the R package bpca (Faria and Demetrio

2008).

121

Results and Discussion

The intertidal thrombolites found at Highborne Cay, The Bahamas are large

unlaminated carbonate structures that result from the interactions of a metabolically

active microbial mat community and its environment (Dupraz et al., 2009; Planavsky et

al., 2009; Mobberley et al., 2012; Mobberley et al. 2013). In this study, we examined the

underlying gene expression associated with the thrombolitic microbial mat at midday,

the peak time point for photosynthesis (Figure 4-1B-D; Visscher and Stolz, 2005).

Metatranscriptomic libraries were generated throughout the vertical profile of the

thrombolitic mats. Using microelectrode profiling of oxygen productivity, the mats were

sectioned into three discrete zones including an oxic, transitional and anoxic zone within

the thrombolitic mat (Figure 4-1E). Additionally, the metagenome of the top 9 mm of the

thrombolite mat was partially sequenced to assess the functional gene complexity of the

community and to serve as a reference map for the gene expression data. This study

represents the first metatranscriptomic analysis of a modern microbialite and has

increased our understanding of the underlying metabolic pathways within these

unlaminated thrombolitic mat ecosystems.

Overview of Thrombolitic Mat Sequencing Libraries

Metatranscriptomic libraries were generated for the oxic (Zone 1; 0 – 3 mm),

transitional (Zone 2; 3 – 5 mm), and anoxic (Zone 3; 5 – 9 mm) zones within the

thrombolitic mat (Figure 4-1D). In previous studies, attempts to remove rRNA gene

transcripts through selective hybridization or enzymatic reactions resulted in a wide

range of residual rRNA in the sequenced metatranscriptome libraries (Poretsky et al.,

2009; He et al., 2010; Stewart et al., 2010; Burow et al., 2013). In this study, the total

RNA extracts from each zone were either directly amplified (Tot. RNA) or underwent an

122

additional MicrobeExpress mRNA enrichment step (mRNA) performed prior to

amplification to assess biases of mRNA enrichment. For each of the

metatranscriptomes, between 32 and 39 million high quality reads were recovered, with

the majority of them classified as ribosomal rRNA (Table 4-1). Due to technical

difficulties in sequencing the mRNA-enriched library, only a total RNA library represents

the anoxic Zone 3. Across all metatranscriptomic samples, the majority of recovered

ribosomal rRNA reads were similar to large subunit rRNA (i.e., 23S and 28S) which has

been observed in a previous marine study (Stewart et al., 2010). The mRNA enrichment

procedure reduced rRNA levels by 15% and changed the relative taxon abundances of

recovered rRNA transcripts compared to the total RNA (Figure 4-2). Additionally, mRNA

enrichment did not increase the percent of protein-encoding sequences annotated

(Table 4-1), suggesting that this procedure did not enhance the sequencing effort. Due

to the lack of a matching mRNA enrichment sample for Zone 3, the rest of the

metatranscriptomic analysis will focus on the three metatranscriptomic libraries

generated with total RNA.

To provide a comparison to the metatranscriptomic libraries, the metagenome

was sequenced as previously described in Mobberley et al., (2013). In this study, the sequences were assembled into contiguous reads (contigs) using the procedures outlined in materials and methods (Table 4-1). Of the 103 million metagenomic reads only 1.6 million reads were assembled in contigs longer than 200 bp. This low recovery of assembled metagenomic reads was not surprising given that high diversity of organisms (> 800 ecotypes) found in the thrombolitic mats (Myshrall et al., 2010;

Mobberley et al., 2012; Edgcomb et al., 2013b; Mobberley et al., 2013). Although there

123

was a low percentage of assembly, 61% of the MetaGeneMark predicted protein-

encoded genes in the metagenome were annotated, with similar taxonomic richness to

that recovered from the unassembled metagenome (Figure 4-2B; Mobberley et al.,

2013). This higher annotation rate was likely the result from longer protein-encoding contigs (> 200 bp) versus read lengths of 100 bp in the metatranscriptomes (Thomas et al., 2012).

Active Members of the Thrombolitic Mat Community During Peak Photosynthetic Activity

Taxonomic diversity in the thrombolitic mat community has been previously

assessed through 16S and 18S rRNA gene surveys (Myshrall et al., 2010; Mobberley et

al. 2012), 18S rRNA transcript amplicon analysis (Edgcomb et al., 2013b), and

unassembled metagenome (Mobberley et al., 2013). In this study, community gene

expression profiling was used to capture a more comprehensive picture of the active

population within the thrombolitic mats providing the first spatial distribution of

taxonomic diversity within a lithifying community.

Taxonomic assignment of rRNA and mRNA transcripts indicated that the

Cyanobacteria and Proteobacteria were the most active phyla in the thrombolitic mats of

Highborne Cay (Figure 4-2; 4-3A) supporting prior 16S rRNA gene diversity

(Baumgartner et al., 2009; Mobberley et al., 2012) and biogeochemical analyses

(Planavsky et al., 2009; Myshrall et al., 2010). Although the percent of total bacterial

transcripts within the three zones remained constant, the overall abundance of

Cyanobacteria decreased with depth, whereas the abundance of transcripts from non-

cyanobacterial phyla such as Proteobacteria, Firmicutes, Bacteroidetes, and

Planctomycetes increased within depth (Figure 4-2). When considering the distribution

124

of rRNA transcripts from the 20 most abundant bacteria, the cyanobacterial orders

Chroococcales, heterocystous Nostocales, and filamentous Oscillatoriales were

enriched throughout the thrombolitic mat (Figure 4-2A; 4-2A). Due to paucity of

representative cyanobacterial genomes in the databases a higher resolution of the

taxonomic identity in the cyanobacteria was lacking.

Although eukaryotes are part of the thrombolitic mat community (Myshrall et al.,

2010; Bernhard et al., 2013; Edgcomb et al., 2013b), our analysis of community gene

expression showed that they are not as transcriptionally active within the thrombolites

compared to the bacteria at midday (Figure 4-2; 4-3B). The range of abundant active

eukaryotic transcripts recovered included a variety of marine worms and arthropods

(Metazoans), algae (Rhodophytes, Viridiplantae), protists (Alveolata, Ciliates), diatoms

(stramenopiles), Fungi, and amoebas and foraminifera (Rhizaria) (Figure 4-2). Within

these dominant eukaryotic groups, rRNA transcript abundances of the photosynthetic

and heterotrophic organisms were enriched in the top 5 mm of the mat (Zones 1 and 2)

with protists, photosynthetic amoeba, and algae having similar distribution throughout

the mat profile (Figure 4-2A; 4-3B). This distribution likely reflects the aerobic nature of

these organisms due to the oxic environment. Similar eukaryotic richness was found in

a study of amplified 18S rRNA transcripts (Edgcomb et al., 2013b) and 18S rRNA genes

(Myshrall et al., 2013), however, the differences in relative abundances of taxa likely reflect biases in gene marker-based studies. For example, active foraminifera have been previously documented in the top 10 mm of thrombolitic mats (Bernhard et al.,

2013, Edgcomb et al., 2013b), however these organisms made up less than 1% of transcript reads within our depth profile. Conversely, metazoan transcripts from

125

nematodes, marine worms, and arthropods were abundant in our metatranscriptome libraries, which is consistent with infaunal organism counts (Tarhan et al., 2013) and

18S rRNA libraries (Myshrall et al., 2010) in the HBC thrombolitic mats. Together, these results suggest that eukaryotes are active members of the thrombolitic mat community, however future studies need to address the extent of their influence in mat metabolism and carbonate precipitation patterns. For example, in the adjacent stromatolitic mats of

Highborne Cay, diatoms result in an increase sediment accretion through trapping and binding (Bowlin et al., 2012).

In addition to bacteria, archaea have been identified within the lithifying microbial mat communities, albeit at lower relative diversity and abundances (Burns et al., 2004,

Couradeau et al., 2011; Khodadad and Foster, 2012; Mobberley et al., 2012). In the thrombolite metatranscriptome, archaeal transcripts ranged from 0.35% in the oxic top 3 mm up to 1.3% in the bottom 5mm of the mat (Figure 4-2A; 4-3C). The high abundance of ammonia-oxidizing Thaumarchaea and halophilic Euryarchaeota was previously detected in a 16S rRNA gene survey (Mobberley et al., 2012). Interestingly, a number of rRNA transcripts similar to several methanogenic archaea were also recovered; these organisms are usually not detected in 16S rRNA gene censuses (Burns et al., 2004;

Goh et al., 2009; Mobberley et al., 2012). Protein-encoding reads recovered from the anoxic zone indicated the presence of metabolically active methanogenic

Euryarchaeota and sulfate-reducing Crenarchaeota within this lower anoxic portion of the mat (Figure 4-2B). Overall, the active bacteria, archaeal and eukaryotic populations in the thrombolitic mat community showed a discrete spatial zonation within the

126

thrombolitic mats that, although were different taxonomically, were spatially distributed

similar to the laminated stromatolitic mats.

Functional Annotation of Gene Transcript Abundance Revealed Spatial Gradients of Gene Expression within the Thrombolitic Mat

Previous studies have shown that thrombolitic mat consortia generate gradients of oxygen, dissolved inorganic carbon (Myshrall et al. 2010), hydrogen sulfide

(Edgcomb et al., 2013b), and carbon substrate utilization (Mobberley et al., 2013) that are linked to the diel cycle as has seen in other microbialitic systems (Dupraz et al.,

2009). In this study, metatranscriptomic profiling was used to determine the spatial distribution of functional gene expression at midday during the peak of photosynthesis

(Figure 4-4A). To preserve the taxonomic affiliation of functional genes, MEGAN was used to parse the RefSeq annotations such that around 50 to 60% of our annotated reads were assigned to a specific taxonomy and gene function (Figure 4-2B; 4-4B). The

SEED subsystem and KEGG metabolic pathways were used for in-depth analysis of thrombolitic mat gene function. A caveat of this approach, however, is that genes can be assigned to more than one subsystem or pathway. For example, the molecular chaperone GroEL is part of the Protein Metabolism and Virulence subsystems, leading to the number of reads assigned to pathways and subsystems not being a one to one ratio with the number of reads annotated.

Within the thrombolitic mat metatranscriptomes most of the recovered, annotated transcripts were bacterial (61.72-68.28%) in origin where as the remaining were either eukaryotic (30.32-33.44%), archaeal (0.35-1.01%), or viral (0.38-1.5%) (Figure 4-4B).

Most eukaryotic transcripts were either hypothetical proteins or could not be assigned to a subsystem, and the remaining eukaryotic functional genes reads comprised 1-2% of

127

annotated SEED reads. These transcripts were primarily respiration and protein synthesis genes, suggesting that heterotrophic eukaryotes contribute to mat metabolic cycling through consumption of bacterial biomass (Figure 4-2B; 4-5). Based on the protein-encoding genes, archaea made up an increasing percent of the metatranscriptomic libraries with depth indicating they may represent an active portion of the deeper mat community (Figure 4-2B; 4-4B). Compared to the eukaryotic transcripts, the archaeal mRNA transcripts had greater metabolic diversity suggesting that archaea may be involved in a wider range metabolic activities within the thrombolitic mat (Figure 4-6). Interestingly, viral proteins transcripts were recovered throughout the mat with the highest concentration in the top 3 mm (Zone 1). Most of these read were similar to structural proteins from cyanophage (Short and Suttle, 2005).

With regard to the bacterial transcripts, large portions of the assembled metagenomic and metatranscriptomic thrombolitic libraries included cellular housekeeping genes involved in protein, RNA, and amino acid metabolisms. These transcripts were enriched in the bottom 5 mm of the mat (43%) relative to the top zones

(22-28%)(Figure 4-4A) with the taxonomic distribution mirroring that for all functional reads (Figure 4-2B). These subsystems also account for most of the differences in seen between the thrombolitic metagenomic contigs and upper mat zones (Figure 4-

7A). The results suggest at midday there is a high level of biomass turnover through protein biosynthesis and degradation, transcriptional regulation, and amino acid cycling.

Increased synthesis and cell cycling with depth in the mat may reflect a potentially increased activity level relative to the oxic and transitional zones.

128

The high number of cyanobacterial photosynthetic gene transcripts (Figure 4-4A;

4-8) reinforces previous findings that oxygenic photosynthesis drives thrombolitic mat

metabolism (Planavsky et al., 2009; Myshrall et al., 2010; Mobberley et al. 2013).

Although the recovered photosynthesis transcripts only made up 2% of the

metagenomic contigs, they represented greater than 34% of the transcripts in the top

two zones of the mat and 8% of the reads in the bottom anoxic Zone 3 (Figure 4-4A).

There was a wide taxonomic distribution of photosystem genes and pigments

transcripts with Nostoc spp., Synechococcus spp., and Trichodesmium spp. being

highly active throughout the mat, and others such as the Anabaena,

Thermosynechococcus, and deeply rooted Gloeobacter having a more limited activity at

depth (Figure 4-8). Very few bacterial reaction center transcripts for anoxygenic

photosynthesis from purple non-sulfur bacteria were recovered suggesting the taxa are

not active phototrophs at midday. Cyanophage-derived psbD gene transcripts were also

recovered although they represented a very small component of the transcriptomic

library (Lindell et al., 2007). Photosynthetic eukaryotes were recovered at low levels,

with diatoms and algae transcripts recovered from the Zone 1 library derived from the

top 3 mm of the thrombolitic mat (Figure 4-8).

Typically in microbial mat systems, cyanobacteria are the primary producers

through their coupling of photosynthesis and carbon fixation (Paerl et al., 2001; Jahnke et al., 2004;Visscher and Stolz, 2005; Planavsky et al., 2009). The 1,5-bisphosphate carboxylase oxygenase (RuBisCO) gene transcripts recovered from the thrombolite spatial metatranscriptome profiles indicate that during midday, at peak photosynthesis, a range of organisms was actively fixing inorganic carbon through the Calvin-Benson-

129

Bassham cycle (Figure 4-8). All active cyanobacteria in the thrombolitic mat with the

exception of unclassified Chroococcales used photosynthetic energy to fix inorganic

carbon. Interestingly, at midday several other bacterial taxa were actively transcribing

RuBisCO with Alphaproteobacteria (e.g. Rhizobiales) producing the highest number of

transcripts in oxic Zone 1 (Figure 4-8; 4-9). The Calvin-Benson-Bassham cycle is a

functional carbon fixation pathway in several bacterial anoxygenic phototrophs and

chemolithoautotrophs (Berg, 2011), which have previously been identified as prominent

members of the thrombolitic mat community (Mobberley et al., 2012; Mobberley et al.,

2013). Additionally, RuBisCO transcripts similar to those from Methanosarcinaceae

were recovered at a low abundance suggesting archaea may also be actively fixing

carbon at midday (Figure 4-8; 4-9). Since only 10 methanogenesis related genes were

recovered in all five metatranscriptomic libraries sample (data not shown) it is likely that

methanogenesis is not a significant metabolism within the top 9 mm of the thrombolitic

mat at midday.

In addition to photosynthesis and carbon fixation another key metabolism in the

lithifying microbial mats is nitrogen fixation (Steppe et al., 2001; Falcon et al., 2007;

Beltran et al., 2012). Bacterial and archaeal nitrogenase (nif) transcripts were well represented throughout the three thrombolitic mat zones comprising nearly 5% of the annotated reads in Zone 1 and 3% in Zones 2 and 3 (Figure 4-8). Heterocyst-forming cyanobacteria capable of nitrogen fixation in the presence of oxygen were abundant in

Zone 1 and included several Nostocales. Additionally, non-heterocystous nitrogen fixing cyanobacteria, such as Cyanothece spp., Synechococcus spp., and Trichodesmium spp. were identified within the thrombolitic mat at midday. However, the highest number

130

of nif transcripts within the thrombolitic mats was derived from Proteobacteria,

particularly those similar to known nitrogen fixing Rhizobiales and other purple non-

sulfur bacteria (Figure 4-8; 4-10). This diversity of bacterial nif transcripts detected in the metatranscriptomes was higher than detected in previous nif gene surveys of microbialites (Steppe et al., 2001; Falcon et al., 2007; Beltran et al., 2012). Additionally, we recovered between 3-5% of the nif transcripts from methanogenic Euryarchaeota

(Dos Santos et al., 2012; Offre et al., 2013), which is the first documentation of archaeal nitrogen fixation in a lithifying microbial mat. The presence of a robust community of nitrogen fixing organisms during peak photosynthetic activity was not observed in the adjacent stromatolites (Steppe et al., 2001), but has been documented in lithifying systems dominated by heterocystous ecotypes (Falcon et al., 2007; Beltran et al.,

2012). Different populations of nitrogen fixers may drive differences in temporal and spatial gene expression that may potentially impact carbonate precipitation.

Respiratory processes in microbialitic systems consume a significant portion of photosynthesis-derived biomass and oxygen and can play a role in carbonate dissolution (Paerl et al. 2001; Visscher and Stolz, 2005; Baumgartner et al., 2009;

Dupraz et al., 2009; Planavsky et al., 2009; Myshrall et al., 2010; Gallagher et al.,

2012). Genes involved in oxidative phosphorylation, including dehydrogenases, cytochrome oxidases, and ATPases were used as proxies for aerobic respiration. Only a few transcripts from the anaerobic respiration processes sulfate reduction, denitrification, and methanogenesis were recovered in our midday metatranscriptomic libraries (data not shown), which likely results from these metabolisms occurring at levels below our detection ability in this study. The profiles of oxidative phosphorylation

131

gene transcripts indicated a diverse aerobic community at midday that decreased in

relative abundance with depth (Figure 4-8; 4-11). Bacteria comprised between 80-85% of the total respiration transcripts with Actinobacteria, Cyanobacteria, Firmicutes,

Alphaproteobacteria, and Gammaproteobacteria having the highest activities (Figure 4-

8).

Additionally, carbohydrate metabolisms were examined to establish differences in sugar uptake and usage between the three distinct thrombolitic mat zones (Figure 4-

12). The total number transcripts involved in sugar processing decreased with depth in the thrombolitic mat but there were spatial differences in relative abundances of sugar metabolism subsystems (Figure 4-12A). Saccharide processing (i.e. mono-, di-, poly-) transcripts from Cyanobacteria, Alpha- and Gammaproteobacteria, and Chloroflexi were dominant in all three zones while transcripts for genes involved organic acid and sugar alcohol metabolism from non-cyanobacterial phyla were enriched in the anoxic Zone 3

(Figure 4-12A,B). The high number of oxidative phosphorylation and sugar metabolism transcripts from Alphaproteobacteria combined with few photosystem gene transcripts suggests that these organisms are undergoing organoheterotrophy at midday during peak photosynthetic activity. This analysis indicated that specific communities of respiring phototrophs and heterotrophs were distributed throughout the thrombolitic mat community. These findings are consistent with substrate utilization experiments in thrombolitic and stromatolitic mats that indicate spatial differences in heterotrophic metabolisms (Decho et al., 2005; Khodadad and Foster, 2012; Mobberley et al. 2013).

The impact of this population on thrombolitic carbonate mineralization, however, needs to be explored further.

132

In addition to functional annotation of the assembled metagenome and metatranscriptomes, the key metabolisms and subsystems that may drive differences between the three zones of the thrombolitic mat community was examined (Figure 4-

7A). To predict patterns of SEED subsystem distributions, a principle coordinate of analysis using the MG-RAST annotation pipeline was used (Figure 4-7B). The results indicated that the subsystems Photosynthesis, RNA metabolism, and Protein metabolism appeared to drive the differences between the metatranscriptomic libraries.

Thrombolite metagenomic contigs and unassembled metagenomic reads did not cluster together tightly indicating differences in the subsystems abundances. For example, Cell division and cell cycle, and Phosphorus, Potassium, and Sulfur metabolism subsystems were a higher portion of the assembled metagenome. These differences may have resulted from the assembly procedure capturing the dominant metabolisms from abundant organisms within the thrombolitic mat, whereas the unassembled metagenomic reads represented the metabolic potential of the whole community.

Interestingly, the unassembled metagenomes from the unlithified Guerrero Negro mats systems and geographically distinct lithifying mats had more similar distribution of subsystem abundance than to either the thrombolite metagenomes or metatranscriptome (Figure 4-7B). This difference can be explained by the similar abundance of gene assigned to Cofactors, vitamins, prosthetic groups, and pigment subsystem (CVpgP) suggesting that metabolic sensing and pigment production are shared by microbial mat metagenomes. Alternatively, this difference could be due to the low metagenomic coverage of the microbial mat metagenomes used in this comparison as they were sequenced from Sanger shotgun libraries or 454 pyrosequencing, which

133

have significantly smaller sequence output than Illumina. These results also reinforce

the need for additional metatranscriptomic studies in other microbialitic systems to

determine differential gene expression over spatial and temporal scales, and how they

ultimately lead to the deposition of carbonate within the thrombolitic mat communities.

Conclusions

Our results revealed distinct spatial zones of metabolic activity within the clotted, unlaminated thrombolitic mats of Highborne Cay, The Bahamas. The dominance of photosystem, RuBisCO, and nitrogen fixation transcripts from coccoid Chroococcales, heterocystous Nostocales, and filamentous Oscillatoriales within the top 5 mm of the thrombolitic mats supports previous biogeochemical (Myshrall et al., 2010; Planavsky et al., 2009) and molecular work (Mobberley et al., 2012; Mobberley et al., 2013) that photosynthesis is a major metabolism in these communities. The detection of carbon fixation transcripts from photosynthetic eukaryotes and archaea indicated that primary producers other than cyanobacteria play an important role in the thrombolitic metabolism, however, their role in influencing carbonate precipitation need to be further explored. Active bacterial heterotrophs including Actinobacteria, Bacteroidetes,

Firmicutes, Alphaproteobacteria, and Gammaproteobacteria were enriched deeper in the thrombolitic mat. Many of these deep mat bacterial transcripts were related to protein and nucleic acid metabolism suggesting a possible higher cellular turnover at in the more anoxic zones of the mat. Many of these bacteria, along with heterotrophic metazoans and protists were most active in the top 3mm of the mat suggesting these organisms are consuming newly created photosynthetic biomass. The role of the eukaryotic heterotrophs in influencing and shaping carbonate precipitation within the thrombolitic mats remains to be investigated. Interestingly, our analysis showed that

134

little bacterial sulfate reduction, which is known to promote carbonate precipitation through remineralization in microbialitic systems (Visscher et al., 1998, Nitti et al., 2012,

Gallagher et al., 2013), was detected at very low levels at midday although known sulfate reducing bacteria and archaea were active members of the thrombolitic mat community. Combined with the previous thrombolite metagenome analysis (Mobberley et al., 2013) these results support the need for additional assessment of sulfate- reduction genes throughout the diel cycle.

This study represents the first functional metatranscriptomic analysis of a lithifying microbial mat ecosystem. Spatial gene expression profile revealed that discrete zones of microbial activities occur within the unlaminated, clotted thrombolitic mat. This metatranscriptome analysis highlighted important metabolic pathways occurring at midday during peak photosynthetic activity. This work provides the foundation for future comparative gene expression studies that will address differences in microbial metabolism throughout the diel cycle and how they may influence carbonate precipitation in thrombolitic mats.

135

Table 4-1. Summary of analysis and annotation of the metagenomic and metatranscriptomic thrombolitic mat libraries. No. rRNA Non rRNA Annotated depth protein- samples Sequence gene gene proteinsg profile encoding readsc readsd reads (% of (mm) genesf (%)e (%) predicted) (%) 127 90343 DNA contigsa 0 - 9.0 1590080 221365 134662 (60.8) (<1.0) (>99.0) 24424465 8590236 6493754 1001850 Zone 1 Tot. RNA 0 - 3.0 33014701 (73.8) (26.2) (19.6) (15.4) 20333226 11941815 9087212 1379396 Zone 1 mRNA 0 - 3.0 32275041 (63.9) (36.1) (28.6) (15.2) 24020842 8221201 5941954 Zone 2 Tot. RNAb 3.0 - 5.0 32242043 637344 (10.7) (77.0) (23.0) (19.0) b 23842465 14825235 10656793 1099843 Zone 2 mRNA 3.0 - 5.0 38667700 (61.6) (38.4) (27.5) (10.3) 27958917 10476762 8049230 1108969 Zone 3 Tot. RNAb 5.0 - 9.0 38435679 (73.0) (27.0) (21.0) (13.8) aMetagenomic contigs assembled from pair-end Illumina sequences. bLibraries that were pair-end Illumina sequenced with only read 1 used for comparative sequence analysis. cNumber of Illumina reads retained after quality filtering with Prinseq. dNumber of ribosomal reads identified by blastn (e-value < 10-5) against the SILVA SSU/LSU V111 database. ePercent of recovered reads based on those retained after quality filtering. f Number of protein encoding genes predicted by FragGeneScan with 1.0% error for metatranscriptomes or MetaGeneMark for metagenomic contigs. gProteins annotated by BLAT to the NCBI nr-database (e-value <10-3).

136 A B

C D

Figure 4-1. The thrombolites of Highborne Cay, The Bahamas. A) Thrombolite platforms forming in the intertidal zone. Bar = 50 cm. B) Cross-section of the clotted carbonate macrostructure of a thrombolite. The red box indicates the area of the surface mat that was sampled. Bar = 2 cm. C) Surface view of the thrombolitic mat. D). Cross-section of the thrombolitic mat delineating the three zones examined in this study. Zone 1 consisted of the top 3 mm of the mat. Zone 2 was between 3 and 5 mm below the mat surface. Zone 3 was the bottom 5-9 mm of the mat. Bar = 2 mm. E) Vertical oxygen productivity profile taken at 12:30 pm.

137

0-9 mm MG-DNA

0-3 mm Tot. RNA

0-3 mm mRNA

3-5 mm Tot. RNA Mat Samples

3-5 mm mRNA

5-9 mm Tot. RNA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of rRNA genes B

0-9 mm MG-DNA

0-3 mm Tot. RNA

0-3 mm mRNA

3-5 mm Tot. RNA Mat Samples

3-5 mm mRNA

5-9 mm Tot. RNA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of mRNA genes

Actinobacteria Firmicutes Alveolata Crenarchaeota Acidobacteria Planctomycetes Cryptophyta Euryarchaeota Bacteroidetes Alphaproteobacteria Euglenozoa Thaumarchaeota Chloroflexi Betaproteobacteria Glaucocystophyceae other Archaea Chroococcales Deltaproteobacteria Haptophyceae Nostocales Epsilonproteobacteria Fungi Oscilliatoriales Gammaproteobacteria Metazoa Pleurocapsales Zetaproteobacteria Parabasalia Prochlorales Verrucomicrobia Rhizaria Stigonematales other Bacteria Rhodophyta unclass. cyanobacteria stramenopiles Chlorophyta Streptophyta other Eukarya

Figure 4-2. Relative abundance of dominant taxa within the thrombolitic mat metagenome and metatranscriptomes. Dominant taxa were considered in each sample as those phyla that comprised at least 1% of the total number of annotated reads in each Domain. Those phyla that made up less than 1% were grouped into “other”. A) rRNA fraction classified based on similarity to SILVA version 111 LSU and SSU libraries. Due to their high abundances across the samples, cyanobacteria was classified at the order level and proteobacteria was classified at the class level. B) Protein-encoding fraction assigned by MEGAN 5 based RefSeq taxonomy.

138

A B Bacteria Taxon Accession 0-3 mm 3-5 mm 5-9 mm Eukaryotic Taxon Accession 0-3 mm 3-5 mm 5-9 mm

Acaryochloris CP000828 Alveolata Sterkiella FJ545743 Chroococcales CP001842 Kryptoperidinium GU591328 Cyanothece CP001291 Fungi Blastobotrys Z50840 Cyanothece AAXW01000002 Glaucocystophceae Cyanophora U30821 Cyanothece CP000806 Platynereis EF117897

Subsection I Chroococcales Cyanothece CP001701 Mesobuthus FJ948787

Cyanothece * CP002198 Metazoa Herdmania X53538 Synecheococcus CP000951 Xiphinema AY580056 Raphidiopsis ACYB01000022 Rhizaria Paulinella CP000815 Nostoc * CP001037 Porphyra U38804

Trichormus CP002059 Gracilaria AY673996

Nostocales

Subsection IV Aphanizomenon JF768743 Grateloupia * HM767138 Rhodophyta Aphanizomenon JF768745 Palmaria Z18289 Subsection III Spirulina AM709631 Phaeodactylum ABQD01000052 Oscilliatoriales Arthrospira JN831265 Phaeodactylum EF553458

Prochloron AGRF01000040 Synedra JQ088178 stramenopiles Prochlorothrix AM709625 Odontella Z67753

Subsection I Prochlorales Prochlorococcus FW306008 Oltmannsiellopsis * HE610120 Subsection V Stigonematales Fischerella AM709634 Funaria X80212 Nicotiana Z00044 Į3URWHREDFWHULD Rhodospirillaceae GU474870 Viridiplante C Archaea Taxon Accession 0-3 mm 3-5 mm 5-9 mm

Crenarchaea Pyrobaculum CP003316 Methanomicrobiales AY714848 Halobacteriales DQ078753 Methanosarcina AE010299 Methanoculleus CP000562

Euryarchaea Methanocaldococcus CP002009 Methanothermococcus CP002792 Thaumarchaeaota HE574568 Thaumarchaeota HE574571 Marine Group I JQ227158 Marine Group I JQ228178 Marine Group I EU283423 Marine Group I U71114

Thaumarchaea Marine Group I U71117 Marine Group I GU137354 Marine Group I GU137382 Marine Group I JN592014 Nitrosopumilus DQ085097 Nitrosopumilus CP000866 Nitrosopumilus JQ346765

Figure 4-3. Relative abundance of the 20 most abundant taxa per phylogenetic domain based on expressed ribosomal RNA transcripts from the Total RNA extracts. A) Bacteria, B) Eukarya, C) Archaea. The SILVA taxonomic scheme was used for Bacteria and Archaea. The EMBL taxonomic scheme was used for Eukaryotes. Each taxa was classified at the deepest taxonomic level down to genus. The size of the circle reflects the number of rRNA transcripts recruited to each rRNA gene from each domain. The following are the ranges for each domain from smallest number of reads represented to the largest number of reads: Bacteria (84,534-641,515); Eukaryotes (86-144,214); Archaea (10- 3,788). Asterisks (*) indicate the presence of a particular taxon recovered from rRNA reads in the assembled metagenome.

139

A Photosynthesis 1 Protein Metabolism 2 Carbohydrates Respiration 3 Virulence 4 RNA Metabolism 5 Stress Response Amino Acids and Derivatives 6 Cofactors, Vitamins, Prosthetic Groups, Pigments Nitrogen Metabolism Clustering-based subsystems Cell Division and Cell Cycle DNA Metabolism Fatty Acids, Lipids, and Isoprenoids Cell Wall and Capsule Phosphorus Metabolism Nucleosides and Nucleotides Phages, Prophages, Transposable elements Motility and Chemotaxis Regulation and Cell signaling Sulfur Metabolism Membrane Transport Miscellaneous Secondary Metabolism Metabolism of Aromatic Compounds Potassium metabolism B Dormancy and Sporulation 1 2 3 4 5 6 0 - 9 mm 0 - 3 mm 0 - 3 mm 3 - 5 mm 3 - 5 mm 5 - 9 mm MG-DN A Tot. RNA mRNA Tot. RNA mRNA Tot. RNA Archaea 0.83% 0.35% 0.60% 0.53% 0.42% 1.01% Bacteria 92.14% 64.70% 70.38% 61.72% 54.03% 68.28% Eukarya 6.77% 33.44% 26.07% 36.60% 44.61% 30.32% Virus 0.14% 1.50% 2.93% 1.10% 0.93% 0.38%

Figure 4-4. Functional gene expression within a spatial profile of a thrombolitic mat. A) Relative abundance of SEED subsystems (Level 1) based on RefSeq annotations of protein-encoding reads. The subsystems in the legend are ordered based on overall abundance within the assembled metagenome (1) and the metatranscriptomes (2-6). B) Differences in community composition based on taxonomic assignment of all RefSeq annotations for each sequenced library. Domain-level differences are given as percent of those RefSeq annotations that were assigned to a specific taxonomy.

140

Amino Acids and Derivatives Carbohydrates 0-9 mm MG-DNA Cell Wall and Capsule Clustering-based subsystems Cofactors, Vitamins, Prosthetic Groups, Pigments DN A Metabolism Fatty Acids, Lipids, and Isoprenoids Membrane Transport 0-3 mm Tot.RNA Metabolism of Aromatic Compounds Miscellaneous Nitrogen Metabolism Nucleosides and Nucleotides Phosphorus Metabolism

Mat Samples Photosynthesis Protein Metabolism 3-5 mm Tot. RNA Respiration RNA Metabolism Secondary Metabolism Stress Response Virulence

5-9 mm Tot.RNA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of Eukaryotic SEED subsystems reads

Figure 4-5. Relative abundance of eukaryotic SEED subsystem reads in the assembled metagenome and metatranscriptomes. Number of reads classified as eukarya: 0-9 mm MG-DNA (89); 0-3 mm Tot. RNA (1190); 3-5 mm Tot. RNA (661); 5-9 mm Tot. RNA (451).

141

Amino Acids and Derivatives Carbohydrates 0-9 mm MG-DNA Cell Division and Cell Cycle Cell Wall and Capsule Clustering-based subsystems Cofactors, Vitamins, Prosthetic Groups, Pigments DNA Metabolism Fatty Acids, Lipids, and Isoprenoids

0-3 mm Tot.RNA Membrane Transport Metabolism of Aromatic Compounds Miscellaneous Nitrogen Metabolism

Mat Samples Nucleosides and Nucleotides Phage 3-5 mm Tot. RNA Phosphorus Metabolism Potassium metabolism Protein Metabolism Respiration RNA Metabolism Secondary Metabolism 5-9 mm Tot.RNA Stress Response Sulfur Metabolism Virulence

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of Archaeal SEED subsystem reads

Figure 4-6. Relative abundance of archaeal SEED subsystem reads in the assembled metagenome and metatranscriptomes. Number of reads classified as archaea: 0-9 mm MG-DNA (82); 0-3 mm Tot. RNA (732); 3-5 mm Tot. RNA (366); 5-9 mm Tot. RNA (806).

142

Protein Metabolism RNA Metabolism 0 Ɣ Ɣ Photosynthesis Virulence í DNA Metabolism LQFLSDO&RPSRQHQW

P r íPP0*FRQWLJV Ɣ

í íPPTRW51$ Ɣ íPPP51$ íPPTRW51$ íPPP51$

í íPPTRW51$ í í í 0 PrLQFLSDO&RPSRQHQW B ƔƔ

Ɣ ƔƔ Photosynthesis

&93J3 0

Amino Acid Metabolism

í Protein Metabolism LQFLSDO&RPSRQHQW RNA Metabolism P r íPP0*FRQWLJV íPP0* Ɣ íPPTRW51$ *XHUUHUR1HJUR Ɣ HBC T\SHVWURPDWROLWH í íPPP51$ íPPTRW51$ HBC T\SHVWURPDWROLWH íPPP51$ Ɣ Pozas Azules thrombolite íPPTRW51$ Rio Mesquites stromatolite í í 0 PrLQFLSDO&RPSRQHQW

Figure 4-7. Principle component analysis of functional gene abundances of thrombolitic mat metatranscriptomes based on SEED subsystems (Level 1) of thrombolitic metatranscriptomes and assembled metagenome. The red biplot lines represent the top 5 subsystems explaining the differences between the samples included in the analysis. (A) Comparison of thrombolitic mat metatranscriptomes and assembled metagenome from this study based on MEGAN 5 annotation (e-value 10-3). The first two principal components account for 99.5% of the variation between samples. (B) Comparison to metagenomes of previously sequenced lithifying (HBC stromatolites, Cuatros Ciénegas microbialites) and non-lithifying (Guerrero Negro) microbial mat communities based on MG-RAST annotation (e-value 10-5). CVPgP represents “Cofactors,Vitamins, Prosthetic Groups, Pigments” subsystem. The first two principal components account for 81.5% of the variation between samples.

143

Photosystems Oxidative Pigments Rubisco nif Phosphorylation

MG 1 2 3 MG 1 2 3 MG 1 2 3 MG 1 2 3 Crenarchaeota Archaea Euryarchaeota Actinobacteria Aquificae 8 Bacteroidetes Chlorobi Chlamydiae Verrucomicrobia 6 Chloroflexi Gloeobacter Anabaena Nostoc 4

Trichormus Cyanobacteria Acaryochloris Cyanothece

Number of reads log(n+1) reads of Number Microcystis 2 Synechococcus Synechocystis Thermosynechococcus unclassified Chroococcales 0 Trichodesmium Prochlorococcus Deferribacteres 'HLQRFRFFXVí7KHrmus Acidobacteria Firmicutes Nitrospirae Planctomycetes Alphaproteobacteria Betaproteobacteria Deltaproteobacteria Epsilonproteobacteria Gammaproteobacteria Spirochaetes 7enericutes 7KHrmotogae Alveolata Amoebozoa Euglenozoa

Heterolobosea Eukarya Choanoflagellida Fungi Metazoa Stramenopiles Bacillariophyta Chlorophyta Streptophyta

Figure 4-8. Organisms involved in energy transformations within the thrombolitic mat during peak photosynthesis (12 PM). Functional annotation was carried out on the RefSeq annotations using the KEGG pathways database. Read abundance was normalized by log(n+1). Photosystem genes and antenna- pigments represent oxygenic photosynthesis. Ribulose-1,5-bisphosphate carboxylase oxygenase (RuBisCO) genes were used to examine carbon fixation. Nitrogenase (nif) genes were used to represent nitrogen fixation occurring within the thrombolitic mat. Genes involved in oxidative phosphorylation are used as a proxy for aerobic respiration.

144

Methanomicrobia Actinobacteria Chroococcales 0-9 mm MG-DNA Gloeobacteria Nostocales Oscillatoriales Prochlorales Deinococcus-Thermus Firmicutes

0-3 mm Tot. RNA Rhizobiales other Alphaproteobacteria Burkholderiales other Betaproteobacteria Chromatiales

Mat Samples other Gammaproteobacteria Verrucomicrobia 3-5 mm Tot. RNA Chlorophyta Streptophyta

5-9mm Tot. RNA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of Rubisco reads

Figure 4-9. Relative taxa abundance of RuBisCO reads based on KEGG pathway annotation in the assembled metagenome and metatranscriptomes. Archaea and Bacteria are shown at the phyla level except for Cyanobacteria (Order) and Proteobacteria (Class). For each class of Proteobacteria, the most abundant order is named. Eukaryotes are shown at the kingdom or phyla level. Number of reads annotated as RuBisCO genes: 0-9 mm MG-DNA (10); 0-3 mm Tot. RNA (2073); 3-5 mm Tot. RNA (875); 5-9 mm Tot. RNA (596).

145

Methanobacteria Methanococci Methanomicrobia 0-9 mm MG-DNA Actinobacteria Aquificae Chlorobi Chloroflexi Chroococcales Nostocales Oscillatoriales 0-3 mm Tot. RNA Deferribacteres Firmicutes Nitrospirae Rhizobiales other Alphaproteobacteria Mat Samples Burkholderiales 3-5 mm Tot. RNA other Betaproteobacteria Desulfuromonadales other Deltaproteobacteria Campylobacterales Chromatiales other Gammaproteobacteria Verrucomicrobia 5-9 mm Tot. RNA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of nif reads Figure 4-10. Relative taxa abundance of nif reads based on KEGG pathway annotation in the assembled metagenome and metatranscriptomes. Archaea and Bacteria are shown at the phyla level except for Cyanobacteria (Order) and Proteobacteria (Class). For each class of Proteobacteria, the most abundant order is named. Eukaryotes are shown at the kingdom or phyla level. Number of reads annotated as nif genes: 0-9 mm MG-DNA (13); 0-3 mm Tot. RNA (3573); 3-5 mm Tot. RNA (832); 5-9 mm Tot. RNA (1362).

146

Thermoprotei Halobacteria Methanomicrobia other Euryarchaeota 0-9 mm MG-DNA Acidobacteria Aquificae Bacteroidetes Chlorobi Chloroflexi Gloeobacteria Nostocales Chroococcales 0-3 mm Tot.RNA Oscillatoriales Prochlorales Planctomycetes Rhizobiales other Alphaproteobacteria Burkholderiales other Betaproteobacteria Mat Samples Myxococcales other Deltaproteobacteria 3-5 mm Tot. RNA Campylobacterales other Epsilonproteobacteria Alteromonadales other Gammaproteobacteria Spirochaetes Tenericutes Thermotogae Verrucomicrobia 5-9 mm Tot. RNA Alveolata Choanoflagellida Euglenozoa Fungi Metazoa 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Stramenopiles Chlorophyta Percent of oxidative phosphorylation reads Streptophyta

Figure 4-11. Relative taxa abundance of oxidative phosphorylation reads based on the KEGG pathway in the assembled metagenome and metatranscriptomes. Archaea and Bacteria are shown at the phyla level except for Cyanobacteria (Order) and Proteobacteria (Class). For each class of Proteobacteria, the most abundant order is named. Eukaryotes are shown at the kingdom or phyla level. Number of reads annotated as genes involved in oxidative phosphorylation: 0-9 mm MG-DNA (389); 0-3 mm Tot. RNA (9427); 3-5 mm Tot. RNA (3816); 5-9 mm Tot. RNA (5096).

147

Monosaccharides Di- & oligosaccharides 0-9 mm MG-DNA Polysaccharides Aminosugars Organic acids Sugar alcohols Glycoside hydrolases 0-3 mm Tot. RNA Mat Samples

3-5 mm Tot. RNA

5-9 mm Tot. RNA

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of sugar metabolism reads B Crenarchaeota Euryarchaeota Thaumarchaeota 0-9 mm MG-DNA Acidobacteria Actinobacteria Aquificae Bacteroidetes Chlorobi Chlamydiae Chloroflexi 0-3 mm Tot. RNA Gloeobacteria Nostocales Chroococcales Oscillatoriales Prochlorales

Mat Samples Deinococcus-Thermus 3-5 mm Tot. RNA Firmicutes Planctomycetes Alphaproteobacteria Betaproteobacteria Deltaproteobacteria Epsilonproteobacteria Gammaproteobacteria 5-9 mm Tot. RNA Spirochaetes Thermotogae Verrucomicrobia Metazoa 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of sugar metabolism reads

Figure 4-12. Relative abundance of SEED carbohydrate subsystem reads associated with sugar metabolism in the assembled metagenome and metatranscriptomes. (A) Breakdown of SEED Carbohydrate subsystem (Level 2). Number of reads annotated as genes involved in sugar metabolism: 0-9 mm MG-DNA (232); 0-3 mm Tot. RNA (879); 3-5 mm Tot. RNA (456); 5-9 mm Tot. RNA (678). (B) Taxa relative abundance of sugar metabolism reads. Bacteria are shown at the Phyla level except for Cyanobacteria (Order) and Proteobacteria (Class).

148

CHAPTER 5 SUMMARY OF RESEARCH

For over 3.5 billion years, microbialites have played a critical role in global geochemical cycling and sequestration of carbon (Awramik, 1971; Kasting and Howard,

2006; Dupraz et al., 2009). These complex microbial ecosystems have a propensity for the trapping, binding, and precipitation of calcium carbonate to form both laminated

(stromatolites) and unlaminated (thrombolites) organosedimentary structures (Kennard and James, 1986; Burne and More, 1987). Biomineralization in microbialites is a direct result of interactions between microbial mats and their surrounding environment and represent an unexplored area of microbial and functional gene diversity associated with carbonate precipitation. The goal of my dissertation was to understand the metabolic processes that contribute to the formation and function of the modern thrombolitic lithifying mats found in Highborne Cay, The Bahamas. The underlying genetic pathways in this model lithifying system were investigated through (1) identifying the microbial diversity within the marine thrombolitic mats; (2) delineating the functional gene potential of the thrombolitic mats; (3) examining spatial patterns of the thrombolitic mat metatranscriptome.

To delineate the bacterial, cyanobacterial, and archaeal lineages of the

Highborne Cay thrombolitic mats, a 16S rRNA gene survey was performed using the

454 pyrosequencing platform. The coverage estimates from the bacterial libraries

(>80%) and cyanobacterial and archaeal libraries (>95%) suggest that this approach captured the majority of the microbial diversity within all mat types. The bacterial libraries contained similar richness of phyla, dominated by Alphaproteobacteria and

Cyanobacteria, however the distribution of taxa varied between the types thrombolitic

149

mats which likely reflects differences in environmental variables between the four mat

types. This finding indicated that bacterial populations were more diverse than

previously determined by 16S rRNA gene clone libraries (Myshrall et al., 2010).

Based on the 16S rRNA gene analyses, the major taxonomic differences

between the thrombolitic communities were within Cyanobacteria and Archaea. The

button mats, which are the most abundant and productive thrombolitic mats, contained

the highest diversity of Cyanobacteria including an enrichment of Dichothrix spp.; a key

organism associated with carbonate precipitation in the thrombolitic mats (Planavsky et

al., 2009). This organism was unique to the HBC thrombolitic mats and was not found in

the adjacent HBC stromatolites. The overall archaeal abundance was low, but there

were differences in Thaumarchaeota and Euryarchaeota diversity between the

thrombolitic mats with button mats containing the most diverse archaea population.

Taken together these analyses indicate that the HBC thrombolitic mats contained

complex consortiums of microorganisms that were distinct from the HBC stromatolitic

mats. These differences as well as the differences between the thrombolitic mat types,

likely reflect of niche differentiation due to environmental parameters such as wave

action and sediment burial. Furthermore this analysis suggests that the intertidal four

thrombolitic mat types may represent successive stages of thrombolite development

controlled by burial. This study represented the first in-depth diversity analysis of

thrombolitic mat communities using high-throughput pyrosequencing and provided a foundation for understanding the interactions and metabolisms of these complex microbial communities. These findings were published in Environmental Microbiology in

January 2012 (Mobberley et al., 2012).

150

To assess the metabolic potential of the thrombolitic mats, the thrombolitic mat metagenome was coupled to community level phenotypic microarrays to establish the range of molecular and metabolic pathways found within these ecosystems. The thrombolitic mat metagenome was generated from whole mat (0-9 mm). Taxonomic analysis of annotated protein features revealed the majority of the reads were similar to

Bacteria (91.7%) although Archaea (0.8%) and Eukaryota (7.4%) were also detected.

The thrombolitic metagenome contained all of the major metabolic pathways that are hypothesized to influence carbonate mineralization. Cyanobacterial molecular pathways, mainly from nitrogen fixing Chroococcales and Nostocales, comprised 38.2% of the total reads, thus reinforcing the concept of cyanobacteria being dominant organisms in lithifying microbial mat systems. These cyanobacterial pathways included oxygenic photosynthesis, carbon fixation, nitrogen fixation, and EPS formation. The relative abundance of photosynthesis genes across all mats was well conserved in comparison to other microbial mat metagenomes, whereas the abundance of respiration gene reads was different in the thrombolitic mat metagenomes compared to the other lithifying ecosystems. Although a variety of aerobic heterotrophs including

Proteobacteria, Bacteroidetes, Planctomycetes were detected, very few anaerobic metabolisms, such as bacterial sulfate reduction and archaeal methanogenesis, were found in the thrombolite metagenome. Even though there were limited numbers of archaeal gene reads recovered, these organisms could potentially be contributing ammonia oxidation and carbon cycling within the thrombolitic mat. Eukaryotic phototrophs including algae, diatoms, and protists were shown to have a small contribution to photosynthesis while arthropods, worms, and nematodes are likely to be

151

consuming bacterially produced EPS and other products. The metagenomic analysis

supported previous biogeochemical and microbial diversity studies (Planavsky et al.,

2009; Myshrall et al., 2010; Mobberley et al. 2012) that indicated that the thrombolitic

mat microbial consortia were capable of a wide range of metabolic activities.

The metabolic capabilities of the thrombolitic mat were gauged through the use

of carbon, nitrogen, sulfur, and phosphorous substrate utilization microarrays to

generate spatial profiles of metabolite utilization. Living mat samples were sectioned

into three discrete horizontal zones based on oxygen productivity profiles: oxic (0-3

mm), transitional (3-5 mm), anoxic (5-9 mm). Results indicated that the upper 3 mm

utilized the greatest number of substrates compared to the 3-5 mm and 5-9 mm zones.

This overall pattern suggests that the top 5 mm of the thrombolitic mat contains a more

metabolically diverse and active community than the deeper, anoxic zone of the mat.

This study provides evidence of hot spots of metabolic activity occurring within the

depth profile of Highborne Cay thrombolites. It also suggests that despite the lack of

mineral layering, as seen in the stromatolites, within the thrombolite there are vertical

gradients of metabolic activity. Together with the metagenome, the metabolic profiling

provided the foundation for examining the active role microorganisms play in

coordinating metabolisms that lead to mineralization. These finding were published in

Photosynthesis Research in July 2013 (Mobberley et al., 2013).

To examine patterns of gene expression within the thrombolitic mats a

metatranscriptome profile of thrombolitic mats was generated from samples collected

midday at the peak of solar activity. Total community RNA was extracted from mats that

had been sectioned into three discrete horizontal layers based on oxygen productivity

152

profiles: oxic (0-3 mm), transitional (3-5 mm), anoxic (5-9 mm). As with the thrombolitic

mat metagenome sequencing, most of the transcripts from the thrombolitic mat zones

were from Bacteria (68.28 – 61.72%) however Eukaryota transcripts made up between

30.32 and 33.4% of the annotated reads indicating eukaryotic organisms have greater activity during midday than in suggested by the metabolic potential (8-10% of the metagenomic reads). The metatranscriptome libraries also showed that although

Archaea comprised a low percentage of the community, they were enriched at depth suggesting increased activity in the deeper layers of the thrombolitic mat. We also detected numerous cyanophage in the oxic layers, which is where their cyanobacterial hosts were most active.

Cyanobacterial photosynthetic activity was the highest in oxic surface layer and decreased with depth. This finding along with the recovery of RuBisCO transcripts from

coccoid Chroococcales, heterocystous Nostocales, and filamentous Oscillatoriales

within the top 5 mm of the thrombolitic mats supports previous biogeochemical

(Myshrall et al., 2009; Planavsky et al., 2009) and molecular work (Mobberley et al.,

2012; Mobberley et al., 2013) that photosynthesis is a major metabolism driving the biogeochemical cycling within these communities. The high numbers of nif transcripts from Cyanobacteria and purple non-sulfur bacteria (Alphaproteobacteria) throughout the mat profile suggests that day-time nitrogen fixation is occurring in the oxic zones. Active bacterial heterotrophs including Actinobacteria, Bacteroidetes, Firmicutes,

Alphaproteobacteria, and Gammaproteobacteria were enriched deeper in the thrombolitic mat while heterotrophic metazoans and protists were most active in the oxic, top 3 mm of the mat suggesting differential distribution of organisms that are

153

consuming photosynthetic biomass. Interestingly, at midday we detected very few bacterial transcripts associated with sulfate reduction, a metabolism that in known to promote carbonate precipitation, although known sulfate reducing bacteria and archaea were active members of the thrombolitic mat community suggesting an alternative metabolism at the peak solar levels. The community gene expression profiling captured by this approach provides a more comprehensive picture of metabolically active

Bacteria, Archaea, and Eukaryota populations within the thrombolitic mat compared to ribosomal rRNA gene surveys and metagenomic sequencing. This study represents first metatranscriptomic analysis of a lithifying microbial mat ecosystem.

In order to understanding the metabolisms that contribute to thrombolite formation and function it is important to understand the underlying genetics. The results of this dissertation have positively impacted the state of knowledge in lithifying microbial communities in three distinct areas. First, combining high-throughput pyrosequencing and morphological analyses has delineated high levels of microbial diversity within the thrombolitic mats than previously known. Second, metagenomics and community-level physiological assays delineated the functional metabolic potential of these lithifying ecosystems. Third, metatranscriptomics revealed distinctive spatial gradients of metabolisms associated with carbonate precipitation within the thrombolitic mats.

Together, this research represents the first in-depth genetic analysis of thrombolitic mat metabolisms and has increased our understanding of the active role microorganisms play within lithifying systems. This dissertation work provides the functional genetic foundation for future work that seeks to link genetic and molecular mechanisms of carbonate precipitation in lithifying mat communities.

154

LIST OF REFERENCES

Abed RM, Garcia-Pichel F, Hernández-Mariné M. (2002). Polyphasic characterization of benthic, moderately halophilic, moderately thermophilic cyanobacteria with very thin trichomes and the proposal of Halomicronema excentricum gen. nov., spp. nov. Arch Microbiol 177: 361-370.

Allen M, Goh F, Burns B, Neilan B. (2009). Bacterial, archaeal and eukaryotic diversity of smooth and pustular microbial mat communities in the hypersaline lagoon of Shark Bay. Geobiology 7: 82-96.

Allen MA, Goh F, Leuko S, Echigo A, Mizuki T, Usami R, et al. (2008). Haloferax elongans sp. nov. and Haloferax mucosum sp. nov., isolated from microbial mats from Hamelin Pool, Shark Bay, Australia. Int J Syst Evol Microbiol 58: 798-802.

Anderson SL, McIntosh L. (1991). Light-activated heterotrophic growth of the cyanobacterium Synechocystis spp. strain PCC 6803: a blue-light-requiring process. J Bacteriol 173: 2761-2767.

Andres MS, Reid RP. (2006). Growth morphologies of modern marine stromatolites: a case study from Highborne Cay, Bahamas. Sediment Geol 185: 319-328.

Andres MS, Sumner DY, Reid RP, Swart PK. (2006). Isotopic fingerprints of microbial respiration in aragonite from Bahamian stromatolites. Geology 34: 973-976.

Arp G, Helms G, Karlinska K, Schumann G, Reimer A, Reitner J, et al. (2012). Photosynthesis versus exopolymer degradation in the formation of microbialites on the atoll of Kiritimati, Republic of Kiribati, Central Pacific. Geomicrobiol J 29: 29-65.

Arp G, Reimer A, Reitner J. (2001). Photosynthesis-induced biofilm calcification and calcium concentrations in Phanerozoic oceans. Science 292: 1701-1704.

Awramik SM. (1971). Precambrian columnar stromatolite diversity: reflection of metazoan appearance. Science 174: 825-827.

Barns SM, Fundyga RE, Jeffries MW, Pace NR. (1994). Remarkable archaeal diversity detected in a Yellowstone National Park hot spring environment. Proc Natl Acad Sci USA 91: 1609-1613.

Baumgartner LK, Dupraz C, Buckley DH, Spear JR, Pace NR, Visscher PT. (2009). Microbial species richness and metabolic activities in hypersaline microbial mats: insight into biosignature formation through lithification. Astrobiology 9: 861-874.

Baumgartner LK, Reid RP, Dupraz C, Decho AW, Buckley DH, Spear JR, et al. (2006). Sulfate reducing bacteria in microbial mats: changing paradigms, new discoveries. Sediment Geol 185: 131-145.

155

Beltrán Y, Centeno CM, García-Oliva F, Legendre P, Falcón LI. (2012). N2 fixation rates and associated diversity (nifH) of micro-bialite and mat-forming consortia from different aquatic environments in Mexico. Aquat Microb Ecol 67: 15-24.

Bernhard JM, Edgcomb VP, Visscher PT, McIntyre-Wressnig A, Summons RE, Bouxsein ML, et al. (2013). Insights into foraminiferal influences on microfabrics of microbialites at Highborne Cay, Bahamas. Proc Natl Acad Sci USA 110: 9830- 9834.

Berrendero E, Perona E, Mateo P. (2008). Genetic and morphological characterization of Rivularia and Calothrix (Nostocales, Cyanobacteria) from running water. Int J Syst Evol Microbiol 58: 447-460.

Bhaya D, Grossman AR, Steunou AS, Khuri N, Cohan FM, Hamamura N, et al. (2007). Population level functional diversity in a microbial community revealed by comparative genomic and metagenomic analyses. ISME J 1: 703-713.

Billi D, Friedmann EI, Hofer KG, Caiola MG, Ocampo-Friedmann R. (2000). Ionizing- radiation resistance in the desiccation-tolerant cyanobacterium Chroococcidiopsis. Appl Environ Microbiol 66: 1489-1492.

Bosak T, Liang B, Wu T, Templer S, Evans A, Vali H, et al. (2012). Cyanobacterial diversity and activity in modern conical microbialites. Geobiology 10:384-401.

Bosak T, Greene S, Newman D. (2007). A likely role for anoxygenic photosynthetic microbes in the formation of ancient stromatolites. Geobiology 5: 119-126.

Bothe H, Schmitz O, Yates MG, Newton WE. (2010). Nitrogen fixation and hydrogen metabolism in cyanobacteria. Microbiology and molecular biology reviews 74: 529-551.

Bowlin EM, Klaus JS, Foster JS, Andres MS, Custals L, Reid RP. (2012). Environmental controls on microbial community cycling in modern marine stromatolites. Sediment Geol 263: 45-55.

Braissant O, Decho AW, Dupraz C, Glunk C, Przekop KM, Visscher PT. (2007). Exopolymeric substances of sulfate‐reducing bacteria: Interactions with calcium at alkaline pH and implication for formation of carbonate minerals. Geobiology 5: 401-411.

Breitbart M, Hoare A, Nitti A, Siefert J, Haynes M, Dinsdale E, et al. (2009). Metagenomic and stable isotopic analyses of modern freshwater microbialites in Cuatro Ciénegas, Mexico. Environ Microbiol 11: 16-34.

Brochier-Armanet C, Boussau B, Gribaldo S, Forterre P. (2008). Mesophilic Crenarchaeota: proposal for a third archaeal phylum, the Thaumarchaeota. Nature Reviews Microbiology 6: 245-252.

156

Burne RV, Moore LS. (1987). Microbialites: organosedimentary deposits of benthic microbial communities. Palaios 2: 241-254.

Burns BP, Gudhka RK, Neilan BA. (2012). Genome sequence of the halophilic archaeon Halococcus hamelinensis. J Bacteriol 194: 2100-2101.

Burns BP, Goh F, Allen M, Neilan BA. (2004). Microbial diversity of extant stromatolites in the hypersaline marine environment of Shark Bay, Australia. Environ Microbiol 6: 1096-1101.

Burow LC, Woebken D, Marshall IP, Lindquist EA, Bebout BM, Prufert-Bebout L, et al. (2012). Anoxic carbon flux in photosynthetic microbial mats as revealed by metatranscriptomics. ISME J 7: 817-829.

Cáceres CE, Tessier AJ. (2003). How long to rest: the ecology of optimal dormancy and environmental constraint. Ecology 84: 1189-1198.

Canfield DE, Des Marais DJ. (1993). Biogeochemical cycles of carbon, sulfur, and free oxygen in a microbial mat. Geochim Cosmochim Acta 57: 3971-3984.

Centeno CM, Legendre P, Beltrán Y, Alcántara‐Hernández RJ, Lidström UE, Ashby MN, et al. (2012). Microbialite genetic diversity and composition relate to environmental variables. FEMS Microbiol Ecol 82: 724-735.

Cole JJ. (1999). Aquatic microbiology for ecosystem scientists: new and recycled paradigms in ecological microbiology. Ecosystems 2: 215-225.

Couradeau E, Benzerara K, Moreira D, Gérard E, Kaźmierczak J, Tavera R, et al. (2011). Prokaryotic and eukaryotic community structure in field and cultured microbialites from the alkaline lake Alchichica (Mexico). PLoS One 6: e28767.

Cox MP, Peterson DA, Biggs PJ. (2010). SolexaQA: At-a-glance quality assessment of Illumina second-generation sequencing data. BMC Bioinformatics 11: e485.

Decho AW, Visscher PT, Reid RP. (2005). Production and cycling of natural microbial exopolymers (EPS) within a marine stromatolite. Palaeogeogr , Palaeoclimatol , Palaeoecol 219: 71-86.

Decho AW. (2000). Microbial biofilms in intertidal systems: an overview. Cont Shelf Res 20: 1257-1273.

Decho AW. (1990). Microbial exopolymer secretions in ocean environments: their role(s) in food webs and marine processes. Oceanogr Mar Biol Annu Rev 28: 73- 153.

DeLong EF. (1992). Archaea in coastal marine environments. Proc Natl Acad Sci U S A 89: 5685-5689.

157

DeSantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, et al. (2006). Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol 72: 5069-5072.

Desnues C, Rodriguez-Brito B, Rayhawk S, Kelley S, Tran T, Haynes M, et al. (2008). Biodiversity and biogeography of phages in modern stromatolites and thrombolites. Nature 452: 340-345.

Dos Santos PC, Fang Z, Mason SW, Setubal JC, Dixon R. (2012). Distribution of nitrogen fixation and nitrogenase-like sequences amongst microbial genomes. BMC Genomics 13: e162.

Dill RF, Shinn EA, Jones AT, Kelly K, Steinen RP. (1986). Giant subtidal stromatolites forming in normal salinity waters. 324: 55-58.

Dillon JG, Tatsumi CM, Tandingan PG, Castenholz RW. (2002). Effect of environmental factors on the synthesis of scytonemin, a UV-screening pigment, in a cyanobacterium (Chroococcidiopsis spp.). Arch Microbiol 177: 322-331.

Dinsdale EA, Edwards RA, Bailey B, Tuba I, Akhter S, McNair K, et al. (2013). Multivariate analysis of functional metagenomes. FGENE 4: e41.

Dravis JJ. (1983). Hardened subtidal stromatolites, Bahamas. Science 219: 385-386.

Dupraz C, Reid RP, Braissant O, Decho AW, Norman RS, Visscher PT. (2009). Processes of carbonate precipitation in modern microbial mats. Earth-Sci Rev 96: 141-162.

Dupraz C, Visscher PT. (2005). Microbial lithification in marine stromatolites and hypersaline mats. Trends Microbiol 13: 429-438.

Dupraz C, Visscher P, Baumgartner L, Reid R. (2004). Microbe–mineral interactions: early carbonate precipitation in a hypersaline lake (Eleuthera Island, Bahamas). Sedimentology 51: 745-765.

Edgcomb VP, Bernhard JM, Beaudoin D, Pruss S, Welander PV, Schubotz F, Mehay S, Gillespie AL, Summons RE. (2013a) Molecular indicators of microbial diversity in oolitic sands of Highborne Cay, Bahamas. Geobiol 11: 234-251.

Edgcomb VP, Bernhard JM, Summons RE, Orsi W, Beaudoin D, Visscher PT. (2013b). Active eukaryotes in microbialites from Highborne Cay, Bahamas, and Hamelin Pool (Shark Bay), Australia. ISME J , doi:doi:10.1038/ismej.2013.130.

Edwards RA, Rodriguez-Brito B, Wegley L, Haynes M, Breitbart M, Peterson DM, et al. (2006). Using pyrosequencing to shed light on deep mine microbial ecology. BMC Genomics 7: e57.

Ehrlich HL. (2002). Geomicrobiology. New York: Marcel Dekker.

158

Engelbrektson A, Kunin V, Wrighton KC, Zvenigorodsky N, Chen F, Ochman H, et al. (2010). Experimental factors affecting PCR-based estimates of microbial species richness and evenness. ISME J 4: 642-647.

Falcón L, Cerritos R, Eguiarte L, Souza V. (2007). Nitrogen fixation in microbial mat and stromatolite communities from Cuatro Ciénegas, Mexico. Microb Ecol 54: 363-373.

Faria JC, Demetrio CGB. (2008). bpca: Bioplot of multivariate data based on principal components analysis. R package version 1.02 edn.

Farias ME, Rascovan N, Toneatii DM, Albarracín VH, Flores MR, Poire DG, et al., (2012). The discovery of stromatolites developing at 3570 m above sea level in a high-altitude volcanic lake Socompa, Argentinean Andes. PLoS One 8: e53497.

Foster JS, Green SJ, Ahrendt SR, Golubic S, Reid RP, Hetherington KL, et al. (2009). Molecular and morphological characterization of cyanobacterial diversity in the stromatolites of Highborne Cay, Bahamas. ISME J 3: 573-587.

Foster JS, Green SJ. (2011). Microbial diversity in modern stromatolites. In: Stromatolites: Interaction of Microbes with Sediments. Springer, pp 383-405.

Frankel RB, Bazylinski DA. (2003). Biologically induced mineralization by bacteria. Reviews in Mineralogy and Geochemistry 54: 95-114.

Franks J, Reid RP, Aspden RJ, Underwood GJ, Paterson DM, Prufert-Bebout L, et al. (2010). Ooid accreting diatom communities from the modern marine stromatolites at Highborne Cay, Bahamas. In: Microbial Mats. Springer, pp 275-285.

Gallagher KL, Braissant O, Kading TJ, Dupraz C, Visscher PT. (2013). Phosphate- related artifacts in carbonate mineralization experiments. Journal of Sedimentary Research 83: 37-49.

Gautret P, Trichet J. (2005). Automicrites in modern cyanobacterial stromatolitic deposits of Rangiroa, Tuamotu Archipelago, French Polynesia: biochemical parameters underlaying their formation. Sediment Geol 178: 55-73.

Giani D, Giani L, Cohen Y, Krumbein WE. (1984). Methanogenesis in the hypersaline Solar Lake (Sinai). FEMS Microbiol Lett 25: 219-224.

Goh F, Allen MA, Leuko S, Kawaguchi T, Decho AW, Burns BP, et al. (2008). Determining the specific microbial populations and their spatial distribution within the stromatolite ecosystem of Shark Bay. ISME J 3: 383-396.

Gomez-Alvarez V, Teal TK, Schmidt TM. (2009). Systematic artifacts in metagenomes from complex microbial communities. ISME J 3: 1314-1317.

159

Good IJ. (1953). The population frequencies of species and the estimation of population parameters. Biometrika 40: 237-264.

Green SJ, Blackford C, Bucki P, Jahnke LL, Prufert-Bebout L. (2008). A salinity and sulfate manipulation of hypersaline microbial mats reveals stasis in the cyanobacterial community structure. ISME J 2: 457-470.

Grootjans A, Van den Ende F, Walsweer A. (1997). The role of microbial mats during primary succession in calcareous dune slacks: an experimental approach. Journal of Coastal Conservation 3: 95-102.

Grotzinger JP, Knoll AH. (1999). Stromatolites in Precambrian carbonates: evolutionary mileposts or environmental dipsticks? Annu Rev Earth Planet Sci 27: 313-358.

Hamady M, Lozupone C, Knight R. (2009). Fast UniFrac: facilitating high-throughput phylogenetic analyses of microbial communities including analysis of pyrosequencing and PhyloChip data. ISME J 4: 17-27.

Hamady M, Walker JJ, Harris JK, Gold NJ, Knight R. (2008). Error-correcting barcoded primers for pyrosequencing hundreds of samples in multiplex. Nature Methods 5: 235-237.

Harris JK, Sahl JW, Castoe TA, Wagner BD, Pollock DD, Spear JR. (2010). Comparison of normalization methods for construction of large, multiplex amplicon pools for next-generation sequencing. Appl Environ Microbiol 76: 3863- 3868.

Haruta S, Kondo M, Nakamura K, Chanchitpricha C, Aiba H, Ishii M, et al. (2004). Succession of a microbial community during stable operation of a semi- continuous garbage-decomposing system. Journal of bioscience and bioengineering 98: 20-27.

Havens KE, Phlips EJ, Cichra MF, Li B. (1998). Light availability as a possible regulator of cyanobacteria species composition in a shallow subtropical lake. Freshwat Biol 39: 547-556.

He S, Wurtzel O, Singh K, Froula JL,Yilmaz S,Tringe SG, et al. (2010). Validation of two ribosomal RNA removal methods for microbial metatranscriptomes. Nature Methods 7: 807-812.

Hongmei J, Aitchison JC, Lacap DC, Peerapornpisal Y, Sompong U, Pointing SB. (2005). Community phylogenetic analysis of moderately thermophilic cyanobacterial mats from China, the Philippines and Thailand. Extremophiles 9: 325-332.

160

Hosoya S, Arunpairojana V, Suwannachart C, Kanjana-Opas A, Yokota A. (2006). Aureispira marina gen. nov., spp. nov., a gliding, arachidonic acid-containing bacterium isolated from the southern coastline of Thailand. Int J Syst Evol Microbiol 56: 2931-2935.

Huse SM, Welch DM, Morrison HG, Sogin ML. (2010). Ironing out the wrinkles in the rare biosphere through improved OTU clustering. Environ Microbiol 12: 1889- 1898.

Huse SM, Huber JA, Morrison HG, Sogin ML, Welch DM. (2007). Accuracy and quality of massively parallel DNA pyrosequencing. Genome Biol 8: e143.

Huson DH, Mitra S, Ruscheweyh H, Weber N, Schuster SC. (2011). Integrative analysis of environmental sequences using MEGAN4. Genome Res 21: 1552-1560.

Jahnke L, Orphan V, Embaye T, Turk K, Kubo M, Summons R, et al. (2008). Lipid biomarker and phylogenetic analyses to reveal archaeal biodiversity and distribution in hypersaline microbial mat and underlying sediment. Geobiology 6: 394-410.

Jahnke LL, Embaye T, Hope J, Turk KA, Van Zuilen M, Des Marais DJ, et al. (2004). Lipid biomarker and carbon isotopic signatures for stromatolite‐forming, microbial mat communities and Phormidium cultures from Yellowstone National Park. Geobiology 2: 31-47.

Jones SE, Lennon JT. (2010). Dormancy contributes to the maintenance of microbial diversity. Proc Natl Acad Sci USA 107: 5881-5886.

Jürgens K, Pernthaler J, Schalla S, Amann R. (1999). Morphological and compositional changes in a planktonic bacterial community in response to enhanced protozoan grazing. Appl Environ Microbiol 65: 1241-1250.

Kah LC, Grotzinger JP. (1992). Early Proterozoic (1.9 Ga) thrombolites of the Rocknest Formation, Northwest Territories, Canada. Palaios 7: 305-315.

Kasting JF, Howard MT. (2006). Atmospheric composition and climate on the early Earth. Philosophical Transactions of the Royal Society B: Biological Sciences 361:1733-1742.

Kawaguchi T, Decho AW. (2000). Biochemical characterization of cyanobacterial extracellular polymers (EPS) from modern marine stromatolites (Bahamas). Preparative Biochemistry and Biotechnology 30: 231-330.

Kennard JM, James NP. (1986). Thrombolites and stromatolites: two distinct types of microbial structures. Palaios 1: 492-503.

161

Khodadad CL, Foster JS. (2012). Metagenomic and metabolic profiling of nonlithifying and lithifying stromatolitic mats of Highborne Cay, The Bahamas. PLoS One 7: e38229.

King GM. (1984). Metabolism of trimethylamine, choline, and glycine betaine by sulfate- reducing and methanogenic bacteria in marine sediments. Appl Environ Microbiol 48: 719-725.

Kolde R. (2013) pheatmap: Pretty Heatmaps. R package version 0.7.4 edn.

Konhauser K, Riding R. (2012). Bacterial Biomineralization. Fundamentals of Geobiology. Edited by Knoll, AH, Canfield, D.E., and Konhauser, KO, John Wiley & Sons, Ltd : 105-130.

Könneke M, Bernhard AE, José R, Walker CB, Waterbury JB, Stahl DA. (2005). Isolation of an autotrophic ammonia-oxidizing marine archaeon. Nature 437: 543- 546.

Kremer B, Kaźmierczak J. (2005). Cyanobacterial mats from Silurian black radiolarian cherts: phototrophic life at the edge of darkness? Journal of Sedimentary Research 75: 897-906.

Krumbein WE, Giele C. (1979). Calcification in a coccoid cyanobacterium associated with the formation of desert stromatolites. Sedimentology 26: 593-604.

Krumbein W, Buchholz H, Franke P, Giani D, Giele C, Wonneberger K. (1979). O2 and H2S coexistence in stromatolites. Naturwissenschaften 66: 381-389.

Kunin V, Engelbrektson A, Ochman H, Hugenholtz P. (2010). Wrinkles in the rare biosphere: pyrosequencing errors can lead to artificial inflation of diversity estimates. Environ Microbiol 12: 118-123.

Kunin V, Raes J, Harris JK, Spear JR, Walker JJ, Ivanova N, et al. (2008). Millimeter- scale genetic gradients and community-level molecular convergence in a hypersaline microbial mat. Molecular Systems Biology 4: 198.

Lane D. (1991). 16S/23S rRNA sequencing. In: Nucleic acid techniques in bacterial systematics. Wiley, pp115-175.

Laval B, Cady SL, Pollack JC, McKay CP, Bird JS, Grotzinger JP, et al. (2000). Modern freshwater microbialite analogues for ancient dendritic reef structures. Nature 407: 626-629.

Leuko S, Goh F, Allen M, Burns B, Walter M, Neilan B. (2007). Analysis of intergenic spacer region length polymorphisms to investigate the halophilic archaeal diversity of stromatolites and microbial mats. Extremophiles 11: 203-210.

162

Ley RE, Harris JK, Wilcox J, Spear JR, Miller SR, Bebout BM, et al. (2006). unexpected diversity and complexity of the Guerrero Negro hypersaline microbial mat. Appl Environ Microbiol 72: 3685-3695.

Lindell D, Jaffe JD, Coleman ML, Futschik ME, Axmann IM, Rector T, et al. (2007). Genome-wide expression dynamics of a marine virus and host reveal features of co-evolution. Nature 449: 83-86.

Liaw A, Weiner M. (2002) Classification and regression by randomforest. R News 3: 18-22.

Liu Z, Lozupone C, Hamady M, Bushman FD, Knight R. (2007). Short pyrosequencing reads suffice for accurate microbial community analysis. Nucleic Acids Res 35: e120.

Lozupone C, Knight R. (2005). UniFrac: a new phylogenetic method for comparing microbial communities. Appl Environ Microbiol 71: 8228-8235.

Ludwig W, Strunk O, Westram R, Richter L, Meier H, Buchner A, et al. (2004). ARB: a software environment for sequence data. Nucleic Acids Res 32: 1363-1371.

Luo C, Tsementzi D, Kyrpides N, Read T, Konstantinidis KT. (2012). Direct comparisons of Illumina vs. Roche 454 sequencing technologies on the same microbial community DNA sample. PLoS One 7: e30087.

Macintyre IG, Prufert‐Bebout L, Reid RP. (2000). The role of endolithic cyanobacteria in the formation of lithified laminae in Bahamian stromatolites. Sedimentology 47: 915-921.

Mackie RI, Sghir A, Gaskins HR. (1999). Developmental microbial ecology of the neonatal gastrointestinal tract. Am J Clin Nutr 69: 1035-1045.

McMurdie PJ, Holmes S. (2013). phyloseq: An R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One 8: e61217.

Meyer F, Paarmann D, D'Souza M, Olson R, Glass EM, Kubal M, et al. (2008). The metagenomics RAST server - a public resource for the automatic phylogenetic and functional analysis of metagenomes. BMC Bioinformatics 9: e386.

Miller SR, Strong AL, Jones KL, Ungerer MC. (2009). Bar-coded pyrosequencing reveals shared bacterial community properties along the temperature gradients of two alkaline hot springs in Yellowstone National Park. Appl Environ Microbiol 75: 4565-4572.

Miyatake T, MacGregor BJ, Boschker HTS. (2009). Linking microbial community function to phylogeny of sulfate-reducing deltaproteobacteria in marine sediments by combining stable isotope probing with magnetic-bead capture hybridization of 16s rrna. Appl Environ Microbiol 75: 4927-4935.

163

Mobberley JM, Khodadad CM, Foster JS. (2013). Metabolic potential of lithifying cyanobacteria-dominated thrombolitic mats. Photosynth Res 118: 125-140.

Mobberley JM, Ortega MC, Foster JS. (2012). Comparative microbial diversity analyses of modern marine thrombolitic mats by barcoded pyrosequencing. Environ Microbiol 14: 82-100.

Morse JW, Arvidson RS, Lüttge A. (2007). Calcium carbonate formation and dissolution. Chem Rev 107: 342-381.

Myshrall K, Mobberley J, Green S, Visscher P, Havemann S, Reid R, et al. (2010). Biogeochemical cycling and microbial diversity in the thrombolitic microbialites of Highborne Cay, Bahamas. Geobiology 8: 337-354.

Nitti A, Daniels CA, Siefert J, Souza V, Hollander D, Breitbart M. (2012). Spatially resolved genomic, stable isotopic, and lipid analyses of a modern freshwater microbialite from Cuatro Ciénegas, Mexico. Astrobiology 12: 685-698.

Nübel U, Garcia-Pichel F, Muyzer G. (1997). PCR primers to amplify 16S rRNA genes from cyanobacteria. Appl Environ Microbiol 63: 3327-3332.

Offre P, Spang A, Schleper C. (2013). Archaea in biogeochemical cycles. Annu Rev Microbiol 67: 437-457.

Oh S, Caro-Quintero A, Tsementzi D, DeLeon-Rodriguez N, Luo C, Poretsky R, et al. (2011). Metagenomic insights into the evolution, function and complexity of the planktonic microbial community of Lake Lanier, a temperate freshwater ecosystem. Appl Environ Microbiol 77: 6000-6011.

Paerl HW, Steppe TF, Reid RP. (2001). Bacterially mediated precipitation in marine stromatolites. Environ Microbiol 3: 123-130.

Papineau D, Walker JJ, Mojzsis SJ, Pace NR. (2005). Composition and structure of microbial communities from stromatolites of Hamelin Pool in Shark Bay, Western Australia. Appl Environ Microbiol 71: 4822-4832.

Paterson DM, Aspden RJ, Visscher PT, Consalvey M, Andres MS, Decho AW, et al. (2008). Light-dependant biostabilisation of sediments by stromatolite assemblages. PLoS One 3: e3176.

Peimbert M, Alcaraz LD, Bonilla-Rosso G, Olmedo-Alvarez G, García-Oliva F, Segovia L, et al. (2012). Comparative metagenomics of two microbial mats at Cuatro Ciénegas Basin I: ancient lessons on how to cope with an environment under severe nutrient stress. Astrobiology 12: 648-658.

Peters JW, Boyd ES, D’Adamo S, Mulder DW, Therien J, Posewitz MC. (2013). Hydrogenases, nitrogenases, anoxia, and H2 production in water-oxidizing phototrophs. In: Algae for Biofuels and Energy. Springer, pp 37-75.

164

Pinckney J, Paerl H, Reid R, Bebout B. (1995). Ecophysiology of stromatolitic microbial mats, Stocking Island, exuma cays, Bahamas. Microb Ecol 29: 19-37.

Planavsky N, Ginsburg RN. (2009). Taphonomy of modern marine Bahamian microbialites. Palaios 24: 5-17.

Planavsky N, Reid R, Lyons T, Myshrall K, Visscher P. (2009). Formation and diagenesis of modern marine calcified cyanobacteria. Geobiology 7: 566-576.

Preston CM, Wu KY, Molinski TF, DeLong EF. (1996). A psychrophilic crenarchaeon inhabits a marine sponge: Cenarchaeum symbiosum gen. nov., spp. nov. Proc Natl Acad Sci USA 93: 6241-6246.

Price MN, Dehal PS, Arkin AP. (2010). FastTree 2–approximately maximum-likelihood trees for large alignments. PLoS One 5: e9490.

Pruesse E, Quast C, Knittel K, Fuchs BM, 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.

R Development_Core_Team. (2010) A language and environment for statistical computing. 3-900051-07-0 edn. R Foundation for Statistical Computing, Vienna, Austria.

Reid RP, Visscher P, Decho A, Stolz JF, Bebout B, Dupraz C, et al. (2000). The role of microbes in accretion, lamination and early lithification of modern marine stromatolites. Nature 406: 989-992.

Riding R. (2011). Microbialites, stromatolites, and thrombolites. Encyclopedia of Geobiology. by J.Reitner and V.Thiel. Springer, Heidelberg :635-654.

Riding R. (2000). Microbial carbonates: the geological record of calcified bacterial–algal mats and biofilms. Sedimentology 47: 179-214.

Rho M, Tang H, Ye Y. (2010). FragGeneScan: predicting genes in short and error-prone reads. Nucleic Acids Res 38: e191-e191.

Rippka R, Castenholz RW, Herdman M. (2001). Phylum Cyanobacteria Section IV. In: Bergey's Manual of Systematic Biology. Springer: New York, pp 562-566.

Robertson CE, Spear JR, Harris JK, Pace NR. (2009). Diversity and stratification of archaea in a hypersaline microbial mat. Appl Environ Microbiol 75: 1801-1810.

Roesch LFW, Lorca GL, Casella G, Giongo A, Naranjo A, Pionzio AM, et al. (2009). Culture-independent identification of gut bacteria correlated with the onset of diabetes in a rat model. ISME J 3: 536-548.

165

Roesch LFW, Camargo FAO, Bento FM, Triplett EW. (2008). Biodiversity of diazotrophic bacteria within the soil, root and stem of field-grown maize. Plant Soil 302: 91-104.

Rusch DB, Halpern AL, Sutton G, Heidelberg KB, Williamson S, Yooseph S, et al. (2007). The Sorcerer II global ocean sampling expedition: northwest Atlantic through eastern tropical Pacific. PLoS Biol 5: e77.

Santos F, Peña A, Nogales B, Soria-Soria E, García del Cura MÁ, González-Martín JA, et al. (2010). Bacterial diversity in dry modern freshwater stromatolites from Ruidera Pools Natural Park, Spain. Syst Appl Microbiol 33: 209-221.

Sato Y, Willis BL, Bourne DG. (2009). Successional changes in bacterial communities during the development of black band disease on the reef coral, Montipora hispida. ISME J 4: 203-214.

Schimel J. (2004). Playing scales in the methane cycle: from microbial ecology to the globe. Proc Natl A.cad Sci U S A 101: 12400-12401.

Schmieder R, Edwards R. (2011). Quality control and preprocessing of metagenomic datasets. Bioinformatics 27: 863-864.

Schloss PD (2009). A High-Throughput DNA Sequence Aligner for Microbial Ecology Studies. PLoS One 12: e8230.

Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. (2009). Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol 75: 7537-7541.

Schloss PD, Handelsman J. (2005). Introducing DOTUR, a computer program for defining operational taxonomic units and estimating species richness. Appl Environ Microbiol 71: 1501-1506.

Selenska-Pobell S. (2002). Diversity and activity in uranium waste piles. Radioact Environ 2: 225-254.

Short CM, Suttle CA. (2005). Nearly identical bacteriophage structural gene sequences are widely distributed in both marine and freshwater environments. Appl Environ Microbiol 71: 480-486.

Smith JM, Green SJ, Kelley CA, Prufert-Bebout L, Bebout BM. (2008). Shifts in methanogen community structure and function associated with long-term manipulation of sulfate and salinity in a hypersaline microbial mat. Environ Microbiol 10: 386-394.

166

Sogin ML, Morrison HG, Huber JA, Welch DM, Huse SM, Neal PR, et al. (2006). Microbial diversity in the deep sea and the underexplored “rare biosphere”. Proc Natl Acad Sci USA 103: 12115-12120.

Stal LJ, van Gemerden H, Krumbein WE. (1985). Structure and development of a benthic marine microbial mat. FEMS Microbiol Lett 31: 111-125.

Steppe TF, Pinckney JL, Dyble J, Paerl HW. (2001). Diazotrophy in modern marine Bahamian stromatolites. Microb Ecol 41: 36-44.

Steven B, Gallegos‐Graves LV, Starkenburg SR, Chain PS, Kuske CR. (2012). Targeted and shotgun metagenomic approaches provide different descriptions of dryland soil microbial communities in a manipulated field study. Environmental Microbiology Reports 4: 248-256.

Stewart FJ, Ottesen EA, DeLong EF. (2010). Development and quantitative analyses of a universal rRNA-subtraction protocol for microbial metatranscriptomics. ISME J 4: 896-907.

Stolz JF. (2000). Structure of microbial mats and biofilms. In: Microbial Sediments. Springer, pp 1-8.

Tarhan LG, Planavsky N, Laumer CE, Stolz JF, Reid RP. (2013). Microbial mat controls on infaunal abundance and diversity in modern marine microbialites. Geobiology doi:10.1111/gbi.12049.

Thomas T, Gilbert J, Meyer F. (2012). Metagenomics- a guide from sampling to data analysis. Microbial Informatics and Experimentation 2: e3.

Van Der Meer MT, Schouten S, Damsté JSS, Ward DM. (2007). Impact of carbon metabolism on 13C signatures of cyanobacteria and green non‐sulfur‐like bacteria inhabiting a microbial mat from an alkaline siliceous hot spring in Yellowstone National Park (USA). Environ Microbiol 9: 482-491.

Visscher PT, Stolz JF. (2005). Microbial mats as bioreactors: populations, processes, and products. Palaeogeogr , Palaeoclimatol , Palaeoecol 219: 87-100.

Visscher PT, Reid RP, Bebout BM. (2000). Microscale observations of sulfate reduction: Correlation of microbial activity with lithified micritic laminae in modern marine stromatolites. Geology 28: 919-922.

Visscher PT, Reid RP, Bebout BM, Hoeft SE, Macintyre IG, Thompson JA. (1998). Formation of lithified micritic laminae in modern marine stromatolites (Bahamas); the role of sulfur cycling. Am Mineral 83: 1482-1493.

167

Walker CB, De la Torre JR, Klotz MG, Urakawa H, Pinel N, Arp DJ, et al. (2010). Nitrosopumilus maritimus genome reveals unique mechanisms for nitrification and autotrophy in globally distributed marine crenarchaea. Proc Natl Acad Sci USA 107: 8818-8823.

Wang Q, Garrity GM, Tiedje JM, Cole JR. (2007). Naive Bayesian classifier for rapid assignment of rRN sequences into the new bacterial taxonomy. Appl Environ Microbiol 73: 5261-5267.

Wang Y, Qian PY. (2009). Conservative fragments in bacterial 16S rRNA genes and primer design for 16S ribosomal DNA amplicons in metagenomic studies. PLoS One 4: e7401.

Weiner S, Dove PM. (2003). An overview of biomineralization processes and the problem of the vital effect. Reviews in Mineralogy and Geochemistry 54:1-29.

Wilke A, Harrison T, Wilkening J, Field D, Glass EM, Kyrpides N, et al. (2012). The M5nr: a novel non-redundant database containing protein sequences and annotations from multiple sources and associated tools. BMC Bioinformatics 13: e141.

Xia Y, Kong Y, Thomsen TR, Halkjaer Nielsen P. (2008). Identification and ecophysiological characterization of epiphytic protein-hydrolyzing saprospiraceae ("Candidatus Epiflobacter" spp.) in activated sludge. Appl Environ Microbiol 74: 2229-2238.

Yannarell AC, Paerl HW. (2007). Effects of salinity and light on organic carbon and nitrogen uptake in a hypersaline microbial mat. FEMS Microbiol Ecol 62: 345- 353.

Zhu W, Lomsadze A, Borodovsky, M. (2010). Ab initio gene identification in metagenome sequences. Nucleic Acids Res 38: e132.

168

BIOGRAPHICAL SKETCH

Jennifer Marie Mobberley was born in Saint Petersburg, FL in 1982. She completed her B.Sc. degree in Microbiology and Cell Science with a minor in Chemistry from the University of Florida in 2004. Jennifer completed her M.Sc. degree in Marine

Science in 2007, where her thesis focused on genomic analysis and characterization of temperate bacteriophages in isolates from the Gulf of Mexico. This work piqued her interest in the functional genetics of complex ecosystems. For this reason, Jennifer accepted a Ph.D. position working on a project examining functional complexity within phototrophic microbial mats under the direction of Dr. Jamie Foster at the University of

Florida. Jennifer was awarded a NASA Graduate Student Research Project Fellowship from 2010-2013 to fund her dissertation work. As a Ph.D. student, she has presented her research at the NASA-Nordic Astrobiology Summer School (2009), International

Society of Microbial Ecology meetings (2010, 2012), Astrobiology Science Conference

(2012), and Florida Branch American Society of Microbiology (2013). Jennifer received her Ph.D. from the University of Florida in the fall of 2013.

169